1 2014-02-18 Jakub Jelinek <jakub@redhat.com>
4 * final.c (SEEN_BB): Remove.
5 (SEEN_NOTE, SEEN_EMITTED): Renumber.
6 (final_scan_insn): Don't force_source_line on second
9 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
12 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
13 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
14 (type_natural_mode): Warn ABI change when %zmm register is not
15 available for AVX512F vector value passing.
17 2014-02-18 Kai Tietz <ktietz@redhat.com>
20 * config/i386/i386.c (ix86_expand_prologue): Use value in
21 rax register as displacement when restoring %r10 or %rax.
22 Fix wrong offset when restoring both registers.
24 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
26 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
27 assertion with conditional return.
29 2014-02-18 Jakub Jelinek <jakub@redhat.com>
30 Uros Bizjak <ubizjak@gmail.com>
33 * config/i386/driver-i386.c (host_detect_local_cpu): If
34 YMM state is not saved by the OS, also clear has_f16c. Move
35 CPUID 0x80000001 handling before YMM state saving checking.
37 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
39 PR rtl-optimization/58960
40 * haifa-sched.c (alloc_global_sched_pressure_data): New,
42 (sched_init): ... here.
43 (free_global_sched_pressure_data): New, factored out from ...
44 (sched_finish): ... here.
45 * sched-int.h (free_global_sched_pressure_data): Declare.
46 * sched-rgn.c (nr_regions_initial): New static global.
47 (haifa_find_rgns): Initialize it.
48 (schedule_region): Disable sched-pressure for the newly
51 2014-02-17 Richard Biener <rguenther@suse.de>
53 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
54 release SSA defs of pattern stmts.
56 2014-02-17 Richard Biener <rguenther@suse.de>
58 * tree-inline.c (expand_call_inline): Release the virtual
59 operand defined by the call we are about to inline.
61 2014-02-17 Richard Biener <rguenther@suse.de>
63 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
65 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
66 Ilya Tocar <ilya.tocar@intel.com>
68 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
69 arguments order in builtin.
70 (_mm512_permutexvar_epi64): Ditto.
71 (_mm512_mask_permutexvar_epi64): Ditto
72 (_mm512_maskz_permutexvar_epi32): Ditto
73 (_mm512_permutexvar_epi32): Ditto
74 (_mm512_mask_permutexvar_epi32): Ditto
76 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
78 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
79 (p8_vmrgow): Likewise.
81 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
83 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
86 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
89 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
90 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
91 into 64-bit and 32-bit moves. On 64-bit moves, add support for
92 using direct move instructions on ISA 2.07. Also adjust
93 instruction length for 64-bit.
94 (mov<mode>_64bit, TFmode/TDmode): Likewise.
95 (mov<mode>_32bit, TFmode/TDmode): Likewise.
97 2014-02-15 Alan Modra <amodra@gmail.com>
101 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
102 find_replacement on parts of insn rtl that might be reloaded.
104 2014-02-15 Richard Biener <rguenther@suse.de>
106 PR tree-optimization/60183
107 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
108 (tree_ssa_phiprop): Calculate and free post-dominators.
110 2014-02-14 Jeff Law <law@redhat.com>
112 PR rtl-optimization/60131
113 * ree.c (get_extended_src_reg): New function.
114 (combine_reaching_defs): Use it rather than assuming location of REG.
115 (find_and_remove_re): Verify first operand of extension is
116 a REG before adding the insns to the copy list.
118 2014-02-14 Roland McGrath <mcgrathr@google.com>
120 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
121 * configure: Regenerated.
122 * config.in: Regenerated.
123 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
124 instead of ASM_SHORT.
126 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
127 Richard Earnshaw <rearnsha@arm.com>
129 PR rtl-optimization/59535
130 * lra-constraints.c (process_alt_operands): Encourage alternative
131 when unassigned pseudo class is superset of the alternative class.
132 (inherit_reload_reg): Don't inherit when optimizing for code size.
133 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
134 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
135 modes not less than 4 for Thumb1.
137 2014-02-14 Kyle McMartin <kyle@redhat.com>
140 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
142 2014-02-14 Richard Biener <rguenther@suse.de>
144 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
145 (get_frame_arg): Drop the assert with langhook types_compatible_p.
146 Do not strip INDIRECT_REFs.
148 2014-02-14 Richard Biener <rguenther@suse.de>
151 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
152 DECL_FUNCTION_SPECIFIC_TARGET.
153 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
154 * tree-streamer-out.c (pack_ts_target_option): Remove.
155 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
156 (write_ts_function_decl_tree_pointers): Do not stream
157 DECL_FUNCTION_SPECIFIC_TARGET.
158 * tree-streamer-in.c (unpack_ts_target_option): Remove.
159 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
160 (lto_input_ts_function_decl_tree_pointers): Do not stream
161 DECL_FUNCTION_SPECIFIC_TARGET.
163 2014-02-14 Jakub Jelinek <jakub@redhat.com>
165 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
166 (get_initial_def_for_induction, vectorizable_induction): Ignore
167 debug stmts when looking for exit_phi.
168 (vectorizable_live_operation): Fix up condition.
170 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
172 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
173 nreverse() because it changes the content of original tree list.
175 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
177 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
178 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
180 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
182 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
183 GNU coding standards.
185 2014-02-13 Jakub Jelinek <jakub@redhat.com>
188 * dwarf2out.c (gen_subprogram_die): Don't call
189 add_calling_convention_attribute if subr_die is old_die.
191 2014-02-13 Sharad Singhai <singhai@google.com>
193 * doc/optinfo.texi: Fix order of nodes.
195 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
197 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
198 operands[2], not operands[3].
200 2014-02-13 Richard Biener <rguenther@suse.de>
203 * doc/install.texi (ISL): Update recommended version to 0.12.2,
204 mention the possibility of an in-tree build.
205 (CLooG): Update recommended version to 0.18.1, mention the
206 possibility of an in-tree build and clarify that the ISL
207 bundled with CLooG does not work.
209 2014-02-13 Jakub Jelinek <jakub@redhat.com>
212 * expr.c (compress_float_constant): If x is a hard register,
213 extend into a pseudo and then move to x.
215 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
217 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
218 caused by bad second argument to warning_at() with -mhotpatch and
219 nested functions (e.g. with gfortran).
221 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
223 * opts.c (option_name): Remove "enabled by default" rider.
225 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
227 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
229 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
230 Uros Bizjak <ubizjak@gmail.com>
233 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
234 * configure: Regenerated.
236 2014-02-12 Richard Biener <rguenther@suse.de>
238 * vec.c (vec_prefix::calculate_allocation): Move as
239 inline variant to vec.h.
240 (vec_prefix::calculate_allocation_1): New out-of-line version.
241 * vec.h (vec_prefix::calculate_allocation_1): Declare.
242 (vec_prefix::m_has_auto_buf): Rename to ...
243 (vec_prefix::m_using_auto_storage): ... this.
244 (vec_prefix::calculate_allocation): Inline the easy cases
245 and dispatch to calculate_allocation_1 which doesn't need the
247 (va_heap::reserve): Use gcc_checking_assert.
248 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
249 m_using_auto_storage.
250 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
252 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
253 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
254 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
256 2014-02-12 Richard Biener <rguenther@suse.de>
258 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
259 when we found a dependence.
261 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
263 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
265 (maybe_fold_stmt): ... into this new function.
266 * omp-low.c (lower_omp): Update comment.
268 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
271 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
274 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
276 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
277 identifiers in comments.
278 (cortexa53_extra_costs): Likewise.
279 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
280 (cortexa7_extra_costs): Likewise.
281 (cortexa12_extra_costs): Likewise.
282 (cortexa15_extra_costs): Likewise.
283 (v7m_extra_costs): Likewise.
285 2014-02-12 Richard Biener <rguenther@suse.de>
288 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
289 of posix_memalign being successful.
290 (lower_stmt): Restrict lowering of posix_memalign to when
291 -ftree-bit-ccp is enabled.
293 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
295 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
297 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
299 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
301 PR rtl-optimization/60116
302 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
303 other_insn once the combination has been validated.
305 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
308 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
310 * ipa-devirt.c: Include demangle.h
311 (odr_violation_reported): New static variable.
312 (add_type_duplicate): Update odr_violations.
313 (maybe_record_node): Add completep parameter; update it.
314 (record_target_from_binfo): Add COMPLETEP parameter;
316 (possible_polymorphic_call_targets_1): Likewise.
317 (struct polymorphic_call_target_d): Add nonconstruction_targets;
318 rename FINAL to COMPLETE.
319 (record_targets_from_bases): Sanity check we found the binfo;
320 fix COMPLETEP updating.
321 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
322 parameter, fix computing of COMPLETEP.
323 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
324 at LTO time do demangling.
325 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
326 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
328 (gimple_get_virt_method_for_binfo): Likewise.
329 * gimple-fold.h (gimple_get_virt_method_for_binfo,
330 gimple_get_virt_method_for_vtable): Update prototypes.
332 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
335 * genautomata.c (add_presence_absence): Fix typo with
336 {final_}presence_list.
338 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
341 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
342 for VSX/Altivec vectors that land in GPR registers.
344 2014-02-11 Richard Henderson <rth@redhat.com>
345 Jakub Jelinek <jakub@redhat.com>
348 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
349 around drhs if type conversion to lacc->type is not useless.
351 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
353 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
355 (cortex-a57.cortex-a53): Likewise.
356 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
358 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
360 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
363 2014-02-11 Renlin Li <Renlin.Li@arm.com>
365 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
366 add_options_for_arm_vfp3.
368 2014-02-11 Jeff Law <law@redhat.com>
371 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
372 object with an undesirable mode.
374 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
377 * config/i386/sol2-9.h: New file.
378 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
379 *-*-solaris2.9*): Use it.
381 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
383 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
384 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
386 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
388 * config/microblaze/microblaze.c: Extend mcpu version format
390 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
392 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
394 2014-02-10 Richard Henderson <rth@redhat.com>
397 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
398 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
399 ms-abi vs -mno-accumulate-outgoing-args.
400 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
401 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
404 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
407 * cfgexpand.c (expand_asm_operands): Attach source location to
408 ASM_INPUT rtx objects.
409 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
411 2014-02-10 Nick Clifton <nickc@redhat.com>
413 * config/mn10300/mn10300.c (popcount): New function.
414 (mn10300_expand_prologue): Include saved registers in stack usage
417 2014-02-10 Jeff Law <law@redhat.com>
420 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
421 when changing the SET_DEST of a prior insn to avoid an input reload.
423 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
425 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
426 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
427 -mcall-openbsd, or -mcall-linux.
428 (CC1_ENDIAN_BIG_SPEC): Remove.
429 (CC1_ENDIAN_LITTLE_SPEC): Remove.
430 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
431 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
432 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
433 and %cc1_endian_default.
434 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
436 2014-02-10 Richard Biener <rguenther@suse.de>
438 PR tree-optimization/60115
439 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
440 MEM_REF handling. Properly verify that the accesses are not
441 out of the objects bound.
443 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
445 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
448 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
450 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
451 proper constants and fix formatting.
452 (possible_polymorphic_call_targets): Fix formatting.
454 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
455 Ilya Tocar <ilya.tocar@intel.com>
457 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
458 (_mm512_loadu_epi32): Renamed into...
459 (_mm512_loadu_si512): This.
460 (_mm512_storeu_epi32): Renamed into...
461 (_mm512_storeu_si512): This.
462 (_mm512_maskz_ceil_ps): Removed.
463 (_mm512_maskz_ceil_pd): Ditto.
464 (_mm512_maskz_floor_ps): Ditto.
465 (_mm512_maskz_floor_pd): Ditto.
466 (_mm512_floor_round_ps): Ditto.
467 (_mm512_floor_round_pd): Ditto.
468 (_mm512_ceil_round_ps): Ditto.
469 (_mm512_ceil_round_pd): Ditto.
470 (_mm512_mask_floor_round_ps): Ditto.
471 (_mm512_mask_floor_round_pd): Ditto.
472 (_mm512_mask_ceil_round_ps): Ditto.
473 (_mm512_mask_ceil_round_pd): Ditto.
474 (_mm512_maskz_floor_round_ps): Ditto.
475 (_mm512_maskz_floor_round_pd): Ditto.
476 (_mm512_maskz_ceil_round_ps): Ditto.
477 (_mm512_maskz_ceil_round_pd): Ditto.
478 (_mm512_expand_pd): Ditto.
479 (_mm512_expand_ps): Ditto.
480 * config/i386/i386.c (ix86_builtins): Remove
481 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
483 * config/i386/predicates.md (const1256_operand): New.
484 (const_1_to_2_operand): Ditto.
485 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
486 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
487 (*avx512pf_gatherpf<mode>sf): Ditto.
488 (avx512pf_gatherpf<mode>df): Ditto.
489 (*avx512pf_gatherpf<mode>df_mask): Ditto.
490 (*avx512pf_gatherpf<mode>df): Ditto.
491 (avx512pf_scatterpf<mode>sf): Ditto.
492 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
493 (*avx512pf_scatterpf<mode>sf): Ditto.
494 (avx512pf_scatterpf<mode>df): Ditto.
495 (*avx512pf_scatterpf<mode>df_mask): Ditto.
496 (*avx512pf_scatterpf<mode>df): Ditto.
497 (avx512f_expand<mode>): Removed.
498 (<shift_insn><mode>3<mask_name>): Change predicate type.
500 2014-02-08 Jakub Jelinek <jakub@redhat.com>
502 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
503 not at the end of datarefs vector use ordered_remove to avoid
504 reordering datarefs vector.
507 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
508 mark local addressable non-static vars as GOVD_PRIVATE
509 instead of GOVD_LOCAL.
510 * omp-low.c (lower_omp_for): Move gimple_bind_vars
511 and BLOCK_VARS of gimple_bind_block to new_stmt rather
515 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
516 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
517 assume_aligned or alloc_align attributes.
518 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
519 arguments. Handle also assume_aligned and alloc_align attributes.
520 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
521 calls to functions with assume_aligned or alloc_align attributes.
522 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
524 2014-02-08 Terry Guo <terry.guo@arm.com>
526 * doc/invoke.texi: Document ARM -march=armv7e-m.
528 2014-02-08 Jakub Jelinek <jakub@redhat.com>
530 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
531 flag on __cilkrts_rethrow builtin.
534 * ipa-cp.c (determine_versionability): Fail at -O0
535 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
536 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
539 2014-02-04 Jakub Jelinek <jakub@redhat.com>
542 * tree-inline.c (copy_forbidden): Fail for
543 __attribute__((optimize (0))) functions.
545 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
547 * varpool.c: Include pointer-set.h.
548 (varpool_remove_unreferenced_decls): Variables in other partitions
549 will not be output; be however careful to not lose information
552 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
554 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
555 lookup in the vtable constructor.
557 2014-02-07 Jeff Law <law@redhat.com>
560 * config/m68k/m68k.md (ashldi_extsi): Turn into a
561 define_insn_and_split.
563 * ipa-inline.c (inline_small_functions): Fix typos.
565 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
567 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
568 (s390_can_use_return_insn): Declare.
569 * config/s390/s390.h (EPILOGUE_USES): Define.
570 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
572 (s390_chunkify_start): Handle return JUMP_LABELs.
573 (s390_early_mach): Emit a main_pool instruction on the entry edge.
574 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
575 (s390_can_use_return_insn): New functions.
576 (s390_fix_long_loop_prediction): Handle conditional returns.
577 (TARGET_SET_UP_BY_PROLOGUE): Define.
578 * config/s390/s390.md (ANY_RETURN): New code iterator.
579 (*creturn, *csimple_return, return, simple_return): New patterns.
581 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
583 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
584 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
585 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
586 REG_CFA_RESTORE list when deciding not to restore a register.
588 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
590 * config/s390/s390.c: Include tree-pass.h and context.h.
591 (s390_early_mach): New function, split out from...
592 (s390_emit_prologue): ...here.
593 (pass_data_s390_early_mach): New pass structure.
594 (pass_s390_early_mach): New class.
595 (s390_option_override): Create and register early_mach pass.
598 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
600 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
601 to match for the exit block.
603 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
605 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
606 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
607 Reject misaligned operands.
609 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
611 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
613 2014-02-07 Richard Biener <rguenther@suse.de>
616 * gimple-low.c (lower_builtin_posix_memalign): New function.
617 (lower_stmt): Call it to lower posix_memalign in a way
618 to make alignment info accessible.
620 2014-02-07 Jakub Jelinek <jakub@redhat.com>
623 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
624 __builtin_setjmp_receiver.
626 2014-02-07 Richard Biener <rguenther@suse.de>
629 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
630 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
631 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
632 Handle BUILT_IN_POSIX_MEMALIGN.
633 (find_func_clobbers): Likewise.
634 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
635 (call_may_clobber_ref_p_1): Likewise.
637 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
640 * ipa-devirt.c (record_target_from_binfo): Remove overactive
643 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
646 * lto-cgraph.c (lto_output_node): Use
647 symtab_get_symbol_partitioning_class.
648 (lto_output_varpool_node): likewise.
649 (symtab_get_symbol_partitioning_class): Move here from
651 * cgraph.h (symbol_partitioning_class): Likewise.
652 (symtab_get_symbol_partitioning_class): Declare.
654 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
656 * ggc.h (ggc_internal_cleared_alloc): New macro.
657 * vec.h (vec_safe_copy): Handle memory stats.
658 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
659 * target-globals.c (save_target_globals): Likewise.
661 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
664 * expr.c (emit_move_resolve_push): Export; be bit more selective
665 on when to clear alias set.
666 * expr.h (emit_move_resolve_push): Declare.
667 * function.h (struct function): Add tail_call_marked.
668 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
669 * config/i386/i386-protos.h (ix86_expand_push): Remove.
670 * config/i386/i386.md (TImode move expander): De not call
672 (FP push expanders): Preserve memory attributes.
673 * config/i386/sse.md (push<mode>1): Remove.
674 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
675 (ix86_expand_push): Remove.
676 * config/i386/mmx.md (push<mode>1): Remove.
678 2014-02-06 Jakub Jelinek <jakub@redhat.com>
680 PR rtl-optimization/60030
681 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
682 lopart with paradoxical subreg before shifting it up by hprec.
684 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
686 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
687 Remove extra newline at end of file.
688 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
689 (arm_issue_rate): Handle cortexa57.
690 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
691 (cortex-a57.cortex-a53): Likewise.
693 2014-02-06 Jakub Jelinek <jakub@redhat.com>
696 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
697 don't record in REG_FRAME_RELATED_EXPR registers not set in that
699 (arm_expand_prologue): Adjust all callers.
700 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
701 info, registers also at the lowest numbered registers side. Use
702 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
706 * var-tracking.c (adjust_mems): Before adding a SET to
707 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
709 2014-02-06 Alan Modra <amodra@gmail.com>
712 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
713 change SDmode to DDmode when lra_in_progress.
715 2014-02-06 Jakub Jelinek <jakub@redhat.com>
718 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
719 free_data_ref on the dr first, and before goto again also set dr
720 to the next dr. For simd_lane_access, free old datarefs[i] before
721 overwriting it. For get_vectype_for_scalar_type failure, don't
722 free_data_ref if simd_lane_access.
724 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
727 * tree.h (opts_for_fn): New inline function.
728 (opt_for_fn): Define.
729 * config/i386/i386.c (ix86_function_regparm): Use
730 opt_for_fn (decl, optimize) instead of optimize.
732 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
734 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
735 for SYMBOL_REF in large memory model.
737 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
739 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
741 (cortex-a57): Likewise.
742 (cortex-a57.cortex-a53): Likewise.
744 2014-02-06 Yury Gribov <y.gribov@samsung.com>
745 Kugan Vivekanandarajah <kuganv@linaro.org>
747 * config/arm/arm.c (arm_vector_alignment_reachable): Check
749 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
751 2014-02-06 Richard Biener <rguenther@suse.de>
753 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
754 set_loop_copy and initialize_original_copy_tables.
756 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
758 * config/aarch64/aarch64-simd.md
759 (aarch64_ashr_simddi): Change QI to SI.
761 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
762 Jakub Jelinek <jakub@redhat.com>
765 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
768 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
770 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
771 CODE_FOR_altivec_vpku[hw]um to
772 CODE_FOR_altivec_vpku[hw]um_direct.
773 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
774 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
775 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
776 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
778 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
780 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
781 generation for -maltivec=be.
782 (altivec_vsumsws): Simplify redundant test.
784 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
786 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
787 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
788 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
789 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
791 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
793 (altivec_vpks<VI_char>ss): Likewise.
794 (altivec_vpks<VI_char>us): Likewise.
795 (altivec_vpku<VI_char>us): Likewise.
796 (altivec_vpku<VI_char>um): Likewise.
797 (altivec_vpku<VI_char>um_direct): New (copy of
798 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
800 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
801 target is little endian and -maltivec=be is not specified.
802 (*altivec_vupkhs<VU_char>_direct): New (copy of
803 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
804 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
805 target is little endian and -maltivec=be is not specified.
806 (*altivec_vupkls<VU_char>_direct): New (copy of
807 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
808 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
809 little endian and -maltivec=be is not specified.
810 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
811 little endian and -maltivec=be is not specified.
813 2014-02-05 Richard Henderson <rth@redhat.com>
816 * combine-stack-adj.c: Revert r206943.
817 * sched-int.h (struct deps_desc): Add last_args_size.
818 * sched-deps.c (init_deps): Initialize it.
819 (sched_analyze_insn): Add OUTPUT dependencies between insns that
820 contain REG_ARGS_SIZE notes.
822 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
824 * lto-cgraph.c (asm_nodes_output): Make global.
825 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
826 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
827 (driver_handle_option): Handle OPT_fwpa.
829 2014-02-05 Jakub Jelinek <jakub@redhat.com>
832 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
833 a comment typo and formatting issue. If odr_hash hasn't been
834 created, return vNULL and set *completep to false.
837 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
838 bb with no successors.
840 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
843 * doc/invoke.texi (-march): Clarify documentation for ARM.
847 2014-02-05 Richard Biener <rguenther@suse.de>
849 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
850 when not vectorizing because of too many alias checks.
851 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
852 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
854 2014-02-05 Nick Clifton <nickc@redhat.com>
856 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
857 accept extended registers in any mode when compiling for the MN10300.
859 2014-02-05 Yury Gribov <y.gribov@samsung.com>
861 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
862 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
863 sanitization attributes.
864 (can_inline_edge_p): Likewise.
865 (sanitize_attrs_match_for_inline_p): New function.
867 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
869 * ipa-prop.c (detect_type_change): Shor circuit testing of
870 type changes on THIS pointer.
872 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
875 * config/pa/pa.c (legitimize_tls_address): Return original address
876 if not passed a SYMBOL_REF rtx.
877 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
879 (pa_emit_move_sequence): Simplify TLS source operands.
880 (pa_legitimate_constant_p): Reject all TLS constants.
881 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
882 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
884 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
886 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
887 groups when we know they are controlled by LTO.
888 * varasm.c (default_binds_local_p_1): If object is in other partition,
889 it will be resolved locally.
891 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
893 * config/host-linux.c (linux_gt_pch_use_address): Don't
894 use SSIZE_MAX because it is not always defined.
896 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
899 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
900 threshold for pseudo splitting.
901 (update_ebb_live_info): Process call argument hard registers and
902 hard registers from insn definition too.
903 (max_small_class_regs_num): New constant.
904 (inherit_in_ebb): Update live hard regs through EBBs. Update
905 reloads_num only for small register classes. Don't split for
908 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
911 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
914 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
916 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
919 2014-02-04 Marek Polacek <polacek@redhat.com>
921 * gdbinit.in (pel): Define.
923 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
925 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
928 2014-02-04 Richard Biener <rguenther@suse.de>
931 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
932 in function context local.
933 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
934 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
935 similar to LTO_imported_decl_ref.
937 2014-02-04 Jakub Jelinek <jakub@redhat.com>
939 PR tree-optimization/60002
940 * cgraphclones.c (build_function_decl_skip_args): Clear
943 PR tree-optimization/60023
944 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
945 false to gsi_replace.
946 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
947 has been in some EH region and vec_stmt could throw, add
948 vec_stmt into the same EH region.
949 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
950 has no lhs, ignore it.
951 * internal-fn.c (expand_MASK_LOAD): Likewise.
954 * tree-inline.c (copy_forbidden): Fail for
955 __attribute__((optimize (0))) functions.
958 * omp-low.c (simd_clone_struct_copy): If from->inbranch
959 is set, copy one less argument.
960 (expand_simd_clones): Don't subtract clone_info->inbranch
961 from simd_clone_struct_alloc argument.
963 PR rtl-optimization/57915
964 * recog.c (simplify_while_replacing): If all unary/binary/relational
965 operation arguments are constant, attempt to simplify those.
968 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
969 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
971 2014-02-04 Richard Biener <rguenther@suse.de>
973 PR tree-optimization/60012
974 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
975 TBAA disambiguation to all DDRs.
977 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
980 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
981 (LINK_SPEC): Use it for -shared, -shared-libgcc.
983 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
986 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
988 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
990 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
991 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
993 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
996 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
997 to figure out targets of polymorphic calls with known decl.
998 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
999 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1000 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1001 (get_polymorphic_call_info): ... here.
1002 (get_polymorphic_call_info_from_invariant): New function.
1004 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1006 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1007 lookup via vtable pointer; check for type consistency
1008 and turn inconsitent facts into UNREACHABLE.
1009 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1010 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1011 type inconsistent querries; return UNREACHABLE instead.
1013 2014-02-03 Richard Henderson <rth@twiddle.net>
1016 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1017 already processed this node.
1018 (normalize_one_pred_1): Pass along mark_set.
1019 (normalize_one_pred): Create and destroy a pointer_set_t.
1020 (normalize_one_pred_chain): Likewise.
1022 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1024 PR gcov-profile/58602
1025 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1027 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1030 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1031 -fno-devirtualize; try to devirtualize by the knowledge of
1032 virtual table pointer given by aggregate propagation.
1033 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1034 (ipa_print_node_jump_functions): Dump also offset that
1035 is relevant for polymorphic calls.
1036 (determine_known_aggregate_parts): Add arg_type parameter; use it
1037 instead of determining the type from pointer type.
1038 (ipa_compute_jump_functions_for_edge): Update call of
1039 determine_known_aggregate_parts.
1040 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1041 (gimple_get_virt_method_for_binfo): ... here; simplify using
1042 vtable_pointer_value_to_vtable.
1043 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1044 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1045 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1046 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1048 (vtable_pointer_value_to_binfo): ... here.
1049 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1051 2014-02-03 Teresa Johnson <tejohnson@google.com>
1053 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1054 redef of outer loop index variable.
1056 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1060 * doc/extend.texi (Function Attributes): Typo.
1062 2014-02-03 Cong Hou <congh@google.com>
1064 PR tree-optimization/60000
1065 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1066 if the vectorized statement is a store. A store statement can only
1067 appear at the end of pattern statements.
1069 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1071 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1072 (ix86_option_override_internal): Default long double to 64-bit for
1073 32-bit Bionic and to 128-bit for 64-bit Bionic.
1075 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1076 TARGET_LONG_DOUBLE_128 is true.
1077 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1079 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1080 (mlong-double-64): Negate -mlong-double-128.
1081 (mlong-double-128): New option.
1083 * config/i386/i386-c.c (ix86_target_macros): Define
1084 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1086 * doc/invoke.texi: Document -mlong-double-128.
1088 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1090 PR rtl-optimization/60024
1091 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1093 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1095 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1097 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1099 PR rtl-optimization/57662
1100 * sel-sched.c (code_motion_path_driver): Do not mark already not
1101 existing blocks in the visiting bitmap.
1103 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1105 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1106 on the insn being emitted.
1108 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1109 Will Deacon <will.deacon@arm.com>
1111 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1113 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1115 * config/arm/arm-tables.opt: Regenerate.
1117 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1119 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1120 for vector types other than V16QImode.
1121 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1122 define_expand, and call altivec_expand_vec_perm_le when producing
1123 code with little endian element order.
1124 (*altivec_vperm_<mode>_internal): New insn having previous
1125 behavior of altivec_vperm_<mode>.
1126 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1127 altivec_expand_vec_perm_le when producing code with little endian
1129 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1130 behavior of altivec_vperm_<mode>_uns.
1132 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1134 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1135 (altivec_vsumsws): Add handling for -maltivec=be with a little
1137 (altivec_vsumsws_direct): New.
1138 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1139 gen_altivec_vsumsws.
1141 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1143 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1144 vtable_pointer_value_to_binfo): New functions.
1145 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1146 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1148 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1150 * config/nios2/nios2.md (load_got_register): Initialize GOT
1151 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1152 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1154 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1156 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1157 preserverd by passthrough, do not propagate the type.
1159 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1161 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1162 (mips_atomic_assign_expand_fenv): New function.
1163 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1165 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1167 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1168 (__builtin_mips_set_fcsr): Likewise.
1169 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1170 MIPS_USI_FTYPE_VOID.
1171 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1172 (mips16_expand_set_fcsr): Likewise.
1173 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1174 (mips16_set_fcsr_stub): Likewise.
1175 (mips16_get_fcsr_one_only_stub): New class.
1176 (mips16_set_fcsr_one_only_stub): Likewise.
1177 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1178 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1179 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1180 (hard_float): New availability predicate.
1181 (mips_builtins): Add get_fcsr and set_fcsr.
1182 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1183 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1184 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1185 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1186 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1189 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1191 * config/mips/mips.c (mips_one_only_stub): New class.
1192 (mips_need_mips16_rdhwr_p): Replace with...
1193 (mips16_rdhwr_stub): ...this new variable.
1194 (mips16_stub_call_address): New function.
1195 (mips16_rdhwr_one_only_stub): New class.
1196 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1197 (mips_output_mips16_rdhwr): Delete.
1198 (mips_finish_stub): New function.
1199 (mips_code_end): Use it to handle rdhwr stubs.
1201 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1204 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1205 when calculating size of integer atomic types.
1207 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1209 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1211 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1213 PR tree-optimization/60003
1214 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1215 * profile.c (branch_prob): Use gimple_call_builtin_p
1216 to check for BUILT_IN_SETJMP_RECEIVER.
1217 * tree-inline.c (copy_bb): Call notice_special_calls.
1219 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1222 * lra-constraints.c (process_alt_operands): Update reload_sum only
1225 2014-01-31 Richard Henderson <rth@redhat.com>
1228 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1229 until after else_eh is processed.
1231 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1233 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1234 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1235 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1236 in smmintrin.h, remove them.
1237 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1238 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1239 * config/i386/i386.md (ROUND_SAE): Fix value.
1240 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1241 (const48_operand): New.
1242 * config/i386/subst.md (round), (round_expand): Use
1243 const_4_or_8_to_11_operand.
1244 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1246 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1248 * config/i386/constraints.md (Yk): Swap meaning with k.
1249 * config/i386/i386.md (movhi_internal): Change Yk to k.
1250 (movqi_internal): Ditto.
1251 (*k<logic><mode>): Ditto.
1254 (kandn<mode>): Ditto.
1255 (*<code>hi_1): Ditto.
1256 (*<code>qi_1): Ditto.
1257 (kxnor<mode>): Ditto.
1258 (kortestzhi): Ditto.
1259 (kortestchi): Ditto.
1261 (*one_cmplhi2_1): Ditto.
1262 (*one_cmplqi2_1): Ditto.
1263 * config/i386/sse.md (): Change k to Yk.
1264 (avx512f_load<mode>_mask): Ditto.
1265 (avx512f_blendm<mode>): Ditto.
1266 (avx512f_store<mode>_mask): Ditto.
1267 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1268 (avx512f_storedqu<mode>_mask): Ditto.
1269 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1271 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1272 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1273 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1274 (avx512f_maskcmp<mode>3): Ditto.
1275 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1276 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1277 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1278 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1279 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1280 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1281 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1282 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1283 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1284 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1285 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1286 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1287 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1288 (vec_extract_lo_<mode>_maskm): Ditto.
1289 (vec_extract_hi_<mode>_maskm): Ditto.
1290 (avx512f_vternlog<mode>_mask): Ditto.
1291 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1292 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1293 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1294 (avx512f_<code>v8div16qi2_mask): Ditto.
1295 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1296 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1297 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1298 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1299 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1300 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1301 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1302 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1303 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1304 (avx512cd_maskb_vec_dupv8di): Ditto.
1305 (avx512cd_maskw_vec_dupv16si): Ditto.
1306 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1307 (avx512f_vpermi2var<mode>3_mask): Ditto.
1308 (avx512f_vpermi2var<mode>3_mask): Ditto.
1309 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1310 (*avx512f_gathersi<mode>): Ditto.
1311 (*avx512f_gathersi<mode>_2): Ditto.
1312 (*avx512f_gatherdi<mode>): Ditto.
1313 (*avx512f_gatherdi<mode>_2): Ditto.
1314 (*avx512f_scattersi<mode>): Ditto.
1315 (*avx512f_scatterdi<mode>): Ditto.
1316 (avx512f_compress<mode>_mask): Ditto.
1317 (avx512f_compressstore<mode>_mask): Ditto.
1318 (avx512f_expand<mode>_mask): Ditto.
1319 * config/i386/subst.md (mask): Change k to Yk.
1320 (mask_scalar_merge): Ditto.
1323 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1325 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1327 2014-01-31 Richard Biener <rguenther@suse.de>
1330 * builtins.c (fold_builtin_memory_op): Make sure to not
1331 use a floating-point mode or a boolean or enumeral type for
1334 2014-01-30 DJ Delorie <dj@redhat.com>
1336 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1337 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1338 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1339 whenever main() has an epilogue.
1341 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1343 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1344 unused variable "field".
1345 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1346 (vsx_mergeh_<mode>): Likewise.
1347 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1348 (altivec_vmrghh): Likewise.
1349 (altivec_vmrghw): Likewise.
1350 (altivec_vmrglb): Likewise.
1351 (altivec_vmrglh): Likewise.
1352 (altivec_vmrglw): Likewise.
1353 (altivec_vspltb): Add missing uses.
1354 (altivec_vsplth): Likewise.
1355 (altivec_vspltw): Likewise.
1356 (altivec_vspltsf): Likewise.
1358 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1361 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1362 frame related instructions.
1364 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1366 PR rtl-optimization/59959
1367 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1368 any reload of register whose subreg is invalid.
1370 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1372 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1373 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1374 Add missing return type - void.
1376 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1378 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1379 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1380 remove element index adjustment for endian (now handled in vsx.md
1382 (altivec_expand_vec_perm_const): Use
1383 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1384 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1385 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1386 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1387 define_expand and a new define_insn *altivec_vspltb_internal;
1388 adjust for -maltivec=be on a little endian target.
1389 (altivec_vspltb_direct): New.
1390 (altivec_vsplth): Divide into a define_expand and a new
1391 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1392 little endian target.
1393 (altivec_vsplth_direct): New.
1394 (altivec_vspltw): Divide into a define_expand and a new
1395 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1396 little endian target.
1397 (altivec_vspltw_direct): New.
1398 (altivec_vspltsf): Divide into a define_expand and a new
1399 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1400 a little endian target.
1402 2014-01-30 Richard Biener <rguenther@suse.de>
1404 PR tree-optimization/59993
1405 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1406 can propagate form the earlier stmt and avoid the transform
1407 when the intermediate result is needed.
1409 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1411 * README.Portability: Fix typo.
1413 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1415 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1416 comparison_operator with ordered_comparison_operator.
1418 2014-01-30 Nick Clifton <nickc@redhat.com>
1420 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1421 Rename to mn10300_store_multiple_regs.
1422 * config/mn10300/mn10300.c: Likewise.
1423 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1424 store_multiple_regs.
1425 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1426 Call mn10300_store_multiple_regs.
1428 2014-01-30 Nick Clifton <nickc@redhat.com>
1429 DJ Delorie <dj@redhat.com>
1431 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1432 %fp 2 to keep registers after it properly word-aligned.
1433 (rl78_alloc_physical_registers_umul): Handle the case where both
1434 input operands are the same.
1436 2014-01-30 Richard Biener <rguenther@suse.de>
1438 PR tree-optimization/59903
1439 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1442 2014-01-30 Jason Merrill <jason@redhat.com>
1445 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1448 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1450 2014-01-30 Richard Biener <rguenther@suse.de>
1452 PR tree-optimization/59951
1453 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1455 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1458 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1459 SFmode to DFmode case.
1461 2014-01-29 DJ Delorie <dj@redhat.com>
1463 * config/msp430/msp430.opt (-minrt): New.
1464 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1466 (ENDFILE_SPEC): Likewise.
1468 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1470 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1471 (estimate_function_body_sizes): Use it.
1473 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1476 * dwarf2out.c (is_cxx_auto): New.
1477 (is_base_type): Use it.
1478 (gen_type_die_with_usage): Likewise.
1480 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1482 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1483 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1484 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1485 -maltivec=be with LE targets.
1486 (vsx_mergeh_<mode>): Likewise.
1487 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1488 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1489 (altivec_vmrghb): Replace with define_expand and new
1490 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1491 (altivec_vmrghb_direct): New define_insn.
1492 (altivec_vmrghh): Replace with define_expand and new
1493 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1494 (altivec_vmrghh_direct): New define_insn.
1495 (altivec_vmrghw): Replace with define_expand and new
1496 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1497 (altivec_vmrghw_direct): New define_insn.
1498 (*altivec_vmrghsf): Adjust for endianness.
1499 (altivec_vmrglb): Replace with define_expand and new
1500 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1501 (altivec_vmrglb_direct): New define_insn.
1502 (altivec_vmrglh): Replace with define_expand and new
1503 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1504 (altivec_vmrglh_direct): New define_insn.
1505 (altivec_vmrglw): Replace with define_expand and new
1506 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1507 (altivec_vmrglw_direct): New define_insn.
1508 (*altivec_vmrglsf): Adjust for endianness.
1509 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1510 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1511 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1512 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1513 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1514 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1515 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1516 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1518 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1520 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1521 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1524 2014-01-29 Richard Biener <rguenther@suse.de>
1526 PR tree-optimization/58742
1527 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1528 associate_pointerplus_align.
1529 (associate_pointerplus_diff): New function.
1530 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1531 and associate_pointerplus_diff.
1533 2014-01-29 Richard Biener <rguenther@suse.de>
1535 * lto-streamer.h (LTO_major_version): Bump to 3.
1536 (LTO_minor_version): Reset to 0.
1538 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1540 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1541 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1542 (arm_file_start): Generate correct asm header for armv7ve.
1543 * config/arm/bpabi.h: Add multilib support for armv7ve.
1544 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1545 and cortex-a15 to armv7ve.
1546 * config/arm/t-aprofile: Add multilib support for armv7ve.
1547 * doc/invoke.texi: Document -march=armv7ve.
1549 2014-01-29 Richard Biener <rguenther@suse.de>
1551 PR tree-optimization/58742
1552 * tree-ssa-forwprop.c (associate_plusminus): Return true
1553 if we changed sth, defer EH cleanup to ...
1554 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1555 (simplify_mult): New function.
1557 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1560 PR tree-optimization/59920
1561 * tree.c (build_common_builtin_nodes): Remove
1562 __builtin_setjmp_dispatcher initialization.
1563 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1564 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1565 instead of gsi_after_labels + manually skipping debug stmts.
1566 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1567 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1568 * tree-inline.c (copy_edges_for_bb): Remove
1569 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1570 argument. Ignore computed_goto_p stmts. Don't call
1571 make_abnormal_goto_edges. If a call might need abnormal edges
1572 for non-local gotos, see if it already has an edge to
1573 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1574 with true argument, don't do anything then, otherwise add
1575 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1576 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1578 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1579 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1580 (lower_stmt): Don't set data->calls_builtin_setjmp.
1581 (lower_builtin_setjmp): Adjust comment.
1582 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1583 * tree-cfg.c (found_computed_goto): Remove.
1584 (factor_computed_gotos): Remove.
1585 (make_goto_expr_edges): Return bool, true for computed gotos.
1586 Don't call make_abnormal_goto_edges.
1587 (build_gimple_cfg): Don't set found_computed_goto, don't call
1588 factor_computed_gotos.
1589 (computed_goto_p): No longer static.
1590 (make_blocks): Don't set found_computed_goto.
1591 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1592 (make_edges): If make_goto_expr_edges returns true, push bb
1593 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1594 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1595 vector. Record mapping between bbs and OpenMP regions if there
1596 are any, adjust make_gimple_omp_edges caller. Call
1597 handle_abnormal_edges.
1598 (make_abnormal_goto_edges): Remove.
1599 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1600 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1601 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1602 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1603 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1604 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1605 filling *region also set *region_idx to (*region)->entry->index.
1608 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1609 For REGs set ORIGINAL_REGNO.
1611 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1613 * doc/md.texi: Mention that a target shouldn't implement
1614 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1617 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1619 PR tree-optimization/59594
1620 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1621 a copy of the datarefs vector rather than the vector itself.
1623 2014-01-28 Jason Merrill <jason@redhat.com>
1626 * dwarf2out.c (auto_die): New static.
1627 (gen_type_die_with_usage): Handle C++1y 'auto'.
1628 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1631 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1634 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1635 (SPEC_X32): Likewise.
1636 (SPEC_64): Likewise.
1637 * config/i386/i386.c (ix86_option_override_internal): Turn off
1638 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1640 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1641 * config/i386/i386.h (TARGET_16BIT): New macro.
1642 (TARGET_16BIT_P): Likewise.
1643 * config/i386/i386.opt: Add m16.
1644 * doc/invoke.texi: Document -m16.
1646 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1648 PR preprocessor/59935
1649 * input.c (location_get_source_line): Bail out on when line number
1650 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1652 2014-01-28 Richard Biener <rguenther@suse.de>
1654 PR tree-optimization/58742
1655 * tree-ssa-forwprop.c (associate_plusminus): Handle
1656 pointer subtraction of the form (T)(P + A) - (T)P.
1658 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1660 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1663 2014-01-28 Richard Biener <rguenther@suse.de>
1666 2014-01-28 Richard Biener <rguenther@suse.de>
1668 PR rtl-optimization/45364
1669 PR rtl-optimization/59890
1670 * var-tracking.c (local_get_addr_clear_given_value): Handle
1671 already cleared slot.
1672 (val_reset): Handle not allocated local_get_addr_cache.
1673 (vt_find_locations): Use post-order on the inverted CFG.
1675 2014-01-28 Richard Biener <rguenther@suse.de>
1677 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1679 2014-01-28 Richard Biener <rguenther@suse.de>
1681 PR rtl-optimization/45364
1682 PR rtl-optimization/59890
1683 * var-tracking.c (local_get_addr_clear_given_value): Handle
1684 already cleared slot.
1685 (val_reset): Handle not allocated local_get_addr_cache.
1686 (vt_find_locations): Use post-order on the inverted CFG.
1688 2014-01-28 Alan Modra <amodra@gmail.com>
1690 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1691 * configure.ac <recursive call for build != host>: Define
1692 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1693 and LD_FOR_BUILD too.
1694 * configure: Regenerate.
1696 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1698 * config/i386/i386.c (get_builtin_code_for_version): Separate
1699 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1700 Broadwell from Haswell.
1702 2014-01-27 Steve Ellcey <sellcey@mips.com>
1704 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1705 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1706 * config/mips/mips.c (mips_option_override): Change setting
1708 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1709 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1710 Change from Mask to Var.
1712 2014-01-27 Jeff Law <law@redhat.com>
1714 * ipa-inline.c (inline_small_functions): Fix typo.
1716 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1718 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1719 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1720 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1721 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1722 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1723 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1724 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1725 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1726 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1727 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1728 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1729 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1730 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1731 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1732 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1733 (_mm512_storeu_epi64): Ditto.
1734 (_mm512_cmpge_epi32_mask): Ditto.
1735 (_mm512_cmpge_epu32_mask): Ditto.
1736 (_mm512_cmpge_epi64_mask): Ditto.
1737 (_mm512_cmpge_epu64_mask): Ditto.
1738 (_mm512_cmple_epi32_mask): Ditto.
1739 (_mm512_cmple_epu32_mask): Ditto.
1740 (_mm512_cmple_epi64_mask): Ditto.
1741 (_mm512_cmple_epu64_mask): Ditto.
1742 (_mm512_cmplt_epi32_mask): Ditto.
1743 (_mm512_cmplt_epu32_mask): Ditto.
1744 (_mm512_cmplt_epi64_mask): Ditto.
1745 (_mm512_cmplt_epu64_mask): Ditto.
1746 (_mm512_cmpneq_epi32_mask): Ditto.
1747 (_mm512_cmpneq_epu32_mask): Ditto.
1748 (_mm512_cmpneq_epi64_mask): Ditto.
1749 (_mm512_cmpneq_epu64_mask): Ditto.
1750 (_mm512_expand_pd): Ditto.
1751 (_mm512_expand_ps): Ditto.
1752 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1753 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1754 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1755 * config/i386/i386.c (ix86_builtins): Add
1756 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1757 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1758 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1759 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1760 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1761 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1762 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1763 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1764 IX86_BUILTIN_PMOVUSQW512_MEM.
1765 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1766 __builtin_ia32_pmovsqd512mem_mask,
1767 __builtin_ia32_pmovqd512mem_mask,
1768 __builtin_ia32_pmovusqw512mem_mask,
1769 __builtin_ia32_pmovsqw512mem_mask,
1770 __builtin_ia32_pmovqw512mem_mask,
1771 __builtin_ia32_pmovusdw512mem_mask,
1772 __builtin_ia32_pmovsdw512mem_mask,
1773 __builtin_ia32_pmovdw512mem_mask,
1774 __builtin_ia32_pmovqb512mem_mask,
1775 __builtin_ia32_pmovusqb512mem_mask,
1776 __builtin_ia32_pmovsqb512mem_mask,
1777 __builtin_ia32_pmovusdb512mem_mask,
1778 __builtin_ia32_pmovsdb512mem_mask,
1779 __builtin_ia32_pmovdb512mem_mask.
1780 (bdesc_args): Add __builtin_ia32_expanddf512,
1781 __builtin_ia32_expandsf512.
1782 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1783 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1784 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1785 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1786 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1787 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1788 (avx512f_<code>v8div16qi2_mask_store): This.
1789 (avx512f_expand<mode>): New.
1791 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1793 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1795 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1796 (_mm512_prefetch_i32scatter_pd): Ditto.
1797 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1798 (_mm512_prefetch_i64scatter_pd): Ditto.
1799 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1800 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1801 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1802 (_mm512_prefetch_i32scatter_ps): Ditto.
1803 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1804 (_mm512_prefetch_i64scatter_ps): Ditto.
1805 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1806 * config/i386/i386-builtin-types.def: Define
1807 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1808 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1809 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1810 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1811 IX86_BUILTIN_SCATTERPFQPD.
1812 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1813 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1814 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1815 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1816 __builtin_ia32_scatterpfqps.
1817 (ix86_expand_builtin): Expand new built-ins.
1818 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1819 fix memory access data type.
1820 (*avx512pf_gatherpf<mode>_mask): Ditto.
1821 (*avx512pf_gatherpf<mode>): Ditto.
1822 (avx512pf_scatterpf<mode>): Ditto.
1823 (*avx512pf_scatterpf<mode>_mask): Ditto.
1824 (*avx512pf_scatterpf<mode>): Ditto.
1825 (GATHER_SCATTER_SF_MEM_MODE): New.
1826 (avx512pf_gatherpf<mode>df): Ditto.
1827 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1828 (*avx512pf_scatterpf<mode>df): Ditto.
1830 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1833 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1834 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1837 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1839 * common/config/arm/arm-common.c
1840 (arm_rewrite_mcpu): Handle multiple names.
1842 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1844 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1846 * gimple-builder.h (create_gimple_tmp): Delete.
1848 2014-01-27 Christian Bruel <christian.bruel@st.com>
1850 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1853 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1855 * config/pa/pa.md (call): Generate indirect long calls to non-local
1856 functions when outputing 32-bit code.
1857 (call_value): Likewise except for special call to buggy powf function.
1859 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1860 portable runtime and PIC indirect calls.
1861 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1862 and PIC call sequences. Use ldo instead of blr to set return register
1863 in PIC call sequence.
1865 2014-01-25 Walter Lee <walt@tilera.com>
1867 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1868 avoid clobbering a live register.
1870 2014-01-25 Walter Lee <walt@tilera.com>
1872 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1873 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1874 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1875 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1877 2014-01-25 Walter Lee <walt@tilera.com>
1879 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1880 arguments on even registers.
1881 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1883 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1884 (BIGGEST_ALIGNMENT): Ditto.
1885 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1887 2014-01-25 Walter Lee <walt@tilera.com>
1889 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1890 insns before bundling.
1891 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1893 2014-01-25 Walter Lee <walt@tilera.com>
1895 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1896 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1897 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1899 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1901 * config/mips/constraints.md (kl): Delete.
1902 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1903 define expands, using...
1904 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1905 instructions for MIPS16.
1906 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1907 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1909 2014-01-25 Walter Lee <walt@tilera.com>
1911 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1915 2014-01-25 Walter Lee <walt@tilera.com>
1917 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1918 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1920 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1922 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1925 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1927 * print-rtl.c (in_call_function_usage): New var.
1928 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1929 EXPR_LIST mode as mode and not as reg note name.
1932 * cfgloopmanip.c (copy_loop_info): If
1933 loop->warned_aggressive_loop_optimizations, make sure
1934 the flag is set in target loop too.
1936 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1938 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1940 * builtins.def: Likewise.
1941 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1942 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1943 * ira.c (ira_setup_eliminable_regset): Likewise.
1944 * omp-low.c (gate_expand_omp): Likewise.
1945 (execute_lower_omp): Likewise.
1946 (diagnose_sb_0): Likewise.
1947 (gate_diagnose_omp_blocks): Likewise.
1948 (simd_clone_clauses_extract): Likewise.
1951 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1953 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1954 correction for little endian...
1955 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1958 2014-01-24 Jeff Law <law@redhat.com>
1960 PR tree-optimization/59919
1961 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1962 for non-returning calls.
1964 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1966 * common/config/aarch64/aarch64-common.c
1967 (aarch64_rewrite_mcpu): Handle multiple names.
1968 * config/aarch64/aarch64.h
1969 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1971 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1973 * input.c (add_file_to_cache_tab): Handle the case where fopen
1976 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1979 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1980 from push operand if code of push isn't PRE_DEC.
1982 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1985 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1986 -mquad-memory-atomic. Update -mquad-memory documentation to say
1987 it is only used for non-atomic loads/stores.
1989 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1990 -mquad-memory or -mquad-memory-atomic switches.
1992 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1993 -mquad-memory-atomic to ISA 2.07 support.
1995 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1996 to separate support of normal quad word memory operations (ldq, stq)
1997 from the atomic quad word memory operations.
1999 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2000 support to separate non-atomic quad word operations from atomic
2001 quad word operations. Disable non-atomic quad word operations in
2002 little endian mode so that we don't have to swap words after the
2003 load and before the store.
2004 (quad_load_store_p): Add comment about atomic quad word support.
2005 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2006 options printed with -mdebug=reg.
2008 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2009 -mquad-memory-atomic as the test for whether we have quad word
2010 atomic instructions.
2011 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2012 or -mp8-vector are used, allow byte/half-word atomic operations.
2014 * config/rs6000/sync.md (load_lockedti): Insure that the address
2015 is a proper indexed or indirect address for the lqarx instruction.
2016 On little endian systems, swap the hi/lo registers after the lqarx
2018 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2019 insure the address is valid for the lqarx instruction.
2020 (store_conditionalti): Insure that the address is a proper indexed
2021 or indirect address for the stqcrx. instruction. On little endian
2022 systems, swap the hi/lo registers before doing the stqcrx.
2024 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2025 insure the address is valid for the stqcrx. instruction.
2027 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2028 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2029 type of quad memory support is available.
2031 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2034 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2035 there is a danger of looping.
2037 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2040 force flag_ira_loop_pressure if set via command line.
2042 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2044 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2045 (ashr_simd): New builtin handling DI mode.
2046 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2047 (aarch64_sshr_simddi): New match pattern.
2048 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2049 (vshrd_n_s64): Likewise.
2050 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2052 2014-01-23 Nick Clifton <nickc@redhat.com>
2054 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2055 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2056 favour of mcu specific scripts.
2057 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2060 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2061 Alex Velenko <Alex.Velenko@arm.com>
2063 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2064 (vaddv_s16): Likewise.
2065 (vaddv_s32): Likewise.
2066 (vaddv_u8): Likewise.
2067 (vaddv_u16): Likewise.
2068 (vaddv_u32): Likewise.
2069 (vaddvq_s8): Likewise.
2070 (vaddvq_s16): Likewise.
2071 (vaddvq_s32): Likewise.
2072 (vaddvq_s64): Likewise.
2073 (vaddvq_u8): Likewise.
2074 (vaddvq_u16): Likewise.
2075 (vaddvq_u32): Likewise.
2076 (vaddvq_u64): Likewise.
2077 (vaddv_f32): Likewise.
2078 (vaddvq_f32): Likewise.
2079 (vaddvq_f64): Likewise.
2080 (vmaxv_f32): Likewise.
2081 (vmaxv_s8): Likewise.
2082 (vmaxv_s16): Likewise.
2083 (vmaxv_s32): Likewise.
2084 (vmaxv_u8): Likewise.
2085 (vmaxv_u16): Likewise.
2086 (vmaxv_u32): Likewise.
2087 (vmaxvq_f32): Likewise.
2088 (vmaxvq_f64): Likewise.
2089 (vmaxvq_s8): Likewise.
2090 (vmaxvq_s16): Likewise.
2091 (vmaxvq_s32): Likewise.
2092 (vmaxvq_u8): Likewise.
2093 (vmaxvq_u16): Likewise.
2094 (vmaxvq_u32): Likewise.
2095 (vmaxnmv_f32): Likewise.
2096 (vmaxnmvq_f32): Likewise.
2097 (vmaxnmvq_f64): Likewise.
2098 (vminv_f32): Likewise.
2099 (vminv_s8): Likewise.
2100 (vminv_s16): Likewise.
2101 (vminv_s32): Likewise.
2102 (vminv_u8): Likewise.
2103 (vminv_u16): Likewise.
2104 (vminv_u32): Likewise.
2105 (vminvq_f32): Likewise.
2106 (vminvq_f64): Likewise.
2107 (vminvq_s8): Likewise.
2108 (vminvq_s16): Likewise.
2109 (vminvq_s32): Likewise.
2110 (vminvq_u8): Likewise.
2111 (vminvq_u16): Likewise.
2112 (vminvq_u32): Likewise.
2113 (vminnmv_f32): Likewise.
2114 (vminnmvq_f32): Likewise.
2115 (vminnmvq_f64): Likewise.
2117 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2119 * config/aarch64/aarch64-simd.md
2120 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2121 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2122 (*aarch64_mul3_elt<mode>): Likewise.
2123 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2124 (*aarch64_mul3_elt_to_64v2df): Likewise.
2125 (*aarch64_mla_elt<mode>): Likewise.
2126 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2127 (*aarch64_mls_elt<mode>): Likewise.
2128 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2129 (*aarch64_fma4_elt<mode>): Likewise.
2130 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2131 (*aarch64_fma4_elt_to_64v2df): Likewise.
2132 (*aarch64_fnma4_elt<mode>): Likewise.
2133 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2134 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2135 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2136 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2137 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2138 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2139 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2140 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2141 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2143 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2145 * config/aarch64/aarch64-simd.md
2146 (aarch64_be_checked_get_lane<mode>): New define_expand.
2147 * config/aarch64/aarch64-simd-builtins.def
2148 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2149 New builtin definition.
2150 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2151 Use new safe be builtin.
2153 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2155 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2157 (aarch64_be_st1<mode>): Likewise.
2158 (aarch_ld1<VALL:mode>): Define_expand modified.
2159 (aarch_st1<VALL:mode>): Likewise.
2160 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2161 (UNSPEC_ST1): Likewise.
2163 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2165 * config/microblaze/microblaze.md: Add trap insn and attribute
2167 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2169 PR preprocessor/58580
2170 * input.h (location_get_source_line): Take an additional line_size
2172 (void diagnostics_file_cache_fini): Declare new function.
2173 * input.c (struct fcache): New type.
2174 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2175 New static constants.
2176 (diagnostic_file_cache_init, total_lines_num)
2177 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2178 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2179 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2180 (get_next_line, read_next_line, goto_next_line, read_line_num):
2181 New static function definitions.
2182 (diagnostic_file_cache_fini): New function.
2183 (location_get_source_line): Take an additional output line_len
2184 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2186 * diagnostic.c (diagnostic_finish): Call
2187 diagnostic_file_cache_fini.
2188 (adjust_line): Take an additional input parameter for the length
2189 of the line, rather than calculating it with strlen.
2190 (diagnostic_show_locus): Adjust the use of
2191 location_get_source_line and adjust_line with respect to their new
2192 signature. While displaying a line now, do not stop at the first
2193 null byte. Rather, display the zero byte as a space and keep
2194 going until we reach the size of the line.
2195 * Makefile.in: Add vec.o to OBJS-libcommon
2197 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2198 Ilya Tocar <ilya.tocar@intel.com>
2200 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2201 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2202 (__builtin_ia32_kmov16): Ditto.
2203 * config/i386/i386.md (UNSPEC_KMOV): New.
2206 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2208 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2209 (_mm512_storeu_si512): Ditto.
2211 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2214 * rtl.h (get_referenced_operands): Declare.
2215 * recog.c (get_referenced_operands): New function.
2216 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2217 operands have been referenced when recording LO_SUM references.
2219 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2221 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2223 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2225 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2226 Enable for generic and recent AMD targets.
2228 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2230 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2231 ARG_SIZE note when adjustment was eliminated.
2233 2014-01-22 Jeff Law <law@redhat.com>
2235 PR tree-optimization/59597
2236 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2237 in file. Accept new argument REGISTERING and use it to modify
2238 dump output appropriately.
2239 (register_jump_thread): Corresponding changes.
2240 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2241 thread paths involving joiner blocks. Add code to dump cancelled
2242 jump threading paths.
2244 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2246 PR rtl-optimization/59477
2247 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2248 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2250 2014-01-22 Tom Tromey <tromey@redhat.com>
2252 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2254 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2256 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2258 PR rtl-optimization/59896
2259 * lra-constraints.c (process_alt_operands): Check unused note for
2260 matched operands of insn with no output reloads.
2262 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2264 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2265 (mips_move_from_gpr_cost): Likewise.
2267 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2269 PR rtl-optimization/59858
2270 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2271 ira_class_hard_regs_num.
2272 (process_alt_operands): Increase reject for dying matched operand.
2274 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2277 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2278 smaller than size, perform several stores or loads and stores
2279 at dst + count - size to store or copy all of size bytes, rather
2280 than just last modesize bytes.
2282 2014-01-20 DJ Delorie <dj@redhat.com>
2284 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2285 that CLOBBERs are REGs before propogating their values.
2287 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2290 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2291 (cgraph_inline_failed_type): New function.
2292 * cgraph.h (DEFCIFCODE): Add type.
2293 (cgraph_inline_failed_type_t): New enum.
2294 (cgraph_inline_failed_type): New prototype.
2295 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2296 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2297 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2298 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2299 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2300 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2301 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2302 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2303 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2304 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2305 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2306 OPTIMIZATION_MISMATCH.
2307 * tree-inline.c (expand_call_inline): Emit errors during
2308 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2310 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2313 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2314 mode attribute in insn output.
2316 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2318 * output.h (output_constant): Delete.
2319 * varasm.c (output_constant): Make private.
2321 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2323 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2325 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2328 * tree.h (fold_builtin_strcat): New prototype.
2329 * builtins.c (fold_builtin_strcat): No longer static. Add len
2330 argument, if non-NULL, don't call c_strlen. Optimize
2331 directly into __builtin_memcpy instead of __builtin_strcpy.
2332 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2333 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2335 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2337 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2338 for SImode_address_operand operands, having only a REG argument.
2340 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2342 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2343 loader name using mbig-endian.
2344 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2346 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2348 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2352 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2354 * config/aarch64/aarch64-protos.h
2355 (aarch64_cannot_change_mode_class_ptr): Declare.
2356 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2357 aarch64_cannot_change_mode_class_ptr): New.
2358 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2359 backend hook aarch64_cannot_change_mode_class.
2361 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2363 * common/config/aarch64/aarch64-common.c
2364 (aarch64_handle_option): Don't handle any option order logic here.
2365 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2366 selected_cpu, warn on architecture version mismatch.
2367 (aarch64_override_options): Fix parsing order for option strings.
2369 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2370 Iain Sandoe <iain@codesourcery.com>
2373 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2374 warning. Amend comment to reflect current functionality.
2376 2014-01-20 Richard Biener <rguenther@suse.de>
2379 * builtins.c (fold_builtin_strcat): Remove case better handled
2380 by tree-ssa-strlen.c.
2382 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2384 * config/aarch64/aarch64.opt
2385 (mcpu, march, mtune): Make case-insensitive.
2387 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2390 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2391 if operands[1] is a REG or ZERO_EXTEND of a REG.
2393 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2395 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2397 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2399 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2400 long non-pic millicode calls.
2402 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2404 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2406 2014-01-19 Kito Cheng <kito@0xlab.org>
2408 * builtins.c (expand_movstr): Check movstr expand done or fail.
2410 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2411 H.J. Lu <hongjiu.lu@intel.com>
2414 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2415 to DImode for zero-extended addresses.
2417 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2419 PR rtl-optimization/57763
2420 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2421 on the new indirect jump_insn and increment LABEL_NUSES (label).
2423 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2427 * config.gcc (x86_archs): New variable.
2428 (x86_64_archs): Likewise.
2429 (x86_cpus): Likewise.
2430 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2431 --with-arch/--with-cpu= options.
2432 Support --with-arch=/--with-cpu={nehalem,westmere,
2433 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2435 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2437 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2438 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2440 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2442 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2444 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2447 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2448 clear cpp_get_options (parse_in)->warn_unused_macros for
2449 ix86_target_macros_internal with cpp_define.
2451 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2453 * jump.c (delete_related_insns): Keep (use (insn))s.
2454 * reorg.c (redundant_insn): Check for barriers too.
2456 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2458 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2460 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2462 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2463 call to $$dyncall when TARGET_LONG_CALLS is true.
2465 2014-01-17 Jeff Law <law@redhat.com>
2467 * ree.c (combine_set_extension): Temporarily disable test for
2468 changing number of hard registers.
2470 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2473 * ipa-inline-analysis.c (inline_free_summary):
2474 Do not free summary of aliases.
2476 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2479 * gimplify.c (gimplify_expr): Use create_tmp_var
2480 instead of create_tmp_var_raw. If cond doesn't have
2481 integral type, don't add the IFN_ANNOTATE builtin at all.
2483 2014-01-17 Martin Jambor <mjambor@suse.cz>
2486 * ipa-cp.c (prev_edge_clone): New variable.
2487 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2488 Also resize prev_edge_clone vector.
2489 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2490 (ipcp_edge_removal_hook): New function.
2491 (ipcp_driver): Register ipcp_edge_removal_hook.
2493 2014-01-17 Andrew Pinski <apinski@cavium.com>
2494 Steve Ellcey <sellcey@mips.com>
2497 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2500 2014-01-17 Jeff Law <law@redhat.com>
2503 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2504 so that pass_ccp runs first.
2506 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2508 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2509 (ix86_adjust_cost): Use !TARGET_XXX.
2510 (do_reorder_for_imul): Likewise.
2511 (swap_top_of_ready_list): Likewise.
2512 (ix86_sched_reorder): Likewise.
2514 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2516 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2517 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2518 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2519 (intel_memset): New. Duplicate slm_memset.
2520 (intel_cost): New. Duplicate slm_cost.
2521 (m_INTEL): New macro.
2522 (processor_target_table): Add "intel".
2523 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2524 with PROCESSOR_INTEL for "intel".
2525 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2526 PROCESSOR_SILVERMONT.
2527 (ix86_issue_rate): Likewise.
2528 (ix86_adjust_cost): Likewise.
2529 (ia32_multipass_dfa_lookahead): Likewise.
2530 (swap_top_of_ready_list): Likewise.
2531 (ix86_sched_reorder): Likewise.
2532 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2533 instead of TARGET_OPT_AGU.
2534 * config/i386/i386.h (TARGET_INTEL): New.
2535 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2536 (processor_type): Add PROCESSOR_INTEL.
2537 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2538 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2540 2014-01-17 Marek Polacek <polacek@redhat.com>
2543 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2546 2014-01-17 Richard Biener <rguenther@suse.de>
2548 PR tree-optimization/46590
2549 * opts.c (default_options_table): Add entries for
2550 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2551 all enabled at -O1 but not for -Og.
2552 * common.opt (fbranch-count-reg): Remove Init(1).
2553 (fmove-loop-invariants): Likewise.
2554 (ftree-pta): Likewise.
2556 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2558 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2559 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2560 decls to at least the GCC 4.8 used alignments.
2563 * tree-nested.c (convert_nonlocal_reference_stmt,
2564 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2565 of GIMPLE_BIND stmts, adjust associated decls.
2567 2014-01-17 Richard Biener <rguenther@suse.de>
2569 PR tree-optimization/46590
2570 * vec.h (vec<>::bseach): New member function implementing
2571 binary search according to C89 bsearch.
2572 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2573 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2574 bitmap pointer again. Make accesses_in_loop a flat array.
2575 (mem_ref_obstack): New global.
2576 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2577 (mark_ref_stored): Likewise.
2578 (ref_indep_loop_p_2): Likewise.
2579 (set_ref_stored_in_loop): New helper function.
2580 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2581 (memref_free): Adjust.
2582 (record_mem_ref_loc): Simplify.
2583 (gather_mem_refs_stmt): Adjust.
2584 (sort_locs_in_loop_postorder_cmp): New function.
2585 (analyze_memory_references): Sort accesses_in_loop after
2586 loop postorder number.
2587 (find_ref_loc_in_loop_cmp): New function.
2588 (for_all_locs_in_loop): Find relevant cluster of locs in
2589 accesses_in_loop and iterate without recursion.
2590 (execute_sm): Avoid uninit warning.
2591 (struct ref_always_accessed): Simplify.
2592 (ref_always_accessed::operator ()): Likewise.
2593 (ref_always_accessed_p): Likewise.
2594 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2595 loop postorder numbers here.
2596 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2599 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2602 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2603 on decls for which assemble_alias has been called.
2605 2014-01-17 Nick Clifton <nickc@redhat.com>
2607 * config/msp430/msp430.opt: (mcpu): New option.
2608 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2609 (msp430_option_override): Parse target_cpu. If the MCU name
2610 matches a generic string, clear target_mcu.
2611 (msp430_attr): Allow numeric interrupt values up to 63.
2612 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2613 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2615 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2617 * config/msp430/msp430.md (popm): Use %J rather than %I.
2618 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2619 (addhi_cy_i): Use immediate_operand for operand 2.
2620 * doc/invoke.texi: Document -mcpu option.
2622 2014-01-17 Richard Biener <rguenther@suse.de>
2624 PR rtl-optimization/38518
2625 * df.h (df_analyze_loop): Declare.
2626 * df-core.c: Include cfgloop.h.
2627 (df_analyze_1): Split out main part of df_analyze.
2628 (df_analyze): Adjust.
2629 (loop_inverted_post_order_compute): New function.
2630 (loop_post_order_compute): Likewise.
2631 (df_analyze_loop): New function avoiding whole-function
2633 * loop-invariant.c (find_defs): Use df_analyze_loop.
2634 (find_invariants): Adjust.
2635 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2637 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2639 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2640 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2642 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2644 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2645 traversal when removing references.
2647 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2650 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2652 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2655 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2656 pushing a reload for an autoinc when we had previously reloaded an
2657 inner part of the address.
2659 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2661 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2663 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2664 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2665 when not giving up or versioning for alias only because of
2667 (vect_analyze_data_ref_dependences): Set to true.
2668 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2670 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2671 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2675 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2678 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2679 operand 0 predicate for gathers, use a new pseudo as subtarget.
2681 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2684 * lra-constraints.c (process_alt_operands): Add printing debug info.
2685 Check absence of input/output reloads for matched operands too.
2687 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2689 PR rtl-optimization/59835
2690 * ira.c (ira_init_register_move_cost): Increase cost for
2693 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2695 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2697 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2700 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2701 non-register objects. Use gen_(high/low)part more consistently.
2704 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2707 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2708 endian support, remove tests for WORDS_BIG_ENDIAN.
2709 (p8_mfvsrd_3_<mode>): Likewise.
2710 (reload_gpr_from_vsx<mode>): Likewise.
2711 (reload_gpr_from_vsxsf): Likewise.
2712 (p8_mfvsrd_4_disf): Likewise.
2714 2014-01-16 Richard Biener <rguenther@suse.de>
2716 PR rtl-optimization/46590
2717 * lcm.c (compute_antinout_edge): Use postorder iteration.
2718 (compute_laterin): Use inverted postorder iteration.
2720 2014-01-16 Nick Clifton <nickc@redhat.com>
2723 * varasm.c (output_constant): Return the number of bytes actually
2725 (output_constructor_array_range): Update the field size with the
2726 number of bytes emitted by output_constant.
2727 (output_constructor_regular_field): Likewise. Also do not
2728 complain if the total number of bytes emitted is now greater
2729 than the expected fieldpos.
2730 * output.h (output_constant): Update prototype and descriptive comment.
2732 2014-01-16 Marek Polacek <polacek@redhat.com>
2735 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2736 it is error_mark_node.
2738 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2740 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2741 VALID_AVX256_REG_OR_OI_MODE.
2743 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2745 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2746 current procedure should be profiled.
2748 2014-01-15 Andrew Pinski <apinski@cavium.com>
2750 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2751 of moving from/to the STACK_REG register class.
2753 2014-01-15 Richard Henderson <rth@redhat.com>
2756 * reginfo.c (global_regs_decl): Globalize.
2757 * rtl.h (global_regs_decl): Declare.
2758 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2759 reserved via global_regs.
2761 2014-01-15 Teresa Johnson <tejohnson@google.com>
2763 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2765 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2767 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2768 and vmulosh rather than call gen_vec_widen_smult_*.
2769 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2770 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2771 (vec_widen_smult_even_v16qi): Likewise.
2772 (vec_widen_umult_even_v8hi): Likewise.
2773 (vec_widen_smult_even_v8hi): Likewise.
2774 (vec_widen_umult_odd_v16qi): Likewise.
2775 (vec_widen_smult_odd_v16qi): Likewise.
2776 (vec_widen_umult_odd_v8hi): Likewise.
2777 (vec_widen_smult_odd_v8hi): Likewise.
2778 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2779 vmuloub rather than call gen_vec_widen_umult_*.
2780 (vec_widen_umult_lo_v16qi): Likewise.
2781 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2782 vmulosb rather than call gen_vec_widen_smult_*.
2783 (vec_widen_smult_lo_v16qi): Likewise.
2784 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2785 rather than call gen_vec_widen_umult_*.
2786 (vec_widen_umult_lo_v8hi): Likewise.
2787 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2788 rather than call gen_vec_widen_smult_*.
2789 (vec_widen_smult_lo_v8hi): Likewise.
2791 2014-01-15 Jeff Law <law@redhat.com>
2793 PR tree-optimization/59747
2794 * ree.c (find_and_remove_re): Properly handle case where a second
2795 eliminated extension requires widening a copy created for elimination
2796 of a prior extension.
2797 (combine_set_extension): Ensure that the number of hard regs needed
2798 for a destination register does not change when we widen it.
2800 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2802 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2803 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2804 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2805 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2806 (avr-*-rtems*): Likewise.
2807 (bfin*-rtems*): Likewise.
2808 (moxie-*-rtems*): Likewise.
2809 (h8300-*-rtems*): Likewise.
2810 (i[34567]86-*-rtems*): Likewise.
2811 (lm32-*-rtems*): Likewise.
2812 (m32r-*-rtems*): Likewise.
2813 (m68k-*-rtems*): Likewise.
2814 (microblaze*-*-rtems*): Likewise.
2815 (mips*-*-rtems*): Likewise.
2816 (powerpc-*-rtems*): Likewise.
2817 (sh-*-rtems*): Likewise.
2818 (sparc-*-rtems*): Likewise.
2819 (sparc64-*-rtems*): Likewise.
2820 (v850-*-rtems*): Likewise.
2821 (m32c-*-rtems*): Likewise.
2823 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2825 PR rtl-optimization/59511
2826 * ira.c (ira_init_register_move_cost): Use memory costs for some
2827 cases of register move cost calculations.
2828 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2829 instead of BB frequency.
2830 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2831 * lra-assigns.c (find_hard_regno_for): Ditto.
2833 2014-01-15 Richard Biener <rguenther@suse.de>
2835 PR tree-optimization/59822
2836 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2837 (vectorizable_load): Use it to hoist defs of uses of invariant
2838 loads out of the loop.
2840 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2841 Kugan Vivekanandarajah <kuganv@linaro.org>
2844 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2847 2014-01-15 Richard Biener <rguenther@suse.de>
2849 PR rtl-optimization/59802
2850 * lcm.c (compute_available): Use inverted postorder to seed
2851 the initial worklist.
2853 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2856 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2857 ADDR_REGS for invalid symrefs in non-PIC code.
2859 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2862 * builtins.c (determine_block_size): Initialize *probable_max_size
2863 even if len_rtx is CONST_INT.
2865 2014-01-14 Andrew Pinski <apinski@cavium.com>
2867 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2868 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2869 (cortexa53_tunings): Likewise.
2870 (aarch64_sched_issue_rate): New function.
2871 (TARGET_SCHED_ISSUE_RATE): Define.
2873 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2875 * ira-costs.c (find_costs_and_classes): Add missed
2876 ira_init_register_move_cost_if_necessary.
2878 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2881 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2883 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2886 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2887 to indicate if type is used for function return value. Warn ABI
2888 change if the vector mode isn't available for function return value.
2889 (ix86_function_arg_advance): Pass false to type_natural_mode.
2890 (ix86_function_arg): Likewise.
2891 (ix86_gimplify_va_arg): Likewise.
2892 (function_arg_32): Don't warn ABI change.
2893 (ix86_function_value): Pass true to type_natural_mode.
2894 (ix86_return_in_memory): Likewise.
2895 (ix86_struct_value_rtx): Removed.
2896 (TARGET_STRUCT_VALUE_RTX): Likewise.
2898 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2900 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2901 converting a conditional jump into a conditional return.
2903 2014-01-14 Richard Biener <rguenther@suse.de>
2905 PR tree-optimization/58921
2906 PR tree-optimization/59006
2907 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2908 hoisting invariant stmts.
2909 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2910 invariant loads on the preheader edge if possible.
2912 2014-01-14 Joey Ye <joey.ye@arm.com>
2914 * doc/plugin.texi (Building GCC plugins): Update to C++.
2916 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2918 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2919 (_mm_rcp28_round_ss): Ditto.
2920 (_mm_rsqrt28_round_sd): Ditto.
2921 (_mm_rsqrt28_round_ss): Ditto.
2922 (_mm_rcp28_sd): Ditto.
2923 (_mm_rcp28_ss): Ditto.
2924 (_mm_rsqrt28_sd): Ditto.
2925 (_mm_rsqrt28_ss): Ditto.
2926 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2927 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2928 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2929 (IX86_BUILTIN_RCP28SD): Ditto.
2930 (IX86_BUILTIN_RCP28SS): Ditto.
2931 (IX86_BUILTIN_RSQRT28SD): Ditto.
2932 (IX86_BUILTIN_RSQRT28SS): Ditto.
2933 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2934 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2935 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2936 (ix86_expand_special_args_builtin): Expand new FTYPE.
2937 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2938 (srcp14<mode>): Make insn unary.
2939 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2940 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2941 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2942 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2943 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2944 Fix rounding: make it SAE only.
2945 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2947 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2949 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2950 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2951 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2952 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2953 (round_saeonly_mask_scalar_operand4): Ditto.
2954 (round_saeonly_mask_scalar_op3): Ditto.
2955 (round_saeonly_mask_scalar_op4): Ditto.
2957 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2959 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2960 Implement -maltivec=be for vec_insert and vec_extract.
2962 2014-01-10 DJ Delorie <dj@redhat.com>
2964 * config/msp430/msp430.md (call_internal): Don't allow memory
2965 references with SP as the base register.
2966 (call_value_internal): Likewise.
2967 * config/msp430/constraints.md (Yc): New. For memory references
2968 that don't use SP as a base register.
2970 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2971 "an integer without a # prefix"
2972 * config/msp430/msp430.md (epilogue_helper): Use it.
2974 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2977 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2978 AVX512F gather builtins.
2979 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2980 on gather decls with INTEGER_TYPE masktype.
2981 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2982 directly into the builtin rather than hoisting it before loop.
2984 PR tree-optimization/59387
2985 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2986 (scev_const_prop): If folded_casts and type has undefined overflow,
2987 use force_gimple_operand instead of force_gimple_operand_gsi and
2988 for each added stmt if it is assign with
2989 arith_code_with_undefined_signed_overflow, call
2990 rewrite_to_defined_overflow.
2991 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2992 gimple-fold.h instead.
2993 (arith_code_with_undefined_signed_overflow,
2994 rewrite_to_defined_overflow): Moved to ...
2995 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2996 rewrite_to_defined_overflow): ... here. No longer static.
2997 Include gimplify-me.h.
2998 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2999 rewrite_to_defined_overflow): New prototypes.
3001 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3003 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3005 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3007 * builtins.c (get_object_alignment_2): Minor tweak.
3008 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3010 2014-01-13 Christian Bruel <christian.bruel@st.com>
3012 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3013 optimized non constant lengths.
3015 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3018 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3019 load as __atomic_load_N if possible.
3021 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3023 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3025 (rs6000_expand_builtin): Adjust call.
3027 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3030 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3031 * config/rs6000/rs6000.c: Include target-globals.h.
3032 (rs6000_set_current_function): Instead of doing target_reinit
3033 unconditionally, use save_target_globals_default_opts and
3034 restore_target_globals.
3036 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3038 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3039 (rs6000_expand_builtin): Handle mffs and mtfsf.
3040 (rs6000_init_builtins): Define mffs and mtfsf.
3041 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3042 (rs6000_mffs): New pattern.
3043 (rs6000_mtfsf): New pattern.
3045 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3047 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3048 Start narrowing with START. Apply candidate-use pair
3049 and check overall cost in narrowing.
3050 (iv_ca_prune): Pass new argument.
3052 2014-01-10 Jeff Law <law@redhat.com>
3055 * ree.c (combine_reaching_defs): Ensure the defining statement
3056 occurs before the extension when optimizing extensions with
3057 different source and destination hard registers.
3059 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3062 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3063 vtables into the type inheritance graph.
3065 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3067 PR rtl-optimization/59754
3068 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3069 modes in the REGNO != REGNO case.
3071 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3073 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3075 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3077 PR tree-optimization/59745
3078 * tree-predcom.c (tree_predictive_commoning_loop): Call
3079 free_affine_expand_cache if giving up because components is NULL.
3081 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3082 GC in payload of target_globals struct instead of allocating them on
3083 the heap and the larger structs separately using GC.
3084 * target-globals.h (struct target_globals): Make regs, hard_regs,
3085 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3086 of GTY((skip)) and change type to void *.
3087 (reset_target_globals): Cast loads from those fields to corresponding
3090 2014-01-10 Steve Ellcey <sellcey@mips.com>
3093 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3094 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3095 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3097 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3100 * aarch64-modes.def (CC_Zmode): New flags mode.
3101 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3102 represents an equality.
3103 (aarch64_get_condition_code): Handle CC_Zmode.
3104 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3106 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3108 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3109 extraction in good case.
3111 2014-01-10 Richard Biener <rguenther@suse.de>
3113 PR tree-optimization/59374
3114 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3115 checking after SLP discovery. Mark stmts not participating
3116 in any SLP instance properly.
3118 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3120 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3121 when handling a SET rtx.
3123 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3125 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3126 (cortex-a57): Likewise.
3127 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3129 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3131 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3132 non-iwmmxt builtins.
3134 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3138 * ipa-devirt.c record_target_from_binfo): Take as argument
3139 stack of binfos and lookup matching one for virtual inheritance.
3140 (possible_polymorphic_call_targets_1): Update.
3142 2014-01-10 Huacai Chen <chenhc@lemote.com>
3144 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3145 kernel strings for Loongson-2E/2F/3A.
3147 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3150 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3151 is_gimple_call before calling gimple_call_internal_p.
3153 2014-01-09 Steve Ellcey <sellcey@mips.com>
3155 * Makefile.in (TREE_FLOW_H): Remove.
3156 (TREE_SSA_H): Add file names from tree-flow.h.
3157 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3158 * tree.h: Remove tree-flow.h reference.
3159 * hash-table.h: Remove tree-flow.h reference.
3160 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3161 reference with tree-ssa-loop.h.
3163 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3165 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3166 default element-order behavior for -maltivec.
3167 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3168 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3169 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3170 when targeting big endian, at least for now.
3171 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3173 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3176 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3177 var satisfies use_register_for_decl, just take into account type
3178 alignment, rather than decl alignment.
3180 PR tree-optimization/59622
3181 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3182 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3183 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3184 Don't devirtualize for inplace at all. For targets.length () == 1,
3185 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3187 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3189 * config/i386/i386.md (cpu): Remove the unused btver1.
3191 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3193 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3195 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3198 * tree-core.h (struct target_globals): New forward declaration.
3199 (struct tree_target_option): Add globals field.
3200 * tree.h (TREE_TARGET_GLOBALS): Define.
3201 (prepare_target_option_nodes_for_pch): New prototype.
3202 * target-globals.h (struct target_globals): Define even if
3204 * tree.c (prepare_target_option_node_for_pch,
3205 prepare_target_option_nodes_for_pch): New functions.
3206 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3207 * config/i386/i386.c: Include target-globals.h.
3208 (ix86_set_current_function): Instead of doing target_reinit
3209 unconditionally, use save_target_globals_default_opts and
3210 restore_target_globals.
3212 2014-01-09 Richard Biener <rguenther@suse.de>
3214 PR tree-optimization/59715
3215 * tree-cfg.h (split_critical_edges): Declare.
3216 * tree-cfg.c (split_critical_edges): Export.
3217 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3219 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3221 * cfgexpand.c (expand_stack_vars): Optionally disable
3222 asan stack protection.
3223 (expand_used_vars): Likewise.
3224 (partition_stack_vars): Likewise.
3225 * asan.c (asan_emit_stack_protection): Optionally disable
3226 after return stack usage.
3227 (instrument_derefs): Optionally disable memory access instrumentation.
3228 (instrument_builtin_call): Likewise.
3229 (instrument_strlen_call): Likewise.
3230 (asan_protect_global): Optionally disable global variables protection.
3231 * doc/invoke.texi: Added doc for new options.
3232 * params.def: Added new options.
3233 * params.h: Likewise.
3235 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3237 PR rtl-optimization/59724
3238 * ifcvt.c (cond_exec_process_if_block): Don't call
3239 flow_find_head_matching_sequence with 0 longest_match.
3240 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3241 non-active insns if !stop_after.
3242 (try_head_merge_bb): Revert 2014-01-07 changes.
3244 2014-01-08 Jeff Law <law@redhat.com>
3246 * ree.c (get_sub_rtx): New function, extracted from...
3247 (merge_def_and_ext): Here.
3248 (combine_reaching_defs): Use get_sub_rtx.
3250 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3252 * cgraph.h (varpool_variable_node): Do not choke on null node.
3254 2014-01-08 Catherine Moore <clm@codesourcery.com>
3256 * config/mips/mips.md (simple_return): Attempt to use JRC
3258 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3260 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3262 PR rtl-optimization/59137
3263 * reorg.c (steal_delay_list_from_target): Call update_block for
3265 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3267 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3269 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3270 two duplicate entries.
3272 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3275 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3277 * config/mips/mips.c (mips_truncated_op_cost): New function.
3278 (mips_rtx_costs): Adjust test for BADDU.
3279 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3281 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3283 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3284 (*baddu_si): ...this new pattern.
3286 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3289 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3291 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3294 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3296 (expand_expr, expand_normal): Adjust.
3297 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3298 inner_reference_p. Use inner_reference_p to expand inner references.
3299 (store_expr): Adjust.
3300 * cfgexpand.c (expand_call_stmt): Adjust.
3302 2014-01-08 Rong Xu <xur@google.com>
3304 * gcov-io.c (gcov_var): Move from gcov-io.h.
3305 (gcov_position): Ditto.
3306 (gcov_is_error): Ditto.
3307 (gcov_rewrite): Ditto.
3308 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3309 only part to libgcc/libgcov.h.
3311 2014-01-08 Marek Polacek <polacek@redhat.com>
3314 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3316 2014-01-08 Marek Polacek <polacek@redhat.com>
3319 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3321 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3323 PR rtl-optimization/59649
3324 * stor-layout.c (get_mode_bounds): For BImode return
3325 0 and STORE_FLAG_VALUE.
3327 2014-01-08 Richard Biener <rguenther@suse.de>
3330 * gimple.h (is_gimple_builtin_call): Remove.
3331 (gimple_builtin_call_types_compatible_p): New.
3332 (gimple_call_builtin_p): New overload.
3333 * gimple.c (is_gimple_builtin_call): Remove.
3334 (validate_call): Rename to ...
3335 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3337 (validate_type): New static function.
3338 (gimple_call_builtin_p): New overload and adjust.
3339 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3340 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3341 (gimple_fold_stmt_to_constant_1): Likewise.
3342 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3344 2014-01-08 Richard Biener <rguenther@suse.de>
3347 * gimplify.c (gimplify_expr): Gimplify register-register type
3348 VIEW_CONVERT_EXPRs to separate stmts.
3350 2014-01-07 Jeff Law <law@redhat.com>
3353 * ree.c (combine_set_extension): Handle case where source
3354 and destination registers in an extension insn are different.
3355 (combine_reaching_defs): Allow source and destination registers
3356 in extension to be different under limited circumstances.
3357 (add_removable_extension): Remove restriction that the
3358 source and destination registers in the extension are the same.
3359 (find_and_remove_re): Emit a copy from the extension's
3360 destination to its source after the defining insn if
3361 the source and destination registers are different.
3364 * ifcvt.c (merge_if_block): If we are merging a block with more than
3365 one successor with a block with no successors, remove any BARRIER
3366 after the second block.
3368 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3370 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3372 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3375 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3376 for 14-bit register offsets when INT14_OK_STRICT is false.
3378 2014-01-07 Roland Stigge <stigge@antcom.de>
3379 Michael Meissner <meissner@linux.vnet.ibm.com>
3382 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3383 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3385 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3387 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3390 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3392 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3393 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3394 rtx is const0_rtx or not.
3396 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3399 * target-globals.c (save_target_globals): Remove this_fn_optab
3401 * toplev.c: Include optabs.h.
3402 (target_reinit): Temporarily restore the global options if another
3403 set of options are in force.
3405 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3407 PR rtl-optimization/58668
3408 * cfgcleanup.c (flow_find_cross_jump): Don't count
3409 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3410 to determine what is counted.
3411 (flow_find_head_matching_sequence): Use active_insn_p to determine
3413 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3415 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3416 determine what is counted.
3418 PR tree-optimization/59643
3419 * tree-predcom.c (split_data_refs_to_components): If one dr is
3420 read and one write, determine_offset fails and the write isn't
3421 in the bad component, just put the read into the bad component.
3423 2014-01-07 Mike Stump <mikestump@comcast.net>
3424 Jakub Jelinek <jakub@redhat.com>
3427 * tree-core.h (struct tree_optimization_option): Change optabs
3428 type from unsigned char * to void *.
3429 * optabs.c (init_tree_optimization_optabs): Adjust
3430 TREE_OPTIMIZATION_OPTABS initialization.
3432 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3435 * config/i386/i386.h (struct machine_function): Add
3436 no_drap_save_restore field.
3437 * config/i386/i386.c (ix86_save_reg): Use
3438 !cfun->machine->no_drap_save_restore instead of
3439 crtl->stack_realign_needed.
3440 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3441 this function clears frame_pointer_needed. Set
3442 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3443 and DRAP reg is needed.
3445 2014-01-06 Marek Polacek <polacek@redhat.com>
3448 * doc/implement-c.texi: Mention that other integer types are
3449 permitted as bit-field types in strictly conforming mode.
3451 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3453 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3456 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3458 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3460 2014-01-06 Martin Jambor <mjambor@suse.cz>
3463 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3465 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3467 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3471 * var-tracking.c (add_stores): Preserve the value of the source even if
3472 we don't record the store.
3474 2014-01-06 Terry Guo <terry.guo@arm.com>
3476 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3478 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3481 * config/darwin.c (darwin_function_section): Adjust return values to
3482 correspond to optimisation changes made in r206070.
3484 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3486 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3487 from prefetch_block tune setting.
3488 (nocona_cost): Correct size of prefetch block to 64.
3490 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3492 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3493 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3494 used to save the static chain register in the computation of the offset
3495 from which the FP registers need to be restored.
3497 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3499 PR tree-optimization/59519
3500 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3501 ICE if get_current_def (current_new_name) is already non-NULL, as long
3502 as it is a phi result of some other phi in *new_exit_bb that has
3505 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3506 or vmovdqu* for misaligned_operand.
3507 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3508 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3509 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3510 aligned_mem for AVX512F masked aligned load and store builtins and for
3513 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3515 PR tree-optimization/59651
3516 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3517 Address range for negative step should be added by TYPE_SIZE_UNIT.
3519 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3521 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3522 overlapping registers also for registers other than the stack pointer.
3524 2014-01-03 Marek Polacek <polacek@redhat.com>
3527 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3530 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3533 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3536 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3537 (push splitter): Use <P:MODE_SIZE> instead of
3538 GET_MODE_SIZE (<P:MODE>mode).
3539 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3540 (mov -1, reg peephole2): Likewise.
3541 * config/i386/sse.md (*mov<mode>_internal,
3542 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3543 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3544 *<code><mode>3, *andnot<mode>3<mask_name>,
3545 <mask_codefor><code><mode>3<mask_name>): Likewise.
3546 * config/i386/subst.md (mask_mode512bit_condition,
3547 sd_mask_mode512bit_condition): Likewise.
3549 2014-01-02 Xinliang David Li <davidxl@google.com>
3551 PR tree-optimization/59303
3552 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3553 (dump_predicates): Better output format.
3554 (pred_equal_p): New function.
3555 (is_neq_relop_p): Ditto.
3556 (is_neq_zero_form_p): Ditto.
3557 (pred_expr_equal_p): Ditto.
3558 (pred_neg_p): Ditto.
3559 (simplify_pred): Ditto.
3560 (simplify_preds_2): Ditto.
3561 (simplify_preds_3): Ditto.
3562 (simplify_preds_4): Ditto.
3563 (simplify_preds): Ditto.
3565 (push_to_worklist): Ditto.
3566 (get_pred_info_from_cmp): Ditto.
3567 (is_degenerated_phi): Ditto.
3568 (normalize_one_pred_1): Ditto.
3569 (normalize_one_pred): Ditto.
3570 (normalize_one_pred_chain): Ditto.
3571 (normalize_preds): Ditto.
3572 (normalize_cond_1): Remove function.
3573 (normalize_cond): Ditto.
3574 (is_gcond_subset_of): Ditto.
3575 (is_subset_of_any): Ditto.
3576 (is_or_set_subset_of): Ditto.
3577 (is_and_set_subset_of): Ditto.
3578 (is_norm_cond_subset_of): Ditto.
3579 (pred_chain_length_cmp): Ditto.
3580 (convert_control_dep_chain_into_preds): Type change.
3581 (find_predicates): Ditto.
3582 (find_def_preds): Ditto.
3583 (destroy_predicates_vecs): Ditto.
3584 (find_matching_predicates_in_rest_chains): Ditto.
3585 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3586 (is_pred_expr_subset): Ditto.
3587 (is_pred_chain_subset_of): Ditto.
3588 (is_included_in): Ditto.
3589 (is_superset_of): Ditto.
3591 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3593 Update copyright years.
3595 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3597 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3598 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3599 config/arc/arc.md, config/arc/arc.opt,
3600 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3601 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3602 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3603 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3604 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3605 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3606 vtable-verify.c, vtable-verify.h: Use the standard form for the
3609 2014-01-02 Tobias Burnus <burnus@net-b.de>
3611 * gcc.c (process_command): Update copyright notice dates.
3612 * gcov-dump.c: Ditto.
3614 * doc/cpp.texi: Bump @copying's copyright year.
3615 * doc/cppinternals.texi: Ditto.
3616 * doc/gcc.texi: Ditto.
3617 * doc/gccint.texi: Ditto.
3618 * doc/gcov.texi: Ditto.
3619 * doc/install.texi: Ditto.
3620 * doc/invoke.texi: Ditto.
3622 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3624 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3626 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3628 * config/i386/sse.md (*mov<mode>_internal): Guard
3629 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3631 PR rtl-optimization/59647
3632 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3633 new_rtx into UNSIGNED_FLOAT rtxes.
3635 Copyright (C) 2014 Free Software Foundation, Inc.
3637 Copying and distribution of this file, with or without modification,
3638 are permitted in any medium without royalty provided the copyright
3639 notice and this notice are preserved.