1 2014-02-20 Richard Biener <rguenther@suse.de>
3 * tree-cfg.c (replace_uses_by): Mark altered BBs before
4 doing the substitution.
5 (verify_gimple_assign_single): Also verify bare MEM_REFs
8 2014-02-20 Martin Jambor <mjambor@suse.cz>
11 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
12 info when checking whether lattices are bottom.
14 2014-02-20 Richard Biener <rguenther@suse.de>
17 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
20 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
23 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale parameter
24 specifying the scaling.
25 (inline_call): Update.
26 (want_inline_recursively): Guard division by zero.
27 (recursive_inlining): Update.
28 * ipa-inline.h (clone_inlined_nodes): Update.
30 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
33 * config/i386/i386.c (classify_argument): Pass structures of size
34 64 bytes or less in register.
36 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
37 Kirill Yukhin <kirill.yukhin@intel.com>
39 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
40 (_mm_rcp28_round_ss): Ditto.
41 (_mm_rsqrt28_round_sd): Ditto.
42 (_mm_rsqrt28_round_ss): Ditto.
43 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
44 (_mm_rcp14_round_ss): Ditto.
45 (_mm_rsqrt14_round_sd): Ditto.
46 (_mm_rsqrt14_round_ss): Ditto.
47 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
48 the first input operand, get rid of match_dup.
49 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
51 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
53 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
54 operand as the first input operand, set type attribute.
55 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
57 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
58 operand as the first input operand, set type attribute.
60 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
62 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
65 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
68 * config/i386/i386.c (construct_container): Remove TFmode check
69 for X86_64_INTEGER_CLASS.
71 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
74 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
75 only when -Wpsabi is enabled.
77 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
80 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
81 passing arrays in registers are the same as for structs, so remove the
82 special case for them.
84 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
86 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
87 destination type, extract only the valid bits if the source type is not
88 integral and has a different mode.
90 2014-02-19 Richard Biener <rguenther@suse.de>
93 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
96 2014-02-19 Richard Biener <rguenther@suse.de>
99 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
100 (ipa_modify_call_arguments): Emit an argument load explicitely and
101 preserve virtual SSA form there and for the replacement call.
102 Do not update SSA form nor free dominance info.
104 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
106 * ipa.c (function_and_variable_visibility): Also clear WEAK
107 flag when disolving COMDAT_GROUP.
109 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
111 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
112 * ipa-prop.c (ipa_set_jf_known_type): Return early when
114 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
115 do more sanity checks.
116 (detect_type_change): Return true when giving up early.
117 (compute_complex_assign_jump_func): Fix type parameter of
119 (compute_complex_ancestor_jump_func): Likewise.
120 (update_jump_functions_after_inlining): Fix updating of
122 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
124 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
126 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
127 inline clones when edge disappears.
129 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
132 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
133 Split 64-bit moves into 2 patterns. Do not allow the use of
134 direct move for TDmode in little endian, since the decimal value
135 has little endian bytes within a word, but the 64-bit pieces are
136 ordered in a big endian fashion, and normal subreg's of TDmode are
138 (mov<mode>_64bit_dm): Likewise.
139 (movtd_64bit_nodm): Likewise.
141 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
143 PR tree-optimization/60174
144 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
145 statement of an SSA_NAME that occurs in an abnormal PHI node.
147 2014-02-18 Jakub Jelinek <jakub@redhat.com>
150 * final.c (SEEN_BB): Remove.
151 (SEEN_NOTE, SEEN_EMITTED): Renumber.
152 (final_scan_insn): Don't force_source_line on second
153 NOTE_INSN_BASIC_BLOCK.
155 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
158 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
159 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
160 (type_natural_mode): Warn ABI change when %zmm register is not
161 available for AVX512F vector value passing.
163 2014-02-18 Kai Tietz <ktietz@redhat.com>
166 * config/i386/i386.c (ix86_expand_prologue): Use value in
167 rax register as displacement when restoring %r10 or %rax.
168 Fix wrong offset when restoring both registers.
170 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
172 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
173 assertion with conditional return.
175 2014-02-18 Jakub Jelinek <jakub@redhat.com>
176 Uros Bizjak <ubizjak@gmail.com>
179 * config/i386/driver-i386.c (host_detect_local_cpu): If
180 YMM state is not saved by the OS, also clear has_f16c. Move
181 CPUID 0x80000001 handling before YMM state saving checking.
183 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
185 PR rtl-optimization/58960
186 * haifa-sched.c (alloc_global_sched_pressure_data): New,
187 factored out from ...
188 (sched_init): ... here.
189 (free_global_sched_pressure_data): New, factored out from ...
190 (sched_finish): ... here.
191 * sched-int.h (free_global_sched_pressure_data): Declare.
192 * sched-rgn.c (nr_regions_initial): New static global.
193 (haifa_find_rgns): Initialize it.
194 (schedule_region): Disable sched-pressure for the newly
197 2014-02-17 Richard Biener <rguenther@suse.de>
199 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
200 release SSA defs of pattern stmts.
202 2014-02-17 Richard Biener <rguenther@suse.de>
204 * tree-inline.c (expand_call_inline): Release the virtual
205 operand defined by the call we are about to inline.
207 2014-02-17 Richard Biener <rguenther@suse.de>
209 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
211 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
212 Ilya Tocar <ilya.tocar@intel.com>
214 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
215 arguments order in builtin.
216 (_mm512_permutexvar_epi64): Ditto.
217 (_mm512_mask_permutexvar_epi64): Ditto
218 (_mm512_maskz_permutexvar_epi32): Ditto
219 (_mm512_permutexvar_epi32): Ditto
220 (_mm512_mask_permutexvar_epi32): Ditto
222 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
224 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
225 (p8_vmrgow): Likewise.
227 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
229 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
232 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
235 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
236 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
237 into 64-bit and 32-bit moves. On 64-bit moves, add support for
238 using direct move instructions on ISA 2.07. Also adjust
239 instruction length for 64-bit.
240 (mov<mode>_64bit, TFmode/TDmode): Likewise.
241 (mov<mode>_32bit, TFmode/TDmode): Likewise.
243 2014-02-15 Alan Modra <amodra@gmail.com>
247 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
248 find_replacement on parts of insn rtl that might be reloaded.
250 2014-02-15 Richard Biener <rguenther@suse.de>
252 PR tree-optimization/60183
253 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
254 (tree_ssa_phiprop): Calculate and free post-dominators.
256 2014-02-14 Jeff Law <law@redhat.com>
258 PR rtl-optimization/60131
259 * ree.c (get_extended_src_reg): New function.
260 (combine_reaching_defs): Use it rather than assuming location of REG.
261 (find_and_remove_re): Verify first operand of extension is
262 a REG before adding the insns to the copy list.
264 2014-02-14 Roland McGrath <mcgrathr@google.com>
266 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
267 * configure: Regenerated.
268 * config.in: Regenerated.
269 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
270 instead of ASM_SHORT.
272 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
273 Richard Earnshaw <rearnsha@arm.com>
275 PR rtl-optimization/59535
276 * lra-constraints.c (process_alt_operands): Encourage alternative
277 when unassigned pseudo class is superset of the alternative class.
278 (inherit_reload_reg): Don't inherit when optimizing for code size.
279 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
280 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
281 modes not less than 4 for Thumb1.
283 2014-02-14 Kyle McMartin <kyle@redhat.com>
286 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
288 2014-02-14 Richard Biener <rguenther@suse.de>
290 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
291 (get_frame_arg): Drop the assert with langhook types_compatible_p.
292 Do not strip INDIRECT_REFs.
294 2014-02-14 Richard Biener <rguenther@suse.de>
297 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
298 DECL_FUNCTION_SPECIFIC_TARGET.
299 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
300 * tree-streamer-out.c (pack_ts_target_option): Remove.
301 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
302 (write_ts_function_decl_tree_pointers): Do not stream
303 DECL_FUNCTION_SPECIFIC_TARGET.
304 * tree-streamer-in.c (unpack_ts_target_option): Remove.
305 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
306 (lto_input_ts_function_decl_tree_pointers): Do not stream
307 DECL_FUNCTION_SPECIFIC_TARGET.
309 2014-02-14 Jakub Jelinek <jakub@redhat.com>
311 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
312 (get_initial_def_for_induction, vectorizable_induction): Ignore
313 debug stmts when looking for exit_phi.
314 (vectorizable_live_operation): Fix up condition.
316 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
318 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
319 nreverse() because it changes the content of original tree list.
321 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
323 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
324 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
326 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
328 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
329 GNU coding standards.
331 2014-02-13 Jakub Jelinek <jakub@redhat.com>
334 * dwarf2out.c (gen_subprogram_die): Don't call
335 add_calling_convention_attribute if subr_die is old_die.
337 2014-02-13 Sharad Singhai <singhai@google.com>
339 * doc/optinfo.texi: Fix order of nodes.
341 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
343 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
344 operands[2], not operands[3].
346 2014-02-13 Richard Biener <rguenther@suse.de>
349 * doc/install.texi (ISL): Update recommended version to 0.12.2,
350 mention the possibility of an in-tree build.
351 (CLooG): Update recommended version to 0.18.1, mention the
352 possibility of an in-tree build and clarify that the ISL
353 bundled with CLooG does not work.
355 2014-02-13 Jakub Jelinek <jakub@redhat.com>
358 * expr.c (compress_float_constant): If x is a hard register,
359 extend into a pseudo and then move to x.
361 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
363 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
364 caused by bad second argument to warning_at() with -mhotpatch and
365 nested functions (e.g. with gfortran).
367 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
369 * opts.c (option_name): Remove "enabled by default" rider.
371 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
373 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
375 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
376 Uros Bizjak <ubizjak@gmail.com>
379 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
380 * configure: Regenerated.
382 2014-02-12 Richard Biener <rguenther@suse.de>
384 * vec.c (vec_prefix::calculate_allocation): Move as
385 inline variant to vec.h.
386 (vec_prefix::calculate_allocation_1): New out-of-line version.
387 * vec.h (vec_prefix::calculate_allocation_1): Declare.
388 (vec_prefix::m_has_auto_buf): Rename to ...
389 (vec_prefix::m_using_auto_storage): ... this.
390 (vec_prefix::calculate_allocation): Inline the easy cases
391 and dispatch to calculate_allocation_1 which doesn't need the
393 (va_heap::reserve): Use gcc_checking_assert.
394 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
395 m_using_auto_storage.
396 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
398 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
399 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
400 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
402 2014-02-12 Richard Biener <rguenther@suse.de>
404 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
405 when we found a dependence.
407 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
409 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
411 (maybe_fold_stmt): ... into this new function.
412 * omp-low.c (lower_omp): Update comment.
414 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
417 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
420 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
422 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
423 identifiers in comments.
424 (cortexa53_extra_costs): Likewise.
425 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
426 (cortexa7_extra_costs): Likewise.
427 (cortexa12_extra_costs): Likewise.
428 (cortexa15_extra_costs): Likewise.
429 (v7m_extra_costs): Likewise.
431 2014-02-12 Richard Biener <rguenther@suse.de>
434 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
435 of posix_memalign being successful.
436 (lower_stmt): Restrict lowering of posix_memalign to when
437 -ftree-bit-ccp is enabled.
439 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
441 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
443 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
445 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
447 PR rtl-optimization/60116
448 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
449 other_insn once the combination has been validated.
451 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
454 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
456 * ipa-devirt.c: Include demangle.h
457 (odr_violation_reported): New static variable.
458 (add_type_duplicate): Update odr_violations.
459 (maybe_record_node): Add completep parameter; update it.
460 (record_target_from_binfo): Add COMPLETEP parameter;
462 (possible_polymorphic_call_targets_1): Likewise.
463 (struct polymorphic_call_target_d): Add nonconstruction_targets;
464 rename FINAL to COMPLETE.
465 (record_targets_from_bases): Sanity check we found the binfo;
466 fix COMPLETEP updating.
467 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
468 parameter, fix computing of COMPLETEP.
469 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
470 at LTO time do demangling.
471 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
472 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
474 (gimple_get_virt_method_for_binfo): Likewise.
475 * gimple-fold.h (gimple_get_virt_method_for_binfo,
476 gimple_get_virt_method_for_vtable): Update prototypes.
478 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
481 * genautomata.c (add_presence_absence): Fix typo with
482 {final_}presence_list.
484 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
487 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
488 for VSX/Altivec vectors that land in GPR registers.
490 2014-02-11 Richard Henderson <rth@redhat.com>
491 Jakub Jelinek <jakub@redhat.com>
494 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
495 around drhs if type conversion to lacc->type is not useless.
497 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
499 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
501 (cortex-a57.cortex-a53): Likewise.
502 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
504 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
506 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
509 2014-02-11 Renlin Li <Renlin.Li@arm.com>
511 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
512 add_options_for_arm_vfp3.
514 2014-02-11 Jeff Law <law@redhat.com>
517 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
518 object with an undesirable mode.
520 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
523 * config/i386/sol2-9.h: New file.
524 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
525 *-*-solaris2.9*): Use it.
527 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
529 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
530 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
532 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
534 * config/microblaze/microblaze.c: Extend mcpu version format
536 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
538 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
540 2014-02-10 Richard Henderson <rth@redhat.com>
543 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
544 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
545 ms-abi vs -mno-accumulate-outgoing-args.
546 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
547 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
550 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
553 * cfgexpand.c (expand_asm_operands): Attach source location to
554 ASM_INPUT rtx objects.
555 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
557 2014-02-10 Nick Clifton <nickc@redhat.com>
559 * config/mn10300/mn10300.c (popcount): New function.
560 (mn10300_expand_prologue): Include saved registers in stack usage
563 2014-02-10 Jeff Law <law@redhat.com>
566 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
567 when changing the SET_DEST of a prior insn to avoid an input reload.
569 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
571 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
572 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
573 -mcall-openbsd, or -mcall-linux.
574 (CC1_ENDIAN_BIG_SPEC): Remove.
575 (CC1_ENDIAN_LITTLE_SPEC): Remove.
576 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
577 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
578 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
579 and %cc1_endian_default.
580 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
582 2014-02-10 Richard Biener <rguenther@suse.de>
584 PR tree-optimization/60115
585 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
586 MEM_REF handling. Properly verify that the accesses are not
587 out of the objects bound.
589 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
591 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
594 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
596 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
597 proper constants and fix formatting.
598 (possible_polymorphic_call_targets): Fix formatting.
600 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
601 Ilya Tocar <ilya.tocar@intel.com>
603 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
604 (_mm512_loadu_epi32): Renamed into...
605 (_mm512_loadu_si512): This.
606 (_mm512_storeu_epi32): Renamed into...
607 (_mm512_storeu_si512): This.
608 (_mm512_maskz_ceil_ps): Removed.
609 (_mm512_maskz_ceil_pd): Ditto.
610 (_mm512_maskz_floor_ps): Ditto.
611 (_mm512_maskz_floor_pd): Ditto.
612 (_mm512_floor_round_ps): Ditto.
613 (_mm512_floor_round_pd): Ditto.
614 (_mm512_ceil_round_ps): Ditto.
615 (_mm512_ceil_round_pd): Ditto.
616 (_mm512_mask_floor_round_ps): Ditto.
617 (_mm512_mask_floor_round_pd): Ditto.
618 (_mm512_mask_ceil_round_ps): Ditto.
619 (_mm512_mask_ceil_round_pd): Ditto.
620 (_mm512_maskz_floor_round_ps): Ditto.
621 (_mm512_maskz_floor_round_pd): Ditto.
622 (_mm512_maskz_ceil_round_ps): Ditto.
623 (_mm512_maskz_ceil_round_pd): Ditto.
624 (_mm512_expand_pd): Ditto.
625 (_mm512_expand_ps): Ditto.
626 * config/i386/i386.c (ix86_builtins): Remove
627 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
629 * config/i386/predicates.md (const1256_operand): New.
630 (const_1_to_2_operand): Ditto.
631 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
632 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
633 (*avx512pf_gatherpf<mode>sf): Ditto.
634 (avx512pf_gatherpf<mode>df): Ditto.
635 (*avx512pf_gatherpf<mode>df_mask): Ditto.
636 (*avx512pf_gatherpf<mode>df): Ditto.
637 (avx512pf_scatterpf<mode>sf): Ditto.
638 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
639 (*avx512pf_scatterpf<mode>sf): Ditto.
640 (avx512pf_scatterpf<mode>df): Ditto.
641 (*avx512pf_scatterpf<mode>df_mask): Ditto.
642 (*avx512pf_scatterpf<mode>df): Ditto.
643 (avx512f_expand<mode>): Removed.
644 (<shift_insn><mode>3<mask_name>): Change predicate type.
646 2014-02-08 Jakub Jelinek <jakub@redhat.com>
648 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
649 not at the end of datarefs vector use ordered_remove to avoid
650 reordering datarefs vector.
653 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
654 mark local addressable non-static vars as GOVD_PRIVATE
655 instead of GOVD_LOCAL.
656 * omp-low.c (lower_omp_for): Move gimple_bind_vars
657 and BLOCK_VARS of gimple_bind_block to new_stmt rather
661 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
662 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
663 assume_aligned or alloc_align attributes.
664 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
665 arguments. Handle also assume_aligned and alloc_align attributes.
666 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
667 calls to functions with assume_aligned or alloc_align attributes.
668 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
670 2014-02-08 Terry Guo <terry.guo@arm.com>
672 * doc/invoke.texi: Document ARM -march=armv7e-m.
674 2014-02-08 Jakub Jelinek <jakub@redhat.com>
676 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
677 flag on __cilkrts_rethrow builtin.
680 * ipa-cp.c (determine_versionability): Fail at -O0
681 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
682 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
685 2014-02-04 Jakub Jelinek <jakub@redhat.com>
688 * tree-inline.c (copy_forbidden): Fail for
689 __attribute__((optimize (0))) functions.
691 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
693 * varpool.c: Include pointer-set.h.
694 (varpool_remove_unreferenced_decls): Variables in other partitions
695 will not be output; be however careful to not lose information
698 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
700 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
701 lookup in the vtable constructor.
703 2014-02-07 Jeff Law <law@redhat.com>
706 * config/m68k/m68k.md (ashldi_extsi): Turn into a
707 define_insn_and_split.
709 * ipa-inline.c (inline_small_functions): Fix typos.
711 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
713 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
714 (s390_can_use_return_insn): Declare.
715 * config/s390/s390.h (EPILOGUE_USES): Define.
716 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
718 (s390_chunkify_start): Handle return JUMP_LABELs.
719 (s390_early_mach): Emit a main_pool instruction on the entry edge.
720 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
721 (s390_can_use_return_insn): New functions.
722 (s390_fix_long_loop_prediction): Handle conditional returns.
723 (TARGET_SET_UP_BY_PROLOGUE): Define.
724 * config/s390/s390.md (ANY_RETURN): New code iterator.
725 (*creturn, *csimple_return, return, simple_return): New patterns.
727 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
729 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
730 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
731 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
732 REG_CFA_RESTORE list when deciding not to restore a register.
734 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
736 * config/s390/s390.c: Include tree-pass.h and context.h.
737 (s390_early_mach): New function, split out from...
738 (s390_emit_prologue): ...here.
739 (pass_data_s390_early_mach): New pass structure.
740 (pass_s390_early_mach): New class.
741 (s390_option_override): Create and register early_mach pass.
744 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
746 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
747 to match for the exit block.
749 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
751 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
752 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
753 Reject misaligned operands.
755 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
757 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
759 2014-02-07 Richard Biener <rguenther@suse.de>
762 * gimple-low.c (lower_builtin_posix_memalign): New function.
763 (lower_stmt): Call it to lower posix_memalign in a way
764 to make alignment info accessible.
766 2014-02-07 Jakub Jelinek <jakub@redhat.com>
769 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
770 __builtin_setjmp_receiver.
772 2014-02-07 Richard Biener <rguenther@suse.de>
775 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
776 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
777 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
778 Handle BUILT_IN_POSIX_MEMALIGN.
779 (find_func_clobbers): Likewise.
780 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
781 (call_may_clobber_ref_p_1): Likewise.
783 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
786 * ipa-devirt.c (record_target_from_binfo): Remove overactive
789 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
792 * lto-cgraph.c (lto_output_node): Use
793 symtab_get_symbol_partitioning_class.
794 (lto_output_varpool_node): likewise.
795 (symtab_get_symbol_partitioning_class): Move here from
797 * cgraph.h (symbol_partitioning_class): Likewise.
798 (symtab_get_symbol_partitioning_class): Declare.
800 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
802 * ggc.h (ggc_internal_cleared_alloc): New macro.
803 * vec.h (vec_safe_copy): Handle memory stats.
804 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
805 * target-globals.c (save_target_globals): Likewise.
807 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
810 * expr.c (emit_move_resolve_push): Export; be bit more selective
811 on when to clear alias set.
812 * expr.h (emit_move_resolve_push): Declare.
813 * function.h (struct function): Add tail_call_marked.
814 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
815 * config/i386/i386-protos.h (ix86_expand_push): Remove.
816 * config/i386/i386.md (TImode move expander): De not call
818 (FP push expanders): Preserve memory attributes.
819 * config/i386/sse.md (push<mode>1): Remove.
820 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
821 (ix86_expand_push): Remove.
822 * config/i386/mmx.md (push<mode>1): Remove.
824 2014-02-06 Jakub Jelinek <jakub@redhat.com>
826 PR rtl-optimization/60030
827 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
828 lopart with paradoxical subreg before shifting it up by hprec.
830 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
832 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
833 Remove extra newline at end of file.
834 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
835 (arm_issue_rate): Handle cortexa57.
836 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
837 (cortex-a57.cortex-a53): Likewise.
839 2014-02-06 Jakub Jelinek <jakub@redhat.com>
842 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
843 don't record in REG_FRAME_RELATED_EXPR registers not set in that
845 (arm_expand_prologue): Adjust all callers.
846 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
847 info, registers also at the lowest numbered registers side. Use
848 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
852 * var-tracking.c (adjust_mems): Before adding a SET to
853 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
855 2014-02-06 Alan Modra <amodra@gmail.com>
858 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
859 change SDmode to DDmode when lra_in_progress.
861 2014-02-06 Jakub Jelinek <jakub@redhat.com>
864 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
865 free_data_ref on the dr first, and before goto again also set dr
866 to the next dr. For simd_lane_access, free old datarefs[i] before
867 overwriting it. For get_vectype_for_scalar_type failure, don't
868 free_data_ref if simd_lane_access.
870 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
873 * tree.h (opts_for_fn): New inline function.
874 (opt_for_fn): Define.
875 * config/i386/i386.c (ix86_function_regparm): Use
876 opt_for_fn (decl, optimize) instead of optimize.
878 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
880 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
881 for SYMBOL_REF in large memory model.
883 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
885 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
887 (cortex-a57): Likewise.
888 (cortex-a57.cortex-a53): Likewise.
890 2014-02-06 Yury Gribov <y.gribov@samsung.com>
891 Kugan Vivekanandarajah <kuganv@linaro.org>
893 * config/arm/arm.c (arm_vector_alignment_reachable): Check
895 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
897 2014-02-06 Richard Biener <rguenther@suse.de>
899 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
900 set_loop_copy and initialize_original_copy_tables.
902 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
904 * config/aarch64/aarch64-simd.md
905 (aarch64_ashr_simddi): Change QI to SI.
907 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
908 Jakub Jelinek <jakub@redhat.com>
911 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
914 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
916 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
917 CODE_FOR_altivec_vpku[hw]um to
918 CODE_FOR_altivec_vpku[hw]um_direct.
919 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
920 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
921 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
922 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
924 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
926 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
927 generation for -maltivec=be.
928 (altivec_vsumsws): Simplify redundant test.
930 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
932 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
933 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
934 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
935 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
937 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
939 (altivec_vpks<VI_char>ss): Likewise.
940 (altivec_vpks<VI_char>us): Likewise.
941 (altivec_vpku<VI_char>us): Likewise.
942 (altivec_vpku<VI_char>um): Likewise.
943 (altivec_vpku<VI_char>um_direct): New (copy of
944 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
946 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
947 target is little endian and -maltivec=be is not specified.
948 (*altivec_vupkhs<VU_char>_direct): New (copy of
949 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
950 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
951 target is little endian and -maltivec=be is not specified.
952 (*altivec_vupkls<VU_char>_direct): New (copy of
953 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
954 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
955 little endian and -maltivec=be is not specified.
956 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
957 little endian and -maltivec=be is not specified.
959 2014-02-05 Richard Henderson <rth@redhat.com>
962 * combine-stack-adj.c: Revert r206943.
963 * sched-int.h (struct deps_desc): Add last_args_size.
964 * sched-deps.c (init_deps): Initialize it.
965 (sched_analyze_insn): Add OUTPUT dependencies between insns that
966 contain REG_ARGS_SIZE notes.
968 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
970 * lto-cgraph.c (asm_nodes_output): Make global.
971 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
972 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
973 (driver_handle_option): Handle OPT_fwpa.
975 2014-02-05 Jakub Jelinek <jakub@redhat.com>
978 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
979 a comment typo and formatting issue. If odr_hash hasn't been
980 created, return vNULL and set *completep to false.
983 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
984 bb with no successors.
986 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
989 * doc/invoke.texi (-march): Clarify documentation for ARM.
993 2014-02-05 Richard Biener <rguenther@suse.de>
995 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
996 when not vectorizing because of too many alias checks.
997 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
998 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1000 2014-02-05 Nick Clifton <nickc@redhat.com>
1002 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1003 accept extended registers in any mode when compiling for the MN10300.
1005 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1007 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1008 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1009 sanitization attributes.
1010 (can_inline_edge_p): Likewise.
1011 (sanitize_attrs_match_for_inline_p): New function.
1013 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1015 * ipa-prop.c (detect_type_change): Shor circuit testing of
1016 type changes on THIS pointer.
1018 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1021 * config/pa/pa.c (legitimize_tls_address): Return original address
1022 if not passed a SYMBOL_REF rtx.
1023 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1025 (pa_emit_move_sequence): Simplify TLS source operands.
1026 (pa_legitimate_constant_p): Reject all TLS constants.
1027 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1028 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1030 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1032 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1033 groups when we know they are controlled by LTO.
1034 * varasm.c (default_binds_local_p_1): If object is in other partition,
1035 it will be resolved locally.
1037 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1039 * config/host-linux.c (linux_gt_pch_use_address): Don't
1040 use SSIZE_MAX because it is not always defined.
1042 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1045 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1046 threshold for pseudo splitting.
1047 (update_ebb_live_info): Process call argument hard registers and
1048 hard registers from insn definition too.
1049 (max_small_class_regs_num): New constant.
1050 (inherit_in_ebb): Update live hard regs through EBBs. Update
1051 reloads_num only for small register classes. Don't split for
1054 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1057 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1060 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1062 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1065 2014-02-04 Marek Polacek <polacek@redhat.com>
1067 * gdbinit.in (pel): Define.
1069 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1071 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1074 2014-02-04 Richard Biener <rguenther@suse.de>
1077 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1078 in function context local.
1079 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1080 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1081 similar to LTO_imported_decl_ref.
1083 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1085 PR tree-optimization/60002
1086 * cgraphclones.c (build_function_decl_skip_args): Clear
1089 PR tree-optimization/60023
1090 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1091 false to gsi_replace.
1092 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1093 has been in some EH region and vec_stmt could throw, add
1094 vec_stmt into the same EH region.
1095 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1096 has no lhs, ignore it.
1097 * internal-fn.c (expand_MASK_LOAD): Likewise.
1100 * tree-inline.c (copy_forbidden): Fail for
1101 __attribute__((optimize (0))) functions.
1104 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1105 is set, copy one less argument.
1106 (expand_simd_clones): Don't subtract clone_info->inbranch
1107 from simd_clone_struct_alloc argument.
1109 PR rtl-optimization/57915
1110 * recog.c (simplify_while_replacing): If all unary/binary/relational
1111 operation arguments are constant, attempt to simplify those.
1114 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1115 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1117 2014-02-04 Richard Biener <rguenther@suse.de>
1119 PR tree-optimization/60012
1120 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1121 TBAA disambiguation to all DDRs.
1123 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1126 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1127 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1129 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1132 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1134 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1136 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1137 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1139 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1142 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1143 to figure out targets of polymorphic calls with known decl.
1144 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1145 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1146 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1147 (get_polymorphic_call_info): ... here.
1148 (get_polymorphic_call_info_from_invariant): New function.
1150 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1152 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1153 lookup via vtable pointer; check for type consistency
1154 and turn inconsitent facts into UNREACHABLE.
1155 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1156 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1157 type inconsistent querries; return UNREACHABLE instead.
1159 2014-02-03 Richard Henderson <rth@twiddle.net>
1162 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1163 already processed this node.
1164 (normalize_one_pred_1): Pass along mark_set.
1165 (normalize_one_pred): Create and destroy a pointer_set_t.
1166 (normalize_one_pred_chain): Likewise.
1168 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1170 PR gcov-profile/58602
1171 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1173 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1176 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1177 -fno-devirtualize; try to devirtualize by the knowledge of
1178 virtual table pointer given by aggregate propagation.
1179 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1180 (ipa_print_node_jump_functions): Dump also offset that
1181 is relevant for polymorphic calls.
1182 (determine_known_aggregate_parts): Add arg_type parameter; use it
1183 instead of determining the type from pointer type.
1184 (ipa_compute_jump_functions_for_edge): Update call of
1185 determine_known_aggregate_parts.
1186 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1187 (gimple_get_virt_method_for_binfo): ... here; simplify using
1188 vtable_pointer_value_to_vtable.
1189 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1190 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1191 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1192 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1194 (vtable_pointer_value_to_binfo): ... here.
1195 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1197 2014-02-03 Teresa Johnson <tejohnson@google.com>
1199 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1200 redef of outer loop index variable.
1202 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1206 * doc/extend.texi (Function Attributes): Typo.
1208 2014-02-03 Cong Hou <congh@google.com>
1210 PR tree-optimization/60000
1211 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1212 if the vectorized statement is a store. A store statement can only
1213 appear at the end of pattern statements.
1215 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1217 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1218 (ix86_option_override_internal): Default long double to 64-bit for
1219 32-bit Bionic and to 128-bit for 64-bit Bionic.
1221 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1222 TARGET_LONG_DOUBLE_128 is true.
1223 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1225 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1226 (mlong-double-64): Negate -mlong-double-128.
1227 (mlong-double-128): New option.
1229 * config/i386/i386-c.c (ix86_target_macros): Define
1230 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1232 * doc/invoke.texi: Document -mlong-double-128.
1234 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1236 PR rtl-optimization/60024
1237 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1239 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1241 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1243 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1245 PR rtl-optimization/57662
1246 * sel-sched.c (code_motion_path_driver): Do not mark already not
1247 existing blocks in the visiting bitmap.
1249 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1251 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1252 on the insn being emitted.
1254 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1255 Will Deacon <will.deacon@arm.com>
1257 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1259 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1261 * config/arm/arm-tables.opt: Regenerate.
1263 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1265 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1266 for vector types other than V16QImode.
1267 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1268 define_expand, and call altivec_expand_vec_perm_le when producing
1269 code with little endian element order.
1270 (*altivec_vperm_<mode>_internal): New insn having previous
1271 behavior of altivec_vperm_<mode>.
1272 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1273 altivec_expand_vec_perm_le when producing code with little endian
1275 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1276 behavior of altivec_vperm_<mode>_uns.
1278 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1280 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1281 (altivec_vsumsws): Add handling for -maltivec=be with a little
1283 (altivec_vsumsws_direct): New.
1284 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1285 gen_altivec_vsumsws.
1287 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1289 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1290 vtable_pointer_value_to_binfo): New functions.
1291 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1292 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1294 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1296 * config/nios2/nios2.md (load_got_register): Initialize GOT
1297 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1298 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1300 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1302 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1303 preserverd by passthrough, do not propagate the type.
1305 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1307 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1308 (mips_atomic_assign_expand_fenv): New function.
1309 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1311 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1313 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1314 (__builtin_mips_set_fcsr): Likewise.
1315 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1316 MIPS_USI_FTYPE_VOID.
1317 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1318 (mips16_expand_set_fcsr): Likewise.
1319 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1320 (mips16_set_fcsr_stub): Likewise.
1321 (mips16_get_fcsr_one_only_stub): New class.
1322 (mips16_set_fcsr_one_only_stub): Likewise.
1323 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1324 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1325 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1326 (hard_float): New availability predicate.
1327 (mips_builtins): Add get_fcsr and set_fcsr.
1328 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1329 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1330 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1331 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1332 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1335 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1337 * config/mips/mips.c (mips_one_only_stub): New class.
1338 (mips_need_mips16_rdhwr_p): Replace with...
1339 (mips16_rdhwr_stub): ...this new variable.
1340 (mips16_stub_call_address): New function.
1341 (mips16_rdhwr_one_only_stub): New class.
1342 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1343 (mips_output_mips16_rdhwr): Delete.
1344 (mips_finish_stub): New function.
1345 (mips_code_end): Use it to handle rdhwr stubs.
1347 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1350 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1351 when calculating size of integer atomic types.
1353 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1355 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1357 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1359 PR tree-optimization/60003
1360 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1361 * profile.c (branch_prob): Use gimple_call_builtin_p
1362 to check for BUILT_IN_SETJMP_RECEIVER.
1363 * tree-inline.c (copy_bb): Call notice_special_calls.
1365 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1368 * lra-constraints.c (process_alt_operands): Update reload_sum only
1371 2014-01-31 Richard Henderson <rth@redhat.com>
1374 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1375 until after else_eh is processed.
1377 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1379 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1380 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1381 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1382 in smmintrin.h, remove them.
1383 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1384 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1385 * config/i386/i386.md (ROUND_SAE): Fix value.
1386 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1387 (const48_operand): New.
1388 * config/i386/subst.md (round), (round_expand): Use
1389 const_4_or_8_to_11_operand.
1390 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1392 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1394 * config/i386/constraints.md (Yk): Swap meaning with k.
1395 * config/i386/i386.md (movhi_internal): Change Yk to k.
1396 (movqi_internal): Ditto.
1397 (*k<logic><mode>): Ditto.
1400 (kandn<mode>): Ditto.
1401 (*<code>hi_1): Ditto.
1402 (*<code>qi_1): Ditto.
1403 (kxnor<mode>): Ditto.
1404 (kortestzhi): Ditto.
1405 (kortestchi): Ditto.
1407 (*one_cmplhi2_1): Ditto.
1408 (*one_cmplqi2_1): Ditto.
1409 * config/i386/sse.md (): Change k to Yk.
1410 (avx512f_load<mode>_mask): Ditto.
1411 (avx512f_blendm<mode>): Ditto.
1412 (avx512f_store<mode>_mask): Ditto.
1413 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1414 (avx512f_storedqu<mode>_mask): Ditto.
1415 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1417 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1418 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1419 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1420 (avx512f_maskcmp<mode>3): Ditto.
1421 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1422 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1423 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1424 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1425 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1426 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1427 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1428 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1429 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1430 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1431 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1432 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1433 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1434 (vec_extract_lo_<mode>_maskm): Ditto.
1435 (vec_extract_hi_<mode>_maskm): Ditto.
1436 (avx512f_vternlog<mode>_mask): Ditto.
1437 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1438 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1439 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1440 (avx512f_<code>v8div16qi2_mask): Ditto.
1441 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1442 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1443 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1444 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1445 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1446 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1447 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1448 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1449 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1450 (avx512cd_maskb_vec_dupv8di): Ditto.
1451 (avx512cd_maskw_vec_dupv16si): Ditto.
1452 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1453 (avx512f_vpermi2var<mode>3_mask): Ditto.
1454 (avx512f_vpermi2var<mode>3_mask): Ditto.
1455 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1456 (*avx512f_gathersi<mode>): Ditto.
1457 (*avx512f_gathersi<mode>_2): Ditto.
1458 (*avx512f_gatherdi<mode>): Ditto.
1459 (*avx512f_gatherdi<mode>_2): Ditto.
1460 (*avx512f_scattersi<mode>): Ditto.
1461 (*avx512f_scatterdi<mode>): Ditto.
1462 (avx512f_compress<mode>_mask): Ditto.
1463 (avx512f_compressstore<mode>_mask): Ditto.
1464 (avx512f_expand<mode>_mask): Ditto.
1465 * config/i386/subst.md (mask): Change k to Yk.
1466 (mask_scalar_merge): Ditto.
1469 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1471 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1473 2014-01-31 Richard Biener <rguenther@suse.de>
1476 * builtins.c (fold_builtin_memory_op): Make sure to not
1477 use a floating-point mode or a boolean or enumeral type for
1480 2014-01-30 DJ Delorie <dj@redhat.com>
1482 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1483 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1484 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1485 whenever main() has an epilogue.
1487 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1489 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1490 unused variable "field".
1491 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1492 (vsx_mergeh_<mode>): Likewise.
1493 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1494 (altivec_vmrghh): Likewise.
1495 (altivec_vmrghw): Likewise.
1496 (altivec_vmrglb): Likewise.
1497 (altivec_vmrglh): Likewise.
1498 (altivec_vmrglw): Likewise.
1499 (altivec_vspltb): Add missing uses.
1500 (altivec_vsplth): Likewise.
1501 (altivec_vspltw): Likewise.
1502 (altivec_vspltsf): Likewise.
1504 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1507 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1508 frame related instructions.
1510 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1512 PR rtl-optimization/59959
1513 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1514 any reload of register whose subreg is invalid.
1516 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1518 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1519 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1520 Add missing return type - void.
1522 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1524 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1525 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1526 remove element index adjustment for endian (now handled in vsx.md
1528 (altivec_expand_vec_perm_const): Use
1529 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1530 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1531 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1532 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1533 define_expand and a new define_insn *altivec_vspltb_internal;
1534 adjust for -maltivec=be on a little endian target.
1535 (altivec_vspltb_direct): New.
1536 (altivec_vsplth): Divide into a define_expand and a new
1537 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1538 little endian target.
1539 (altivec_vsplth_direct): New.
1540 (altivec_vspltw): Divide into a define_expand and a new
1541 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1542 little endian target.
1543 (altivec_vspltw_direct): New.
1544 (altivec_vspltsf): Divide into a define_expand and a new
1545 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1546 a little endian target.
1548 2014-01-30 Richard Biener <rguenther@suse.de>
1550 PR tree-optimization/59993
1551 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1552 can propagate form the earlier stmt and avoid the transform
1553 when the intermediate result is needed.
1555 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1557 * README.Portability: Fix typo.
1559 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1561 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1562 comparison_operator with ordered_comparison_operator.
1564 2014-01-30 Nick Clifton <nickc@redhat.com>
1566 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1567 Rename to mn10300_store_multiple_regs.
1568 * config/mn10300/mn10300.c: Likewise.
1569 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1570 store_multiple_regs.
1571 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1572 Call mn10300_store_multiple_regs.
1574 2014-01-30 Nick Clifton <nickc@redhat.com>
1575 DJ Delorie <dj@redhat.com>
1577 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1578 %fp 2 to keep registers after it properly word-aligned.
1579 (rl78_alloc_physical_registers_umul): Handle the case where both
1580 input operands are the same.
1582 2014-01-30 Richard Biener <rguenther@suse.de>
1584 PR tree-optimization/59903
1585 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1588 2014-01-30 Jason Merrill <jason@redhat.com>
1591 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1594 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1596 2014-01-30 Richard Biener <rguenther@suse.de>
1598 PR tree-optimization/59951
1599 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1601 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1604 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1605 SFmode to DFmode case.
1607 2014-01-29 DJ Delorie <dj@redhat.com>
1609 * config/msp430/msp430.opt (-minrt): New.
1610 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1612 (ENDFILE_SPEC): Likewise.
1614 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1616 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1617 (estimate_function_body_sizes): Use it.
1619 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1622 * dwarf2out.c (is_cxx_auto): New.
1623 (is_base_type): Use it.
1624 (gen_type_die_with_usage): Likewise.
1626 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1628 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1629 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1630 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1631 -maltivec=be with LE targets.
1632 (vsx_mergeh_<mode>): Likewise.
1633 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1634 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1635 (altivec_vmrghb): Replace with define_expand and new
1636 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1637 (altivec_vmrghb_direct): New define_insn.
1638 (altivec_vmrghh): Replace with define_expand and new
1639 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1640 (altivec_vmrghh_direct): New define_insn.
1641 (altivec_vmrghw): Replace with define_expand and new
1642 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1643 (altivec_vmrghw_direct): New define_insn.
1644 (*altivec_vmrghsf): Adjust for endianness.
1645 (altivec_vmrglb): Replace with define_expand and new
1646 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1647 (altivec_vmrglb_direct): New define_insn.
1648 (altivec_vmrglh): Replace with define_expand and new
1649 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1650 (altivec_vmrglh_direct): New define_insn.
1651 (altivec_vmrglw): Replace with define_expand and new
1652 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1653 (altivec_vmrglw_direct): New define_insn.
1654 (*altivec_vmrglsf): Adjust for endianness.
1655 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1656 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1657 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1658 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1659 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1660 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1661 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1662 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1664 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1666 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1667 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1670 2014-01-29 Richard Biener <rguenther@suse.de>
1672 PR tree-optimization/58742
1673 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1674 associate_pointerplus_align.
1675 (associate_pointerplus_diff): New function.
1676 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1677 and associate_pointerplus_diff.
1679 2014-01-29 Richard Biener <rguenther@suse.de>
1681 * lto-streamer.h (LTO_major_version): Bump to 3.
1682 (LTO_minor_version): Reset to 0.
1684 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1686 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1687 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1688 (arm_file_start): Generate correct asm header for armv7ve.
1689 * config/arm/bpabi.h: Add multilib support for armv7ve.
1690 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1691 and cortex-a15 to armv7ve.
1692 * config/arm/t-aprofile: Add multilib support for armv7ve.
1693 * doc/invoke.texi: Document -march=armv7ve.
1695 2014-01-29 Richard Biener <rguenther@suse.de>
1697 PR tree-optimization/58742
1698 * tree-ssa-forwprop.c (associate_plusminus): Return true
1699 if we changed sth, defer EH cleanup to ...
1700 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1701 (simplify_mult): New function.
1703 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1706 PR tree-optimization/59920
1707 * tree.c (build_common_builtin_nodes): Remove
1708 __builtin_setjmp_dispatcher initialization.
1709 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1710 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1711 instead of gsi_after_labels + manually skipping debug stmts.
1712 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1713 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1714 * tree-inline.c (copy_edges_for_bb): Remove
1715 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1716 argument. Ignore computed_goto_p stmts. Don't call
1717 make_abnormal_goto_edges. If a call might need abnormal edges
1718 for non-local gotos, see if it already has an edge to
1719 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1720 with true argument, don't do anything then, otherwise add
1721 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1722 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1724 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1725 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1726 (lower_stmt): Don't set data->calls_builtin_setjmp.
1727 (lower_builtin_setjmp): Adjust comment.
1728 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1729 * tree-cfg.c (found_computed_goto): Remove.
1730 (factor_computed_gotos): Remove.
1731 (make_goto_expr_edges): Return bool, true for computed gotos.
1732 Don't call make_abnormal_goto_edges.
1733 (build_gimple_cfg): Don't set found_computed_goto, don't call
1734 factor_computed_gotos.
1735 (computed_goto_p): No longer static.
1736 (make_blocks): Don't set found_computed_goto.
1737 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1738 (make_edges): If make_goto_expr_edges returns true, push bb
1739 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1740 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1741 vector. Record mapping between bbs and OpenMP regions if there
1742 are any, adjust make_gimple_omp_edges caller. Call
1743 handle_abnormal_edges.
1744 (make_abnormal_goto_edges): Remove.
1745 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1746 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1747 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1748 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1749 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1750 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1751 filling *region also set *region_idx to (*region)->entry->index.
1754 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1755 For REGs set ORIGINAL_REGNO.
1757 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1759 * doc/md.texi: Mention that a target shouldn't implement
1760 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1763 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1765 PR tree-optimization/59594
1766 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1767 a copy of the datarefs vector rather than the vector itself.
1769 2014-01-28 Jason Merrill <jason@redhat.com>
1772 * dwarf2out.c (auto_die): New static.
1773 (gen_type_die_with_usage): Handle C++1y 'auto'.
1774 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1777 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1780 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1781 (SPEC_X32): Likewise.
1782 (SPEC_64): Likewise.
1783 * config/i386/i386.c (ix86_option_override_internal): Turn off
1784 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1786 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1787 * config/i386/i386.h (TARGET_16BIT): New macro.
1788 (TARGET_16BIT_P): Likewise.
1789 * config/i386/i386.opt: Add m16.
1790 * doc/invoke.texi: Document -m16.
1792 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1794 PR preprocessor/59935
1795 * input.c (location_get_source_line): Bail out on when line number
1796 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1798 2014-01-28 Richard Biener <rguenther@suse.de>
1800 PR tree-optimization/58742
1801 * tree-ssa-forwprop.c (associate_plusminus): Handle
1802 pointer subtraction of the form (T)(P + A) - (T)P.
1804 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1806 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1809 2014-01-28 Richard Biener <rguenther@suse.de>
1812 2014-01-28 Richard Biener <rguenther@suse.de>
1814 PR rtl-optimization/45364
1815 PR rtl-optimization/59890
1816 * var-tracking.c (local_get_addr_clear_given_value): Handle
1817 already cleared slot.
1818 (val_reset): Handle not allocated local_get_addr_cache.
1819 (vt_find_locations): Use post-order on the inverted CFG.
1821 2014-01-28 Richard Biener <rguenther@suse.de>
1823 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1825 2014-01-28 Richard Biener <rguenther@suse.de>
1827 PR rtl-optimization/45364
1828 PR rtl-optimization/59890
1829 * var-tracking.c (local_get_addr_clear_given_value): Handle
1830 already cleared slot.
1831 (val_reset): Handle not allocated local_get_addr_cache.
1832 (vt_find_locations): Use post-order on the inverted CFG.
1834 2014-01-28 Alan Modra <amodra@gmail.com>
1836 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1837 * configure.ac <recursive call for build != host>: Define
1838 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1839 and LD_FOR_BUILD too.
1840 * configure: Regenerate.
1842 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1844 * config/i386/i386.c (get_builtin_code_for_version): Separate
1845 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1846 Broadwell from Haswell.
1848 2014-01-27 Steve Ellcey <sellcey@mips.com>
1850 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1851 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1852 * config/mips/mips.c (mips_option_override): Change setting
1854 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1855 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1856 Change from Mask to Var.
1858 2014-01-27 Jeff Law <law@redhat.com>
1860 * ipa-inline.c (inline_small_functions): Fix typo.
1862 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1864 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1865 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1866 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1867 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1868 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1869 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1870 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1871 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1872 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1873 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1874 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1875 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1876 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1877 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1878 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1879 (_mm512_storeu_epi64): Ditto.
1880 (_mm512_cmpge_epi32_mask): Ditto.
1881 (_mm512_cmpge_epu32_mask): Ditto.
1882 (_mm512_cmpge_epi64_mask): Ditto.
1883 (_mm512_cmpge_epu64_mask): Ditto.
1884 (_mm512_cmple_epi32_mask): Ditto.
1885 (_mm512_cmple_epu32_mask): Ditto.
1886 (_mm512_cmple_epi64_mask): Ditto.
1887 (_mm512_cmple_epu64_mask): Ditto.
1888 (_mm512_cmplt_epi32_mask): Ditto.
1889 (_mm512_cmplt_epu32_mask): Ditto.
1890 (_mm512_cmplt_epi64_mask): Ditto.
1891 (_mm512_cmplt_epu64_mask): Ditto.
1892 (_mm512_cmpneq_epi32_mask): Ditto.
1893 (_mm512_cmpneq_epu32_mask): Ditto.
1894 (_mm512_cmpneq_epi64_mask): Ditto.
1895 (_mm512_cmpneq_epu64_mask): Ditto.
1896 (_mm512_expand_pd): Ditto.
1897 (_mm512_expand_ps): Ditto.
1898 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1899 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1900 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1901 * config/i386/i386.c (ix86_builtins): Add
1902 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1903 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1904 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1905 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1906 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1907 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1908 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1909 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1910 IX86_BUILTIN_PMOVUSQW512_MEM.
1911 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1912 __builtin_ia32_pmovsqd512mem_mask,
1913 __builtin_ia32_pmovqd512mem_mask,
1914 __builtin_ia32_pmovusqw512mem_mask,
1915 __builtin_ia32_pmovsqw512mem_mask,
1916 __builtin_ia32_pmovqw512mem_mask,
1917 __builtin_ia32_pmovusdw512mem_mask,
1918 __builtin_ia32_pmovsdw512mem_mask,
1919 __builtin_ia32_pmovdw512mem_mask,
1920 __builtin_ia32_pmovqb512mem_mask,
1921 __builtin_ia32_pmovusqb512mem_mask,
1922 __builtin_ia32_pmovsqb512mem_mask,
1923 __builtin_ia32_pmovusdb512mem_mask,
1924 __builtin_ia32_pmovsdb512mem_mask,
1925 __builtin_ia32_pmovdb512mem_mask.
1926 (bdesc_args): Add __builtin_ia32_expanddf512,
1927 __builtin_ia32_expandsf512.
1928 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1929 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1930 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1931 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1932 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1933 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1934 (avx512f_<code>v8div16qi2_mask_store): This.
1935 (avx512f_expand<mode>): New.
1937 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1939 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1941 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1942 (_mm512_prefetch_i32scatter_pd): Ditto.
1943 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1944 (_mm512_prefetch_i64scatter_pd): Ditto.
1945 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1946 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1947 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1948 (_mm512_prefetch_i32scatter_ps): Ditto.
1949 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1950 (_mm512_prefetch_i64scatter_ps): Ditto.
1951 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1952 * config/i386/i386-builtin-types.def: Define
1953 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1954 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1955 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1956 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1957 IX86_BUILTIN_SCATTERPFQPD.
1958 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1959 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1960 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1961 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1962 __builtin_ia32_scatterpfqps.
1963 (ix86_expand_builtin): Expand new built-ins.
1964 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1965 fix memory access data type.
1966 (*avx512pf_gatherpf<mode>_mask): Ditto.
1967 (*avx512pf_gatherpf<mode>): Ditto.
1968 (avx512pf_scatterpf<mode>): Ditto.
1969 (*avx512pf_scatterpf<mode>_mask): Ditto.
1970 (*avx512pf_scatterpf<mode>): Ditto.
1971 (GATHER_SCATTER_SF_MEM_MODE): New.
1972 (avx512pf_gatherpf<mode>df): Ditto.
1973 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1974 (*avx512pf_scatterpf<mode>df): Ditto.
1976 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1979 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1980 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1983 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1985 * common/config/arm/arm-common.c
1986 (arm_rewrite_mcpu): Handle multiple names.
1988 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1990 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1992 * gimple-builder.h (create_gimple_tmp): Delete.
1994 2014-01-27 Christian Bruel <christian.bruel@st.com>
1996 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1999 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2001 * config/pa/pa.md (call): Generate indirect long calls to non-local
2002 functions when outputing 32-bit code.
2003 (call_value): Likewise except for special call to buggy powf function.
2005 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2006 portable runtime and PIC indirect calls.
2007 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2008 and PIC call sequences. Use ldo instead of blr to set return register
2009 in PIC call sequence.
2011 2014-01-25 Walter Lee <walt@tilera.com>
2013 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2014 avoid clobbering a live register.
2016 2014-01-25 Walter Lee <walt@tilera.com>
2018 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2019 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2020 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2021 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2023 2014-01-25 Walter Lee <walt@tilera.com>
2025 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2026 arguments on even registers.
2027 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2029 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2030 (BIGGEST_ALIGNMENT): Ditto.
2031 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2033 2014-01-25 Walter Lee <walt@tilera.com>
2035 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2036 insns before bundling.
2037 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2039 2014-01-25 Walter Lee <walt@tilera.com>
2041 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2042 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2043 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2045 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2047 * config/mips/constraints.md (kl): Delete.
2048 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2049 define expands, using...
2050 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2051 instructions for MIPS16.
2052 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2053 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2055 2014-01-25 Walter Lee <walt@tilera.com>
2057 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2061 2014-01-25 Walter Lee <walt@tilera.com>
2063 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2064 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2066 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2068 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2071 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2073 * print-rtl.c (in_call_function_usage): New var.
2074 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2075 EXPR_LIST mode as mode and not as reg note name.
2078 * cfgloopmanip.c (copy_loop_info): If
2079 loop->warned_aggressive_loop_optimizations, make sure
2080 the flag is set in target loop too.
2082 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2084 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2086 * builtins.def: Likewise.
2087 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2088 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2089 * ira.c (ira_setup_eliminable_regset): Likewise.
2090 * omp-low.c (gate_expand_omp): Likewise.
2091 (execute_lower_omp): Likewise.
2092 (diagnose_sb_0): Likewise.
2093 (gate_diagnose_omp_blocks): Likewise.
2094 (simd_clone_clauses_extract): Likewise.
2097 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2099 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2100 correction for little endian...
2101 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2104 2014-01-24 Jeff Law <law@redhat.com>
2106 PR tree-optimization/59919
2107 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2108 for non-returning calls.
2110 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2112 * common/config/aarch64/aarch64-common.c
2113 (aarch64_rewrite_mcpu): Handle multiple names.
2114 * config/aarch64/aarch64.h
2115 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2117 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2119 * input.c (add_file_to_cache_tab): Handle the case where fopen
2122 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2125 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2126 from push operand if code of push isn't PRE_DEC.
2128 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2131 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2132 -mquad-memory-atomic. Update -mquad-memory documentation to say
2133 it is only used for non-atomic loads/stores.
2135 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2136 -mquad-memory or -mquad-memory-atomic switches.
2138 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2139 -mquad-memory-atomic to ISA 2.07 support.
2141 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2142 to separate support of normal quad word memory operations (ldq, stq)
2143 from the atomic quad word memory operations.
2145 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2146 support to separate non-atomic quad word operations from atomic
2147 quad word operations. Disable non-atomic quad word operations in
2148 little endian mode so that we don't have to swap words after the
2149 load and before the store.
2150 (quad_load_store_p): Add comment about atomic quad word support.
2151 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2152 options printed with -mdebug=reg.
2154 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2155 -mquad-memory-atomic as the test for whether we have quad word
2156 atomic instructions.
2157 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2158 or -mp8-vector are used, allow byte/half-word atomic operations.
2160 * config/rs6000/sync.md (load_lockedti): Insure that the address
2161 is a proper indexed or indirect address for the lqarx instruction.
2162 On little endian systems, swap the hi/lo registers after the lqarx
2164 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2165 insure the address is valid for the lqarx instruction.
2166 (store_conditionalti): Insure that the address is a proper indexed
2167 or indirect address for the stqcrx. instruction. On little endian
2168 systems, swap the hi/lo registers before doing the stqcrx.
2170 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2171 insure the address is valid for the stqcrx. instruction.
2173 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2174 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2175 type of quad memory support is available.
2177 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2180 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2181 there is a danger of looping.
2183 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2185 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2186 force flag_ira_loop_pressure if set via command line.
2188 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2190 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2191 (ashr_simd): New builtin handling DI mode.
2192 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2193 (aarch64_sshr_simddi): New match pattern.
2194 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2195 (vshrd_n_s64): Likewise.
2196 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2198 2014-01-23 Nick Clifton <nickc@redhat.com>
2200 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2201 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2202 favour of mcu specific scripts.
2203 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2206 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2207 Alex Velenko <Alex.Velenko@arm.com>
2209 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2210 (vaddv_s16): Likewise.
2211 (vaddv_s32): Likewise.
2212 (vaddv_u8): Likewise.
2213 (vaddv_u16): Likewise.
2214 (vaddv_u32): Likewise.
2215 (vaddvq_s8): Likewise.
2216 (vaddvq_s16): Likewise.
2217 (vaddvq_s32): Likewise.
2218 (vaddvq_s64): Likewise.
2219 (vaddvq_u8): Likewise.
2220 (vaddvq_u16): Likewise.
2221 (vaddvq_u32): Likewise.
2222 (vaddvq_u64): Likewise.
2223 (vaddv_f32): Likewise.
2224 (vaddvq_f32): Likewise.
2225 (vaddvq_f64): Likewise.
2226 (vmaxv_f32): Likewise.
2227 (vmaxv_s8): Likewise.
2228 (vmaxv_s16): Likewise.
2229 (vmaxv_s32): Likewise.
2230 (vmaxv_u8): Likewise.
2231 (vmaxv_u16): Likewise.
2232 (vmaxv_u32): Likewise.
2233 (vmaxvq_f32): Likewise.
2234 (vmaxvq_f64): Likewise.
2235 (vmaxvq_s8): Likewise.
2236 (vmaxvq_s16): Likewise.
2237 (vmaxvq_s32): Likewise.
2238 (vmaxvq_u8): Likewise.
2239 (vmaxvq_u16): Likewise.
2240 (vmaxvq_u32): Likewise.
2241 (vmaxnmv_f32): Likewise.
2242 (vmaxnmvq_f32): Likewise.
2243 (vmaxnmvq_f64): Likewise.
2244 (vminv_f32): Likewise.
2245 (vminv_s8): Likewise.
2246 (vminv_s16): Likewise.
2247 (vminv_s32): Likewise.
2248 (vminv_u8): Likewise.
2249 (vminv_u16): Likewise.
2250 (vminv_u32): Likewise.
2251 (vminvq_f32): Likewise.
2252 (vminvq_f64): Likewise.
2253 (vminvq_s8): Likewise.
2254 (vminvq_s16): Likewise.
2255 (vminvq_s32): Likewise.
2256 (vminvq_u8): Likewise.
2257 (vminvq_u16): Likewise.
2258 (vminvq_u32): Likewise.
2259 (vminnmv_f32): Likewise.
2260 (vminnmvq_f32): Likewise.
2261 (vminnmvq_f64): Likewise.
2263 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2265 * config/aarch64/aarch64-simd.md
2266 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2267 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2268 (*aarch64_mul3_elt<mode>): Likewise.
2269 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2270 (*aarch64_mul3_elt_to_64v2df): Likewise.
2271 (*aarch64_mla_elt<mode>): Likewise.
2272 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2273 (*aarch64_mls_elt<mode>): Likewise.
2274 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2275 (*aarch64_fma4_elt<mode>): Likewise.
2276 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2277 (*aarch64_fma4_elt_to_64v2df): Likewise.
2278 (*aarch64_fnma4_elt<mode>): Likewise.
2279 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2280 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2281 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2282 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2283 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2284 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2285 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2286 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2287 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2289 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2291 * config/aarch64/aarch64-simd.md
2292 (aarch64_be_checked_get_lane<mode>): New define_expand.
2293 * config/aarch64/aarch64-simd-builtins.def
2294 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2295 New builtin definition.
2296 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2297 Use new safe be builtin.
2299 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2301 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2303 (aarch64_be_st1<mode>): Likewise.
2304 (aarch_ld1<VALL:mode>): Define_expand modified.
2305 (aarch_st1<VALL:mode>): Likewise.
2306 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2307 (UNSPEC_ST1): Likewise.
2309 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2311 * config/microblaze/microblaze.md: Add trap insn and attribute
2313 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2315 PR preprocessor/58580
2316 * input.h (location_get_source_line): Take an additional line_size
2318 (void diagnostics_file_cache_fini): Declare new function.
2319 * input.c (struct fcache): New type.
2320 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2321 New static constants.
2322 (diagnostic_file_cache_init, total_lines_num)
2323 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2324 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2325 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2326 (get_next_line, read_next_line, goto_next_line, read_line_num):
2327 New static function definitions.
2328 (diagnostic_file_cache_fini): New function.
2329 (location_get_source_line): Take an additional output line_len
2330 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2332 * diagnostic.c (diagnostic_finish): Call
2333 diagnostic_file_cache_fini.
2334 (adjust_line): Take an additional input parameter for the length
2335 of the line, rather than calculating it with strlen.
2336 (diagnostic_show_locus): Adjust the use of
2337 location_get_source_line and adjust_line with respect to their new
2338 signature. While displaying a line now, do not stop at the first
2339 null byte. Rather, display the zero byte as a space and keep
2340 going until we reach the size of the line.
2341 * Makefile.in: Add vec.o to OBJS-libcommon
2343 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2344 Ilya Tocar <ilya.tocar@intel.com>
2346 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2347 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2348 (__builtin_ia32_kmov16): Ditto.
2349 * config/i386/i386.md (UNSPEC_KMOV): New.
2352 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2354 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2355 (_mm512_storeu_si512): Ditto.
2357 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2360 * rtl.h (get_referenced_operands): Declare.
2361 * recog.c (get_referenced_operands): New function.
2362 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2363 operands have been referenced when recording LO_SUM references.
2365 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2367 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2369 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2371 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2372 Enable for generic and recent AMD targets.
2374 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2376 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2377 ARG_SIZE note when adjustment was eliminated.
2379 2014-01-22 Jeff Law <law@redhat.com>
2381 PR tree-optimization/59597
2382 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2383 in file. Accept new argument REGISTERING and use it to modify
2384 dump output appropriately.
2385 (register_jump_thread): Corresponding changes.
2386 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2387 thread paths involving joiner blocks. Add code to dump cancelled
2388 jump threading paths.
2390 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2392 PR rtl-optimization/59477
2393 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2394 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2396 2014-01-22 Tom Tromey <tromey@redhat.com>
2398 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2400 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2402 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2404 PR rtl-optimization/59896
2405 * lra-constraints.c (process_alt_operands): Check unused note for
2406 matched operands of insn with no output reloads.
2408 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2410 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2411 (mips_move_from_gpr_cost): Likewise.
2413 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2415 PR rtl-optimization/59858
2416 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2417 ira_class_hard_regs_num.
2418 (process_alt_operands): Increase reject for dying matched operand.
2420 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2423 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2424 smaller than size, perform several stores or loads and stores
2425 at dst + count - size to store or copy all of size bytes, rather
2426 than just last modesize bytes.
2428 2014-01-20 DJ Delorie <dj@redhat.com>
2430 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2431 that CLOBBERs are REGs before propogating their values.
2433 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2436 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2437 (cgraph_inline_failed_type): New function.
2438 * cgraph.h (DEFCIFCODE): Add type.
2439 (cgraph_inline_failed_type_t): New enum.
2440 (cgraph_inline_failed_type): New prototype.
2441 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2442 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2443 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2444 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2445 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2446 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2447 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2448 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2449 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2450 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2451 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2452 OPTIMIZATION_MISMATCH.
2453 * tree-inline.c (expand_call_inline): Emit errors during
2454 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2456 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2459 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2460 mode attribute in insn output.
2462 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2464 * output.h (output_constant): Delete.
2465 * varasm.c (output_constant): Make private.
2467 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2469 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2471 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2474 * tree.h (fold_builtin_strcat): New prototype.
2475 * builtins.c (fold_builtin_strcat): No longer static. Add len
2476 argument, if non-NULL, don't call c_strlen. Optimize
2477 directly into __builtin_memcpy instead of __builtin_strcpy.
2478 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2479 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2481 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2483 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2484 for SImode_address_operand operands, having only a REG argument.
2486 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2488 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2489 loader name using mbig-endian.
2490 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2492 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2494 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2498 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2500 * config/aarch64/aarch64-protos.h
2501 (aarch64_cannot_change_mode_class_ptr): Declare.
2502 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2503 aarch64_cannot_change_mode_class_ptr): New.
2504 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2505 backend hook aarch64_cannot_change_mode_class.
2507 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2509 * common/config/aarch64/aarch64-common.c
2510 (aarch64_handle_option): Don't handle any option order logic here.
2511 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2512 selected_cpu, warn on architecture version mismatch.
2513 (aarch64_override_options): Fix parsing order for option strings.
2515 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2516 Iain Sandoe <iain@codesourcery.com>
2519 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2520 warning. Amend comment to reflect current functionality.
2522 2014-01-20 Richard Biener <rguenther@suse.de>
2525 * builtins.c (fold_builtin_strcat): Remove case better handled
2526 by tree-ssa-strlen.c.
2528 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2530 * config/aarch64/aarch64.opt
2531 (mcpu, march, mtune): Make case-insensitive.
2533 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2536 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2537 if operands[1] is a REG or ZERO_EXTEND of a REG.
2539 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2541 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2543 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2545 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2546 long non-pic millicode calls.
2548 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2550 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2552 2014-01-19 Kito Cheng <kito@0xlab.org>
2554 * builtins.c (expand_movstr): Check movstr expand done or fail.
2556 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2557 H.J. Lu <hongjiu.lu@intel.com>
2560 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2561 to DImode for zero-extended addresses.
2563 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2565 PR rtl-optimization/57763
2566 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2567 on the new indirect jump_insn and increment LABEL_NUSES (label).
2569 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2573 * config.gcc (x86_archs): New variable.
2574 (x86_64_archs): Likewise.
2575 (x86_cpus): Likewise.
2576 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2577 --with-arch/--with-cpu= options.
2578 Support --with-arch=/--with-cpu={nehalem,westmere,
2579 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2581 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2583 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2584 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2586 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2588 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2590 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2593 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2594 clear cpp_get_options (parse_in)->warn_unused_macros for
2595 ix86_target_macros_internal with cpp_define.
2597 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2599 * jump.c (delete_related_insns): Keep (use (insn))s.
2600 * reorg.c (redundant_insn): Check for barriers too.
2602 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2604 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2606 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2608 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2609 call to $$dyncall when TARGET_LONG_CALLS is true.
2611 2014-01-17 Jeff Law <law@redhat.com>
2613 * ree.c (combine_set_extension): Temporarily disable test for
2614 changing number of hard registers.
2616 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2619 * ipa-inline-analysis.c (inline_free_summary):
2620 Do not free summary of aliases.
2622 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2625 * gimplify.c (gimplify_expr): Use create_tmp_var
2626 instead of create_tmp_var_raw. If cond doesn't have
2627 integral type, don't add the IFN_ANNOTATE builtin at all.
2629 2014-01-17 Martin Jambor <mjambor@suse.cz>
2632 * ipa-cp.c (prev_edge_clone): New variable.
2633 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2634 Also resize prev_edge_clone vector.
2635 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2636 (ipcp_edge_removal_hook): New function.
2637 (ipcp_driver): Register ipcp_edge_removal_hook.
2639 2014-01-17 Andrew Pinski <apinski@cavium.com>
2640 Steve Ellcey <sellcey@mips.com>
2643 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2646 2014-01-17 Jeff Law <law@redhat.com>
2649 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2650 so that pass_ccp runs first.
2652 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2654 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2655 (ix86_adjust_cost): Use !TARGET_XXX.
2656 (do_reorder_for_imul): Likewise.
2657 (swap_top_of_ready_list): Likewise.
2658 (ix86_sched_reorder): Likewise.
2660 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2662 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2663 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2664 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2665 (intel_memset): New. Duplicate slm_memset.
2666 (intel_cost): New. Duplicate slm_cost.
2667 (m_INTEL): New macro.
2668 (processor_target_table): Add "intel".
2669 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2670 with PROCESSOR_INTEL for "intel".
2671 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2672 PROCESSOR_SILVERMONT.
2673 (ix86_issue_rate): Likewise.
2674 (ix86_adjust_cost): Likewise.
2675 (ia32_multipass_dfa_lookahead): Likewise.
2676 (swap_top_of_ready_list): Likewise.
2677 (ix86_sched_reorder): Likewise.
2678 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2679 instead of TARGET_OPT_AGU.
2680 * config/i386/i386.h (TARGET_INTEL): New.
2681 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2682 (processor_type): Add PROCESSOR_INTEL.
2683 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2684 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2686 2014-01-17 Marek Polacek <polacek@redhat.com>
2689 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2692 2014-01-17 Richard Biener <rguenther@suse.de>
2694 PR tree-optimization/46590
2695 * opts.c (default_options_table): Add entries for
2696 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2697 all enabled at -O1 but not for -Og.
2698 * common.opt (fbranch-count-reg): Remove Init(1).
2699 (fmove-loop-invariants): Likewise.
2700 (ftree-pta): Likewise.
2702 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2704 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2705 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2706 decls to at least the GCC 4.8 used alignments.
2709 * tree-nested.c (convert_nonlocal_reference_stmt,
2710 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2711 of GIMPLE_BIND stmts, adjust associated decls.
2713 2014-01-17 Richard Biener <rguenther@suse.de>
2715 PR tree-optimization/46590
2716 * vec.h (vec<>::bseach): New member function implementing
2717 binary search according to C89 bsearch.
2718 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2719 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2720 bitmap pointer again. Make accesses_in_loop a flat array.
2721 (mem_ref_obstack): New global.
2722 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2723 (mark_ref_stored): Likewise.
2724 (ref_indep_loop_p_2): Likewise.
2725 (set_ref_stored_in_loop): New helper function.
2726 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2727 (memref_free): Adjust.
2728 (record_mem_ref_loc): Simplify.
2729 (gather_mem_refs_stmt): Adjust.
2730 (sort_locs_in_loop_postorder_cmp): New function.
2731 (analyze_memory_references): Sort accesses_in_loop after
2732 loop postorder number.
2733 (find_ref_loc_in_loop_cmp): New function.
2734 (for_all_locs_in_loop): Find relevant cluster of locs in
2735 accesses_in_loop and iterate without recursion.
2736 (execute_sm): Avoid uninit warning.
2737 (struct ref_always_accessed): Simplify.
2738 (ref_always_accessed::operator ()): Likewise.
2739 (ref_always_accessed_p): Likewise.
2740 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2741 loop postorder numbers here.
2742 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2745 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2748 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2749 on decls for which assemble_alias has been called.
2751 2014-01-17 Nick Clifton <nickc@redhat.com>
2753 * config/msp430/msp430.opt: (mcpu): New option.
2754 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2755 (msp430_option_override): Parse target_cpu. If the MCU name
2756 matches a generic string, clear target_mcu.
2757 (msp430_attr): Allow numeric interrupt values up to 63.
2758 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2759 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2761 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2763 * config/msp430/msp430.md (popm): Use %J rather than %I.
2764 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2765 (addhi_cy_i): Use immediate_operand for operand 2.
2766 * doc/invoke.texi: Document -mcpu option.
2768 2014-01-17 Richard Biener <rguenther@suse.de>
2770 PR rtl-optimization/38518
2771 * df.h (df_analyze_loop): Declare.
2772 * df-core.c: Include cfgloop.h.
2773 (df_analyze_1): Split out main part of df_analyze.
2774 (df_analyze): Adjust.
2775 (loop_inverted_post_order_compute): New function.
2776 (loop_post_order_compute): Likewise.
2777 (df_analyze_loop): New function avoiding whole-function
2779 * loop-invariant.c (find_defs): Use df_analyze_loop.
2780 (find_invariants): Adjust.
2781 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2783 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2785 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2786 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2788 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2790 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2791 traversal when removing references.
2793 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2796 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2798 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2801 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2802 pushing a reload for an autoinc when we had previously reloaded an
2803 inner part of the address.
2805 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2807 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2809 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2810 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2811 when not giving up or versioning for alias only because of
2813 (vect_analyze_data_ref_dependences): Set to true.
2814 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2816 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2817 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2821 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2824 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2825 operand 0 predicate for gathers, use a new pseudo as subtarget.
2827 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2830 * lra-constraints.c (process_alt_operands): Add printing debug info.
2831 Check absence of input/output reloads for matched operands too.
2833 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2835 PR rtl-optimization/59835
2836 * ira.c (ira_init_register_move_cost): Increase cost for
2839 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2841 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2843 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2846 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2847 non-register objects. Use gen_(high/low)part more consistently.
2850 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2853 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2854 endian support, remove tests for WORDS_BIG_ENDIAN.
2855 (p8_mfvsrd_3_<mode>): Likewise.
2856 (reload_gpr_from_vsx<mode>): Likewise.
2857 (reload_gpr_from_vsxsf): Likewise.
2858 (p8_mfvsrd_4_disf): Likewise.
2860 2014-01-16 Richard Biener <rguenther@suse.de>
2862 PR rtl-optimization/46590
2863 * lcm.c (compute_antinout_edge): Use postorder iteration.
2864 (compute_laterin): Use inverted postorder iteration.
2866 2014-01-16 Nick Clifton <nickc@redhat.com>
2869 * varasm.c (output_constant): Return the number of bytes actually
2871 (output_constructor_array_range): Update the field size with the
2872 number of bytes emitted by output_constant.
2873 (output_constructor_regular_field): Likewise. Also do not
2874 complain if the total number of bytes emitted is now greater
2875 than the expected fieldpos.
2876 * output.h (output_constant): Update prototype and descriptive comment.
2878 2014-01-16 Marek Polacek <polacek@redhat.com>
2881 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2882 it is error_mark_node.
2884 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2886 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2887 VALID_AVX256_REG_OR_OI_MODE.
2889 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2891 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2892 current procedure should be profiled.
2894 2014-01-15 Andrew Pinski <apinski@cavium.com>
2896 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2897 of moving from/to the STACK_REG register class.
2899 2014-01-15 Richard Henderson <rth@redhat.com>
2902 * reginfo.c (global_regs_decl): Globalize.
2903 * rtl.h (global_regs_decl): Declare.
2904 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2905 reserved via global_regs.
2907 2014-01-15 Teresa Johnson <tejohnson@google.com>
2909 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2911 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2913 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2914 and vmulosh rather than call gen_vec_widen_smult_*.
2915 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2916 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2917 (vec_widen_smult_even_v16qi): Likewise.
2918 (vec_widen_umult_even_v8hi): Likewise.
2919 (vec_widen_smult_even_v8hi): Likewise.
2920 (vec_widen_umult_odd_v16qi): Likewise.
2921 (vec_widen_smult_odd_v16qi): Likewise.
2922 (vec_widen_umult_odd_v8hi): Likewise.
2923 (vec_widen_smult_odd_v8hi): Likewise.
2924 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2925 vmuloub rather than call gen_vec_widen_umult_*.
2926 (vec_widen_umult_lo_v16qi): Likewise.
2927 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2928 vmulosb rather than call gen_vec_widen_smult_*.
2929 (vec_widen_smult_lo_v16qi): Likewise.
2930 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2931 rather than call gen_vec_widen_umult_*.
2932 (vec_widen_umult_lo_v8hi): Likewise.
2933 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2934 rather than call gen_vec_widen_smult_*.
2935 (vec_widen_smult_lo_v8hi): Likewise.
2937 2014-01-15 Jeff Law <law@redhat.com>
2939 PR tree-optimization/59747
2940 * ree.c (find_and_remove_re): Properly handle case where a second
2941 eliminated extension requires widening a copy created for elimination
2942 of a prior extension.
2943 (combine_set_extension): Ensure that the number of hard regs needed
2944 for a destination register does not change when we widen it.
2946 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2948 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2949 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2950 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2951 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2952 (avr-*-rtems*): Likewise.
2953 (bfin*-rtems*): Likewise.
2954 (moxie-*-rtems*): Likewise.
2955 (h8300-*-rtems*): Likewise.
2956 (i[34567]86-*-rtems*): Likewise.
2957 (lm32-*-rtems*): Likewise.
2958 (m32r-*-rtems*): Likewise.
2959 (m68k-*-rtems*): Likewise.
2960 (microblaze*-*-rtems*): Likewise.
2961 (mips*-*-rtems*): Likewise.
2962 (powerpc-*-rtems*): Likewise.
2963 (sh-*-rtems*): Likewise.
2964 (sparc-*-rtems*): Likewise.
2965 (sparc64-*-rtems*): Likewise.
2966 (v850-*-rtems*): Likewise.
2967 (m32c-*-rtems*): Likewise.
2969 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2971 PR rtl-optimization/59511
2972 * ira.c (ira_init_register_move_cost): Use memory costs for some
2973 cases of register move cost calculations.
2974 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2975 instead of BB frequency.
2976 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2977 * lra-assigns.c (find_hard_regno_for): Ditto.
2979 2014-01-15 Richard Biener <rguenther@suse.de>
2981 PR tree-optimization/59822
2982 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2983 (vectorizable_load): Use it to hoist defs of uses of invariant
2984 loads out of the loop.
2986 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2987 Kugan Vivekanandarajah <kuganv@linaro.org>
2990 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2993 2014-01-15 Richard Biener <rguenther@suse.de>
2995 PR rtl-optimization/59802
2996 * lcm.c (compute_available): Use inverted postorder to seed
2997 the initial worklist.
2999 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3002 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3003 ADDR_REGS for invalid symrefs in non-PIC code.
3005 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3008 * builtins.c (determine_block_size): Initialize *probable_max_size
3009 even if len_rtx is CONST_INT.
3011 2014-01-14 Andrew Pinski <apinski@cavium.com>
3013 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3014 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3015 (cortexa53_tunings): Likewise.
3016 (aarch64_sched_issue_rate): New function.
3017 (TARGET_SCHED_ISSUE_RATE): Define.
3019 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3021 * ira-costs.c (find_costs_and_classes): Add missed
3022 ira_init_register_move_cost_if_necessary.
3024 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3027 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3029 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3032 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3033 to indicate if type is used for function return value. Warn ABI
3034 change if the vector mode isn't available for function return value.
3035 (ix86_function_arg_advance): Pass false to type_natural_mode.
3036 (ix86_function_arg): Likewise.
3037 (ix86_gimplify_va_arg): Likewise.
3038 (function_arg_32): Don't warn ABI change.
3039 (ix86_function_value): Pass true to type_natural_mode.
3040 (ix86_return_in_memory): Likewise.
3041 (ix86_struct_value_rtx): Removed.
3042 (TARGET_STRUCT_VALUE_RTX): Likewise.
3044 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3046 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3047 converting a conditional jump into a conditional return.
3049 2014-01-14 Richard Biener <rguenther@suse.de>
3051 PR tree-optimization/58921
3052 PR tree-optimization/59006
3053 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3054 hoisting invariant stmts.
3055 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3056 invariant loads on the preheader edge if possible.
3058 2014-01-14 Joey Ye <joey.ye@arm.com>
3060 * doc/plugin.texi (Building GCC plugins): Update to C++.
3062 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3064 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3065 (_mm_rcp28_round_ss): Ditto.
3066 (_mm_rsqrt28_round_sd): Ditto.
3067 (_mm_rsqrt28_round_ss): Ditto.
3068 (_mm_rcp28_sd): Ditto.
3069 (_mm_rcp28_ss): Ditto.
3070 (_mm_rsqrt28_sd): Ditto.
3071 (_mm_rsqrt28_ss): Ditto.
3072 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3073 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3074 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3075 (IX86_BUILTIN_RCP28SD): Ditto.
3076 (IX86_BUILTIN_RCP28SS): Ditto.
3077 (IX86_BUILTIN_RSQRT28SD): Ditto.
3078 (IX86_BUILTIN_RSQRT28SS): Ditto.
3079 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3080 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3081 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3082 (ix86_expand_special_args_builtin): Expand new FTYPE.
3083 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3084 (srcp14<mode>): Make insn unary.
3085 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3086 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3087 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3088 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3089 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3090 Fix rounding: make it SAE only.
3091 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3093 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3095 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3096 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3097 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3098 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3099 (round_saeonly_mask_scalar_operand4): Ditto.
3100 (round_saeonly_mask_scalar_op3): Ditto.
3101 (round_saeonly_mask_scalar_op4): Ditto.
3103 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3105 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3106 Implement -maltivec=be for vec_insert and vec_extract.
3108 2014-01-10 DJ Delorie <dj@redhat.com>
3110 * config/msp430/msp430.md (call_internal): Don't allow memory
3111 references with SP as the base register.
3112 (call_value_internal): Likewise.
3113 * config/msp430/constraints.md (Yc): New. For memory references
3114 that don't use SP as a base register.
3116 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3117 "an integer without a # prefix"
3118 * config/msp430/msp430.md (epilogue_helper): Use it.
3120 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3123 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3124 AVX512F gather builtins.
3125 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3126 on gather decls with INTEGER_TYPE masktype.
3127 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3128 directly into the builtin rather than hoisting it before loop.
3130 PR tree-optimization/59387
3131 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3132 (scev_const_prop): If folded_casts and type has undefined overflow,
3133 use force_gimple_operand instead of force_gimple_operand_gsi and
3134 for each added stmt if it is assign with
3135 arith_code_with_undefined_signed_overflow, call
3136 rewrite_to_defined_overflow.
3137 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3138 gimple-fold.h instead.
3139 (arith_code_with_undefined_signed_overflow,
3140 rewrite_to_defined_overflow): Moved to ...
3141 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3142 rewrite_to_defined_overflow): ... here. No longer static.
3143 Include gimplify-me.h.
3144 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3145 rewrite_to_defined_overflow): New prototypes.
3147 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3149 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3151 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3153 * builtins.c (get_object_alignment_2): Minor tweak.
3154 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3156 2014-01-13 Christian Bruel <christian.bruel@st.com>
3158 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3159 optimized non constant lengths.
3161 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3164 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3165 load as __atomic_load_N if possible.
3167 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3169 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3171 (rs6000_expand_builtin): Adjust call.
3173 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3176 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3177 * config/rs6000/rs6000.c: Include target-globals.h.
3178 (rs6000_set_current_function): Instead of doing target_reinit
3179 unconditionally, use save_target_globals_default_opts and
3180 restore_target_globals.
3182 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3184 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3185 (rs6000_expand_builtin): Handle mffs and mtfsf.
3186 (rs6000_init_builtins): Define mffs and mtfsf.
3187 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3188 (rs6000_mffs): New pattern.
3189 (rs6000_mtfsf): New pattern.
3191 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3193 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3194 Start narrowing with START. Apply candidate-use pair
3195 and check overall cost in narrowing.
3196 (iv_ca_prune): Pass new argument.
3198 2014-01-10 Jeff Law <law@redhat.com>
3201 * ree.c (combine_reaching_defs): Ensure the defining statement
3202 occurs before the extension when optimizing extensions with
3203 different source and destination hard registers.
3205 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3208 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3209 vtables into the type inheritance graph.
3211 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3213 PR rtl-optimization/59754
3214 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3215 modes in the REGNO != REGNO case.
3217 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3219 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3221 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3223 PR tree-optimization/59745
3224 * tree-predcom.c (tree_predictive_commoning_loop): Call
3225 free_affine_expand_cache if giving up because components is NULL.
3227 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3228 GC in payload of target_globals struct instead of allocating them on
3229 the heap and the larger structs separately using GC.
3230 * target-globals.h (struct target_globals): Make regs, hard_regs,
3231 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3232 of GTY((skip)) and change type to void *.
3233 (reset_target_globals): Cast loads from those fields to corresponding
3236 2014-01-10 Steve Ellcey <sellcey@mips.com>
3239 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3240 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3241 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3243 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3246 * aarch64-modes.def (CC_Zmode): New flags mode.
3247 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3248 represents an equality.
3249 (aarch64_get_condition_code): Handle CC_Zmode.
3250 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3252 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3254 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3255 extraction in good case.
3257 2014-01-10 Richard Biener <rguenther@suse.de>
3259 PR tree-optimization/59374
3260 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3261 checking after SLP discovery. Mark stmts not participating
3262 in any SLP instance properly.
3264 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3266 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3267 when handling a SET rtx.
3269 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3271 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3272 (cortex-a57): Likewise.
3273 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3275 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3277 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3278 non-iwmmxt builtins.
3280 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3284 * ipa-devirt.c record_target_from_binfo): Take as argument
3285 stack of binfos and lookup matching one for virtual inheritance.
3286 (possible_polymorphic_call_targets_1): Update.
3288 2014-01-10 Huacai Chen <chenhc@lemote.com>
3290 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3291 kernel strings for Loongson-2E/2F/3A.
3293 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3296 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3297 is_gimple_call before calling gimple_call_internal_p.
3299 2014-01-09 Steve Ellcey <sellcey@mips.com>
3301 * Makefile.in (TREE_FLOW_H): Remove.
3302 (TREE_SSA_H): Add file names from tree-flow.h.
3303 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3304 * tree.h: Remove tree-flow.h reference.
3305 * hash-table.h: Remove tree-flow.h reference.
3306 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3307 reference with tree-ssa-loop.h.
3309 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3311 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3312 default element-order behavior for -maltivec.
3313 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3314 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3315 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3316 when targeting big endian, at least for now.
3317 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3319 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3322 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3323 var satisfies use_register_for_decl, just take into account type
3324 alignment, rather than decl alignment.
3326 PR tree-optimization/59622
3327 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3328 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3329 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3330 Don't devirtualize for inplace at all. For targets.length () == 1,
3331 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3333 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3335 * config/i386/i386.md (cpu): Remove the unused btver1.
3337 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3339 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3341 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3344 * tree-core.h (struct target_globals): New forward declaration.
3345 (struct tree_target_option): Add globals field.
3346 * tree.h (TREE_TARGET_GLOBALS): Define.
3347 (prepare_target_option_nodes_for_pch): New prototype.
3348 * target-globals.h (struct target_globals): Define even if
3350 * tree.c (prepare_target_option_node_for_pch,
3351 prepare_target_option_nodes_for_pch): New functions.
3352 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3353 * config/i386/i386.c: Include target-globals.h.
3354 (ix86_set_current_function): Instead of doing target_reinit
3355 unconditionally, use save_target_globals_default_opts and
3356 restore_target_globals.
3358 2014-01-09 Richard Biener <rguenther@suse.de>
3360 PR tree-optimization/59715
3361 * tree-cfg.h (split_critical_edges): Declare.
3362 * tree-cfg.c (split_critical_edges): Export.
3363 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3365 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3367 * cfgexpand.c (expand_stack_vars): Optionally disable
3368 asan stack protection.
3369 (expand_used_vars): Likewise.
3370 (partition_stack_vars): Likewise.
3371 * asan.c (asan_emit_stack_protection): Optionally disable
3372 after return stack usage.
3373 (instrument_derefs): Optionally disable memory access instrumentation.
3374 (instrument_builtin_call): Likewise.
3375 (instrument_strlen_call): Likewise.
3376 (asan_protect_global): Optionally disable global variables protection.
3377 * doc/invoke.texi: Added doc for new options.
3378 * params.def: Added new options.
3379 * params.h: Likewise.
3381 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3383 PR rtl-optimization/59724
3384 * ifcvt.c (cond_exec_process_if_block): Don't call
3385 flow_find_head_matching_sequence with 0 longest_match.
3386 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3387 non-active insns if !stop_after.
3388 (try_head_merge_bb): Revert 2014-01-07 changes.
3390 2014-01-08 Jeff Law <law@redhat.com>
3392 * ree.c (get_sub_rtx): New function, extracted from...
3393 (merge_def_and_ext): Here.
3394 (combine_reaching_defs): Use get_sub_rtx.
3396 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3398 * cgraph.h (varpool_variable_node): Do not choke on null node.
3400 2014-01-08 Catherine Moore <clm@codesourcery.com>
3402 * config/mips/mips.md (simple_return): Attempt to use JRC
3404 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3406 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3408 PR rtl-optimization/59137
3409 * reorg.c (steal_delay_list_from_target): Call update_block for
3411 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3413 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3415 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3416 two duplicate entries.
3418 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3421 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3423 * config/mips/mips.c (mips_truncated_op_cost): New function.
3424 (mips_rtx_costs): Adjust test for BADDU.
3425 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3427 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3429 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3430 (*baddu_si): ...this new pattern.
3432 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3435 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3437 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3440 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3442 (expand_expr, expand_normal): Adjust.
3443 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3444 inner_reference_p. Use inner_reference_p to expand inner references.
3445 (store_expr): Adjust.
3446 * cfgexpand.c (expand_call_stmt): Adjust.
3448 2014-01-08 Rong Xu <xur@google.com>
3450 * gcov-io.c (gcov_var): Move from gcov-io.h.
3451 (gcov_position): Ditto.
3452 (gcov_is_error): Ditto.
3453 (gcov_rewrite): Ditto.
3454 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3455 only part to libgcc/libgcov.h.
3457 2014-01-08 Marek Polacek <polacek@redhat.com>
3460 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3462 2014-01-08 Marek Polacek <polacek@redhat.com>
3465 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3467 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3469 PR rtl-optimization/59649
3470 * stor-layout.c (get_mode_bounds): For BImode return
3471 0 and STORE_FLAG_VALUE.
3473 2014-01-08 Richard Biener <rguenther@suse.de>
3476 * gimple.h (is_gimple_builtin_call): Remove.
3477 (gimple_builtin_call_types_compatible_p): New.
3478 (gimple_call_builtin_p): New overload.
3479 * gimple.c (is_gimple_builtin_call): Remove.
3480 (validate_call): Rename to ...
3481 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3483 (validate_type): New static function.
3484 (gimple_call_builtin_p): New overload and adjust.
3485 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3486 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3487 (gimple_fold_stmt_to_constant_1): Likewise.
3488 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3490 2014-01-08 Richard Biener <rguenther@suse.de>
3493 * gimplify.c (gimplify_expr): Gimplify register-register type
3494 VIEW_CONVERT_EXPRs to separate stmts.
3496 2014-01-07 Jeff Law <law@redhat.com>
3499 * ree.c (combine_set_extension): Handle case where source
3500 and destination registers in an extension insn are different.
3501 (combine_reaching_defs): Allow source and destination registers
3502 in extension to be different under limited circumstances.
3503 (add_removable_extension): Remove restriction that the
3504 source and destination registers in the extension are the same.
3505 (find_and_remove_re): Emit a copy from the extension's
3506 destination to its source after the defining insn if
3507 the source and destination registers are different.
3510 * ifcvt.c (merge_if_block): If we are merging a block with more than
3511 one successor with a block with no successors, remove any BARRIER
3512 after the second block.
3514 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3516 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3518 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3521 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3522 for 14-bit register offsets when INT14_OK_STRICT is false.
3524 2014-01-07 Roland Stigge <stigge@antcom.de>
3525 Michael Meissner <meissner@linux.vnet.ibm.com>
3528 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3529 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3531 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3533 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3536 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3538 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3539 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3540 rtx is const0_rtx or not.
3542 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3545 * target-globals.c (save_target_globals): Remove this_fn_optab
3547 * toplev.c: Include optabs.h.
3548 (target_reinit): Temporarily restore the global options if another
3549 set of options are in force.
3551 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3553 PR rtl-optimization/58668
3554 * cfgcleanup.c (flow_find_cross_jump): Don't count
3555 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3556 to determine what is counted.
3557 (flow_find_head_matching_sequence): Use active_insn_p to determine
3559 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3561 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3562 determine what is counted.
3564 PR tree-optimization/59643
3565 * tree-predcom.c (split_data_refs_to_components): If one dr is
3566 read and one write, determine_offset fails and the write isn't
3567 in the bad component, just put the read into the bad component.
3569 2014-01-07 Mike Stump <mikestump@comcast.net>
3570 Jakub Jelinek <jakub@redhat.com>
3573 * tree-core.h (struct tree_optimization_option): Change optabs
3574 type from unsigned char * to void *.
3575 * optabs.c (init_tree_optimization_optabs): Adjust
3576 TREE_OPTIMIZATION_OPTABS initialization.
3578 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3581 * config/i386/i386.h (struct machine_function): Add
3582 no_drap_save_restore field.
3583 * config/i386/i386.c (ix86_save_reg): Use
3584 !cfun->machine->no_drap_save_restore instead of
3585 crtl->stack_realign_needed.
3586 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3587 this function clears frame_pointer_needed. Set
3588 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3589 and DRAP reg is needed.
3591 2014-01-06 Marek Polacek <polacek@redhat.com>
3594 * doc/implement-c.texi: Mention that other integer types are
3595 permitted as bit-field types in strictly conforming mode.
3597 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3599 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3602 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3604 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3606 2014-01-06 Martin Jambor <mjambor@suse.cz>
3609 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3611 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3613 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3617 * var-tracking.c (add_stores): Preserve the value of the source even if
3618 we don't record the store.
3620 2014-01-06 Terry Guo <terry.guo@arm.com>
3622 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3624 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3627 * config/darwin.c (darwin_function_section): Adjust return values to
3628 correspond to optimisation changes made in r206070.
3630 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3632 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3633 from prefetch_block tune setting.
3634 (nocona_cost): Correct size of prefetch block to 64.
3636 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3638 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3639 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3640 used to save the static chain register in the computation of the offset
3641 from which the FP registers need to be restored.
3643 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3645 PR tree-optimization/59519
3646 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3647 ICE if get_current_def (current_new_name) is already non-NULL, as long
3648 as it is a phi result of some other phi in *new_exit_bb that has
3651 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3652 or vmovdqu* for misaligned_operand.
3653 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3654 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3655 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3656 aligned_mem for AVX512F masked aligned load and store builtins and for
3659 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3661 PR tree-optimization/59651
3662 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3663 Address range for negative step should be added by TYPE_SIZE_UNIT.
3665 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3667 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3668 overlapping registers also for registers other than the stack pointer.
3670 2014-01-03 Marek Polacek <polacek@redhat.com>
3673 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3676 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3679 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3682 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3683 (push splitter): Use <P:MODE_SIZE> instead of
3684 GET_MODE_SIZE (<P:MODE>mode).
3685 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3686 (mov -1, reg peephole2): Likewise.
3687 * config/i386/sse.md (*mov<mode>_internal,
3688 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3689 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3690 *<code><mode>3, *andnot<mode>3<mask_name>,
3691 <mask_codefor><code><mode>3<mask_name>): Likewise.
3692 * config/i386/subst.md (mask_mode512bit_condition,
3693 sd_mask_mode512bit_condition): Likewise.
3695 2014-01-02 Xinliang David Li <davidxl@google.com>
3697 PR tree-optimization/59303
3698 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3699 (dump_predicates): Better output format.
3700 (pred_equal_p): New function.
3701 (is_neq_relop_p): Ditto.
3702 (is_neq_zero_form_p): Ditto.
3703 (pred_expr_equal_p): Ditto.
3704 (pred_neg_p): Ditto.
3705 (simplify_pred): Ditto.
3706 (simplify_preds_2): Ditto.
3707 (simplify_preds_3): Ditto.
3708 (simplify_preds_4): Ditto.
3709 (simplify_preds): Ditto.
3711 (push_to_worklist): Ditto.
3712 (get_pred_info_from_cmp): Ditto.
3713 (is_degenerated_phi): Ditto.
3714 (normalize_one_pred_1): Ditto.
3715 (normalize_one_pred): Ditto.
3716 (normalize_one_pred_chain): Ditto.
3717 (normalize_preds): Ditto.
3718 (normalize_cond_1): Remove function.
3719 (normalize_cond): Ditto.
3720 (is_gcond_subset_of): Ditto.
3721 (is_subset_of_any): Ditto.
3722 (is_or_set_subset_of): Ditto.
3723 (is_and_set_subset_of): Ditto.
3724 (is_norm_cond_subset_of): Ditto.
3725 (pred_chain_length_cmp): Ditto.
3726 (convert_control_dep_chain_into_preds): Type change.
3727 (find_predicates): Ditto.
3728 (find_def_preds): Ditto.
3729 (destroy_predicates_vecs): Ditto.
3730 (find_matching_predicates_in_rest_chains): Ditto.
3731 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3732 (is_pred_expr_subset): Ditto.
3733 (is_pred_chain_subset_of): Ditto.
3734 (is_included_in): Ditto.
3735 (is_superset_of): Ditto.
3737 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3739 Update copyright years.
3741 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3743 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3744 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3745 config/arc/arc.md, config/arc/arc.opt,
3746 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3747 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3748 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3749 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3750 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3751 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3752 vtable-verify.c, vtable-verify.h: Use the standard form for the
3755 2014-01-02 Tobias Burnus <burnus@net-b.de>
3757 * gcc.c (process_command): Update copyright notice dates.
3758 * gcov-dump.c: Ditto.
3760 * doc/cpp.texi: Bump @copying's copyright year.
3761 * doc/cppinternals.texi: Ditto.
3762 * doc/gcc.texi: Ditto.
3763 * doc/gccint.texi: Ditto.
3764 * doc/gcov.texi: Ditto.
3765 * doc/install.texi: Ditto.
3766 * doc/invoke.texi: Ditto.
3768 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3770 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3772 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3774 * config/i386/sse.md (*mov<mode>_internal): Guard
3775 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3777 PR rtl-optimization/59647
3778 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3779 new_rtx into UNSIGNED_FLOAT rtxes.
3781 Copyright (C) 2014 Free Software Foundation, Inc.
3783 Copying and distribution of this file, with or without modification,
3784 are permitted in any medium without royalty provided the copyright
3785 notice and this notice are preserved.