1 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
4 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
5 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
6 (type_natural_mode): Warn ABI change when %zmm register is not
7 available for AVX512F vector value passing.
9 2014-02-18 Kai Tietz <ktietz@redhat.com>
12 * config/i386/i386.c (ix86_expand_prologue): Use value in
13 rax register as displacement when restoring %r10 or %rax.
14 Fix wrong offset when restoring both registers.
16 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
18 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
19 assertion with conditional return.
21 2014-02-18 Jakub Jelinek <jakub@redhat.com>
22 Uros Bizjak <ubizjak@gmail.com>
25 * config/i386/driver-i386.c (host_detect_local_cpu): If
26 YMM state is not saved by the OS, also clear has_f16c. Move
27 CPUID 0x80000001 handling before YMM state saving checking.
29 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
31 PR rtl-optimization/58960
32 * haifa-sched.c (alloc_global_sched_pressure_data): New,
34 (sched_init): ... here.
35 (free_global_sched_pressure_data): New, factored out from ...
36 (sched_finish): ... here.
37 * sched-int.h (free_global_sched_pressure_data): Declare.
38 * sched-rgn.c (nr_regions_initial): New static global.
39 (haifa_find_rgns): Initialize it.
40 (schedule_region): Disable sched-pressure for the newly
43 2014-02-17 Richard Biener <rguenther@suse.de>
45 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
46 release SSA defs of pattern stmts.
48 2014-02-17 Richard Biener <rguenther@suse.de>
50 * tree-inline.c (expand_call_inline): Release the virtual
51 operand defined by the call we are about to inline.
53 2014-02-17 Richard Biener <rguenther@suse.de>
55 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
57 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
58 Ilya Tocar <ilya.tocar@intel.com>
60 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
61 arguments order in builtin.
62 (_mm512_permutexvar_epi64): Ditto.
63 (_mm512_mask_permutexvar_epi64): Ditto
64 (_mm512_maskz_permutexvar_epi32): Ditto
65 (_mm512_permutexvar_epi32): Ditto
66 (_mm512_mask_permutexvar_epi32): Ditto
68 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
70 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
71 (p8_vmrgow): Likewise.
73 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
75 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
78 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
81 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
82 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
83 into 64-bit and 32-bit moves. On 64-bit moves, add support for
84 using direct move instructions on ISA 2.07. Also adjust
85 instruction length for 64-bit.
86 (mov<mode>_64bit, TFmode/TDmode): Likewise.
87 (mov<mode>_32bit, TFmode/TDmode): Likewise.
89 2014-02-15 Alan Modra <amodra@gmail.com>
93 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
94 find_replacement on parts of insn rtl that might be reloaded.
96 2014-02-15 Richard Biener <rguenther@suse.de>
98 PR tree-optimization/60183
99 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
100 (tree_ssa_phiprop): Calculate and free post-dominators.
102 2014-02-14 Jeff Law <law@redhat.com>
104 PR rtl-optimization/60131
105 * ree.c (get_extended_src_reg): New function.
106 (combine_reaching_defs): Use it rather than assuming location of REG.
107 (find_and_remove_re): Verify first operand of extension is
108 a REG before adding the insns to the copy list.
110 2014-02-14 Roland McGrath <mcgrathr@google.com>
112 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
113 * configure: Regenerated.
114 * config.in: Regenerated.
115 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
116 instead of ASM_SHORT.
118 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
119 Richard Earnshaw <rearnsha@arm.com>
121 PR rtl-optimization/59535
122 * lra-constraints.c (process_alt_operands): Encourage alternative
123 when unassigned pseudo class is superset of the alternative class.
124 (inherit_reload_reg): Don't inherit when optimizing for code size.
125 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
126 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
127 modes not less than 4 for Thumb1.
129 2014-02-14 Kyle McMartin <kyle@redhat.com>
132 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
134 2014-02-14 Richard Biener <rguenther@suse.de>
136 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
137 (get_frame_arg): Drop the assert with langhook types_compatible_p.
138 Do not strip INDIRECT_REFs.
140 2014-02-14 Richard Biener <rguenther@suse.de>
143 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
144 DECL_FUNCTION_SPECIFIC_TARGET.
145 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
146 * tree-streamer-out.c (pack_ts_target_option): Remove.
147 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
148 (write_ts_function_decl_tree_pointers): Do not stream
149 DECL_FUNCTION_SPECIFIC_TARGET.
150 * tree-streamer-in.c (unpack_ts_target_option): Remove.
151 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
152 (lto_input_ts_function_decl_tree_pointers): Do not stream
153 DECL_FUNCTION_SPECIFIC_TARGET.
155 2014-02-14 Jakub Jelinek <jakub@redhat.com>
157 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
158 (get_initial_def_for_induction, vectorizable_induction): Ignore
159 debug stmts when looking for exit_phi.
160 (vectorizable_live_operation): Fix up condition.
162 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
164 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
165 nreverse() because it changes the content of original tree list.
167 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
169 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
170 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
172 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
174 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
175 GNU coding standards.
177 2014-02-13 Jakub Jelinek <jakub@redhat.com>
180 * dwarf2out.c (gen_subprogram_die): Don't call
181 add_calling_convention_attribute if subr_die is old_die.
183 2014-02-13 Sharad Singhai <singhai@google.com>
185 * doc/optinfo.texi: Fix order of nodes.
187 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
189 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
190 operands[2], not operands[3].
192 2014-02-13 Richard Biener <rguenther@suse.de>
195 * doc/install.texi (ISL): Update recommended version to 0.12.2,
196 mention the possibility of an in-tree build.
197 (CLooG): Update recommended version to 0.18.1, mention the
198 possibility of an in-tree build and clarify that the ISL
199 bundled with CLooG does not work.
201 2014-02-13 Jakub Jelinek <jakub@redhat.com>
204 * expr.c (compress_float_constant): If x is a hard register,
205 extend into a pseudo and then move to x.
207 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
209 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
210 caused by bad second argument to warning_at() with -mhotpatch and
211 nested functions (e.g. with gfortran).
213 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
215 * opts.c (option_name): Remove "enabled by default" rider.
217 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
219 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
221 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
222 Uros Bizjak <ubizjak@gmail.com>
225 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
226 * configure: Regenerated.
228 2014-02-12 Richard Biener <rguenther@suse.de>
230 * vec.c (vec_prefix::calculate_allocation): Move as
231 inline variant to vec.h.
232 (vec_prefix::calculate_allocation_1): New out-of-line version.
233 * vec.h (vec_prefix::calculate_allocation_1): Declare.
234 (vec_prefix::m_has_auto_buf): Rename to ...
235 (vec_prefix::m_using_auto_storage): ... this.
236 (vec_prefix::calculate_allocation): Inline the easy cases
237 and dispatch to calculate_allocation_1 which doesn't need the
239 (va_heap::reserve): Use gcc_checking_assert.
240 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
241 m_using_auto_storage.
242 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
244 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
245 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
246 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
248 2014-02-12 Richard Biener <rguenther@suse.de>
250 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
251 when we found a dependence.
253 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
255 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
257 (maybe_fold_stmt): ... into this new function.
258 * omp-low.c (lower_omp): Update comment.
260 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
263 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
266 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
268 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
269 identifiers in comments.
270 (cortexa53_extra_costs): Likewise.
271 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
272 (cortexa7_extra_costs): Likewise.
273 (cortexa12_extra_costs): Likewise.
274 (cortexa15_extra_costs): Likewise.
275 (v7m_extra_costs): Likewise.
277 2014-02-12 Richard Biener <rguenther@suse.de>
280 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
281 of posix_memalign being successful.
282 (lower_stmt): Restrict lowering of posix_memalign to when
283 -ftree-bit-ccp is enabled.
285 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
287 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
289 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
291 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
293 PR rtl-optimization/60116
294 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
295 other_insn once the combination has been validated.
297 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
300 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
302 * ipa-devirt.c: Include demangle.h
303 (odr_violation_reported): New static variable.
304 (add_type_duplicate): Update odr_violations.
305 (maybe_record_node): Add completep parameter; update it.
306 (record_target_from_binfo): Add COMPLETEP parameter;
308 (possible_polymorphic_call_targets_1): Likewise.
309 (struct polymorphic_call_target_d): Add nonconstruction_targets;
310 rename FINAL to COMPLETE.
311 (record_targets_from_bases): Sanity check we found the binfo;
312 fix COMPLETEP updating.
313 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
314 parameter, fix computing of COMPLETEP.
315 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
316 at LTO time do demangling.
317 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
318 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
320 (gimple_get_virt_method_for_binfo): Likewise.
321 * gimple-fold.h (gimple_get_virt_method_for_binfo,
322 gimple_get_virt_method_for_vtable): Update prototypes.
324 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
327 * genautomata.c (add_presence_absence): Fix typo with
328 {final_}presence_list.
330 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
333 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
334 for VSX/Altivec vectors that land in GPR registers.
336 2014-02-11 Richard Henderson <rth@redhat.com>
337 Jakub Jelinek <jakub@redhat.com>
340 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
341 around drhs if type conversion to lacc->type is not useless.
343 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
345 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
347 (cortex-a57.cortex-a53): Likewise.
348 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
350 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
352 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
355 2014-02-11 Renlin Li <Renlin.Li@arm.com>
357 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
358 add_options_for_arm_vfp3.
360 2014-02-11 Jeff Law <law@redhat.com>
363 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
364 object with an undesirable mode.
366 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
369 * config/i386/sol2-9.h: New file.
370 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
371 *-*-solaris2.9*): Use it.
373 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
375 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
376 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
378 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
380 * config/microblaze/microblaze.c: Extend mcpu version format
382 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
384 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
386 2014-02-10 Richard Henderson <rth@redhat.com>
389 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
390 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
391 ms-abi vs -mno-accumulate-outgoing-args.
392 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
393 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
396 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
399 * cfgexpand.c (expand_asm_operands): Attach source location to
400 ASM_INPUT rtx objects.
401 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
403 2014-02-10 Nick Clifton <nickc@redhat.com>
405 * config/mn10300/mn10300.c (popcount): New function.
406 (mn10300_expand_prologue): Include saved registers in stack usage
409 2014-02-10 Jeff Law <law@redhat.com>
412 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
413 when changing the SET_DEST of a prior insn to avoid an input reload.
415 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
417 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
418 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
419 -mcall-openbsd, or -mcall-linux.
420 (CC1_ENDIAN_BIG_SPEC): Remove.
421 (CC1_ENDIAN_LITTLE_SPEC): Remove.
422 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
423 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
424 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
425 and %cc1_endian_default.
426 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
428 2014-02-10 Richard Biener <rguenther@suse.de>
430 PR tree-optimization/60115
431 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
432 MEM_REF handling. Properly verify that the accesses are not
433 out of the objects bound.
435 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
437 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
440 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
442 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
443 proper constants and fix formatting.
444 (possible_polymorphic_call_targets): Fix formatting.
446 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
447 Ilya Tocar <ilya.tocar@intel.com>
449 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
450 (_mm512_loadu_epi32): Renamed into...
451 (_mm512_loadu_si512): This.
452 (_mm512_storeu_epi32): Renamed into...
453 (_mm512_storeu_si512): This.
454 (_mm512_maskz_ceil_ps): Removed.
455 (_mm512_maskz_ceil_pd): Ditto.
456 (_mm512_maskz_floor_ps): Ditto.
457 (_mm512_maskz_floor_pd): Ditto.
458 (_mm512_floor_round_ps): Ditto.
459 (_mm512_floor_round_pd): Ditto.
460 (_mm512_ceil_round_ps): Ditto.
461 (_mm512_ceil_round_pd): Ditto.
462 (_mm512_mask_floor_round_ps): Ditto.
463 (_mm512_mask_floor_round_pd): Ditto.
464 (_mm512_mask_ceil_round_ps): Ditto.
465 (_mm512_mask_ceil_round_pd): Ditto.
466 (_mm512_maskz_floor_round_ps): Ditto.
467 (_mm512_maskz_floor_round_pd): Ditto.
468 (_mm512_maskz_ceil_round_ps): Ditto.
469 (_mm512_maskz_ceil_round_pd): Ditto.
470 (_mm512_expand_pd): Ditto.
471 (_mm512_expand_ps): Ditto.
472 * config/i386/i386.c (ix86_builtins): Remove
473 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
475 * config/i386/predicates.md (const1256_operand): New.
476 (const_1_to_2_operand): Ditto.
477 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
478 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
479 (*avx512pf_gatherpf<mode>sf): Ditto.
480 (avx512pf_gatherpf<mode>df): Ditto.
481 (*avx512pf_gatherpf<mode>df_mask): Ditto.
482 (*avx512pf_gatherpf<mode>df): Ditto.
483 (avx512pf_scatterpf<mode>sf): Ditto.
484 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
485 (*avx512pf_scatterpf<mode>sf): Ditto.
486 (avx512pf_scatterpf<mode>df): Ditto.
487 (*avx512pf_scatterpf<mode>df_mask): Ditto.
488 (*avx512pf_scatterpf<mode>df): Ditto.
489 (avx512f_expand<mode>): Removed.
490 (<shift_insn><mode>3<mask_name>): Change predicate type.
492 2014-02-08 Jakub Jelinek <jakub@redhat.com>
494 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
495 not at the end of datarefs vector use ordered_remove to avoid
496 reordering datarefs vector.
499 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
500 mark local addressable non-static vars as GOVD_PRIVATE
501 instead of GOVD_LOCAL.
502 * omp-low.c (lower_omp_for): Move gimple_bind_vars
503 and BLOCK_VARS of gimple_bind_block to new_stmt rather
507 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
508 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
509 assume_aligned or alloc_align attributes.
510 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
511 arguments. Handle also assume_aligned and alloc_align attributes.
512 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
513 calls to functions with assume_aligned or alloc_align attributes.
514 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
516 2014-02-08 Terry Guo <terry.guo@arm.com>
518 * doc/invoke.texi: Document ARM -march=armv7e-m.
520 2014-02-08 Jakub Jelinek <jakub@redhat.com>
522 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
523 flag on __cilkrts_rethrow builtin.
526 * ipa-cp.c (determine_versionability): Fail at -O0
527 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
528 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
531 2014-02-04 Jakub Jelinek <jakub@redhat.com>
534 * tree-inline.c (copy_forbidden): Fail for
535 __attribute__((optimize (0))) functions.
537 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
539 * varpool.c: Include pointer-set.h.
540 (varpool_remove_unreferenced_decls): Variables in other partitions
541 will not be output; be however careful to not lose information
544 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
546 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
547 lookup in the vtable constructor.
549 2014-02-07 Jeff Law <law@redhat.com>
552 * config/m68k/m68k.md (ashldi_extsi): Turn into a
553 define_insn_and_split.
555 * ipa-inline.c (inline_small_functions): Fix typos.
557 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
559 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
560 (s390_can_use_return_insn): Declare.
561 * config/s390/s390.h (EPILOGUE_USES): Define.
562 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
564 (s390_chunkify_start): Handle return JUMP_LABELs.
565 (s390_early_mach): Emit a main_pool instruction on the entry edge.
566 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
567 (s390_can_use_return_insn): New functions.
568 (s390_fix_long_loop_prediction): Handle conditional returns.
569 (TARGET_SET_UP_BY_PROLOGUE): Define.
570 * config/s390/s390.md (ANY_RETURN): New code iterator.
571 (*creturn, *csimple_return, return, simple_return): New patterns.
573 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
575 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
576 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
577 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
578 REG_CFA_RESTORE list when deciding not to restore a register.
580 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
582 * config/s390/s390.c: Include tree-pass.h and context.h.
583 (s390_early_mach): New function, split out from...
584 (s390_emit_prologue): ...here.
585 (pass_data_s390_early_mach): New pass structure.
586 (pass_s390_early_mach): New class.
587 (s390_option_override): Create and register early_mach pass.
590 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
592 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
593 to match for the exit block.
595 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
597 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
598 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
599 Reject misaligned operands.
601 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
603 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
605 2014-02-07 Richard Biener <rguenther@suse.de>
608 * gimple-low.c (lower_builtin_posix_memalign): New function.
609 (lower_stmt): Call it to lower posix_memalign in a way
610 to make alignment info accessible.
612 2014-02-07 Jakub Jelinek <jakub@redhat.com>
615 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
616 __builtin_setjmp_receiver.
618 2014-02-07 Richard Biener <rguenther@suse.de>
621 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
622 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
623 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
624 Handle BUILT_IN_POSIX_MEMALIGN.
625 (find_func_clobbers): Likewise.
626 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
627 (call_may_clobber_ref_p_1): Likewise.
629 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
632 * ipa-devirt.c (record_target_from_binfo): Remove overactive
635 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
638 * lto-cgraph.c (lto_output_node): Use
639 symtab_get_symbol_partitioning_class.
640 (lto_output_varpool_node): likewise.
641 (symtab_get_symbol_partitioning_class): Move here from
643 * cgraph.h (symbol_partitioning_class): Likewise.
644 (symtab_get_symbol_partitioning_class): Declare.
646 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
648 * ggc.h (ggc_internal_cleared_alloc): New macro.
649 * vec.h (vec_safe_copy): Handle memory stats.
650 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
651 * target-globals.c (save_target_globals): Likewise.
653 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
656 * expr.c (emit_move_resolve_push): Export; be bit more selective
657 on when to clear alias set.
658 * expr.h (emit_move_resolve_push): Declare.
659 * function.h (struct function): Add tail_call_marked.
660 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
661 * config/i386/i386-protos.h (ix86_expand_push): Remove.
662 * config/i386/i386.md (TImode move expander): De not call
664 (FP push expanders): Preserve memory attributes.
665 * config/i386/sse.md (push<mode>1): Remove.
666 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
667 (ix86_expand_push): Remove.
668 * config/i386/mmx.md (push<mode>1): Remove.
670 2014-02-06 Jakub Jelinek <jakub@redhat.com>
672 PR rtl-optimization/60030
673 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
674 lopart with paradoxical subreg before shifting it up by hprec.
676 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
678 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
679 Remove extra newline at end of file.
680 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
681 (arm_issue_rate): Handle cortexa57.
682 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
683 (cortex-a57.cortex-a53): Likewise.
685 2014-02-06 Jakub Jelinek <jakub@redhat.com>
688 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
689 don't record in REG_FRAME_RELATED_EXPR registers not set in that
691 (arm_expand_prologue): Adjust all callers.
692 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
693 info, registers also at the lowest numbered registers side. Use
694 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
698 * var-tracking.c (adjust_mems): Before adding a SET to
699 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
701 2014-02-06 Alan Modra <amodra@gmail.com>
704 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
705 change SDmode to DDmode when lra_in_progress.
707 2014-02-06 Jakub Jelinek <jakub@redhat.com>
710 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
711 free_data_ref on the dr first, and before goto again also set dr
712 to the next dr. For simd_lane_access, free old datarefs[i] before
713 overwriting it. For get_vectype_for_scalar_type failure, don't
714 free_data_ref if simd_lane_access.
716 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
719 * tree.h (opts_for_fn): New inline function.
720 (opt_for_fn): Define.
721 * config/i386/i386.c (ix86_function_regparm): Use
722 opt_for_fn (decl, optimize) instead of optimize.
724 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
726 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
727 for SYMBOL_REF in large memory model.
729 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
731 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
733 (cortex-a57): Likewise.
734 (cortex-a57.cortex-a53): Likewise.
736 2014-02-06 Yury Gribov <y.gribov@samsung.com>
737 Kugan Vivekanandarajah <kuganv@linaro.org>
739 * config/arm/arm.c (arm_vector_alignment_reachable): Check
741 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
743 2014-02-06 Richard Biener <rguenther@suse.de>
745 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
746 set_loop_copy and initialize_original_copy_tables.
748 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
750 * config/aarch64/aarch64-simd.md
751 (aarch64_ashr_simddi): Change QI to SI.
753 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
754 Jakub Jelinek <jakub@redhat.com>
757 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
760 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
762 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
763 CODE_FOR_altivec_vpku[hw]um to
764 CODE_FOR_altivec_vpku[hw]um_direct.
765 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
766 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
767 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
768 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
770 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
772 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
773 generation for -maltivec=be.
774 (altivec_vsumsws): Simplify redundant test.
776 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
778 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
779 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
780 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
781 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
783 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
785 (altivec_vpks<VI_char>ss): Likewise.
786 (altivec_vpks<VI_char>us): Likewise.
787 (altivec_vpku<VI_char>us): Likewise.
788 (altivec_vpku<VI_char>um): Likewise.
789 (altivec_vpku<VI_char>um_direct): New (copy of
790 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
792 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
793 target is little endian and -maltivec=be is not specified.
794 (*altivec_vupkhs<VU_char>_direct): New (copy of
795 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
796 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
797 target is little endian and -maltivec=be is not specified.
798 (*altivec_vupkls<VU_char>_direct): New (copy of
799 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
800 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
801 little endian and -maltivec=be is not specified.
802 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
803 little endian and -maltivec=be is not specified.
805 2014-02-05 Richard Henderson <rth@redhat.com>
808 * combine-stack-adj.c: Revert r206943.
809 * sched-int.h (struct deps_desc): Add last_args_size.
810 * sched-deps.c (init_deps): Initialize it.
811 (sched_analyze_insn): Add OUTPUT dependencies between insns that
812 contain REG_ARGS_SIZE notes.
814 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
816 * lto-cgraph.c (asm_nodes_output): Make global.
817 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
818 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
819 (driver_handle_option): Handle OPT_fwpa.
821 2014-02-05 Jakub Jelinek <jakub@redhat.com>
824 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
825 a comment typo and formatting issue. If odr_hash hasn't been
826 created, return vNULL and set *completep to false.
829 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
830 bb with no successors.
832 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
835 * doc/invoke.texi (-march): Clarify documentation for ARM.
839 2014-02-05 Richard Biener <rguenther@suse.de>
841 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
842 when not vectorizing because of too many alias checks.
843 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
844 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
846 2014-02-05 Nick Clifton <nickc@redhat.com>
848 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
849 accept extended registers in any mode when compiling for the MN10300.
851 2014-02-05 Yury Gribov <y.gribov@samsung.com>
853 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
854 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
855 sanitization attributes.
856 (can_inline_edge_p): Likewise.
857 (sanitize_attrs_match_for_inline_p): New function.
859 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
861 * ipa-prop.c (detect_type_change): Shor circuit testing of
862 type changes on THIS pointer.
864 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
867 * config/pa/pa.c (legitimize_tls_address): Return original address
868 if not passed a SYMBOL_REF rtx.
869 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
871 (pa_emit_move_sequence): Simplify TLS source operands.
872 (pa_legitimate_constant_p): Reject all TLS constants.
873 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
874 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
876 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
878 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
879 groups when we know they are controlled by LTO.
880 * varasm.c (default_binds_local_p_1): If object is in other partition,
881 it will be resolved locally.
883 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
885 * config/host-linux.c (linux_gt_pch_use_address): Don't
886 use SSIZE_MAX because it is not always defined.
888 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
891 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
892 threshold for pseudo splitting.
893 (update_ebb_live_info): Process call argument hard registers and
894 hard registers from insn definition too.
895 (max_small_class_regs_num): New constant.
896 (inherit_in_ebb): Update live hard regs through EBBs. Update
897 reloads_num only for small register classes. Don't split for
900 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
903 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
906 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
908 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
911 2014-02-04 Marek Polacek <polacek@redhat.com>
913 * gdbinit.in (pel): Define.
915 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
917 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
920 2014-02-04 Richard Biener <rguenther@suse.de>
923 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
924 in function context local.
925 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
926 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
927 similar to LTO_imported_decl_ref.
929 2014-02-04 Jakub Jelinek <jakub@redhat.com>
931 PR tree-optimization/60002
932 * cgraphclones.c (build_function_decl_skip_args): Clear
935 PR tree-optimization/60023
936 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
937 false to gsi_replace.
938 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
939 has been in some EH region and vec_stmt could throw, add
940 vec_stmt into the same EH region.
941 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
942 has no lhs, ignore it.
943 * internal-fn.c (expand_MASK_LOAD): Likewise.
946 * tree-inline.c (copy_forbidden): Fail for
947 __attribute__((optimize (0))) functions.
950 * omp-low.c (simd_clone_struct_copy): If from->inbranch
951 is set, copy one less argument.
952 (expand_simd_clones): Don't subtract clone_info->inbranch
953 from simd_clone_struct_alloc argument.
955 PR rtl-optimization/57915
956 * recog.c (simplify_while_replacing): If all unary/binary/relational
957 operation arguments are constant, attempt to simplify those.
960 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
961 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
963 2014-02-04 Richard Biener <rguenther@suse.de>
965 PR tree-optimization/60012
966 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
967 TBAA disambiguation to all DDRs.
969 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
972 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
973 (LINK_SPEC): Use it for -shared, -shared-libgcc.
975 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
978 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
980 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
982 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
983 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
985 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
988 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
989 to figure out targets of polymorphic calls with known decl.
990 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
991 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
992 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
993 (get_polymorphic_call_info): ... here.
994 (get_polymorphic_call_info_from_invariant): New function.
996 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
998 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
999 lookup via vtable pointer; check for type consistency
1000 and turn inconsitent facts into UNREACHABLE.
1001 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1002 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1003 type inconsistent querries; return UNREACHABLE instead.
1005 2014-02-03 Richard Henderson <rth@twiddle.net>
1008 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1009 already processed this node.
1010 (normalize_one_pred_1): Pass along mark_set.
1011 (normalize_one_pred): Create and destroy a pointer_set_t.
1012 (normalize_one_pred_chain): Likewise.
1014 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1016 PR gcov-profile/58602
1017 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1019 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1022 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1023 -fno-devirtualize; try to devirtualize by the knowledge of
1024 virtual table pointer given by aggregate propagation.
1025 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1026 (ipa_print_node_jump_functions): Dump also offset that
1027 is relevant for polymorphic calls.
1028 (determine_known_aggregate_parts): Add arg_type parameter; use it
1029 instead of determining the type from pointer type.
1030 (ipa_compute_jump_functions_for_edge): Update call of
1031 determine_known_aggregate_parts.
1032 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1033 (gimple_get_virt_method_for_binfo): ... here; simplify using
1034 vtable_pointer_value_to_vtable.
1035 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1036 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1037 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1038 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1040 (vtable_pointer_value_to_binfo): ... here.
1041 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1043 2014-02-03 Teresa Johnson <tejohnson@google.com>
1045 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1046 redef of outer loop index variable.
1048 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1052 * doc/extend.texi (Function Attributes): Typo.
1054 2014-02-03 Cong Hou <congh@google.com>
1056 PR tree-optimization/60000
1057 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1058 if the vectorized statement is a store. A store statement can only
1059 appear at the end of pattern statements.
1061 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1063 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1064 (ix86_option_override_internal): Default long double to 64-bit for
1065 32-bit Bionic and to 128-bit for 64-bit Bionic.
1067 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1068 TARGET_LONG_DOUBLE_128 is true.
1069 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1071 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1072 (mlong-double-64): Negate -mlong-double-128.
1073 (mlong-double-128): New option.
1075 * config/i386/i386-c.c (ix86_target_macros): Define
1076 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1078 * doc/invoke.texi: Document -mlong-double-128.
1080 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1082 PR rtl-optimization/60024
1083 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1085 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1087 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1089 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1091 PR rtl-optimization/57662
1092 * sel-sched.c (code_motion_path_driver): Do not mark already not
1093 existing blocks in the visiting bitmap.
1095 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1097 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1098 on the insn being emitted.
1100 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1101 Will Deacon <will.deacon@arm.com>
1103 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1105 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1107 * config/arm/arm-tables.opt: Regenerate.
1109 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1111 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1112 for vector types other than V16QImode.
1113 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1114 define_expand, and call altivec_expand_vec_perm_le when producing
1115 code with little endian element order.
1116 (*altivec_vperm_<mode>_internal): New insn having previous
1117 behavior of altivec_vperm_<mode>.
1118 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1119 altivec_expand_vec_perm_le when producing code with little endian
1121 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1122 behavior of altivec_vperm_<mode>_uns.
1124 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1126 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1127 (altivec_vsumsws): Add handling for -maltivec=be with a little
1129 (altivec_vsumsws_direct): New.
1130 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1131 gen_altivec_vsumsws.
1133 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1135 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1136 vtable_pointer_value_to_binfo): New functions.
1137 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1138 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1140 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1142 * config/nios2/nios2.md (load_got_register): Initialize GOT
1143 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1144 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1146 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1148 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1149 preserverd by passthrough, do not propagate the type.
1151 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1153 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1154 (mips_atomic_assign_expand_fenv): New function.
1155 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1157 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1159 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1160 (__builtin_mips_set_fcsr): Likewise.
1161 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1162 MIPS_USI_FTYPE_VOID.
1163 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1164 (mips16_expand_set_fcsr): Likewise.
1165 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1166 (mips16_set_fcsr_stub): Likewise.
1167 (mips16_get_fcsr_one_only_stub): New class.
1168 (mips16_set_fcsr_one_only_stub): Likewise.
1169 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1170 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1171 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1172 (hard_float): New availability predicate.
1173 (mips_builtins): Add get_fcsr and set_fcsr.
1174 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1175 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1176 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1177 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1178 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1181 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1183 * config/mips/mips.c (mips_one_only_stub): New class.
1184 (mips_need_mips16_rdhwr_p): Replace with...
1185 (mips16_rdhwr_stub): ...this new variable.
1186 (mips16_stub_call_address): New function.
1187 (mips16_rdhwr_one_only_stub): New class.
1188 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1189 (mips_output_mips16_rdhwr): Delete.
1190 (mips_finish_stub): New function.
1191 (mips_code_end): Use it to handle rdhwr stubs.
1193 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1196 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1197 when calculating size of integer atomic types.
1199 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1201 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1203 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1205 PR tree-optimization/60003
1206 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1207 * profile.c (branch_prob): Use gimple_call_builtin_p
1208 to check for BUILT_IN_SETJMP_RECEIVER.
1209 * tree-inline.c (copy_bb): Call notice_special_calls.
1211 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1214 * lra-constraints.c (process_alt_operands): Update reload_sum only
1217 2014-01-31 Richard Henderson <rth@redhat.com>
1220 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1221 until after else_eh is processed.
1223 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1225 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1226 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1227 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1228 in smmintrin.h, remove them.
1229 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1230 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1231 * config/i386/i386.md (ROUND_SAE): Fix value.
1232 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1233 (const48_operand): New.
1234 * config/i386/subst.md (round), (round_expand): Use
1235 const_4_or_8_to_11_operand.
1236 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1238 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1240 * config/i386/constraints.md (Yk): Swap meaning with k.
1241 * config/i386/i386.md (movhi_internal): Change Yk to k.
1242 (movqi_internal): Ditto.
1243 (*k<logic><mode>): Ditto.
1246 (kandn<mode>): Ditto.
1247 (*<code>hi_1): Ditto.
1248 (*<code>qi_1): Ditto.
1249 (kxnor<mode>): Ditto.
1250 (kortestzhi): Ditto.
1251 (kortestchi): Ditto.
1253 (*one_cmplhi2_1): Ditto.
1254 (*one_cmplqi2_1): Ditto.
1255 * config/i386/sse.md (): Change k to Yk.
1256 (avx512f_load<mode>_mask): Ditto.
1257 (avx512f_blendm<mode>): Ditto.
1258 (avx512f_store<mode>_mask): Ditto.
1259 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1260 (avx512f_storedqu<mode>_mask): Ditto.
1261 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1263 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1264 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1265 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1266 (avx512f_maskcmp<mode>3): Ditto.
1267 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1268 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1269 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1270 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1271 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1272 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1273 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1274 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1275 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1276 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1277 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1278 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1279 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1280 (vec_extract_lo_<mode>_maskm): Ditto.
1281 (vec_extract_hi_<mode>_maskm): Ditto.
1282 (avx512f_vternlog<mode>_mask): Ditto.
1283 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1284 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1285 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1286 (avx512f_<code>v8div16qi2_mask): Ditto.
1287 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1288 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1289 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1290 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1291 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1292 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1293 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1294 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1295 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1296 (avx512cd_maskb_vec_dupv8di): Ditto.
1297 (avx512cd_maskw_vec_dupv16si): Ditto.
1298 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1299 (avx512f_vpermi2var<mode>3_mask): Ditto.
1300 (avx512f_vpermi2var<mode>3_mask): Ditto.
1301 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1302 (*avx512f_gathersi<mode>): Ditto.
1303 (*avx512f_gathersi<mode>_2): Ditto.
1304 (*avx512f_gatherdi<mode>): Ditto.
1305 (*avx512f_gatherdi<mode>_2): Ditto.
1306 (*avx512f_scattersi<mode>): Ditto.
1307 (*avx512f_scatterdi<mode>): Ditto.
1308 (avx512f_compress<mode>_mask): Ditto.
1309 (avx512f_compressstore<mode>_mask): Ditto.
1310 (avx512f_expand<mode>_mask): Ditto.
1311 * config/i386/subst.md (mask): Change k to Yk.
1312 (mask_scalar_merge): Ditto.
1315 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1317 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1319 2014-01-31 Richard Biener <rguenther@suse.de>
1322 * builtins.c (fold_builtin_memory_op): Make sure to not
1323 use a floating-point mode or a boolean or enumeral type for
1326 2014-01-30 DJ Delorie <dj@redhat.com>
1328 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1329 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1330 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1331 whenever main() has an epilogue.
1333 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1335 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1336 unused variable "field".
1337 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1338 (vsx_mergeh_<mode>): Likewise.
1339 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1340 (altivec_vmrghh): Likewise.
1341 (altivec_vmrghw): Likewise.
1342 (altivec_vmrglb): Likewise.
1343 (altivec_vmrglh): Likewise.
1344 (altivec_vmrglw): Likewise.
1345 (altivec_vspltb): Add missing uses.
1346 (altivec_vsplth): Likewise.
1347 (altivec_vspltw): Likewise.
1348 (altivec_vspltsf): Likewise.
1350 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1353 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1354 frame related instructions.
1356 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1358 PR rtl-optimization/59959
1359 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1360 any reload of register whose subreg is invalid.
1362 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1364 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1365 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1366 Add missing return type - void.
1368 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1370 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1371 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1372 remove element index adjustment for endian (now handled in vsx.md
1374 (altivec_expand_vec_perm_const): Use
1375 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1376 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1377 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1378 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1379 define_expand and a new define_insn *altivec_vspltb_internal;
1380 adjust for -maltivec=be on a little endian target.
1381 (altivec_vspltb_direct): New.
1382 (altivec_vsplth): Divide into a define_expand and a new
1383 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1384 little endian target.
1385 (altivec_vsplth_direct): New.
1386 (altivec_vspltw): Divide into a define_expand and a new
1387 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1388 little endian target.
1389 (altivec_vspltw_direct): New.
1390 (altivec_vspltsf): Divide into a define_expand and a new
1391 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1392 a little endian target.
1394 2014-01-30 Richard Biener <rguenther@suse.de>
1396 PR tree-optimization/59993
1397 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1398 can propagate form the earlier stmt and avoid the transform
1399 when the intermediate result is needed.
1401 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1403 * README.Portability: Fix typo.
1405 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1407 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1408 comparison_operator with ordered_comparison_operator.
1410 2014-01-30 Nick Clifton <nickc@redhat.com>
1412 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1413 Rename to mn10300_store_multiple_regs.
1414 * config/mn10300/mn10300.c: Likewise.
1415 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1416 store_multiple_regs.
1417 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1418 Call mn10300_store_multiple_regs.
1420 2014-01-30 Nick Clifton <nickc@redhat.com>
1421 DJ Delorie <dj@redhat.com>
1423 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1424 %fp 2 to keep registers after it properly word-aligned.
1425 (rl78_alloc_physical_registers_umul): Handle the case where both
1426 input operands are the same.
1428 2014-01-30 Richard Biener <rguenther@suse.de>
1430 PR tree-optimization/59903
1431 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1434 2014-01-30 Jason Merrill <jason@redhat.com>
1437 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1440 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1442 2014-01-30 Richard Biener <rguenther@suse.de>
1444 PR tree-optimization/59951
1445 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1447 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1450 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1451 SFmode to DFmode case.
1453 2014-01-29 DJ Delorie <dj@redhat.com>
1455 * config/msp430/msp430.opt (-minrt): New.
1456 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1458 (ENDFILE_SPEC): Likewise.
1460 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1462 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1463 (estimate_function_body_sizes): Use it.
1465 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1468 * dwarf2out.c (is_cxx_auto): New.
1469 (is_base_type): Use it.
1470 (gen_type_die_with_usage): Likewise.
1472 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1474 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1475 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1476 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1477 -maltivec=be with LE targets.
1478 (vsx_mergeh_<mode>): Likewise.
1479 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1480 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1481 (altivec_vmrghb): Replace with define_expand and new
1482 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1483 (altivec_vmrghb_direct): New define_insn.
1484 (altivec_vmrghh): Replace with define_expand and new
1485 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1486 (altivec_vmrghh_direct): New define_insn.
1487 (altivec_vmrghw): Replace with define_expand and new
1488 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1489 (altivec_vmrghw_direct): New define_insn.
1490 (*altivec_vmrghsf): Adjust for endianness.
1491 (altivec_vmrglb): Replace with define_expand and new
1492 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1493 (altivec_vmrglb_direct): New define_insn.
1494 (altivec_vmrglh): Replace with define_expand and new
1495 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1496 (altivec_vmrglh_direct): New define_insn.
1497 (altivec_vmrglw): Replace with define_expand and new
1498 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1499 (altivec_vmrglw_direct): New define_insn.
1500 (*altivec_vmrglsf): Adjust for endianness.
1501 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1502 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1503 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1504 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1505 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1506 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1507 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1508 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1510 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1512 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1513 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1516 2014-01-29 Richard Biener <rguenther@suse.de>
1518 PR tree-optimization/58742
1519 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1520 associate_pointerplus_align.
1521 (associate_pointerplus_diff): New function.
1522 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1523 and associate_pointerplus_diff.
1525 2014-01-29 Richard Biener <rguenther@suse.de>
1527 * lto-streamer.h (LTO_major_version): Bump to 3.
1528 (LTO_minor_version): Reset to 0.
1530 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1532 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1533 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1534 (arm_file_start): Generate correct asm header for armv7ve.
1535 * config/arm/bpabi.h: Add multilib support for armv7ve.
1536 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1537 and cortex-a15 to armv7ve.
1538 * config/arm/t-aprofile: Add multilib support for armv7ve.
1539 * doc/invoke.texi: Document -march=armv7ve.
1541 2014-01-29 Richard Biener <rguenther@suse.de>
1543 PR tree-optimization/58742
1544 * tree-ssa-forwprop.c (associate_plusminus): Return true
1545 if we changed sth, defer EH cleanup to ...
1546 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1547 (simplify_mult): New function.
1549 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1552 PR tree-optimization/59920
1553 * tree.c (build_common_builtin_nodes): Remove
1554 __builtin_setjmp_dispatcher initialization.
1555 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1556 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1557 instead of gsi_after_labels + manually skipping debug stmts.
1558 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1559 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1560 * tree-inline.c (copy_edges_for_bb): Remove
1561 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1562 argument. Ignore computed_goto_p stmts. Don't call
1563 make_abnormal_goto_edges. If a call might need abnormal edges
1564 for non-local gotos, see if it already has an edge to
1565 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1566 with true argument, don't do anything then, otherwise add
1567 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1568 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1570 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1571 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1572 (lower_stmt): Don't set data->calls_builtin_setjmp.
1573 (lower_builtin_setjmp): Adjust comment.
1574 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1575 * tree-cfg.c (found_computed_goto): Remove.
1576 (factor_computed_gotos): Remove.
1577 (make_goto_expr_edges): Return bool, true for computed gotos.
1578 Don't call make_abnormal_goto_edges.
1579 (build_gimple_cfg): Don't set found_computed_goto, don't call
1580 factor_computed_gotos.
1581 (computed_goto_p): No longer static.
1582 (make_blocks): Don't set found_computed_goto.
1583 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1584 (make_edges): If make_goto_expr_edges returns true, push bb
1585 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1586 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1587 vector. Record mapping between bbs and OpenMP regions if there
1588 are any, adjust make_gimple_omp_edges caller. Call
1589 handle_abnormal_edges.
1590 (make_abnormal_goto_edges): Remove.
1591 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1592 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1593 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1594 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1595 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1596 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1597 filling *region also set *region_idx to (*region)->entry->index.
1600 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1601 For REGs set ORIGINAL_REGNO.
1603 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1605 * doc/md.texi: Mention that a target shouldn't implement
1606 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1609 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1611 PR tree-optimization/59594
1612 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1613 a copy of the datarefs vector rather than the vector itself.
1615 2014-01-28 Jason Merrill <jason@redhat.com>
1618 * dwarf2out.c (auto_die): New static.
1619 (gen_type_die_with_usage): Handle C++1y 'auto'.
1620 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1623 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1626 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1627 (SPEC_X32): Likewise.
1628 (SPEC_64): Likewise.
1629 * config/i386/i386.c (ix86_option_override_internal): Turn off
1630 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1632 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1633 * config/i386/i386.h (TARGET_16BIT): New macro.
1634 (TARGET_16BIT_P): Likewise.
1635 * config/i386/i386.opt: Add m16.
1636 * doc/invoke.texi: Document -m16.
1638 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1640 PR preprocessor/59935
1641 * input.c (location_get_source_line): Bail out on when line number
1642 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1644 2014-01-28 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/58742
1647 * tree-ssa-forwprop.c (associate_plusminus): Handle
1648 pointer subtraction of the form (T)(P + A) - (T)P.
1650 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1652 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1655 2014-01-28 Richard Biener <rguenther@suse.de>
1658 2014-01-28 Richard Biener <rguenther@suse.de>
1660 PR rtl-optimization/45364
1661 PR rtl-optimization/59890
1662 * var-tracking.c (local_get_addr_clear_given_value): Handle
1663 already cleared slot.
1664 (val_reset): Handle not allocated local_get_addr_cache.
1665 (vt_find_locations): Use post-order on the inverted CFG.
1667 2014-01-28 Richard Biener <rguenther@suse.de>
1669 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1671 2014-01-28 Richard Biener <rguenther@suse.de>
1673 PR rtl-optimization/45364
1674 PR rtl-optimization/59890
1675 * var-tracking.c (local_get_addr_clear_given_value): Handle
1676 already cleared slot.
1677 (val_reset): Handle not allocated local_get_addr_cache.
1678 (vt_find_locations): Use post-order on the inverted CFG.
1680 2014-01-28 Alan Modra <amodra@gmail.com>
1682 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1683 * configure.ac <recursive call for build != host>: Define
1684 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1685 and LD_FOR_BUILD too.
1686 * configure: Regenerate.
1688 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1690 * config/i386/i386.c (get_builtin_code_for_version): Separate
1691 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1692 Broadwell from Haswell.
1694 2014-01-27 Steve Ellcey <sellcey@mips.com>
1696 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1697 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1698 * config/mips/mips.c (mips_option_override): Change setting
1700 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1701 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1702 Change from Mask to Var.
1704 2014-01-27 Jeff Law <law@redhat.com>
1706 * ipa-inline.c (inline_small_functions): Fix typo.
1708 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1710 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1711 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1712 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1713 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1714 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1715 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1716 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1717 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1718 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1719 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1720 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1721 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1722 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1723 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1724 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1725 (_mm512_storeu_epi64): Ditto.
1726 (_mm512_cmpge_epi32_mask): Ditto.
1727 (_mm512_cmpge_epu32_mask): Ditto.
1728 (_mm512_cmpge_epi64_mask): Ditto.
1729 (_mm512_cmpge_epu64_mask): Ditto.
1730 (_mm512_cmple_epi32_mask): Ditto.
1731 (_mm512_cmple_epu32_mask): Ditto.
1732 (_mm512_cmple_epi64_mask): Ditto.
1733 (_mm512_cmple_epu64_mask): Ditto.
1734 (_mm512_cmplt_epi32_mask): Ditto.
1735 (_mm512_cmplt_epu32_mask): Ditto.
1736 (_mm512_cmplt_epi64_mask): Ditto.
1737 (_mm512_cmplt_epu64_mask): Ditto.
1738 (_mm512_cmpneq_epi32_mask): Ditto.
1739 (_mm512_cmpneq_epu32_mask): Ditto.
1740 (_mm512_cmpneq_epi64_mask): Ditto.
1741 (_mm512_cmpneq_epu64_mask): Ditto.
1742 (_mm512_expand_pd): Ditto.
1743 (_mm512_expand_ps): Ditto.
1744 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1745 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1746 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1747 * config/i386/i386.c (ix86_builtins): Add
1748 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1749 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1750 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1751 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1752 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1753 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1754 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1755 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1756 IX86_BUILTIN_PMOVUSQW512_MEM.
1757 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1758 __builtin_ia32_pmovsqd512mem_mask,
1759 __builtin_ia32_pmovqd512mem_mask,
1760 __builtin_ia32_pmovusqw512mem_mask,
1761 __builtin_ia32_pmovsqw512mem_mask,
1762 __builtin_ia32_pmovqw512mem_mask,
1763 __builtin_ia32_pmovusdw512mem_mask,
1764 __builtin_ia32_pmovsdw512mem_mask,
1765 __builtin_ia32_pmovdw512mem_mask,
1766 __builtin_ia32_pmovqb512mem_mask,
1767 __builtin_ia32_pmovusqb512mem_mask,
1768 __builtin_ia32_pmovsqb512mem_mask,
1769 __builtin_ia32_pmovusdb512mem_mask,
1770 __builtin_ia32_pmovsdb512mem_mask,
1771 __builtin_ia32_pmovdb512mem_mask.
1772 (bdesc_args): Add __builtin_ia32_expanddf512,
1773 __builtin_ia32_expandsf512.
1774 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1775 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1776 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1777 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1778 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1779 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1780 (avx512f_<code>v8div16qi2_mask_store): This.
1781 (avx512f_expand<mode>): New.
1783 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1785 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1787 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1788 (_mm512_prefetch_i32scatter_pd): Ditto.
1789 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1790 (_mm512_prefetch_i64scatter_pd): Ditto.
1791 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1792 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1793 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1794 (_mm512_prefetch_i32scatter_ps): Ditto.
1795 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1796 (_mm512_prefetch_i64scatter_ps): Ditto.
1797 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1798 * config/i386/i386-builtin-types.def: Define
1799 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1800 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1801 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1802 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1803 IX86_BUILTIN_SCATTERPFQPD.
1804 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1805 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1806 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1807 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1808 __builtin_ia32_scatterpfqps.
1809 (ix86_expand_builtin): Expand new built-ins.
1810 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1811 fix memory access data type.
1812 (*avx512pf_gatherpf<mode>_mask): Ditto.
1813 (*avx512pf_gatherpf<mode>): Ditto.
1814 (avx512pf_scatterpf<mode>): Ditto.
1815 (*avx512pf_scatterpf<mode>_mask): Ditto.
1816 (*avx512pf_scatterpf<mode>): Ditto.
1817 (GATHER_SCATTER_SF_MEM_MODE): New.
1818 (avx512pf_gatherpf<mode>df): Ditto.
1819 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1820 (*avx512pf_scatterpf<mode>df): Ditto.
1822 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1825 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1826 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1829 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1831 * common/config/arm/arm-common.c
1832 (arm_rewrite_mcpu): Handle multiple names.
1834 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1836 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1838 * gimple-builder.h (create_gimple_tmp): Delete.
1840 2014-01-27 Christian Bruel <christian.bruel@st.com>
1842 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1845 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1847 * config/pa/pa.md (call): Generate indirect long calls to non-local
1848 functions when outputing 32-bit code.
1849 (call_value): Likewise except for special call to buggy powf function.
1851 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1852 portable runtime and PIC indirect calls.
1853 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1854 and PIC call sequences. Use ldo instead of blr to set return register
1855 in PIC call sequence.
1857 2014-01-25 Walter Lee <walt@tilera.com>
1859 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1860 avoid clobbering a live register.
1862 2014-01-25 Walter Lee <walt@tilera.com>
1864 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1865 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1866 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1867 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1869 2014-01-25 Walter Lee <walt@tilera.com>
1871 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1872 arguments on even registers.
1873 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1875 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1876 (BIGGEST_ALIGNMENT): Ditto.
1877 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1879 2014-01-25 Walter Lee <walt@tilera.com>
1881 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1882 insns before bundling.
1883 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1885 2014-01-25 Walter Lee <walt@tilera.com>
1887 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1888 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1889 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1891 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1893 * config/mips/constraints.md (kl): Delete.
1894 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1895 define expands, using...
1896 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1897 instructions for MIPS16.
1898 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1899 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1901 2014-01-25 Walter Lee <walt@tilera.com>
1903 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1907 2014-01-25 Walter Lee <walt@tilera.com>
1909 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1910 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1912 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1914 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1917 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1919 * print-rtl.c (in_call_function_usage): New var.
1920 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1921 EXPR_LIST mode as mode and not as reg note name.
1924 * cfgloopmanip.c (copy_loop_info): If
1925 loop->warned_aggressive_loop_optimizations, make sure
1926 the flag is set in target loop too.
1928 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1930 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1932 * builtins.def: Likewise.
1933 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1934 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1935 * ira.c (ira_setup_eliminable_regset): Likewise.
1936 * omp-low.c (gate_expand_omp): Likewise.
1937 (execute_lower_omp): Likewise.
1938 (diagnose_sb_0): Likewise.
1939 (gate_diagnose_omp_blocks): Likewise.
1940 (simd_clone_clauses_extract): Likewise.
1943 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1945 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1946 correction for little endian...
1947 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1950 2014-01-24 Jeff Law <law@redhat.com>
1952 PR tree-optimization/59919
1953 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1954 for non-returning calls.
1956 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1958 * common/config/aarch64/aarch64-common.c
1959 (aarch64_rewrite_mcpu): Handle multiple names.
1960 * config/aarch64/aarch64.h
1961 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1963 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1965 * input.c (add_file_to_cache_tab): Handle the case where fopen
1968 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1971 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1972 from push operand if code of push isn't PRE_DEC.
1974 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1977 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1978 -mquad-memory-atomic. Update -mquad-memory documentation to say
1979 it is only used for non-atomic loads/stores.
1981 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1982 -mquad-memory or -mquad-memory-atomic switches.
1984 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1985 -mquad-memory-atomic to ISA 2.07 support.
1987 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1988 to separate support of normal quad word memory operations (ldq, stq)
1989 from the atomic quad word memory operations.
1991 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1992 support to separate non-atomic quad word operations from atomic
1993 quad word operations. Disable non-atomic quad word operations in
1994 little endian mode so that we don't have to swap words after the
1995 load and before the store.
1996 (quad_load_store_p): Add comment about atomic quad word support.
1997 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1998 options printed with -mdebug=reg.
2000 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2001 -mquad-memory-atomic as the test for whether we have quad word
2002 atomic instructions.
2003 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2004 or -mp8-vector are used, allow byte/half-word atomic operations.
2006 * config/rs6000/sync.md (load_lockedti): Insure that the address
2007 is a proper indexed or indirect address for the lqarx instruction.
2008 On little endian systems, swap the hi/lo registers after the lqarx
2010 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2011 insure the address is valid for the lqarx instruction.
2012 (store_conditionalti): Insure that the address is a proper indexed
2013 or indirect address for the stqcrx. instruction. On little endian
2014 systems, swap the hi/lo registers before doing the stqcrx.
2016 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2017 insure the address is valid for the stqcrx. instruction.
2019 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2020 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2021 type of quad memory support is available.
2023 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2026 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2027 there is a danger of looping.
2029 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2031 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2032 force flag_ira_loop_pressure if set via command line.
2034 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2036 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2037 (ashr_simd): New builtin handling DI mode.
2038 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2039 (aarch64_sshr_simddi): New match pattern.
2040 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2041 (vshrd_n_s64): Likewise.
2042 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2044 2014-01-23 Nick Clifton <nickc@redhat.com>
2046 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2047 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2048 favour of mcu specific scripts.
2049 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2052 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2053 Alex Velenko <Alex.Velenko@arm.com>
2055 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2056 (vaddv_s16): Likewise.
2057 (vaddv_s32): Likewise.
2058 (vaddv_u8): Likewise.
2059 (vaddv_u16): Likewise.
2060 (vaddv_u32): Likewise.
2061 (vaddvq_s8): Likewise.
2062 (vaddvq_s16): Likewise.
2063 (vaddvq_s32): Likewise.
2064 (vaddvq_s64): Likewise.
2065 (vaddvq_u8): Likewise.
2066 (vaddvq_u16): Likewise.
2067 (vaddvq_u32): Likewise.
2068 (vaddvq_u64): Likewise.
2069 (vaddv_f32): Likewise.
2070 (vaddvq_f32): Likewise.
2071 (vaddvq_f64): Likewise.
2072 (vmaxv_f32): Likewise.
2073 (vmaxv_s8): Likewise.
2074 (vmaxv_s16): Likewise.
2075 (vmaxv_s32): Likewise.
2076 (vmaxv_u8): Likewise.
2077 (vmaxv_u16): Likewise.
2078 (vmaxv_u32): Likewise.
2079 (vmaxvq_f32): Likewise.
2080 (vmaxvq_f64): Likewise.
2081 (vmaxvq_s8): Likewise.
2082 (vmaxvq_s16): Likewise.
2083 (vmaxvq_s32): Likewise.
2084 (vmaxvq_u8): Likewise.
2085 (vmaxvq_u16): Likewise.
2086 (vmaxvq_u32): Likewise.
2087 (vmaxnmv_f32): Likewise.
2088 (vmaxnmvq_f32): Likewise.
2089 (vmaxnmvq_f64): Likewise.
2090 (vminv_f32): Likewise.
2091 (vminv_s8): Likewise.
2092 (vminv_s16): Likewise.
2093 (vminv_s32): Likewise.
2094 (vminv_u8): Likewise.
2095 (vminv_u16): Likewise.
2096 (vminv_u32): Likewise.
2097 (vminvq_f32): Likewise.
2098 (vminvq_f64): Likewise.
2099 (vminvq_s8): Likewise.
2100 (vminvq_s16): Likewise.
2101 (vminvq_s32): Likewise.
2102 (vminvq_u8): Likewise.
2103 (vminvq_u16): Likewise.
2104 (vminvq_u32): Likewise.
2105 (vminnmv_f32): Likewise.
2106 (vminnmvq_f32): Likewise.
2107 (vminnmvq_f64): Likewise.
2109 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2111 * config/aarch64/aarch64-simd.md
2112 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2113 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2114 (*aarch64_mul3_elt<mode>): Likewise.
2115 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2116 (*aarch64_mul3_elt_to_64v2df): Likewise.
2117 (*aarch64_mla_elt<mode>): Likewise.
2118 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2119 (*aarch64_mls_elt<mode>): Likewise.
2120 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2121 (*aarch64_fma4_elt<mode>): Likewise.
2122 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2123 (*aarch64_fma4_elt_to_64v2df): Likewise.
2124 (*aarch64_fnma4_elt<mode>): Likewise.
2125 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2126 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2127 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2128 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2129 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2130 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2131 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2132 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2133 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2135 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2137 * config/aarch64/aarch64-simd.md
2138 (aarch64_be_checked_get_lane<mode>): New define_expand.
2139 * config/aarch64/aarch64-simd-builtins.def
2140 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2141 New builtin definition.
2142 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2143 Use new safe be builtin.
2145 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2147 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2149 (aarch64_be_st1<mode>): Likewise.
2150 (aarch_ld1<VALL:mode>): Define_expand modified.
2151 (aarch_st1<VALL:mode>): Likewise.
2152 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2153 (UNSPEC_ST1): Likewise.
2155 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2157 * config/microblaze/microblaze.md: Add trap insn and attribute
2159 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2161 PR preprocessor/58580
2162 * input.h (location_get_source_line): Take an additional line_size
2164 (void diagnostics_file_cache_fini): Declare new function.
2165 * input.c (struct fcache): New type.
2166 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2167 New static constants.
2168 (diagnostic_file_cache_init, total_lines_num)
2169 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2170 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2171 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2172 (get_next_line, read_next_line, goto_next_line, read_line_num):
2173 New static function definitions.
2174 (diagnostic_file_cache_fini): New function.
2175 (location_get_source_line): Take an additional output line_len
2176 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2178 * diagnostic.c (diagnostic_finish): Call
2179 diagnostic_file_cache_fini.
2180 (adjust_line): Take an additional input parameter for the length
2181 of the line, rather than calculating it with strlen.
2182 (diagnostic_show_locus): Adjust the use of
2183 location_get_source_line and adjust_line with respect to their new
2184 signature. While displaying a line now, do not stop at the first
2185 null byte. Rather, display the zero byte as a space and keep
2186 going until we reach the size of the line.
2187 * Makefile.in: Add vec.o to OBJS-libcommon
2189 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2190 Ilya Tocar <ilya.tocar@intel.com>
2192 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2193 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2194 (__builtin_ia32_kmov16): Ditto.
2195 * config/i386/i386.md (UNSPEC_KMOV): New.
2198 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2200 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2201 (_mm512_storeu_si512): Ditto.
2203 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2206 * rtl.h (get_referenced_operands): Declare.
2207 * recog.c (get_referenced_operands): New function.
2208 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2209 operands have been referenced when recording LO_SUM references.
2211 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2213 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2215 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2217 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2218 Enable for generic and recent AMD targets.
2220 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2222 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2223 ARG_SIZE note when adjustment was eliminated.
2225 2014-01-22 Jeff Law <law@redhat.com>
2227 PR tree-optimization/59597
2228 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2229 in file. Accept new argument REGISTERING and use it to modify
2230 dump output appropriately.
2231 (register_jump_thread): Corresponding changes.
2232 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2233 thread paths involving joiner blocks. Add code to dump cancelled
2234 jump threading paths.
2236 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2238 PR rtl-optimization/59477
2239 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2240 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2242 2014-01-22 Tom Tromey <tromey@redhat.com>
2244 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2246 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2248 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2250 PR rtl-optimization/59896
2251 * lra-constraints.c (process_alt_operands): Check unused note for
2252 matched operands of insn with no output reloads.
2254 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2256 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2257 (mips_move_from_gpr_cost): Likewise.
2259 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2261 PR rtl-optimization/59858
2262 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2263 ira_class_hard_regs_num.
2264 (process_alt_operands): Increase reject for dying matched operand.
2266 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2269 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2270 smaller than size, perform several stores or loads and stores
2271 at dst + count - size to store or copy all of size bytes, rather
2272 than just last modesize bytes.
2274 2014-01-20 DJ Delorie <dj@redhat.com>
2276 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2277 that CLOBBERs are REGs before propogating their values.
2279 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2282 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2283 (cgraph_inline_failed_type): New function.
2284 * cgraph.h (DEFCIFCODE): Add type.
2285 (cgraph_inline_failed_type_t): New enum.
2286 (cgraph_inline_failed_type): New prototype.
2287 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2288 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2289 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2290 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2291 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2292 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2293 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2294 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2295 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2296 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2297 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2298 OPTIMIZATION_MISMATCH.
2299 * tree-inline.c (expand_call_inline): Emit errors during
2300 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2302 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2305 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2306 mode attribute in insn output.
2308 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2310 * output.h (output_constant): Delete.
2311 * varasm.c (output_constant): Make private.
2313 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2315 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2317 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2320 * tree.h (fold_builtin_strcat): New prototype.
2321 * builtins.c (fold_builtin_strcat): No longer static. Add len
2322 argument, if non-NULL, don't call c_strlen. Optimize
2323 directly into __builtin_memcpy instead of __builtin_strcpy.
2324 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2325 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2327 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2329 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2330 for SImode_address_operand operands, having only a REG argument.
2332 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2334 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2335 loader name using mbig-endian.
2336 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2338 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2340 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2344 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2346 * config/aarch64/aarch64-protos.h
2347 (aarch64_cannot_change_mode_class_ptr): Declare.
2348 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2349 aarch64_cannot_change_mode_class_ptr): New.
2350 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2351 backend hook aarch64_cannot_change_mode_class.
2353 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2355 * common/config/aarch64/aarch64-common.c
2356 (aarch64_handle_option): Don't handle any option order logic here.
2357 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2358 selected_cpu, warn on architecture version mismatch.
2359 (aarch64_override_options): Fix parsing order for option strings.
2361 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2362 Iain Sandoe <iain@codesourcery.com>
2365 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2366 warning. Amend comment to reflect current functionality.
2368 2014-01-20 Richard Biener <rguenther@suse.de>
2371 * builtins.c (fold_builtin_strcat): Remove case better handled
2372 by tree-ssa-strlen.c.
2374 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2376 * config/aarch64/aarch64.opt
2377 (mcpu, march, mtune): Make case-insensitive.
2379 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2382 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2383 if operands[1] is a REG or ZERO_EXTEND of a REG.
2385 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2387 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2389 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2391 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2392 long non-pic millicode calls.
2394 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2396 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2398 2014-01-19 Kito Cheng <kito@0xlab.org>
2400 * builtins.c (expand_movstr): Check movstr expand done or fail.
2402 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2403 H.J. Lu <hongjiu.lu@intel.com>
2406 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2407 to DImode for zero-extended addresses.
2409 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2411 PR rtl-optimization/57763
2412 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2413 on the new indirect jump_insn and increment LABEL_NUSES (label).
2415 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2419 * config.gcc (x86_archs): New variable.
2420 (x86_64_archs): Likewise.
2421 (x86_cpus): Likewise.
2422 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2423 --with-arch/--with-cpu= options.
2424 Support --with-arch=/--with-cpu={nehalem,westmere,
2425 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2427 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2429 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2430 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2432 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2434 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2436 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2439 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2440 clear cpp_get_options (parse_in)->warn_unused_macros for
2441 ix86_target_macros_internal with cpp_define.
2443 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2445 * jump.c (delete_related_insns): Keep (use (insn))s.
2446 * reorg.c (redundant_insn): Check for barriers too.
2448 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2450 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2452 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2454 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2455 call to $$dyncall when TARGET_LONG_CALLS is true.
2457 2014-01-17 Jeff Law <law@redhat.com>
2459 * ree.c (combine_set_extension): Temporarily disable test for
2460 changing number of hard registers.
2462 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2465 * ipa-inline-analysis.c (inline_free_summary):
2466 Do not free summary of aliases.
2468 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2471 * gimplify.c (gimplify_expr): Use create_tmp_var
2472 instead of create_tmp_var_raw. If cond doesn't have
2473 integral type, don't add the IFN_ANNOTATE builtin at all.
2475 2014-01-17 Martin Jambor <mjambor@suse.cz>
2478 * ipa-cp.c (prev_edge_clone): New variable.
2479 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2480 Also resize prev_edge_clone vector.
2481 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2482 (ipcp_edge_removal_hook): New function.
2483 (ipcp_driver): Register ipcp_edge_removal_hook.
2485 2014-01-17 Andrew Pinski <apinski@cavium.com>
2486 Steve Ellcey <sellcey@mips.com>
2489 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2492 2014-01-17 Jeff Law <law@redhat.com>
2495 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2496 so that pass_ccp runs first.
2498 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2500 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2501 (ix86_adjust_cost): Use !TARGET_XXX.
2502 (do_reorder_for_imul): Likewise.
2503 (swap_top_of_ready_list): Likewise.
2504 (ix86_sched_reorder): Likewise.
2506 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2508 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2509 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2510 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2511 (intel_memset): New. Duplicate slm_memset.
2512 (intel_cost): New. Duplicate slm_cost.
2513 (m_INTEL): New macro.
2514 (processor_target_table): Add "intel".
2515 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2516 with PROCESSOR_INTEL for "intel".
2517 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2518 PROCESSOR_SILVERMONT.
2519 (ix86_issue_rate): Likewise.
2520 (ix86_adjust_cost): Likewise.
2521 (ia32_multipass_dfa_lookahead): Likewise.
2522 (swap_top_of_ready_list): Likewise.
2523 (ix86_sched_reorder): Likewise.
2524 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2525 instead of TARGET_OPT_AGU.
2526 * config/i386/i386.h (TARGET_INTEL): New.
2527 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2528 (processor_type): Add PROCESSOR_INTEL.
2529 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2530 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2532 2014-01-17 Marek Polacek <polacek@redhat.com>
2535 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2538 2014-01-17 Richard Biener <rguenther@suse.de>
2540 PR tree-optimization/46590
2541 * opts.c (default_options_table): Add entries for
2542 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2543 all enabled at -O1 but not for -Og.
2544 * common.opt (fbranch-count-reg): Remove Init(1).
2545 (fmove-loop-invariants): Likewise.
2546 (ftree-pta): Likewise.
2548 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2550 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2551 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2552 decls to at least the GCC 4.8 used alignments.
2555 * tree-nested.c (convert_nonlocal_reference_stmt,
2556 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2557 of GIMPLE_BIND stmts, adjust associated decls.
2559 2014-01-17 Richard Biener <rguenther@suse.de>
2561 PR tree-optimization/46590
2562 * vec.h (vec<>::bseach): New member function implementing
2563 binary search according to C89 bsearch.
2564 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2565 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2566 bitmap pointer again. Make accesses_in_loop a flat array.
2567 (mem_ref_obstack): New global.
2568 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2569 (mark_ref_stored): Likewise.
2570 (ref_indep_loop_p_2): Likewise.
2571 (set_ref_stored_in_loop): New helper function.
2572 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2573 (memref_free): Adjust.
2574 (record_mem_ref_loc): Simplify.
2575 (gather_mem_refs_stmt): Adjust.
2576 (sort_locs_in_loop_postorder_cmp): New function.
2577 (analyze_memory_references): Sort accesses_in_loop after
2578 loop postorder number.
2579 (find_ref_loc_in_loop_cmp): New function.
2580 (for_all_locs_in_loop): Find relevant cluster of locs in
2581 accesses_in_loop and iterate without recursion.
2582 (execute_sm): Avoid uninit warning.
2583 (struct ref_always_accessed): Simplify.
2584 (ref_always_accessed::operator ()): Likewise.
2585 (ref_always_accessed_p): Likewise.
2586 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2587 loop postorder numbers here.
2588 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2591 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2594 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2595 on decls for which assemble_alias has been called.
2597 2014-01-17 Nick Clifton <nickc@redhat.com>
2599 * config/msp430/msp430.opt: (mcpu): New option.
2600 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2601 (msp430_option_override): Parse target_cpu. If the MCU name
2602 matches a generic string, clear target_mcu.
2603 (msp430_attr): Allow numeric interrupt values up to 63.
2604 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2605 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2607 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2609 * config/msp430/msp430.md (popm): Use %J rather than %I.
2610 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2611 (addhi_cy_i): Use immediate_operand for operand 2.
2612 * doc/invoke.texi: Document -mcpu option.
2614 2014-01-17 Richard Biener <rguenther@suse.de>
2616 PR rtl-optimization/38518
2617 * df.h (df_analyze_loop): Declare.
2618 * df-core.c: Include cfgloop.h.
2619 (df_analyze_1): Split out main part of df_analyze.
2620 (df_analyze): Adjust.
2621 (loop_inverted_post_order_compute): New function.
2622 (loop_post_order_compute): Likewise.
2623 (df_analyze_loop): New function avoiding whole-function
2625 * loop-invariant.c (find_defs): Use df_analyze_loop.
2626 (find_invariants): Adjust.
2627 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2629 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2631 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2632 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2634 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2636 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2637 traversal when removing references.
2639 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2642 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2644 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2647 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2648 pushing a reload for an autoinc when we had previously reloaded an
2649 inner part of the address.
2651 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2653 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2655 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2656 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2657 when not giving up or versioning for alias only because of
2659 (vect_analyze_data_ref_dependences): Set to true.
2660 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2662 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2663 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2667 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2670 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2671 operand 0 predicate for gathers, use a new pseudo as subtarget.
2673 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2676 * lra-constraints.c (process_alt_operands): Add printing debug info.
2677 Check absence of input/output reloads for matched operands too.
2679 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2681 PR rtl-optimization/59835
2682 * ira.c (ira_init_register_move_cost): Increase cost for
2685 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2687 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2689 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2692 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2693 non-register objects. Use gen_(high/low)part more consistently.
2696 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2699 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2700 endian support, remove tests for WORDS_BIG_ENDIAN.
2701 (p8_mfvsrd_3_<mode>): Likewise.
2702 (reload_gpr_from_vsx<mode>): Likewise.
2703 (reload_gpr_from_vsxsf): Likewise.
2704 (p8_mfvsrd_4_disf): Likewise.
2706 2014-01-16 Richard Biener <rguenther@suse.de>
2708 PR rtl-optimization/46590
2709 * lcm.c (compute_antinout_edge): Use postorder iteration.
2710 (compute_laterin): Use inverted postorder iteration.
2712 2014-01-16 Nick Clifton <nickc@redhat.com>
2715 * varasm.c (output_constant): Return the number of bytes actually
2717 (output_constructor_array_range): Update the field size with the
2718 number of bytes emitted by output_constant.
2719 (output_constructor_regular_field): Likewise. Also do not
2720 complain if the total number of bytes emitted is now greater
2721 than the expected fieldpos.
2722 * output.h (output_constant): Update prototype and descriptive comment.
2724 2014-01-16 Marek Polacek <polacek@redhat.com>
2727 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2728 it is error_mark_node.
2730 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2732 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2733 VALID_AVX256_REG_OR_OI_MODE.
2735 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2737 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2738 current procedure should be profiled.
2740 2014-01-15 Andrew Pinski <apinski@cavium.com>
2742 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2743 of moving from/to the STACK_REG register class.
2745 2014-01-15 Richard Henderson <rth@redhat.com>
2748 * reginfo.c (global_regs_decl): Globalize.
2749 * rtl.h (global_regs_decl): Declare.
2750 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2751 reserved via global_regs.
2753 2014-01-15 Teresa Johnson <tejohnson@google.com>
2755 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2757 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2759 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2760 and vmulosh rather than call gen_vec_widen_smult_*.
2761 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2762 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2763 (vec_widen_smult_even_v16qi): Likewise.
2764 (vec_widen_umult_even_v8hi): Likewise.
2765 (vec_widen_smult_even_v8hi): Likewise.
2766 (vec_widen_umult_odd_v16qi): Likewise.
2767 (vec_widen_smult_odd_v16qi): Likewise.
2768 (vec_widen_umult_odd_v8hi): Likewise.
2769 (vec_widen_smult_odd_v8hi): Likewise.
2770 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2771 vmuloub rather than call gen_vec_widen_umult_*.
2772 (vec_widen_umult_lo_v16qi): Likewise.
2773 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2774 vmulosb rather than call gen_vec_widen_smult_*.
2775 (vec_widen_smult_lo_v16qi): Likewise.
2776 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2777 rather than call gen_vec_widen_umult_*.
2778 (vec_widen_umult_lo_v8hi): Likewise.
2779 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2780 rather than call gen_vec_widen_smult_*.
2781 (vec_widen_smult_lo_v8hi): Likewise.
2783 2014-01-15 Jeff Law <law@redhat.com>
2785 PR tree-optimization/59747
2786 * ree.c (find_and_remove_re): Properly handle case where a second
2787 eliminated extension requires widening a copy created for elimination
2788 of a prior extension.
2789 (combine_set_extension): Ensure that the number of hard regs needed
2790 for a destination register does not change when we widen it.
2792 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2794 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2795 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2796 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2797 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2798 (avr-*-rtems*): Likewise.
2799 (bfin*-rtems*): Likewise.
2800 (moxie-*-rtems*): Likewise.
2801 (h8300-*-rtems*): Likewise.
2802 (i[34567]86-*-rtems*): Likewise.
2803 (lm32-*-rtems*): Likewise.
2804 (m32r-*-rtems*): Likewise.
2805 (m68k-*-rtems*): Likewise.
2806 (microblaze*-*-rtems*): Likewise.
2807 (mips*-*-rtems*): Likewise.
2808 (powerpc-*-rtems*): Likewise.
2809 (sh-*-rtems*): Likewise.
2810 (sparc-*-rtems*): Likewise.
2811 (sparc64-*-rtems*): Likewise.
2812 (v850-*-rtems*): Likewise.
2813 (m32c-*-rtems*): Likewise.
2815 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2817 PR rtl-optimization/59511
2818 * ira.c (ira_init_register_move_cost): Use memory costs for some
2819 cases of register move cost calculations.
2820 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2821 instead of BB frequency.
2822 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2823 * lra-assigns.c (find_hard_regno_for): Ditto.
2825 2014-01-15 Richard Biener <rguenther@suse.de>
2827 PR tree-optimization/59822
2828 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2829 (vectorizable_load): Use it to hoist defs of uses of invariant
2830 loads out of the loop.
2832 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2833 Kugan Vivekanandarajah <kuganv@linaro.org>
2836 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2839 2014-01-15 Richard Biener <rguenther@suse.de>
2841 PR rtl-optimization/59802
2842 * lcm.c (compute_available): Use inverted postorder to seed
2843 the initial worklist.
2845 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2848 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2849 ADDR_REGS for invalid symrefs in non-PIC code.
2851 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2854 * builtins.c (determine_block_size): Initialize *probable_max_size
2855 even if len_rtx is CONST_INT.
2857 2014-01-14 Andrew Pinski <apinski@cavium.com>
2859 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2860 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2861 (cortexa53_tunings): Likewise.
2862 (aarch64_sched_issue_rate): New function.
2863 (TARGET_SCHED_ISSUE_RATE): Define.
2865 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2867 * ira-costs.c (find_costs_and_classes): Add missed
2868 ira_init_register_move_cost_if_necessary.
2870 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2873 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2875 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2878 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2879 to indicate if type is used for function return value. Warn ABI
2880 change if the vector mode isn't available for function return value.
2881 (ix86_function_arg_advance): Pass false to type_natural_mode.
2882 (ix86_function_arg): Likewise.
2883 (ix86_gimplify_va_arg): Likewise.
2884 (function_arg_32): Don't warn ABI change.
2885 (ix86_function_value): Pass true to type_natural_mode.
2886 (ix86_return_in_memory): Likewise.
2887 (ix86_struct_value_rtx): Removed.
2888 (TARGET_STRUCT_VALUE_RTX): Likewise.
2890 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2892 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2893 converting a conditional jump into a conditional return.
2895 2014-01-14 Richard Biener <rguenther@suse.de>
2897 PR tree-optimization/58921
2898 PR tree-optimization/59006
2899 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2900 hoisting invariant stmts.
2901 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2902 invariant loads on the preheader edge if possible.
2904 2014-01-14 Joey Ye <joey.ye@arm.com>
2906 * doc/plugin.texi (Building GCC plugins): Update to C++.
2908 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2910 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2911 (_mm_rcp28_round_ss): Ditto.
2912 (_mm_rsqrt28_round_sd): Ditto.
2913 (_mm_rsqrt28_round_ss): Ditto.
2914 (_mm_rcp28_sd): Ditto.
2915 (_mm_rcp28_ss): Ditto.
2916 (_mm_rsqrt28_sd): Ditto.
2917 (_mm_rsqrt28_ss): Ditto.
2918 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2919 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2920 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2921 (IX86_BUILTIN_RCP28SD): Ditto.
2922 (IX86_BUILTIN_RCP28SS): Ditto.
2923 (IX86_BUILTIN_RSQRT28SD): Ditto.
2924 (IX86_BUILTIN_RSQRT28SS): Ditto.
2925 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2926 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2927 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2928 (ix86_expand_special_args_builtin): Expand new FTYPE.
2929 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2930 (srcp14<mode>): Make insn unary.
2931 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2932 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2933 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2934 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2935 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2936 Fix rounding: make it SAE only.
2937 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2939 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2941 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2942 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2943 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2944 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2945 (round_saeonly_mask_scalar_operand4): Ditto.
2946 (round_saeonly_mask_scalar_op3): Ditto.
2947 (round_saeonly_mask_scalar_op4): Ditto.
2949 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2951 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2952 Implement -maltivec=be for vec_insert and vec_extract.
2954 2014-01-10 DJ Delorie <dj@redhat.com>
2956 * config/msp430/msp430.md (call_internal): Don't allow memory
2957 references with SP as the base register.
2958 (call_value_internal): Likewise.
2959 * config/msp430/constraints.md (Yc): New. For memory references
2960 that don't use SP as a base register.
2962 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2963 "an integer without a # prefix"
2964 * config/msp430/msp430.md (epilogue_helper): Use it.
2966 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2969 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2970 AVX512F gather builtins.
2971 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2972 on gather decls with INTEGER_TYPE masktype.
2973 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2974 directly into the builtin rather than hoisting it before loop.
2976 PR tree-optimization/59387
2977 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2978 (scev_const_prop): If folded_casts and type has undefined overflow,
2979 use force_gimple_operand instead of force_gimple_operand_gsi and
2980 for each added stmt if it is assign with
2981 arith_code_with_undefined_signed_overflow, call
2982 rewrite_to_defined_overflow.
2983 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2984 gimple-fold.h instead.
2985 (arith_code_with_undefined_signed_overflow,
2986 rewrite_to_defined_overflow): Moved to ...
2987 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2988 rewrite_to_defined_overflow): ... here. No longer static.
2989 Include gimplify-me.h.
2990 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2991 rewrite_to_defined_overflow): New prototypes.
2993 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2995 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2997 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2999 * builtins.c (get_object_alignment_2): Minor tweak.
3000 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3002 2014-01-13 Christian Bruel <christian.bruel@st.com>
3004 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3005 optimized non constant lengths.
3007 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3010 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3011 load as __atomic_load_N if possible.
3013 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3015 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3017 (rs6000_expand_builtin): Adjust call.
3019 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3022 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3023 * config/rs6000/rs6000.c: Include target-globals.h.
3024 (rs6000_set_current_function): Instead of doing target_reinit
3025 unconditionally, use save_target_globals_default_opts and
3026 restore_target_globals.
3028 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3030 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3031 (rs6000_expand_builtin): Handle mffs and mtfsf.
3032 (rs6000_init_builtins): Define mffs and mtfsf.
3033 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3034 (rs6000_mffs): New pattern.
3035 (rs6000_mtfsf): New pattern.
3037 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3039 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3040 Start narrowing with START. Apply candidate-use pair
3041 and check overall cost in narrowing.
3042 (iv_ca_prune): Pass new argument.
3044 2014-01-10 Jeff Law <law@redhat.com>
3047 * ree.c (combine_reaching_defs): Ensure the defining statement
3048 occurs before the extension when optimizing extensions with
3049 different source and destination hard registers.
3051 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3054 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3055 vtables into the type inheritance graph.
3057 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3059 PR rtl-optimization/59754
3060 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3061 modes in the REGNO != REGNO case.
3063 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3065 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3067 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3069 PR tree-optimization/59745
3070 * tree-predcom.c (tree_predictive_commoning_loop): Call
3071 free_affine_expand_cache if giving up because components is NULL.
3073 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3074 GC in payload of target_globals struct instead of allocating them on
3075 the heap and the larger structs separately using GC.
3076 * target-globals.h (struct target_globals): Make regs, hard_regs,
3077 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3078 of GTY((skip)) and change type to void *.
3079 (reset_target_globals): Cast loads from those fields to corresponding
3082 2014-01-10 Steve Ellcey <sellcey@mips.com>
3085 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3086 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3087 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3089 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3092 * aarch64-modes.def (CC_Zmode): New flags mode.
3093 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3094 represents an equality.
3095 (aarch64_get_condition_code): Handle CC_Zmode.
3096 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3098 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3100 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3101 extraction in good case.
3103 2014-01-10 Richard Biener <rguenther@suse.de>
3105 PR tree-optimization/59374
3106 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3107 checking after SLP discovery. Mark stmts not participating
3108 in any SLP instance properly.
3110 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3112 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3113 when handling a SET rtx.
3115 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3117 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3118 (cortex-a57): Likewise.
3119 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3121 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3123 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3124 non-iwmmxt builtins.
3126 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3130 * ipa-devirt.c record_target_from_binfo): Take as argument
3131 stack of binfos and lookup matching one for virtual inheritance.
3132 (possible_polymorphic_call_targets_1): Update.
3134 2014-01-10 Huacai Chen <chenhc@lemote.com>
3136 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3137 kernel strings for Loongson-2E/2F/3A.
3139 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3142 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3143 is_gimple_call before calling gimple_call_internal_p.
3145 2014-01-09 Steve Ellcey <sellcey@mips.com>
3147 * Makefile.in (TREE_FLOW_H): Remove.
3148 (TREE_SSA_H): Add file names from tree-flow.h.
3149 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3150 * tree.h: Remove tree-flow.h reference.
3151 * hash-table.h: Remove tree-flow.h reference.
3152 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3153 reference with tree-ssa-loop.h.
3155 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3157 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3158 default element-order behavior for -maltivec.
3159 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3160 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3161 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3162 when targeting big endian, at least for now.
3163 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3165 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3168 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3169 var satisfies use_register_for_decl, just take into account type
3170 alignment, rather than decl alignment.
3172 PR tree-optimization/59622
3173 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3174 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3175 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3176 Don't devirtualize for inplace at all. For targets.length () == 1,
3177 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3179 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3181 * config/i386/i386.md (cpu): Remove the unused btver1.
3183 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3185 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3187 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3190 * tree-core.h (struct target_globals): New forward declaration.
3191 (struct tree_target_option): Add globals field.
3192 * tree.h (TREE_TARGET_GLOBALS): Define.
3193 (prepare_target_option_nodes_for_pch): New prototype.
3194 * target-globals.h (struct target_globals): Define even if
3196 * tree.c (prepare_target_option_node_for_pch,
3197 prepare_target_option_nodes_for_pch): New functions.
3198 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3199 * config/i386/i386.c: Include target-globals.h.
3200 (ix86_set_current_function): Instead of doing target_reinit
3201 unconditionally, use save_target_globals_default_opts and
3202 restore_target_globals.
3204 2014-01-09 Richard Biener <rguenther@suse.de>
3206 PR tree-optimization/59715
3207 * tree-cfg.h (split_critical_edges): Declare.
3208 * tree-cfg.c (split_critical_edges): Export.
3209 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3211 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3213 * cfgexpand.c (expand_stack_vars): Optionally disable
3214 asan stack protection.
3215 (expand_used_vars): Likewise.
3216 (partition_stack_vars): Likewise.
3217 * asan.c (asan_emit_stack_protection): Optionally disable
3218 after return stack usage.
3219 (instrument_derefs): Optionally disable memory access instrumentation.
3220 (instrument_builtin_call): Likewise.
3221 (instrument_strlen_call): Likewise.
3222 (asan_protect_global): Optionally disable global variables protection.
3223 * doc/invoke.texi: Added doc for new options.
3224 * params.def: Added new options.
3225 * params.h: Likewise.
3227 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3229 PR rtl-optimization/59724
3230 * ifcvt.c (cond_exec_process_if_block): Don't call
3231 flow_find_head_matching_sequence with 0 longest_match.
3232 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3233 non-active insns if !stop_after.
3234 (try_head_merge_bb): Revert 2014-01-07 changes.
3236 2014-01-08 Jeff Law <law@redhat.com>
3238 * ree.c (get_sub_rtx): New function, extracted from...
3239 (merge_def_and_ext): Here.
3240 (combine_reaching_defs): Use get_sub_rtx.
3242 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3244 * cgraph.h (varpool_variable_node): Do not choke on null node.
3246 2014-01-08 Catherine Moore <clm@codesourcery.com>
3248 * config/mips/mips.md (simple_return): Attempt to use JRC
3250 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3252 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3254 PR rtl-optimization/59137
3255 * reorg.c (steal_delay_list_from_target): Call update_block for
3257 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3259 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3261 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3262 two duplicate entries.
3264 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3267 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3269 * config/mips/mips.c (mips_truncated_op_cost): New function.
3270 (mips_rtx_costs): Adjust test for BADDU.
3271 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3273 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3275 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3276 (*baddu_si): ...this new pattern.
3278 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3281 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3283 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3286 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3288 (expand_expr, expand_normal): Adjust.
3289 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3290 inner_reference_p. Use inner_reference_p to expand inner references.
3291 (store_expr): Adjust.
3292 * cfgexpand.c (expand_call_stmt): Adjust.
3294 2014-01-08 Rong Xu <xur@google.com>
3296 * gcov-io.c (gcov_var): Move from gcov-io.h.
3297 (gcov_position): Ditto.
3298 (gcov_is_error): Ditto.
3299 (gcov_rewrite): Ditto.
3300 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3301 only part to libgcc/libgcov.h.
3303 2014-01-08 Marek Polacek <polacek@redhat.com>
3306 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3308 2014-01-08 Marek Polacek <polacek@redhat.com>
3311 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3313 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3315 PR rtl-optimization/59649
3316 * stor-layout.c (get_mode_bounds): For BImode return
3317 0 and STORE_FLAG_VALUE.
3319 2014-01-08 Richard Biener <rguenther@suse.de>
3322 * gimple.h (is_gimple_builtin_call): Remove.
3323 (gimple_builtin_call_types_compatible_p): New.
3324 (gimple_call_builtin_p): New overload.
3325 * gimple.c (is_gimple_builtin_call): Remove.
3326 (validate_call): Rename to ...
3327 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3329 (validate_type): New static function.
3330 (gimple_call_builtin_p): New overload and adjust.
3331 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3332 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3333 (gimple_fold_stmt_to_constant_1): Likewise.
3334 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3336 2014-01-08 Richard Biener <rguenther@suse.de>
3339 * gimplify.c (gimplify_expr): Gimplify register-register type
3340 VIEW_CONVERT_EXPRs to separate stmts.
3342 2014-01-07 Jeff Law <law@redhat.com>
3345 * ree.c (combine_set_extension): Handle case where source
3346 and destination registers in an extension insn are different.
3347 (combine_reaching_defs): Allow source and destination registers
3348 in extension to be different under limited circumstances.
3349 (add_removable_extension): Remove restriction that the
3350 source and destination registers in the extension are the same.
3351 (find_and_remove_re): Emit a copy from the extension's
3352 destination to its source after the defining insn if
3353 the source and destination registers are different.
3356 * ifcvt.c (merge_if_block): If we are merging a block with more than
3357 one successor with a block with no successors, remove any BARRIER
3358 after the second block.
3360 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3362 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3364 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3367 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3368 for 14-bit register offsets when INT14_OK_STRICT is false.
3370 2014-01-07 Roland Stigge <stigge@antcom.de>
3371 Michael Meissner <meissner@linux.vnet.ibm.com>
3374 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3375 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3377 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3379 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3382 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3384 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3385 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3386 rtx is const0_rtx or not.
3388 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3391 * target-globals.c (save_target_globals): Remove this_fn_optab
3393 * toplev.c: Include optabs.h.
3394 (target_reinit): Temporarily restore the global options if another
3395 set of options are in force.
3397 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3399 PR rtl-optimization/58668
3400 * cfgcleanup.c (flow_find_cross_jump): Don't count
3401 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3402 to determine what is counted.
3403 (flow_find_head_matching_sequence): Use active_insn_p to determine
3405 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3407 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3408 determine what is counted.
3410 PR tree-optimization/59643
3411 * tree-predcom.c (split_data_refs_to_components): If one dr is
3412 read and one write, determine_offset fails and the write isn't
3413 in the bad component, just put the read into the bad component.
3415 2014-01-07 Mike Stump <mikestump@comcast.net>
3416 Jakub Jelinek <jakub@redhat.com>
3419 * tree-core.h (struct tree_optimization_option): Change optabs
3420 type from unsigned char * to void *.
3421 * optabs.c (init_tree_optimization_optabs): Adjust
3422 TREE_OPTIMIZATION_OPTABS initialization.
3424 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3427 * config/i386/i386.h (struct machine_function): Add
3428 no_drap_save_restore field.
3429 * config/i386/i386.c (ix86_save_reg): Use
3430 !cfun->machine->no_drap_save_restore instead of
3431 crtl->stack_realign_needed.
3432 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3433 this function clears frame_pointer_needed. Set
3434 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3435 and DRAP reg is needed.
3437 2014-01-06 Marek Polacek <polacek@redhat.com>
3440 * doc/implement-c.texi: Mention that other integer types are
3441 permitted as bit-field types in strictly conforming mode.
3443 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3445 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3448 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3450 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3452 2014-01-06 Martin Jambor <mjambor@suse.cz>
3455 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3457 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3459 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3463 * var-tracking.c (add_stores): Preserve the value of the source even if
3464 we don't record the store.
3466 2014-01-06 Terry Guo <terry.guo@arm.com>
3468 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3470 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3473 * config/darwin.c (darwin_function_section): Adjust return values to
3474 correspond to optimisation changes made in r206070.
3476 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3478 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3479 from prefetch_block tune setting.
3480 (nocona_cost): Correct size of prefetch block to 64.
3482 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3484 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3485 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3486 used to save the static chain register in the computation of the offset
3487 from which the FP registers need to be restored.
3489 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3491 PR tree-optimization/59519
3492 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3493 ICE if get_current_def (current_new_name) is already non-NULL, as long
3494 as it is a phi result of some other phi in *new_exit_bb that has
3497 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3498 or vmovdqu* for misaligned_operand.
3499 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3500 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3501 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3502 aligned_mem for AVX512F masked aligned load and store builtins and for
3505 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3507 PR tree-optimization/59651
3508 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3509 Address range for negative step should be added by TYPE_SIZE_UNIT.
3511 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3513 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3514 overlapping registers also for registers other than the stack pointer.
3516 2014-01-03 Marek Polacek <polacek@redhat.com>
3519 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3522 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3525 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3528 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3529 (push splitter): Use <P:MODE_SIZE> instead of
3530 GET_MODE_SIZE (<P:MODE>mode).
3531 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3532 (mov -1, reg peephole2): Likewise.
3533 * config/i386/sse.md (*mov<mode>_internal,
3534 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3535 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3536 *<code><mode>3, *andnot<mode>3<mask_name>,
3537 <mask_codefor><code><mode>3<mask_name>): Likewise.
3538 * config/i386/subst.md (mask_mode512bit_condition,
3539 sd_mask_mode512bit_condition): Likewise.
3541 2014-01-02 Xinliang David Li <davidxl@google.com>
3543 PR tree-optimization/59303
3544 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3545 (dump_predicates): Better output format.
3546 (pred_equal_p): New function.
3547 (is_neq_relop_p): Ditto.
3548 (is_neq_zero_form_p): Ditto.
3549 (pred_expr_equal_p): Ditto.
3550 (pred_neg_p): Ditto.
3551 (simplify_pred): Ditto.
3552 (simplify_preds_2): Ditto.
3553 (simplify_preds_3): Ditto.
3554 (simplify_preds_4): Ditto.
3555 (simplify_preds): Ditto.
3557 (push_to_worklist): Ditto.
3558 (get_pred_info_from_cmp): Ditto.
3559 (is_degenerated_phi): Ditto.
3560 (normalize_one_pred_1): Ditto.
3561 (normalize_one_pred): Ditto.
3562 (normalize_one_pred_chain): Ditto.
3563 (normalize_preds): Ditto.
3564 (normalize_cond_1): Remove function.
3565 (normalize_cond): Ditto.
3566 (is_gcond_subset_of): Ditto.
3567 (is_subset_of_any): Ditto.
3568 (is_or_set_subset_of): Ditto.
3569 (is_and_set_subset_of): Ditto.
3570 (is_norm_cond_subset_of): Ditto.
3571 (pred_chain_length_cmp): Ditto.
3572 (convert_control_dep_chain_into_preds): Type change.
3573 (find_predicates): Ditto.
3574 (find_def_preds): Ditto.
3575 (destroy_predicates_vecs): Ditto.
3576 (find_matching_predicates_in_rest_chains): Ditto.
3577 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3578 (is_pred_expr_subset): Ditto.
3579 (is_pred_chain_subset_of): Ditto.
3580 (is_included_in): Ditto.
3581 (is_superset_of): Ditto.
3583 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3585 Update copyright years.
3587 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3589 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3590 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3591 config/arc/arc.md, config/arc/arc.opt,
3592 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3593 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3594 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3595 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3596 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3597 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3598 vtable-verify.c, vtable-verify.h: Use the standard form for the
3601 2014-01-02 Tobias Burnus <burnus@net-b.de>
3603 * gcc.c (process_command): Update copyright notice dates.
3604 * gcov-dump.c: Ditto.
3606 * doc/cpp.texi: Bump @copying's copyright year.
3607 * doc/cppinternals.texi: Ditto.
3608 * doc/gcc.texi: Ditto.
3609 * doc/gccint.texi: Ditto.
3610 * doc/gcov.texi: Ditto.
3611 * doc/install.texi: Ditto.
3612 * doc/invoke.texi: Ditto.
3614 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3616 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3618 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3620 * config/i386/sse.md (*mov<mode>_internal): Guard
3621 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3623 PR rtl-optimization/59647
3624 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3625 new_rtx into UNSIGNED_FLOAT rtxes.
3627 Copyright (C) 2014 Free Software Foundation, Inc.
3629 Copying and distribution of this file, with or without modification,
3630 are permitted in any medium without royalty provided the copyright
3631 notice and this notice are preserved.