1 2014-03-20 Bernd Schmidt <bernds@codesourcery.com>
3 Mostly by Michael Zolotukhin:
4 * lto-wrapper.c (OFFLOAD_FUNC_TABLE_SECTION_NAME,
5 OFFLOAD_TARGET_NAMES_ENV): New defines.
6 (offload_names): New static variable.
7 (free_array_of_ptrs, parse_env_var, access_check,
8 prepare_target_image, replace_special_characters,
9 compile_images_for_openmp_targets): New static functions.
10 (run_gcc): Determine whether offload sections are present. If so,
11 run compile_images_for_openmp_targets and print the image names.
13 * Makefile.in (COLLECT2_OBJS): Add collect-utils.o.
14 (LTO_WRAPPER_OBJS): New variable.
15 (lto-wrapper$(exeext)): Use it.
16 * collect2.c: Include "collect-utils.h".
17 (vflag, debug): Remove variables. Rename all uses of vflag to
19 (at_file_supplied): No longer static.
20 (tool_name): New variable.
21 (do_wait, fork_execute, maybe_unlink): Don't declare.
22 (tool_cleanup): Renamed from collect_atexit. New arg from_signal.
23 (collect_atexit): New static function.
24 (notice): Remove function.
25 (handler): Just call tool_cleanup.
26 (maybe_run_lto_and_relink, main, do_dsymutil): Add new arg to
28 (collect_wait, do_wait, collect_execute): Remove functions.
29 (maybe_unlink): No longer static.
30 * collect2.h (vflag, debug): Don't declare.
31 (at_file_supplied): Declare.
32 * collect-utils.c (utils_cleanup): New arg from_signal. All callers
34 (collect_execute): Replace with implementation from collect2, plus a
35 new arg use_atfile. All callers changed.
36 (collect_wait): Replace with implementation from collect2.
37 (maybe_unlink_file): Remove function.
38 (fork_execute): Replace with implementation from collect2, plus a
39 new arg use_atfile. All callers changed.
40 (do_wait): Add call to utils_cleanup to the error path.
41 * collect-utils.h (collect_execute, fork_execute, utils_cleanup,
42 tool_cleanup): Adjust declarations.
43 * lto-wrapper.c (tool_cleanup): Add unused bool argument.
44 * tlink.c: Include "collect-utils.h".
45 (tlink_execute): New arg use_atfile. All callers changed.
46 (tlink_init, tlink_execute): Remove declarations.
48 * Makefile.in (ALL_HOST_BACKEND_OBJS): Add collect-utils.o.
49 (lto-wrapper$(exeext)): Link with collect-utils.o.
50 * collect-utils.c: New file.
51 * collect-utils.h: New file.
52 * lto-wrapper.c: Include "collect-utils.h".
53 (args_name): Delete variable.
54 (tool_name): New variable.
55 (tool_cleanup): New function.
56 (maybe_unlink): Renamed from maybe_unlink_file. All callers changed.
57 (lto_wrapper_cleanup, fatal_signal, fatal, fatal_perror,
58 collect_execute, collect_wait, fork_execute): Remove functions.
60 Mostly by Michael Zolotukhin:
61 * omp-low.c: Include "common/common-target.h".
62 (expand_omp_target): Pass in address of __OPENMP_TARGET__.
63 (add_decls_addresses_to_decl_constructor, omp_finish_file): New
65 * omp-low.h (omp_finish_file): Declare.
66 * toplev.c: Include "omp-low.h".
67 (compile_file): Call omp_finish_file.
68 * target.def (record_offload_symbol): New hook.
69 * doc/tm.texi.in (TARGET_RECORD_OFFLOAD_SYMBOL): Add.
70 * doc/tm.texi: Regenerate.
71 * configure.ac (ENABLE_OFFLOADING): Define if we have offload_targets.
72 * configure: Regenerate.
73 * config.in: Regenerate.
75 * config/darwin.c: Include "lto-section-names.h".
76 (LTO_SEGMENT_NAME): Don't define.
77 * config/i386/winnt.c: Include "lto-section-names.h".
78 * lto-streamer.c: Include "lto-section-names.h".
79 * lto-streamer.h (LTO_SECTION_NAME_PREFIX, OMP_SECTION_NAME_PREFIX):
81 (section_name_prefix): Don't declare.
82 * lto-wrapper.c: Include "lto-section-names.h".
83 (LTO_SECTION_NAME_PREFIX): Don't define.
84 * lto-section-names.h: New file.
85 * cgraphunit.c: Include "lto-section-names.h".
87 * configure.ac (real_target_noncanonical, tool_prefix,
88 accel_dir_suffix, offload_targets): Compute new variables.
89 (--enable-as-accelerator-for, --enable-as-accelerator,
90 --enable-offload-targets): New options.
91 * configure: Regenerate.
92 * config.in: Regenerate.
93 * Makefile.in (real_target_noncanonical, tool_prefix,
94 accel_dir_suffix, enable_as_accelerator): New variables substituted
96 (libsubdir, libexecsubdir, GCC_TARGET_INSTALL_NAME): Tweak for the
97 possibility of being configured as an offload compiler.
98 (install_driver): Likewise.
99 (DRIVER_DEFINES): Pass new defines DEFAULT_REAL_TARGET_MACHINE and
101 * gcc.c (spec_host_machine, accel_dir_suffix): New variables.
102 (process_command): Tweak path construction for the possibility
103 of being configured as an offload compiler.
104 (main): Likewise. Look up specs in just_machine_suffix only if not
105 ACCEL_COMPILER. Construct OFFLOAD_TARGET_NAMES environment variable
106 if we have OFFLOAD_TARGETS.
108 2014-03-05 Jakub Jelinek <jakub@redhat.com>
111 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
112 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
115 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
116 (TM_H): Add x86-tune.def.
118 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
120 * config/aarch64/aarch64.c (generic_tunings):
121 Use cortexa57_extra_costs.
123 2014-03-05 Jakub Jelinek <jakub@redhat.com>
126 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
127 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
128 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
131 2014-03-04 Heiher <r@hev.cc>
133 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
134 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
136 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
138 * config/i386/predicates.md (const2356_operand): Change to ...
139 (const2367_operand): ... this.
140 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
142 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
143 (*avx512pf_scatterpf<mode>sf): Ditto.
144 (avx512pf_scatterpf<mode>df): Ditto.
145 (*avx512pf_scatterpf<mode>df_mask): Ditto.
146 (*avx512pf_scatterpf<mode>df): Ditto.
147 * config/i386/i386.c (ix86_expand_builtin): Update
148 incorrect hint operand error message.
150 2014-03-04 Richard Biener <rguenther@suse.de>
152 * lto-section-in.c (lto_get_section_data): Fix const cast.
154 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
156 * tree-streamer.c (record_common_node): Assert we don't record
157 nodes with type double.
158 (preload_common_node): Skip type double, complex double and
159 double pointer since it is now frontend dependent due to
160 fshort-double option.
162 2014-03-04 Richard Biener <rguenther@suse.de>
165 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
166 (lto_input_toplevel_asms): Likewise.
167 * lto-section-in.c (lto_get_section_data): Instead do it here
170 2014-03-04 Richard Biener <rguenther@suse.de>
172 PR tree-optimization/60382
173 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
174 dead PHIs a reduction.
176 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
178 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
180 (_mm_prefetch): Move out of GCC target("sse") pragma.
181 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
182 GCC target("prfchw") pragma.
183 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
185 * config/i386/i386.c (ix86_option_override_internal): Enable
186 -mprfchw with -mprefetchwt1.
188 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
190 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
193 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
195 * opts.h (CL_PCH_IGNORE): Define.
196 * targhooks.c (option_affects_pch_p):
197 Return false for options that have CL_PCH_IGNORE set.
198 * opt-functions.awk: Process PchIgnore.
199 * doc/options.texi: Document PchIgnore.
201 * config/arc/arc.opt (misize): Add PchIgnore property.
203 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
205 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
206 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
207 constraint on constants to permit them being loaded into
208 GENERAL_REGS or BASE_REGS.
210 2014-03-03 Nick Clifton <nickc@redhat.com>
212 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
213 anti-cacnonical alternatives.
214 (negandhi3_real): New pattern.
215 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
217 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
219 * config/avr/avr-mcus.def: Remove atxmega16x1.
220 * config/avr/avr-tables.opt: Regenerate.
221 * config/avr/t-multilib: Regenerate.
222 * doc/avr-mmcu.texi: Regenerate.
224 2014-03-03 Tobias Grosser <tobias@grosser.es>
225 Mircea Namolaru <mircea.namolaru@inria.fr>
227 PR tree-optimization/58028
228 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
231 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
233 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
234 not handled by recognizers.
236 2014-03-03 Jakub Jelinek <jakub@redhat.com>
239 * function.c (expand_function_end): Don't emit
240 clobber_return_register sequence if clobber_after is a BARRIER.
241 * cfgexpand.c (construct_exit_block): Append instructions before
242 return_label to prev_bb.
244 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
246 * config/rs6000/constraints.md: Document reserved use of "wc".
248 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
251 * ipa.c (function_and_variable_visibility): When dissolving comdat
252 group, also set all symbols to local.
254 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
259 2013-12-14 Jan Hubicka <jh@suse.cz>
261 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
263 2014-03-02 Jon Beniston <jon@beniston.com>
269 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
270 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
271 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
272 (simple_return, *simple_return): New patterns
273 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
274 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
276 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
278 * dwarf2out.c (gen_subprogram_die): Tidy.
280 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
283 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
284 (*mov_t_msb_neg_negc): ... this new insn.
286 2014-02-28 Jason Merrill <jason@redhat.com>
289 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
292 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
295 * dwarf2out.c (decltype_auto_die): New static.
296 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
297 (gen_type_die_with_usage): Handle 'decltype(auto)'.
298 (is_cxx_auto): Likewise.
300 2014-02-28 Ian Bolton <ian.bolton@arm.com>
302 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
303 we are not using general regs only.
305 2014-02-28 Richard Biener <rguenther@suse.de>
308 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
309 previous fix and only allow to remove trivial pre-headers
310 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
311 (remove_forwarder_block): Properly update the latch of a loop.
313 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
316 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
317 (cselib_preserved_hash_table): New.
318 (preserve_constants_and_equivs): Move preserved vals to it.
319 (cselib_find_slot): Look it up first.
320 (cselib_init): Initialize it.
321 (cselib_finish): Release it.
322 (dump_cselib_table): Dump it.
324 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
327 * cselib.c (remove_useless_values): Skip to avoid quadratic
328 behavior if the condition moved from...
329 (cselib_process_insn): ... here holds.
331 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
334 * var-tracking.c (vt_initialize): Apply the same condition to
335 preserve the CFA base value.
337 2014-02-28 Joey Ye <joey.ye@arm.com>
340 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
341 if reload in progress or completed.
343 2014-02-28 Tobias Burnus <burnus@net-b.de>
346 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
349 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
351 * doc/tm.texi.in (Condition Code Status): Update documention for
352 relative locations of cc0-setter and cc0-user.
354 2014-02-27 Jeff Law <law@redhat.com>
356 PR rtl-optimization/52714
357 * combine.c (try_combine): When splitting an unrecognized PARALLEL
358 into two independent simple sets, if I3 is a jump, ensure the
359 pattern we place into I3 is a (set (pc) ...).
361 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
362 Jeff Law <law@redhat.com>
364 PR rtl-optimization/49847
365 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
366 are in different blocks.
367 * doc/tm.texi (Condition Code Status): Update documention for
368 relative locations of cc0-setter and cc0-user.
370 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
373 * lra.c (lra_emit_add): Check SUBREG too.
375 2014-02-27 Andreas Schwab <schwab@suse.de>
377 * config/m68k/m68k.c (m68k_option_override): Disable
378 -flive-range-shrinkage for classic m68k.
379 (m68k_override_options_after_change): Likewise.
381 2014-02-27 Marek Polacek <polacek@redhat.com>
384 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
385 -Wmaybe-uninitialized.
387 2014-02-27 Alan Modra <amodra@gmail.com>
390 * reload1.c (emit_input_reload_insns): When reload_override_in,
391 set old to rl->in_reg when rl->in_reg is a subreg.
393 2014-02-26 Richard Biener <rguenther@suse.de>
396 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
398 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
400 * common/config/i386/predicates.md (const1256_operand): Remove.
401 (const2356_operand): New.
402 (const_1_to_2_operand): Remove.
403 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
404 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
405 (*avx512pf_gatherpf<mode>sf): Ditto.
406 (avx512pf_gatherpf<mode>df): Ditto.
407 (*avx512pf_gatherpf<mode>df_mask): Ditto.
408 (*avx512pf_gatherpf<mode>df): Ditto.
409 (avx512pf_scatterpf<mode>sf): Ditto.
410 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
411 (*avx512pf_scatterpf<mode>sf): Ditto.
412 (avx512pf_scatterpf<mode>df): Ditto.
413 (*avx512pf_scatterpf<mode>df_mask): Ditto.
414 (*avx512pf_scatterpf<mode>df): Ditto.
415 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
417 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
419 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
420 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
421 (_mm512_mask_testn_epi64_mask): Move to ...
422 * config/i386/avx512cdintrin.h: Here.
423 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
424 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
425 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
426 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
427 TARGET_AVX512F from TARGET_AVX512CD.
429 2014-02-26 Richard Biener <rguenther@suse.de>
432 * ipa.c (walk_polymorphic_call_targets): Properly guard
433 call to inline_update_overall_summary.
435 2014-02-26 Bin Cheng <bin.cheng@arm.com>
438 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
439 and latches only if requested. Fix latch if it is removed.
440 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
441 LOOPS_HAVE_PREHEADERS.
443 2014-02-25 Andrew Pinski <apinski@cavium.com>
445 * builtins.c (expand_builtin_thread_pointer): Create a new target
446 when the target is NULL.
448 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
450 PR rtl-optimization/60317
451 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
452 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
453 * lra-assigns.c: Include params.h.
454 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
455 other reload pseudos considerations.
457 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
459 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
460 to use canonical form for nor<mode>3.
462 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
465 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
468 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
470 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
471 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
472 (ix86_handle_option): Handle OPT_mprefetchwt1.
473 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
474 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
476 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
477 OPTION_MASK_ISA_PREFETCHWT1.
478 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
479 (PTA_PREFETCHWT1): New.
480 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
481 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
482 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
483 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
484 (*prefetch_avx512pf_<mode>_: Change into ...
485 (*prefetch_prefetchwt1_<mode>: This.
486 * config/i386/i386.opt (mprefetchwt1): New.
487 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
488 (_mm_prefetch): Handle intent to write.
489 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
491 2014-02-25 Richard Biener <rguenther@suse.de>
494 * emit-rtl.c (mem_attrs_htab): Remove.
495 (mem_attrs_htab_hash): Likewise.
496 (mem_attrs_htab_eq): Likewise.
497 (set_mem_attrs): Always allocate new mem-attrs when something changed.
498 (init_emit_once): Do not allocate mem_attrs_htab.
500 2014-02-25 Richard Biener <rguenther@suse.de>
503 * lto-opts.c (lto_write_options): Output non-explicit conservative
504 -fwrapv, -fno-trapv and -fno-strict-overflow.
505 * lto-wrapper.c (merge_and_complain): Handle merging those options.
506 (run_gcc): And pass them through.
508 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
510 * sel-sched.c (calculate_new_fences): New parameter ptime.
511 Calculate it as a maximum over all fence cycles.
512 (sel_sched_region_2): Adjust the call to calculate_new_fences.
513 Print the final schedule timing when sched_verbose.
515 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
517 PR rtl-optimization/60292
518 * sel-sched.c (fill_vec_av_set): Do not reset target availability
519 bit fot the fence instruction.
521 2014-02-24 Alangi Derick <alangiderick@gmail.com>
523 * calls.h: Fix typo in comment.
525 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
527 * config/pa/pa.c (pa_output_move_double): Don't valididate when
528 adjusting offsetable addresses.
530 2014-02-24 Guozhi Wei <carrot@google.com>
532 * sparseset.h (sparseset_pop): Fix the wrong index.
534 2014-02-24 Walter Lee <walt@tilera.com>
536 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
537 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
539 * common/config/tilegx/tilegx-common.c
540 (TARGET_DEFAULT_TARGET_FLAGS): Define.
541 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
543 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
544 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
545 (tilegx_gimplify_va_arg_expr): Handle big endian.
546 (tilegx_expand_unaligned_load): Ditto.
547 (tilegx_expand_unaligned_store): Ditto.
548 (TARGET_RETURN_IN_MSB): New.
549 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
550 (TARGET_ENDIAN_DEFAULT): New.
551 (TARGET_BIG_ENDIAN): Handle big endian.
552 (BYTES_BIG_ENDIAN): Ditto.
553 (WORDS_BIG_ENDIAN): Ditto.
554 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
557 * config/tilegx/tilegx.md (extv): Handle big endian.
560 (insn_st<n>_add<bitsuffix>): Ditto.
561 (insn_stnt<n>): Ditto.
562 (insn_stnt<n>_add<bitsuffix>):Ditto.
563 (vec_interleave_highv8qi): Handle big endian.
564 (vec_interleave_highv8qi_be): New.
565 (vec_interleave_highv8qi_le): New.
566 (insn_v1int_h): Handle big endian.
567 (vec_interleave_lowv8qi): Handle big endian.
568 (vec_interleave_lowv8qi_be): New.
569 (vec_interleave_lowv8qi_le): New.
570 (insn_v1int_l): Handle big endian.
571 (vec_interleave_highv4hi): Handle big endian.
572 (vec_interleave_highv4hi_be): New.
573 (vec_interleave_highv4hi_le): New.
574 (insn_v2int_h): Handle big endian.
575 (vec_interleave_lowv4hi): Handle big endian.
576 (vec_interleave_lowv4hi_be): New.
577 (vec_interleave_lowv4hi_le): New.
578 (insn_v2int_l): Handle big endian.
579 (vec_interleave_highv2si): Handle big endian.
580 (vec_interleave_highv2si_be): New.
581 (vec_interleave_highv2si_le): New.
582 (insn_v4int_h): Handle big endian.
583 (vec_interleave_lowv2si): Handle big endian.
584 (vec_interleave_lowv2si_be): New.
585 (vec_interleave_lowv2si_le): New.
586 (insn_v4int_l): Handle big endian.
587 * config/tilegx/tilegx.opt (mbig-endian): New option.
588 (mlittle-endian): New option.
589 * doc/install.texi: Document tilegxbe-linux.
590 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
592 2014-02-24 Martin Jambor <mjambor@suse.cz>
595 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
596 there are no parameter descriptors.
598 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
600 PR rtl-optimization/60268
601 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
602 initialization to ...
603 (sched_rgn_init): ... here.
604 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
606 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
608 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
611 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
613 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
616 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
618 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
619 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
621 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
623 * config/microblaze/predicates.md: Add cmp_op predicate.
624 * config/microblaze/microblaze.md: Add branch_compare instruction
625 which uses cmp_op predicate and emits cmp insn before branch.
626 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
627 to microblaze_expand_conditional_branch and consolidate logic.
628 (microblaze_expand_conditional_branch): emit branch_compare
629 insn instead of handling cmp op separate from branch insn.
631 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
633 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
636 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
638 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
639 define_insn with define_expand and new define_insn
640 *altivec_lve<VI_char>x_internal.
641 (altivec_stve<VI_char>x): Replace define_insn with define_expand
642 and new define_insn *altivec_stve<VI_char>x_internal.
643 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
645 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
647 (altivec_expand_stvex_be): New function.
649 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
651 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
652 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
653 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
654 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
656 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
659 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
660 instead of emit_move_insn.
662 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
664 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
666 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
669 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
671 * config/rs6000/altivec.md (altivec_lvxl): Rename as
672 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
673 (altivec_lvxl_<mode>): New define_expand incorporating
674 -maltivec=be semantics where needed.
675 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
676 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
677 semantics where needed.
678 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
679 (altivec_stvx_<mode>): New define_expand incorporating
680 -maltivec=be semantics where needed.
681 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
682 VM2 iterator instead of V4SI.
683 (altivec_stvxl_<mode>): New define_expand incorporating
684 -maltivec=be semantics where needed.
685 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
686 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
687 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
688 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
689 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
690 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
691 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
692 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
693 ALTIVEC_BUILTIN_STVXL.
694 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
695 (altivec_expand_stvx_be): Likewise.
696 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
697 (altivec_expand_lvx_be): Likewise.
698 (altivec_expand_stvx_be): Likewise.
699 (altivec_expand_builtin): Add cases for
700 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
701 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
702 (altivec_init_builtins): Add definitions for
703 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
704 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
706 2014-02-21 Catherine Moore <clm@codesourcery.com>
708 * doc/invoke.texi (mvirt, mno-virt): Document.
709 * config/mips/mips.opt (mvirt): New option.
710 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
712 2014-02-21 Richard Biener <rguenther@suse.de>
714 PR tree-optimization/60276
715 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
716 (STMT_VINFO_MIN_NEG_DIST): New macro.
717 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
718 STMT_VINFO_MIN_NEG_DIST.
719 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
720 made for negative dependence distances still hold.
722 2014-02-21 Richard Biener <rguenther@suse.de>
725 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
726 DECL_INITIAL for globals not in the current function context.
728 2014-02-21 Jakub Jelinek <jakub@redhat.com>
730 PR tree-optimization/56490
731 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
732 * tree-ssa-uninit.c: Include params.h.
733 (compute_control_dep_chain): Add num_calls argument, return false
734 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
735 num_calls to recursive call.
736 (find_predicates): Change dep_chain into normal array,
737 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
738 variable and adjust compute_control_dep_chain caller.
739 (find_def_preds): Likewise.
741 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
743 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
744 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
746 2014-02-21 Nick Clifton <nickc@redhat.com>
748 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
750 (popqi1): Add mode to pre_dec.
753 2014-02-21 Jakub Jelinek <jakub@redhat.com>
755 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
756 mode for mask of V8SFmode permutation.
758 2014-02-20 Richard Henderson <rth@redhat.com>
761 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
762 a new pseudo for OLDVAL.
764 2014-02-20 Jakub Jelinek <jakub@redhat.com>
767 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
768 gen_reg_rtx if d->testing_p.
769 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
770 if d->testing_p and we will certainly return true.
771 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
774 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
776 * emit-rtl.c (gen_reg_rtx): Assert that
777 crtl->emit.regno_pointer_align_length is non-zero.
779 2014-02-20 Richard Henderson <rth@redhat.com>
782 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
783 on failure the store back into EXPECT.
785 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
786 Sandra Loosemore <sandra@codesourcery.com>
788 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
789 * config/nios2/nios2.c (nios2_function_profiler): Add
790 -fPIC (flag_pic == 2) support.
791 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
792 (nios2_large_offset_p): New function.
793 (nios2_unspec_reloc_p): Move up position, update to use
794 nios2_large_offset_p.
795 (nios2_unspec_address): Remove function.
796 (nios2_unspec_offset): New function.
797 (nios2_large_got_address): New function.
798 (nios2_got_address): Add large offset support.
799 (nios2_legitimize_tls_address): Update usage of removed and new
801 (nios2_symbol_binds_local_p): New function.
802 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
803 (nios2_legitimize_address): Update to use nios2_large_offset_p.
804 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
805 (nios2_print_operand): Merge H/L processing, add hiadj/lo
806 processing for (const (unspec ...)).
807 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
809 2014-02-20 Richard Biener <rguenther@suse.de>
811 * tree-cfg.c (replace_uses_by): Mark altered BBs before
812 doing the substitution.
813 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
815 2014-02-20 Martin Jambor <mjambor@suse.cz>
818 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
819 info when checking whether lattices are bottom.
821 2014-02-20 Richard Biener <rguenther@suse.de>
824 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
827 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
830 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
831 parameter specifying the scaling.
832 (inline_call): Update.
833 (want_inline_recursively): Guard division by zero.
834 (recursive_inlining): Update.
835 * ipa-inline.h (clone_inlined_nodes): Update.
837 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
840 * config/i386/i386.c (classify_argument): Pass structures of size
841 64 bytes or less in register.
843 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
844 Kirill Yukhin <kirill.yukhin@intel.com>
846 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
847 (_mm_rcp28_round_ss): Ditto.
848 (_mm_rsqrt28_round_sd): Ditto.
849 (_mm_rsqrt28_round_ss): Ditto.
850 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
851 (_mm_rcp14_round_ss): Ditto.
852 (_mm_rsqrt14_round_sd): Ditto.
853 (_mm_rsqrt14_round_ss): Ditto.
854 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
855 the first input operand, get rid of match_dup.
856 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
858 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
860 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
861 operand as the first input operand, set type attribute.
862 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
864 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
865 operand as the first input operand, set type attribute.
867 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
869 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
872 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
875 * config/i386/i386.c (construct_container): Remove TFmode check
876 for X86_64_INTEGER_CLASS.
878 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
881 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
882 only when -Wpsabi is enabled.
884 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
887 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
888 passing arrays in registers are the same as for structs, so remove the
889 special case for them.
891 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
893 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
894 destination type, extract only the valid bits if the source type is not
895 integral and has a different mode.
897 2014-02-19 Richard Biener <rguenther@suse.de>
900 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
903 2014-02-19 Richard Biener <rguenther@suse.de>
906 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
907 (ipa_modify_call_arguments): Emit an argument load explicitely and
908 preserve virtual SSA form there and for the replacement call.
909 Do not update SSA form nor free dominance info.
911 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
913 * ipa.c (function_and_variable_visibility): Also clear WEAK
914 flag when disolving COMDAT_GROUP.
916 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
918 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
919 * ipa-prop.c (ipa_set_jf_known_type): Return early when
921 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
922 do more sanity checks.
923 (detect_type_change): Return true when giving up early.
924 (compute_complex_assign_jump_func): Fix type parameter of
926 (compute_complex_ancestor_jump_func): Likewise.
927 (update_jump_functions_after_inlining): Fix updating of
929 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
931 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
933 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
934 inline clones when edge disappears.
936 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
939 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
940 Split 64-bit moves into 2 patterns. Do not allow the use of
941 direct move for TDmode in little endian, since the decimal value
942 has little endian bytes within a word, but the 64-bit pieces are
943 ordered in a big endian fashion, and normal subreg's of TDmode are
945 (mov<mode>_64bit_dm): Likewise.
946 (movtd_64bit_nodm): Likewise.
948 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
950 PR tree-optimization/60174
951 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
952 statement of an SSA_NAME that occurs in an abnormal PHI node.
954 2014-02-18 Jakub Jelinek <jakub@redhat.com>
957 * final.c (SEEN_BB): Remove.
958 (SEEN_NOTE, SEEN_EMITTED): Renumber.
959 (final_scan_insn): Don't force_source_line on second
960 NOTE_INSN_BASIC_BLOCK.
962 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
965 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
966 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
967 (type_natural_mode): Warn ABI change when %zmm register is not
968 available for AVX512F vector value passing.
970 2014-02-18 Kai Tietz <ktietz@redhat.com>
973 * config/i386/i386.c (ix86_expand_prologue): Use value in
974 rax register as displacement when restoring %r10 or %rax.
975 Fix wrong offset when restoring both registers.
977 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
979 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
980 assertion with conditional return.
982 2014-02-18 Jakub Jelinek <jakub@redhat.com>
983 Uros Bizjak <ubizjak@gmail.com>
986 * config/i386/driver-i386.c (host_detect_local_cpu): If
987 YMM state is not saved by the OS, also clear has_f16c. Move
988 CPUID 0x80000001 handling before YMM state saving checking.
990 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
992 PR rtl-optimization/58960
993 * haifa-sched.c (alloc_global_sched_pressure_data): New,
994 factored out from ...
995 (sched_init): ... here.
996 (free_global_sched_pressure_data): New, factored out from ...
997 (sched_finish): ... here.
998 * sched-int.h (free_global_sched_pressure_data): Declare.
999 * sched-rgn.c (nr_regions_initial): New static global.
1000 (haifa_find_rgns): Initialize it.
1001 (schedule_region): Disable sched-pressure for the newly
1004 2014-02-17 Richard Biener <rguenther@suse.de>
1006 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1007 release SSA defs of pattern stmts.
1009 2014-02-17 Richard Biener <rguenther@suse.de>
1011 * tree-inline.c (expand_call_inline): Release the virtual
1012 operand defined by the call we are about to inline.
1014 2014-02-17 Richard Biener <rguenther@suse.de>
1016 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1018 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1019 Ilya Tocar <ilya.tocar@intel.com>
1021 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1022 arguments order in builtin.
1023 (_mm512_permutexvar_epi64): Ditto.
1024 (_mm512_mask_permutexvar_epi64): Ditto
1025 (_mm512_maskz_permutexvar_epi32): Ditto
1026 (_mm512_permutexvar_epi32): Ditto
1027 (_mm512_mask_permutexvar_epi32): Ditto
1029 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1031 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1032 (p8_vmrgow): Likewise.
1034 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1036 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1039 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1042 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1043 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1044 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1045 using direct move instructions on ISA 2.07. Also adjust
1046 instruction length for 64-bit.
1047 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1048 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1050 2014-02-15 Alan Modra <amodra@gmail.com>
1054 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1055 find_replacement on parts of insn rtl that might be reloaded.
1057 2014-02-15 Richard Biener <rguenther@suse.de>
1059 PR tree-optimization/60183
1060 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1061 (tree_ssa_phiprop): Calculate and free post-dominators.
1063 2014-02-14 Jeff Law <law@redhat.com>
1065 PR rtl-optimization/60131
1066 * ree.c (get_extended_src_reg): New function.
1067 (combine_reaching_defs): Use it rather than assuming location of REG.
1068 (find_and_remove_re): Verify first operand of extension is
1069 a REG before adding the insns to the copy list.
1071 2014-02-14 Roland McGrath <mcgrathr@google.com>
1073 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1074 * configure: Regenerated.
1075 * config.in: Regenerated.
1076 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1077 instead of ASM_SHORT.
1079 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1080 Richard Earnshaw <rearnsha@arm.com>
1082 PR rtl-optimization/59535
1083 * lra-constraints.c (process_alt_operands): Encourage alternative
1084 when unassigned pseudo class is superset of the alternative class.
1085 (inherit_reload_reg): Don't inherit when optimizing for code size.
1086 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1087 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1088 modes not less than 4 for Thumb1.
1090 2014-02-14 Kyle McMartin <kyle@redhat.com>
1093 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1095 2014-02-14 Richard Biener <rguenther@suse.de>
1097 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1098 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1099 Do not strip INDIRECT_REFs.
1101 2014-02-14 Richard Biener <rguenther@suse.de>
1104 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1105 DECL_FUNCTION_SPECIFIC_TARGET.
1106 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1107 * tree-streamer-out.c (pack_ts_target_option): Remove.
1108 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1109 (write_ts_function_decl_tree_pointers): Do not stream
1110 DECL_FUNCTION_SPECIFIC_TARGET.
1111 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1112 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1113 (lto_input_ts_function_decl_tree_pointers): Do not stream
1114 DECL_FUNCTION_SPECIFIC_TARGET.
1116 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1118 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1119 (get_initial_def_for_induction, vectorizable_induction): Ignore
1120 debug stmts when looking for exit_phi.
1121 (vectorizable_live_operation): Fix up condition.
1123 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1125 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1126 nreverse() because it changes the content of original tree list.
1128 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1130 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1131 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1133 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1135 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1136 GNU coding standards.
1138 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1141 * dwarf2out.c (gen_subprogram_die): Don't call
1142 add_calling_convention_attribute if subr_die is old_die.
1144 2014-02-13 Sharad Singhai <singhai@google.com>
1146 * doc/optinfo.texi: Fix order of nodes.
1148 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1150 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1151 operands[2], not operands[3].
1153 2014-02-13 Richard Biener <rguenther@suse.de>
1156 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1157 mention the possibility of an in-tree build.
1158 (CLooG): Update recommended version to 0.18.1, mention the
1159 possibility of an in-tree build and clarify that the ISL
1160 bundled with CLooG does not work.
1162 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1165 * expr.c (compress_float_constant): If x is a hard register,
1166 extend into a pseudo and then move to x.
1168 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1170 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1171 caused by bad second argument to warning_at() with -mhotpatch and
1172 nested functions (e.g. with gfortran).
1174 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1176 * opts.c (option_name): Remove "enabled by default" rider.
1178 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1180 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1182 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1183 Uros Bizjak <ubizjak@gmail.com>
1186 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1187 * configure: Regenerated.
1189 2014-02-12 Richard Biener <rguenther@suse.de>
1191 * vec.c (vec_prefix::calculate_allocation): Move as
1192 inline variant to vec.h.
1193 (vec_prefix::calculate_allocation_1): New out-of-line version.
1194 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1195 (vec_prefix::m_has_auto_buf): Rename to ...
1196 (vec_prefix::m_using_auto_storage): ... this.
1197 (vec_prefix::calculate_allocation): Inline the easy cases
1198 and dispatch to calculate_allocation_1 which doesn't need the
1200 (va_heap::reserve): Use gcc_checking_assert.
1201 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1202 m_using_auto_storage.
1203 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1205 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1206 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1207 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1209 2014-02-12 Richard Biener <rguenther@suse.de>
1211 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1212 when we found a dependence.
1214 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1216 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1218 (maybe_fold_stmt): ... into this new function.
1219 * omp-low.c (lower_omp): Update comment.
1221 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1224 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1227 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1229 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1230 identifiers in comments.
1231 (cortexa53_extra_costs): Likewise.
1232 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1233 (cortexa7_extra_costs): Likewise.
1234 (cortexa12_extra_costs): Likewise.
1235 (cortexa15_extra_costs): Likewise.
1236 (v7m_extra_costs): Likewise.
1238 2014-02-12 Richard Biener <rguenther@suse.de>
1241 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1242 of posix_memalign being successful.
1243 (lower_stmt): Restrict lowering of posix_memalign to when
1244 -ftree-bit-ccp is enabled.
1246 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1248 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1250 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1252 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1254 PR rtl-optimization/60116
1255 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1256 other_insn once the combination has been validated.
1258 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1261 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1263 * ipa-devirt.c: Include demangle.h
1264 (odr_violation_reported): New static variable.
1265 (add_type_duplicate): Update odr_violations.
1266 (maybe_record_node): Add completep parameter; update it.
1267 (record_target_from_binfo): Add COMPLETEP parameter;
1268 update it as needed.
1269 (possible_polymorphic_call_targets_1): Likewise.
1270 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1271 rename FINAL to COMPLETE.
1272 (record_targets_from_bases): Sanity check we found the binfo;
1273 fix COMPLETEP updating.
1274 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1275 parameter, fix computing of COMPLETEP.
1276 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1277 at LTO time do demangling.
1278 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1279 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1281 (gimple_get_virt_method_for_binfo): Likewise.
1282 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1283 gimple_get_virt_method_for_vtable): Update prototypes.
1285 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1288 * genautomata.c (add_presence_absence): Fix typo with
1289 {final_}presence_list.
1291 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1294 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1295 for VSX/Altivec vectors that land in GPR registers.
1297 2014-02-11 Richard Henderson <rth@redhat.com>
1298 Jakub Jelinek <jakub@redhat.com>
1301 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1302 around drhs if type conversion to lacc->type is not useless.
1304 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1306 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1308 (cortex-a57.cortex-a53): Likewise.
1309 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1311 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1313 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1316 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1318 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1319 add_options_for_arm_vfp3.
1321 2014-02-11 Jeff Law <law@redhat.com>
1324 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1325 object with an undesirable mode.
1327 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1330 * config/i386/sol2-9.h: New file.
1331 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1332 *-*-solaris2.9*): Use it.
1334 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1336 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1337 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1339 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1341 * config/microblaze/microblaze.c: Extend mcpu version format
1343 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1345 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1347 2014-02-10 Richard Henderson <rth@redhat.com>
1350 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1351 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1352 ms-abi vs -mno-accumulate-outgoing-args.
1353 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1354 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1357 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1360 * cfgexpand.c (expand_asm_operands): Attach source location to
1361 ASM_INPUT rtx objects.
1362 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1364 2014-02-10 Nick Clifton <nickc@redhat.com>
1366 * config/mn10300/mn10300.c (popcount): New function.
1367 (mn10300_expand_prologue): Include saved registers in stack usage
1370 2014-02-10 Jeff Law <law@redhat.com>
1373 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1374 when changing the SET_DEST of a prior insn to avoid an input reload.
1376 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1378 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1379 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1380 -mcall-openbsd, or -mcall-linux.
1381 (CC1_ENDIAN_BIG_SPEC): Remove.
1382 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1383 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1384 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1385 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1386 and %cc1_endian_default.
1387 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1389 2014-02-10 Richard Biener <rguenther@suse.de>
1391 PR tree-optimization/60115
1392 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1393 MEM_REF handling. Properly verify that the accesses are not
1394 out of the objects bound.
1396 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1398 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1401 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1403 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1404 proper constants and fix formatting.
1405 (possible_polymorphic_call_targets): Fix formatting.
1407 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1408 Ilya Tocar <ilya.tocar@intel.com>
1410 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1411 (_mm512_loadu_epi32): Renamed into...
1412 (_mm512_loadu_si512): This.
1413 (_mm512_storeu_epi32): Renamed into...
1414 (_mm512_storeu_si512): This.
1415 (_mm512_maskz_ceil_ps): Removed.
1416 (_mm512_maskz_ceil_pd): Ditto.
1417 (_mm512_maskz_floor_ps): Ditto.
1418 (_mm512_maskz_floor_pd): Ditto.
1419 (_mm512_floor_round_ps): Ditto.
1420 (_mm512_floor_round_pd): Ditto.
1421 (_mm512_ceil_round_ps): Ditto.
1422 (_mm512_ceil_round_pd): Ditto.
1423 (_mm512_mask_floor_round_ps): Ditto.
1424 (_mm512_mask_floor_round_pd): Ditto.
1425 (_mm512_mask_ceil_round_ps): Ditto.
1426 (_mm512_mask_ceil_round_pd): Ditto.
1427 (_mm512_maskz_floor_round_ps): Ditto.
1428 (_mm512_maskz_floor_round_pd): Ditto.
1429 (_mm512_maskz_ceil_round_ps): Ditto.
1430 (_mm512_maskz_ceil_round_pd): Ditto.
1431 (_mm512_expand_pd): Ditto.
1432 (_mm512_expand_ps): Ditto.
1433 * config/i386/i386.c (ix86_builtins): Remove
1434 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1435 (bdesc_args): Ditto.
1436 * config/i386/predicates.md (const1256_operand): New.
1437 (const_1_to_2_operand): Ditto.
1438 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1439 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1440 (*avx512pf_gatherpf<mode>sf): Ditto.
1441 (avx512pf_gatherpf<mode>df): Ditto.
1442 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1443 (*avx512pf_gatherpf<mode>df): Ditto.
1444 (avx512pf_scatterpf<mode>sf): Ditto.
1445 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1446 (*avx512pf_scatterpf<mode>sf): Ditto.
1447 (avx512pf_scatterpf<mode>df): Ditto.
1448 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1449 (*avx512pf_scatterpf<mode>df): Ditto.
1450 (avx512f_expand<mode>): Removed.
1451 (<shift_insn><mode>3<mask_name>): Change predicate type.
1453 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1455 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1456 not at the end of datarefs vector use ordered_remove to avoid
1457 reordering datarefs vector.
1460 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1461 mark local addressable non-static vars as GOVD_PRIVATE
1462 instead of GOVD_LOCAL.
1463 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1464 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1468 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1469 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1470 assume_aligned or alloc_align attributes.
1471 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1472 arguments. Handle also assume_aligned and alloc_align attributes.
1473 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1474 calls to functions with assume_aligned or alloc_align attributes.
1475 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1477 2014-02-08 Terry Guo <terry.guo@arm.com>
1479 * doc/invoke.texi: Document ARM -march=armv7e-m.
1481 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1483 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1484 flag on __cilkrts_rethrow builtin.
1487 * ipa-cp.c (determine_versionability): Fail at -O0
1488 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1489 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1492 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1495 * tree-inline.c (copy_forbidden): Fail for
1496 __attribute__((optimize (0))) functions.
1498 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1500 * varpool.c: Include pointer-set.h.
1501 (varpool_remove_unreferenced_decls): Variables in other partitions
1502 will not be output; be however careful to not lose information
1505 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1507 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1508 lookup in the vtable constructor.
1510 2014-02-07 Jeff Law <law@redhat.com>
1513 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1514 define_insn_and_split.
1516 * ipa-inline.c (inline_small_functions): Fix typos.
1518 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1520 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1521 (s390_can_use_return_insn): Declare.
1522 * config/s390/s390.h (EPILOGUE_USES): Define.
1523 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1525 (s390_chunkify_start): Handle return JUMP_LABELs.
1526 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1527 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1528 (s390_can_use_return_insn): New functions.
1529 (s390_fix_long_loop_prediction): Handle conditional returns.
1530 (TARGET_SET_UP_BY_PROLOGUE): Define.
1531 * config/s390/s390.md (ANY_RETURN): New code iterator.
1532 (*creturn, *csimple_return, return, simple_return): New patterns.
1534 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1536 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1537 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1538 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1539 REG_CFA_RESTORE list when deciding not to restore a register.
1541 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1543 * config/s390/s390.c: Include tree-pass.h and context.h.
1544 (s390_early_mach): New function, split out from...
1545 (s390_emit_prologue): ...here.
1546 (pass_data_s390_early_mach): New pass structure.
1547 (pass_s390_early_mach): New class.
1548 (s390_option_override): Create and register early_mach pass.
1549 Move to end of file.
1551 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1553 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1554 to match for the exit block.
1556 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1558 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1559 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1560 Reject misaligned operands.
1562 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1564 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1566 2014-02-07 Richard Biener <rguenther@suse.de>
1569 * gimple-low.c (lower_builtin_posix_memalign): New function.
1570 (lower_stmt): Call it to lower posix_memalign in a way
1571 to make alignment info accessible.
1573 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1576 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1577 __builtin_setjmp_receiver.
1579 2014-02-07 Richard Biener <rguenther@suse.de>
1582 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1583 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1584 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1585 Handle BUILT_IN_POSIX_MEMALIGN.
1586 (find_func_clobbers): Likewise.
1587 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1588 (call_may_clobber_ref_p_1): Likewise.
1590 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1593 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1596 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1599 * lto-cgraph.c (lto_output_node): Use
1600 symtab_get_symbol_partitioning_class.
1601 (lto_output_varpool_node): likewise.
1602 (symtab_get_symbol_partitioning_class): Move here from
1604 * cgraph.h (symbol_partitioning_class): Likewise.
1605 (symtab_get_symbol_partitioning_class): Declare.
1607 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1609 * ggc.h (ggc_internal_cleared_alloc): New macro.
1610 * vec.h (vec_safe_copy): Handle memory stats.
1611 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1612 * target-globals.c (save_target_globals): Likewise.
1614 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1617 * expr.c (emit_move_resolve_push): Export; be bit more selective
1618 on when to clear alias set.
1619 * expr.h (emit_move_resolve_push): Declare.
1620 * function.h (struct function): Add tail_call_marked.
1621 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1622 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1623 * config/i386/i386.md (TImode move expander): De not call
1625 (FP push expanders): Preserve memory attributes.
1626 * config/i386/sse.md (push<mode>1): Remove.
1627 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1628 (ix86_expand_push): Remove.
1629 * config/i386/mmx.md (push<mode>1): Remove.
1631 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1633 PR rtl-optimization/60030
1634 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1635 lopart with paradoxical subreg before shifting it up by hprec.
1637 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1639 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1640 Remove extra newline at end of file.
1641 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1642 (arm_issue_rate): Handle cortexa57.
1643 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1644 (cortex-a57.cortex-a53): Likewise.
1646 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1649 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1650 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1652 (arm_expand_prologue): Adjust all callers.
1653 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1654 info, registers also at the lowest numbered registers side. Use
1655 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1659 * var-tracking.c (adjust_mems): Before adding a SET to
1660 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1662 2014-02-06 Alan Modra <amodra@gmail.com>
1665 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1666 change SDmode to DDmode when lra_in_progress.
1668 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1671 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1672 free_data_ref on the dr first, and before goto again also set dr
1673 to the next dr. For simd_lane_access, free old datarefs[i] before
1674 overwriting it. For get_vectype_for_scalar_type failure, don't
1675 free_data_ref if simd_lane_access.
1677 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1680 * tree.h (opts_for_fn): New inline function.
1681 (opt_for_fn): Define.
1682 * config/i386/i386.c (ix86_function_regparm): Use
1683 opt_for_fn (decl, optimize) instead of optimize.
1685 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1687 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1688 for SYMBOL_REF in large memory model.
1690 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1692 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1694 (cortex-a57): Likewise.
1695 (cortex-a57.cortex-a53): Likewise.
1697 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1698 Kugan Vivekanandarajah <kuganv@linaro.org>
1700 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1702 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1704 2014-02-06 Richard Biener <rguenther@suse.de>
1706 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1707 set_loop_copy and initialize_original_copy_tables.
1709 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1711 * config/aarch64/aarch64-simd.md
1712 (aarch64_ashr_simddi): Change QI to SI.
1714 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1715 Jakub Jelinek <jakub@redhat.com>
1718 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1721 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1723 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1724 CODE_FOR_altivec_vpku[hw]um to
1725 CODE_FOR_altivec_vpku[hw]um_direct.
1726 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1727 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1728 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1729 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1731 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1733 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1734 generation for -maltivec=be.
1735 (altivec_vsumsws): Simplify redundant test.
1737 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1739 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1740 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1741 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1742 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1743 gen_altivec_vpkuwum.
1744 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1746 (altivec_vpks<VI_char>ss): Likewise.
1747 (altivec_vpks<VI_char>us): Likewise.
1748 (altivec_vpku<VI_char>us): Likewise.
1749 (altivec_vpku<VI_char>um): Likewise.
1750 (altivec_vpku<VI_char>um_direct): New (copy of
1751 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1753 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1754 target is little endian and -maltivec=be is not specified.
1755 (*altivec_vupkhs<VU_char>_direct): New (copy of
1756 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1757 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1758 target is little endian and -maltivec=be is not specified.
1759 (*altivec_vupkls<VU_char>_direct): New (copy of
1760 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1761 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1762 little endian and -maltivec=be is not specified.
1763 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1764 little endian and -maltivec=be is not specified.
1766 2014-02-05 Richard Henderson <rth@redhat.com>
1769 * combine-stack-adj.c: Revert r206943.
1770 * sched-int.h (struct deps_desc): Add last_args_size.
1771 * sched-deps.c (init_deps): Initialize it.
1772 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1773 contain REG_ARGS_SIZE notes.
1775 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1777 * lto-cgraph.c (asm_nodes_output): Make global.
1778 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1779 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1780 (driver_handle_option): Handle OPT_fwpa.
1782 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1785 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1786 a comment typo and formatting issue. If odr_hash hasn't been
1787 created, return vNULL and set *completep to false.
1790 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1791 bb with no successors.
1793 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1796 * doc/invoke.texi (-march): Clarify documentation for ARM.
1800 2014-02-05 Richard Biener <rguenther@suse.de>
1802 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1803 when not vectorizing because of too many alias checks.
1804 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1805 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1807 2014-02-05 Nick Clifton <nickc@redhat.com>
1809 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1810 accept extended registers in any mode when compiling for the MN10300.
1812 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1814 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1815 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1816 sanitization attributes.
1817 (can_inline_edge_p): Likewise.
1818 (sanitize_attrs_match_for_inline_p): New function.
1820 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1822 * ipa-prop.c (detect_type_change): Shor circuit testing of
1823 type changes on THIS pointer.
1825 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1828 * config/pa/pa.c (legitimize_tls_address): Return original address
1829 if not passed a SYMBOL_REF rtx.
1830 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1832 (pa_emit_move_sequence): Simplify TLS source operands.
1833 (pa_legitimate_constant_p): Reject all TLS constants.
1834 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1835 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1837 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1839 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1840 groups when we know they are controlled by LTO.
1841 * varasm.c (default_binds_local_p_1): If object is in other partition,
1842 it will be resolved locally.
1844 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1846 * config/host-linux.c (linux_gt_pch_use_address): Don't
1847 use SSIZE_MAX because it is not always defined.
1849 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1852 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1853 threshold for pseudo splitting.
1854 (update_ebb_live_info): Process call argument hard registers and
1855 hard registers from insn definition too.
1856 (max_small_class_regs_num): New constant.
1857 (inherit_in_ebb): Update live hard regs through EBBs. Update
1858 reloads_num only for small register classes. Don't split for
1861 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1864 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1867 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1869 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1872 2014-02-04 Marek Polacek <polacek@redhat.com>
1874 * gdbinit.in (pel): Define.
1876 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1878 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1881 2014-02-04 Richard Biener <rguenther@suse.de>
1884 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1885 in function context local.
1886 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1887 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1888 similar to LTO_imported_decl_ref.
1890 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1892 PR tree-optimization/60002
1893 * cgraphclones.c (build_function_decl_skip_args): Clear
1896 PR tree-optimization/60023
1897 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1898 false to gsi_replace.
1899 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1900 has been in some EH region and vec_stmt could throw, add
1901 vec_stmt into the same EH region.
1902 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1903 has no lhs, ignore it.
1904 * internal-fn.c (expand_MASK_LOAD): Likewise.
1907 * tree-inline.c (copy_forbidden): Fail for
1908 __attribute__((optimize (0))) functions.
1911 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1912 is set, copy one less argument.
1913 (expand_simd_clones): Don't subtract clone_info->inbranch
1914 from simd_clone_struct_alloc argument.
1916 PR rtl-optimization/57915
1917 * recog.c (simplify_while_replacing): If all unary/binary/relational
1918 operation arguments are constant, attempt to simplify those.
1921 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1922 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1924 2014-02-04 Richard Biener <rguenther@suse.de>
1926 PR tree-optimization/60012
1927 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1928 TBAA disambiguation to all DDRs.
1930 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1933 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1934 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1936 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1939 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1941 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1943 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1944 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1946 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1949 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1950 to figure out targets of polymorphic calls with known decl.
1951 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1952 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1953 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1954 (get_polymorphic_call_info): ... here.
1955 (get_polymorphic_call_info_from_invariant): New function.
1957 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1959 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1960 lookup via vtable pointer; check for type consistency
1961 and turn inconsitent facts into UNREACHABLE.
1962 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1963 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1964 type inconsistent querries; return UNREACHABLE instead.
1966 2014-02-03 Richard Henderson <rth@twiddle.net>
1969 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1970 already processed this node.
1971 (normalize_one_pred_1): Pass along mark_set.
1972 (normalize_one_pred): Create and destroy a pointer_set_t.
1973 (normalize_one_pred_chain): Likewise.
1975 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1977 PR gcov-profile/58602
1978 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1980 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1983 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1984 -fno-devirtualize; try to devirtualize by the knowledge of
1985 virtual table pointer given by aggregate propagation.
1986 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1987 (ipa_print_node_jump_functions): Dump also offset that
1988 is relevant for polymorphic calls.
1989 (determine_known_aggregate_parts): Add arg_type parameter; use it
1990 instead of determining the type from pointer type.
1991 (ipa_compute_jump_functions_for_edge): Update call of
1992 determine_known_aggregate_parts.
1993 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1994 (gimple_get_virt_method_for_binfo): ... here; simplify using
1995 vtable_pointer_value_to_vtable.
1996 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1997 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1998 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1999 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2001 (vtable_pointer_value_to_binfo): ... here.
2002 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2004 2014-02-03 Teresa Johnson <tejohnson@google.com>
2006 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2007 redef of outer loop index variable.
2009 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2013 * doc/extend.texi (Function Attributes): Typo.
2015 2014-02-03 Cong Hou <congh@google.com>
2017 PR tree-optimization/60000
2018 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2019 if the vectorized statement is a store. A store statement can only
2020 appear at the end of pattern statements.
2022 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2024 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2025 (ix86_option_override_internal): Default long double to 64-bit for
2026 32-bit Bionic and to 128-bit for 64-bit Bionic.
2028 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2029 TARGET_LONG_DOUBLE_128 is true.
2030 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2032 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2033 (mlong-double-64): Negate -mlong-double-128.
2034 (mlong-double-128): New option.
2036 * config/i386/i386-c.c (ix86_target_macros): Define
2037 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2039 * doc/invoke.texi: Document -mlong-double-128.
2041 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2043 PR rtl-optimization/60024
2044 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2046 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2048 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2050 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2052 PR rtl-optimization/57662
2053 * sel-sched.c (code_motion_path_driver): Do not mark already not
2054 existing blocks in the visiting bitmap.
2056 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2058 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2059 on the insn being emitted.
2061 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2062 Will Deacon <will.deacon@arm.com>
2064 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2066 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2068 * config/arm/arm-tables.opt: Regenerate.
2070 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2072 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2073 for vector types other than V16QImode.
2074 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2075 define_expand, and call altivec_expand_vec_perm_le when producing
2076 code with little endian element order.
2077 (*altivec_vperm_<mode>_internal): New insn having previous
2078 behavior of altivec_vperm_<mode>.
2079 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2080 altivec_expand_vec_perm_le when producing code with little endian
2082 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2083 behavior of altivec_vperm_<mode>_uns.
2085 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2087 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2088 (altivec_vsumsws): Add handling for -maltivec=be with a little
2090 (altivec_vsumsws_direct): New.
2091 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2092 gen_altivec_vsumsws.
2094 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2096 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2097 vtable_pointer_value_to_binfo): New functions.
2098 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2099 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2101 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2103 * config/nios2/nios2.md (load_got_register): Initialize GOT
2104 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2105 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2107 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2109 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2110 preserverd by passthrough, do not propagate the type.
2112 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2114 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2115 (mips_atomic_assign_expand_fenv): New function.
2116 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2118 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2120 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2121 (__builtin_mips_set_fcsr): Likewise.
2122 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2123 MIPS_USI_FTYPE_VOID.
2124 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2125 (mips16_expand_set_fcsr): Likewise.
2126 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2127 (mips16_set_fcsr_stub): Likewise.
2128 (mips16_get_fcsr_one_only_stub): New class.
2129 (mips16_set_fcsr_one_only_stub): Likewise.
2130 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2131 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2132 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2133 (hard_float): New availability predicate.
2134 (mips_builtins): Add get_fcsr and set_fcsr.
2135 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2136 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2137 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2138 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2139 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2142 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2144 * config/mips/mips.c (mips_one_only_stub): New class.
2145 (mips_need_mips16_rdhwr_p): Replace with...
2146 (mips16_rdhwr_stub): ...this new variable.
2147 (mips16_stub_call_address): New function.
2148 (mips16_rdhwr_one_only_stub): New class.
2149 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2150 (mips_output_mips16_rdhwr): Delete.
2151 (mips_finish_stub): New function.
2152 (mips_code_end): Use it to handle rdhwr stubs.
2154 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2157 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2158 when calculating size of integer atomic types.
2160 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2162 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2164 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2166 PR tree-optimization/60003
2167 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2168 * profile.c (branch_prob): Use gimple_call_builtin_p
2169 to check for BUILT_IN_SETJMP_RECEIVER.
2170 * tree-inline.c (copy_bb): Call notice_special_calls.
2172 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2175 * lra-constraints.c (process_alt_operands): Update reload_sum only
2178 2014-01-31 Richard Henderson <rth@redhat.com>
2181 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2182 until after else_eh is processed.
2184 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2186 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2187 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2188 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2189 in smmintrin.h, remove them.
2190 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2191 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2192 * config/i386/i386.md (ROUND_SAE): Fix value.
2193 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2194 (const48_operand): New.
2195 * config/i386/subst.md (round), (round_expand): Use
2196 const_4_or_8_to_11_operand.
2197 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2199 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2201 * config/i386/constraints.md (Yk): Swap meaning with k.
2202 * config/i386/i386.md (movhi_internal): Change Yk to k.
2203 (movqi_internal): Ditto.
2204 (*k<logic><mode>): Ditto.
2207 (kandn<mode>): Ditto.
2208 (*<code>hi_1): Ditto.
2209 (*<code>qi_1): Ditto.
2210 (kxnor<mode>): Ditto.
2211 (kortestzhi): Ditto.
2212 (kortestchi): Ditto.
2214 (*one_cmplhi2_1): Ditto.
2215 (*one_cmplqi2_1): Ditto.
2216 * config/i386/sse.md (): Change k to Yk.
2217 (avx512f_load<mode>_mask): Ditto.
2218 (avx512f_blendm<mode>): Ditto.
2219 (avx512f_store<mode>_mask): Ditto.
2220 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2221 (avx512f_storedqu<mode>_mask): Ditto.
2222 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2224 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2225 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2226 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2227 (avx512f_maskcmp<mode>3): Ditto.
2228 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2229 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2230 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2231 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2232 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2233 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2234 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2235 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2236 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2237 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2238 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2239 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2240 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2241 (vec_extract_lo_<mode>_maskm): Ditto.
2242 (vec_extract_hi_<mode>_maskm): Ditto.
2243 (avx512f_vternlog<mode>_mask): Ditto.
2244 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2245 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2246 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2247 (avx512f_<code>v8div16qi2_mask): Ditto.
2248 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2249 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2250 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2251 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2252 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2253 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2254 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2255 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2256 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2257 (avx512cd_maskb_vec_dupv8di): Ditto.
2258 (avx512cd_maskw_vec_dupv16si): Ditto.
2259 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2260 (avx512f_vpermi2var<mode>3_mask): Ditto.
2261 (avx512f_vpermi2var<mode>3_mask): Ditto.
2262 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2263 (*avx512f_gathersi<mode>): Ditto.
2264 (*avx512f_gathersi<mode>_2): Ditto.
2265 (*avx512f_gatherdi<mode>): Ditto.
2266 (*avx512f_gatherdi<mode>_2): Ditto.
2267 (*avx512f_scattersi<mode>): Ditto.
2268 (*avx512f_scatterdi<mode>): Ditto.
2269 (avx512f_compress<mode>_mask): Ditto.
2270 (avx512f_compressstore<mode>_mask): Ditto.
2271 (avx512f_expand<mode>_mask): Ditto.
2272 * config/i386/subst.md (mask): Change k to Yk.
2273 (mask_scalar_merge): Ditto.
2276 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2278 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2280 2014-01-31 Richard Biener <rguenther@suse.de>
2283 * builtins.c (fold_builtin_memory_op): Make sure to not
2284 use a floating-point mode or a boolean or enumeral type for
2287 2014-01-30 DJ Delorie <dj@redhat.com>
2289 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2290 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2291 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2292 whenever main() has an epilogue.
2294 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2296 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2297 unused variable "field".
2298 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2299 (vsx_mergeh_<mode>): Likewise.
2300 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2301 (altivec_vmrghh): Likewise.
2302 (altivec_vmrghw): Likewise.
2303 (altivec_vmrglb): Likewise.
2304 (altivec_vmrglh): Likewise.
2305 (altivec_vmrglw): Likewise.
2306 (altivec_vspltb): Add missing uses.
2307 (altivec_vsplth): Likewise.
2308 (altivec_vspltw): Likewise.
2309 (altivec_vspltsf): Likewise.
2311 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2314 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2315 frame related instructions.
2317 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2319 PR rtl-optimization/59959
2320 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2321 any reload of register whose subreg is invalid.
2323 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2325 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2326 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2327 Add missing return type - void.
2329 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2331 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2332 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2333 remove element index adjustment for endian (now handled in vsx.md
2335 (altivec_expand_vec_perm_const): Use
2336 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2337 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2338 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2339 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2340 define_expand and a new define_insn *altivec_vspltb_internal;
2341 adjust for -maltivec=be on a little endian target.
2342 (altivec_vspltb_direct): New.
2343 (altivec_vsplth): Divide into a define_expand and a new
2344 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2345 little endian target.
2346 (altivec_vsplth_direct): New.
2347 (altivec_vspltw): Divide into a define_expand and a new
2348 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2349 little endian target.
2350 (altivec_vspltw_direct): New.
2351 (altivec_vspltsf): Divide into a define_expand and a new
2352 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2353 a little endian target.
2355 2014-01-30 Richard Biener <rguenther@suse.de>
2357 PR tree-optimization/59993
2358 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2359 can propagate form the earlier stmt and avoid the transform
2360 when the intermediate result is needed.
2362 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2364 * README.Portability: Fix typo.
2366 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2368 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2369 comparison_operator with ordered_comparison_operator.
2371 2014-01-30 Nick Clifton <nickc@redhat.com>
2373 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2374 Rename to mn10300_store_multiple_regs.
2375 * config/mn10300/mn10300.c: Likewise.
2376 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2377 store_multiple_regs.
2378 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2379 Call mn10300_store_multiple_regs.
2381 2014-01-30 Nick Clifton <nickc@redhat.com>
2382 DJ Delorie <dj@redhat.com>
2384 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2385 %fp 2 to keep registers after it properly word-aligned.
2386 (rl78_alloc_physical_registers_umul): Handle the case where both
2387 input operands are the same.
2389 2014-01-30 Richard Biener <rguenther@suse.de>
2391 PR tree-optimization/59903
2392 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2395 2014-01-30 Jason Merrill <jason@redhat.com>
2398 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2401 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2403 2014-01-30 Richard Biener <rguenther@suse.de>
2405 PR tree-optimization/59951
2406 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2408 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2411 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2412 SFmode to DFmode case.
2414 2014-01-29 DJ Delorie <dj@redhat.com>
2416 * config/msp430/msp430.opt (-minrt): New.
2417 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2419 (ENDFILE_SPEC): Likewise.
2421 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2423 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2424 (estimate_function_body_sizes): Use it.
2426 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2429 * dwarf2out.c (is_cxx_auto): New.
2430 (is_base_type): Use it.
2431 (gen_type_die_with_usage): Likewise.
2433 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2435 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2436 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2437 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2438 -maltivec=be with LE targets.
2439 (vsx_mergeh_<mode>): Likewise.
2440 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2441 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2442 (altivec_vmrghb): Replace with define_expand and new
2443 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2444 (altivec_vmrghb_direct): New define_insn.
2445 (altivec_vmrghh): Replace with define_expand and new
2446 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2447 (altivec_vmrghh_direct): New define_insn.
2448 (altivec_vmrghw): Replace with define_expand and new
2449 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2450 (altivec_vmrghw_direct): New define_insn.
2451 (*altivec_vmrghsf): Adjust for endianness.
2452 (altivec_vmrglb): Replace with define_expand and new
2453 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2454 (altivec_vmrglb_direct): New define_insn.
2455 (altivec_vmrglh): Replace with define_expand and new
2456 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2457 (altivec_vmrglh_direct): New define_insn.
2458 (altivec_vmrglw): Replace with define_expand and new
2459 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2460 (altivec_vmrglw_direct): New define_insn.
2461 (*altivec_vmrglsf): Adjust for endianness.
2462 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2463 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2464 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2465 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2466 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2467 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2468 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2469 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2471 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2473 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2474 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2477 2014-01-29 Richard Biener <rguenther@suse.de>
2479 PR tree-optimization/58742
2480 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2481 associate_pointerplus_align.
2482 (associate_pointerplus_diff): New function.
2483 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2484 and associate_pointerplus_diff.
2486 2014-01-29 Richard Biener <rguenther@suse.de>
2488 * lto-streamer.h (LTO_major_version): Bump to 3.
2489 (LTO_minor_version): Reset to 0.
2491 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2493 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2494 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2495 (arm_file_start): Generate correct asm header for armv7ve.
2496 * config/arm/bpabi.h: Add multilib support for armv7ve.
2497 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2498 and cortex-a15 to armv7ve.
2499 * config/arm/t-aprofile: Add multilib support for armv7ve.
2500 * doc/invoke.texi: Document -march=armv7ve.
2502 2014-01-29 Richard Biener <rguenther@suse.de>
2504 PR tree-optimization/58742
2505 * tree-ssa-forwprop.c (associate_plusminus): Return true
2506 if we changed sth, defer EH cleanup to ...
2507 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2508 (simplify_mult): New function.
2510 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2513 PR tree-optimization/59920
2514 * tree.c (build_common_builtin_nodes): Remove
2515 __builtin_setjmp_dispatcher initialization.
2516 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2517 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2518 instead of gsi_after_labels + manually skipping debug stmts.
2519 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2520 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2521 * tree-inline.c (copy_edges_for_bb): Remove
2522 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2523 argument. Ignore computed_goto_p stmts. Don't call
2524 make_abnormal_goto_edges. If a call might need abnormal edges
2525 for non-local gotos, see if it already has an edge to
2526 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2527 with true argument, don't do anything then, otherwise add
2528 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2529 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2531 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2532 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2533 (lower_stmt): Don't set data->calls_builtin_setjmp.
2534 (lower_builtin_setjmp): Adjust comment.
2535 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2536 * tree-cfg.c (found_computed_goto): Remove.
2537 (factor_computed_gotos): Remove.
2538 (make_goto_expr_edges): Return bool, true for computed gotos.
2539 Don't call make_abnormal_goto_edges.
2540 (build_gimple_cfg): Don't set found_computed_goto, don't call
2541 factor_computed_gotos.
2542 (computed_goto_p): No longer static.
2543 (make_blocks): Don't set found_computed_goto.
2544 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2545 (make_edges): If make_goto_expr_edges returns true, push bb
2546 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2547 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2548 vector. Record mapping between bbs and OpenMP regions if there
2549 are any, adjust make_gimple_omp_edges caller. Call
2550 handle_abnormal_edges.
2551 (make_abnormal_goto_edges): Remove.
2552 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2553 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2554 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2555 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2556 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2557 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2558 filling *region also set *region_idx to (*region)->entry->index.
2561 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2562 For REGs set ORIGINAL_REGNO.
2564 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2566 * doc/md.texi: Mention that a target shouldn't implement
2567 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2570 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2572 PR tree-optimization/59594
2573 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2574 a copy of the datarefs vector rather than the vector itself.
2576 2014-01-28 Jason Merrill <jason@redhat.com>
2579 * dwarf2out.c (auto_die): New static.
2580 (gen_type_die_with_usage): Handle C++1y 'auto'.
2581 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2584 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2587 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2588 (SPEC_X32): Likewise.
2589 (SPEC_64): Likewise.
2590 * config/i386/i386.c (ix86_option_override_internal): Turn off
2591 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2593 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2594 * config/i386/i386.h (TARGET_16BIT): New macro.
2595 (TARGET_16BIT_P): Likewise.
2596 * config/i386/i386.opt: Add m16.
2597 * doc/invoke.texi: Document -m16.
2599 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2601 PR preprocessor/59935
2602 * input.c (location_get_source_line): Bail out on when line number
2603 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2605 2014-01-28 Richard Biener <rguenther@suse.de>
2607 PR tree-optimization/58742
2608 * tree-ssa-forwprop.c (associate_plusminus): Handle
2609 pointer subtraction of the form (T)(P + A) - (T)P.
2611 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2613 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2616 2014-01-28 Richard Biener <rguenther@suse.de>
2619 2014-01-28 Richard Biener <rguenther@suse.de>
2621 PR rtl-optimization/45364
2622 PR rtl-optimization/59890
2623 * var-tracking.c (local_get_addr_clear_given_value): Handle
2624 already cleared slot.
2625 (val_reset): Handle not allocated local_get_addr_cache.
2626 (vt_find_locations): Use post-order on the inverted CFG.
2628 2014-01-28 Richard Biener <rguenther@suse.de>
2630 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2632 2014-01-28 Richard Biener <rguenther@suse.de>
2634 PR rtl-optimization/45364
2635 PR rtl-optimization/59890
2636 * var-tracking.c (local_get_addr_clear_given_value): Handle
2637 already cleared slot.
2638 (val_reset): Handle not allocated local_get_addr_cache.
2639 (vt_find_locations): Use post-order on the inverted CFG.
2641 2014-01-28 Alan Modra <amodra@gmail.com>
2643 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2644 * configure.ac <recursive call for build != host>: Define
2645 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2646 and LD_FOR_BUILD too.
2647 * configure: Regenerate.
2649 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2651 * config/i386/i386.c (get_builtin_code_for_version): Separate
2652 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2653 Broadwell from Haswell.
2655 2014-01-27 Steve Ellcey <sellcey@mips.com>
2657 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2658 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2659 * config/mips/mips.c (mips_option_override): Change setting
2661 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2662 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2663 Change from Mask to Var.
2665 2014-01-27 Jeff Law <law@redhat.com>
2667 * ipa-inline.c (inline_small_functions): Fix typo.
2669 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2671 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2672 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2673 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2674 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2675 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2676 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2677 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2678 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2679 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2680 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2681 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2682 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2683 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2684 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2685 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2686 (_mm512_storeu_epi64): Ditto.
2687 (_mm512_cmpge_epi32_mask): Ditto.
2688 (_mm512_cmpge_epu32_mask): Ditto.
2689 (_mm512_cmpge_epi64_mask): Ditto.
2690 (_mm512_cmpge_epu64_mask): Ditto.
2691 (_mm512_cmple_epi32_mask): Ditto.
2692 (_mm512_cmple_epu32_mask): Ditto.
2693 (_mm512_cmple_epi64_mask): Ditto.
2694 (_mm512_cmple_epu64_mask): Ditto.
2695 (_mm512_cmplt_epi32_mask): Ditto.
2696 (_mm512_cmplt_epu32_mask): Ditto.
2697 (_mm512_cmplt_epi64_mask): Ditto.
2698 (_mm512_cmplt_epu64_mask): Ditto.
2699 (_mm512_cmpneq_epi32_mask): Ditto.
2700 (_mm512_cmpneq_epu32_mask): Ditto.
2701 (_mm512_cmpneq_epi64_mask): Ditto.
2702 (_mm512_cmpneq_epu64_mask): Ditto.
2703 (_mm512_expand_pd): Ditto.
2704 (_mm512_expand_ps): Ditto.
2705 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2706 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2707 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2708 * config/i386/i386.c (ix86_builtins): Add
2709 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2710 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2711 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2712 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2713 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2714 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2715 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2716 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2717 IX86_BUILTIN_PMOVUSQW512_MEM.
2718 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2719 __builtin_ia32_pmovsqd512mem_mask,
2720 __builtin_ia32_pmovqd512mem_mask,
2721 __builtin_ia32_pmovusqw512mem_mask,
2722 __builtin_ia32_pmovsqw512mem_mask,
2723 __builtin_ia32_pmovqw512mem_mask,
2724 __builtin_ia32_pmovusdw512mem_mask,
2725 __builtin_ia32_pmovsdw512mem_mask,
2726 __builtin_ia32_pmovdw512mem_mask,
2727 __builtin_ia32_pmovqb512mem_mask,
2728 __builtin_ia32_pmovusqb512mem_mask,
2729 __builtin_ia32_pmovsqb512mem_mask,
2730 __builtin_ia32_pmovusdb512mem_mask,
2731 __builtin_ia32_pmovsdb512mem_mask,
2732 __builtin_ia32_pmovdb512mem_mask.
2733 (bdesc_args): Add __builtin_ia32_expanddf512,
2734 __builtin_ia32_expandsf512.
2735 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2736 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2737 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2738 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2739 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2740 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2741 (avx512f_<code>v8div16qi2_mask_store): This.
2742 (avx512f_expand<mode>): New.
2744 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2746 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2748 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2749 (_mm512_prefetch_i32scatter_pd): Ditto.
2750 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2751 (_mm512_prefetch_i64scatter_pd): Ditto.
2752 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2753 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2754 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2755 (_mm512_prefetch_i32scatter_ps): Ditto.
2756 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2757 (_mm512_prefetch_i64scatter_ps): Ditto.
2758 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2759 * config/i386/i386-builtin-types.def: Define
2760 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2761 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2762 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2763 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2764 IX86_BUILTIN_SCATTERPFQPD.
2765 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2766 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2767 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2768 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2769 __builtin_ia32_scatterpfqps.
2770 (ix86_expand_builtin): Expand new built-ins.
2771 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2772 fix memory access data type.
2773 (*avx512pf_gatherpf<mode>_mask): Ditto.
2774 (*avx512pf_gatherpf<mode>): Ditto.
2775 (avx512pf_scatterpf<mode>): Ditto.
2776 (*avx512pf_scatterpf<mode>_mask): Ditto.
2777 (*avx512pf_scatterpf<mode>): Ditto.
2778 (GATHER_SCATTER_SF_MEM_MODE): New.
2779 (avx512pf_gatherpf<mode>df): Ditto.
2780 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2781 (*avx512pf_scatterpf<mode>df): Ditto.
2783 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2786 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2787 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2790 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2792 * common/config/arm/arm-common.c
2793 (arm_rewrite_mcpu): Handle multiple names.
2795 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2797 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2799 * gimple-builder.h (create_gimple_tmp): Delete.
2801 2014-01-27 Christian Bruel <christian.bruel@st.com>
2803 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2806 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2808 * config/pa/pa.md (call): Generate indirect long calls to non-local
2809 functions when outputing 32-bit code.
2810 (call_value): Likewise except for special call to buggy powf function.
2812 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2813 portable runtime and PIC indirect calls.
2814 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2815 and PIC call sequences. Use ldo instead of blr to set return register
2816 in PIC call sequence.
2818 2014-01-25 Walter Lee <walt@tilera.com>
2820 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2821 avoid clobbering a live register.
2823 2014-01-25 Walter Lee <walt@tilera.com>
2825 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2826 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2827 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2828 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2830 2014-01-25 Walter Lee <walt@tilera.com>
2832 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2833 arguments on even registers.
2834 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2836 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2837 (BIGGEST_ALIGNMENT): Ditto.
2838 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2840 2014-01-25 Walter Lee <walt@tilera.com>
2842 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2843 insns before bundling.
2844 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2846 2014-01-25 Walter Lee <walt@tilera.com>
2848 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2849 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2850 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2852 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2854 * config/mips/constraints.md (kl): Delete.
2855 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2856 define expands, using...
2857 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2858 instructions for MIPS16.
2859 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2860 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2862 2014-01-25 Walter Lee <walt@tilera.com>
2864 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2868 2014-01-25 Walter Lee <walt@tilera.com>
2870 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2871 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2873 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2875 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2878 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2880 * print-rtl.c (in_call_function_usage): New var.
2881 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2882 EXPR_LIST mode as mode and not as reg note name.
2885 * cfgloopmanip.c (copy_loop_info): If
2886 loop->warned_aggressive_loop_optimizations, make sure
2887 the flag is set in target loop too.
2889 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2891 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2893 * builtins.def: Likewise.
2894 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2895 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2896 * ira.c (ira_setup_eliminable_regset): Likewise.
2897 * omp-low.c (gate_expand_omp): Likewise.
2898 (execute_lower_omp): Likewise.
2899 (diagnose_sb_0): Likewise.
2900 (gate_diagnose_omp_blocks): Likewise.
2901 (simd_clone_clauses_extract): Likewise.
2904 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2906 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2907 correction for little endian...
2908 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2911 2014-01-24 Jeff Law <law@redhat.com>
2913 PR tree-optimization/59919
2914 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2915 for non-returning calls.
2917 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2919 * common/config/aarch64/aarch64-common.c
2920 (aarch64_rewrite_mcpu): Handle multiple names.
2921 * config/aarch64/aarch64.h
2922 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2924 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2926 * input.c (add_file_to_cache_tab): Handle the case where fopen
2929 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2932 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2933 from push operand if code of push isn't PRE_DEC.
2935 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2938 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2939 -mquad-memory-atomic. Update -mquad-memory documentation to say
2940 it is only used for non-atomic loads/stores.
2942 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2943 -mquad-memory or -mquad-memory-atomic switches.
2945 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2946 -mquad-memory-atomic to ISA 2.07 support.
2948 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2949 to separate support of normal quad word memory operations (ldq, stq)
2950 from the atomic quad word memory operations.
2952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2953 support to separate non-atomic quad word operations from atomic
2954 quad word operations. Disable non-atomic quad word operations in
2955 little endian mode so that we don't have to swap words after the
2956 load and before the store.
2957 (quad_load_store_p): Add comment about atomic quad word support.
2958 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2959 options printed with -mdebug=reg.
2961 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2962 -mquad-memory-atomic as the test for whether we have quad word
2963 atomic instructions.
2964 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2965 or -mp8-vector are used, allow byte/half-word atomic operations.
2967 * config/rs6000/sync.md (load_lockedti): Insure that the address
2968 is a proper indexed or indirect address for the lqarx instruction.
2969 On little endian systems, swap the hi/lo registers after the lqarx
2971 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2972 insure the address is valid for the lqarx instruction.
2973 (store_conditionalti): Insure that the address is a proper indexed
2974 or indirect address for the stqcrx. instruction. On little endian
2975 systems, swap the hi/lo registers before doing the stqcrx.
2977 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2978 insure the address is valid for the stqcrx. instruction.
2980 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2981 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2982 type of quad memory support is available.
2984 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2987 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2988 there is a danger of looping.
2990 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2992 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2993 force flag_ira_loop_pressure if set via command line.
2995 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2997 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2998 (ashr_simd): New builtin handling DI mode.
2999 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3000 (aarch64_sshr_simddi): New match pattern.
3001 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3002 (vshrd_n_s64): Likewise.
3003 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3005 2014-01-23 Nick Clifton <nickc@redhat.com>
3007 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3008 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3009 favour of mcu specific scripts.
3010 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3013 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3014 Alex Velenko <Alex.Velenko@arm.com>
3016 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3017 (vaddv_s16): Likewise.
3018 (vaddv_s32): Likewise.
3019 (vaddv_u8): Likewise.
3020 (vaddv_u16): Likewise.
3021 (vaddv_u32): Likewise.
3022 (vaddvq_s8): Likewise.
3023 (vaddvq_s16): Likewise.
3024 (vaddvq_s32): Likewise.
3025 (vaddvq_s64): Likewise.
3026 (vaddvq_u8): Likewise.
3027 (vaddvq_u16): Likewise.
3028 (vaddvq_u32): Likewise.
3029 (vaddvq_u64): Likewise.
3030 (vaddv_f32): Likewise.
3031 (vaddvq_f32): Likewise.
3032 (vaddvq_f64): Likewise.
3033 (vmaxv_f32): Likewise.
3034 (vmaxv_s8): Likewise.
3035 (vmaxv_s16): Likewise.
3036 (vmaxv_s32): Likewise.
3037 (vmaxv_u8): Likewise.
3038 (vmaxv_u16): Likewise.
3039 (vmaxv_u32): Likewise.
3040 (vmaxvq_f32): Likewise.
3041 (vmaxvq_f64): Likewise.
3042 (vmaxvq_s8): Likewise.
3043 (vmaxvq_s16): Likewise.
3044 (vmaxvq_s32): Likewise.
3045 (vmaxvq_u8): Likewise.
3046 (vmaxvq_u16): Likewise.
3047 (vmaxvq_u32): Likewise.
3048 (vmaxnmv_f32): Likewise.
3049 (vmaxnmvq_f32): Likewise.
3050 (vmaxnmvq_f64): Likewise.
3051 (vminv_f32): Likewise.
3052 (vminv_s8): Likewise.
3053 (vminv_s16): Likewise.
3054 (vminv_s32): Likewise.
3055 (vminv_u8): Likewise.
3056 (vminv_u16): Likewise.
3057 (vminv_u32): Likewise.
3058 (vminvq_f32): Likewise.
3059 (vminvq_f64): Likewise.
3060 (vminvq_s8): Likewise.
3061 (vminvq_s16): Likewise.
3062 (vminvq_s32): Likewise.
3063 (vminvq_u8): Likewise.
3064 (vminvq_u16): Likewise.
3065 (vminvq_u32): Likewise.
3066 (vminnmv_f32): Likewise.
3067 (vminnmvq_f32): Likewise.
3068 (vminnmvq_f64): Likewise.
3070 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3072 * config/aarch64/aarch64-simd.md
3073 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3074 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3075 (*aarch64_mul3_elt<mode>): Likewise.
3076 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3077 (*aarch64_mul3_elt_to_64v2df): Likewise.
3078 (*aarch64_mla_elt<mode>): Likewise.
3079 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3080 (*aarch64_mls_elt<mode>): Likewise.
3081 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3082 (*aarch64_fma4_elt<mode>): Likewise.
3083 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3084 (*aarch64_fma4_elt_to_64v2df): Likewise.
3085 (*aarch64_fnma4_elt<mode>): Likewise.
3086 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3087 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3088 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3089 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3090 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3091 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3092 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3093 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3094 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3096 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3098 * config/aarch64/aarch64-simd.md
3099 (aarch64_be_checked_get_lane<mode>): New define_expand.
3100 * config/aarch64/aarch64-simd-builtins.def
3101 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3102 New builtin definition.
3103 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3104 Use new safe be builtin.
3106 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3108 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3110 (aarch64_be_st1<mode>): Likewise.
3111 (aarch_ld1<VALL:mode>): Define_expand modified.
3112 (aarch_st1<VALL:mode>): Likewise.
3113 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3114 (UNSPEC_ST1): Likewise.
3116 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3118 * config/microblaze/microblaze.md: Add trap insn and attribute
3120 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3122 PR preprocessor/58580
3123 * input.h (location_get_source_line): Take an additional line_size
3125 (void diagnostics_file_cache_fini): Declare new function.
3126 * input.c (struct fcache): New type.
3127 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3128 New static constants.
3129 (diagnostic_file_cache_init, total_lines_num)
3130 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3131 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3132 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3133 (get_next_line, read_next_line, goto_next_line, read_line_num):
3134 New static function definitions.
3135 (diagnostic_file_cache_fini): New function.
3136 (location_get_source_line): Take an additional output line_len
3137 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3139 * diagnostic.c (diagnostic_finish): Call
3140 diagnostic_file_cache_fini.
3141 (adjust_line): Take an additional input parameter for the length
3142 of the line, rather than calculating it with strlen.
3143 (diagnostic_show_locus): Adjust the use of
3144 location_get_source_line and adjust_line with respect to their new
3145 signature. While displaying a line now, do not stop at the first
3146 null byte. Rather, display the zero byte as a space and keep
3147 going until we reach the size of the line.
3148 * Makefile.in: Add vec.o to OBJS-libcommon
3150 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3151 Ilya Tocar <ilya.tocar@intel.com>
3153 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3154 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3155 (__builtin_ia32_kmov16): Ditto.
3156 * config/i386/i386.md (UNSPEC_KMOV): New.
3159 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3161 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3162 (_mm512_storeu_si512): Ditto.
3164 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3167 * rtl.h (get_referenced_operands): Declare.
3168 * recog.c (get_referenced_operands): New function.
3169 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3170 operands have been referenced when recording LO_SUM references.
3172 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3174 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3176 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3178 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3179 Enable for generic and recent AMD targets.
3181 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3183 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3184 ARG_SIZE note when adjustment was eliminated.
3186 2014-01-22 Jeff Law <law@redhat.com>
3188 PR tree-optimization/59597
3189 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3190 in file. Accept new argument REGISTERING and use it to modify
3191 dump output appropriately.
3192 (register_jump_thread): Corresponding changes.
3193 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3194 thread paths involving joiner blocks. Add code to dump cancelled
3195 jump threading paths.
3197 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3199 PR rtl-optimization/59477
3200 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3201 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3203 2014-01-22 Tom Tromey <tromey@redhat.com>
3205 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3207 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3209 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3211 PR rtl-optimization/59896
3212 * lra-constraints.c (process_alt_operands): Check unused note for
3213 matched operands of insn with no output reloads.
3215 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3217 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3218 (mips_move_from_gpr_cost): Likewise.
3220 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3222 PR rtl-optimization/59858
3223 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3224 ira_class_hard_regs_num.
3225 (process_alt_operands): Increase reject for dying matched operand.
3227 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3230 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3231 smaller than size, perform several stores or loads and stores
3232 at dst + count - size to store or copy all of size bytes, rather
3233 than just last modesize bytes.
3235 2014-01-20 DJ Delorie <dj@redhat.com>
3237 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3238 that CLOBBERs are REGs before propogating their values.
3240 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3243 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3244 (cgraph_inline_failed_type): New function.
3245 * cgraph.h (DEFCIFCODE): Add type.
3246 (cgraph_inline_failed_type_t): New enum.
3247 (cgraph_inline_failed_type): New prototype.
3248 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3249 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3250 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3251 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3252 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3253 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3254 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3255 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3256 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3257 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3258 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3259 OPTIMIZATION_MISMATCH.
3260 * tree-inline.c (expand_call_inline): Emit errors during
3261 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3263 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3266 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3267 mode attribute in insn output.
3269 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3271 * output.h (output_constant): Delete.
3272 * varasm.c (output_constant): Make private.
3274 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3276 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3278 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3281 * tree.h (fold_builtin_strcat): New prototype.
3282 * builtins.c (fold_builtin_strcat): No longer static. Add len
3283 argument, if non-NULL, don't call c_strlen. Optimize
3284 directly into __builtin_memcpy instead of __builtin_strcpy.
3285 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3286 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3288 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3290 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3291 for SImode_address_operand operands, having only a REG argument.
3293 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3295 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3296 loader name using mbig-endian.
3297 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3299 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3301 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3305 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3307 * config/aarch64/aarch64-protos.h
3308 (aarch64_cannot_change_mode_class_ptr): Declare.
3309 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3310 aarch64_cannot_change_mode_class_ptr): New.
3311 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3312 backend hook aarch64_cannot_change_mode_class.
3314 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3316 * common/config/aarch64/aarch64-common.c
3317 (aarch64_handle_option): Don't handle any option order logic here.
3318 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3319 selected_cpu, warn on architecture version mismatch.
3320 (aarch64_override_options): Fix parsing order for option strings.
3322 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3323 Iain Sandoe <iain@codesourcery.com>
3326 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3327 warning. Amend comment to reflect current functionality.
3329 2014-01-20 Richard Biener <rguenther@suse.de>
3332 * builtins.c (fold_builtin_strcat): Remove case better handled
3333 by tree-ssa-strlen.c.
3335 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3337 * config/aarch64/aarch64.opt
3338 (mcpu, march, mtune): Make case-insensitive.
3340 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3343 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3344 if operands[1] is a REG or ZERO_EXTEND of a REG.
3346 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3348 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3350 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3352 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3353 long non-pic millicode calls.
3355 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3357 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3359 2014-01-19 Kito Cheng <kito@0xlab.org>
3361 * builtins.c (expand_movstr): Check movstr expand done or fail.
3363 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3364 H.J. Lu <hongjiu.lu@intel.com>
3367 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3368 to DImode for zero-extended addresses.
3370 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3372 PR rtl-optimization/57763
3373 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3374 on the new indirect jump_insn and increment LABEL_NUSES (label).
3376 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3380 * config.gcc (x86_archs): New variable.
3381 (x86_64_archs): Likewise.
3382 (x86_cpus): Likewise.
3383 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3384 --with-arch/--with-cpu= options.
3385 Support --with-arch=/--with-cpu={nehalem,westmere,
3386 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3388 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3390 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3391 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3393 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3395 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3397 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3400 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3401 clear cpp_get_options (parse_in)->warn_unused_macros for
3402 ix86_target_macros_internal with cpp_define.
3404 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3406 * jump.c (delete_related_insns): Keep (use (insn))s.
3407 * reorg.c (redundant_insn): Check for barriers too.
3409 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3411 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3413 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3415 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3416 call to $$dyncall when TARGET_LONG_CALLS is true.
3418 2014-01-17 Jeff Law <law@redhat.com>
3420 * ree.c (combine_set_extension): Temporarily disable test for
3421 changing number of hard registers.
3423 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3426 * ipa-inline-analysis.c (inline_free_summary):
3427 Do not free summary of aliases.
3429 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3432 * gimplify.c (gimplify_expr): Use create_tmp_var
3433 instead of create_tmp_var_raw. If cond doesn't have
3434 integral type, don't add the IFN_ANNOTATE builtin at all.
3436 2014-01-17 Martin Jambor <mjambor@suse.cz>
3439 * ipa-cp.c (prev_edge_clone): New variable.
3440 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3441 Also resize prev_edge_clone vector.
3442 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3443 (ipcp_edge_removal_hook): New function.
3444 (ipcp_driver): Register ipcp_edge_removal_hook.
3446 2014-01-17 Andrew Pinski <apinski@cavium.com>
3447 Steve Ellcey <sellcey@mips.com>
3450 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3453 2014-01-17 Jeff Law <law@redhat.com>
3456 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3457 so that pass_ccp runs first.
3459 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3461 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3462 (ix86_adjust_cost): Use !TARGET_XXX.
3463 (do_reorder_for_imul): Likewise.
3464 (swap_top_of_ready_list): Likewise.
3465 (ix86_sched_reorder): Likewise.
3467 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3469 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3470 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3471 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3472 (intel_memset): New. Duplicate slm_memset.
3473 (intel_cost): New. Duplicate slm_cost.
3474 (m_INTEL): New macro.
3475 (processor_target_table): Add "intel".
3476 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3477 with PROCESSOR_INTEL for "intel".
3478 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3479 PROCESSOR_SILVERMONT.
3480 (ix86_issue_rate): Likewise.
3481 (ix86_adjust_cost): Likewise.
3482 (ia32_multipass_dfa_lookahead): Likewise.
3483 (swap_top_of_ready_list): Likewise.
3484 (ix86_sched_reorder): Likewise.
3485 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3486 instead of TARGET_OPT_AGU.
3487 * config/i386/i386.h (TARGET_INTEL): New.
3488 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3489 (processor_type): Add PROCESSOR_INTEL.
3490 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3491 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3493 2014-01-17 Marek Polacek <polacek@redhat.com>
3496 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3499 2014-01-17 Richard Biener <rguenther@suse.de>
3501 PR tree-optimization/46590
3502 * opts.c (default_options_table): Add entries for
3503 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3504 all enabled at -O1 but not for -Og.
3505 * common.opt (fbranch-count-reg): Remove Init(1).
3506 (fmove-loop-invariants): Likewise.
3507 (ftree-pta): Likewise.
3509 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3511 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3512 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3513 decls to at least the GCC 4.8 used alignments.
3516 * tree-nested.c (convert_nonlocal_reference_stmt,
3517 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3518 of GIMPLE_BIND stmts, adjust associated decls.
3520 2014-01-17 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/46590
3523 * vec.h (vec<>::bseach): New member function implementing
3524 binary search according to C89 bsearch.
3525 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3526 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3527 bitmap pointer again. Make accesses_in_loop a flat array.
3528 (mem_ref_obstack): New global.
3529 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3530 (mark_ref_stored): Likewise.
3531 (ref_indep_loop_p_2): Likewise.
3532 (set_ref_stored_in_loop): New helper function.
3533 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3534 (memref_free): Adjust.
3535 (record_mem_ref_loc): Simplify.
3536 (gather_mem_refs_stmt): Adjust.
3537 (sort_locs_in_loop_postorder_cmp): New function.
3538 (analyze_memory_references): Sort accesses_in_loop after
3539 loop postorder number.
3540 (find_ref_loc_in_loop_cmp): New function.
3541 (for_all_locs_in_loop): Find relevant cluster of locs in
3542 accesses_in_loop and iterate without recursion.
3543 (execute_sm): Avoid uninit warning.
3544 (struct ref_always_accessed): Simplify.
3545 (ref_always_accessed::operator ()): Likewise.
3546 (ref_always_accessed_p): Likewise.
3547 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3548 loop postorder numbers here.
3549 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3552 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3555 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3556 on decls for which assemble_alias has been called.
3558 2014-01-17 Nick Clifton <nickc@redhat.com>
3560 * config/msp430/msp430.opt: (mcpu): New option.
3561 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3562 (msp430_option_override): Parse target_cpu. If the MCU name
3563 matches a generic string, clear target_mcu.
3564 (msp430_attr): Allow numeric interrupt values up to 63.
3565 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3566 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3568 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3570 * config/msp430/msp430.md (popm): Use %J rather than %I.
3571 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3572 (addhi_cy_i): Use immediate_operand for operand 2.
3573 * doc/invoke.texi: Document -mcpu option.
3575 2014-01-17 Richard Biener <rguenther@suse.de>
3577 PR rtl-optimization/38518
3578 * df.h (df_analyze_loop): Declare.
3579 * df-core.c: Include cfgloop.h.
3580 (df_analyze_1): Split out main part of df_analyze.
3581 (df_analyze): Adjust.
3582 (loop_inverted_post_order_compute): New function.
3583 (loop_post_order_compute): Likewise.
3584 (df_analyze_loop): New function avoiding whole-function
3586 * loop-invariant.c (find_defs): Use df_analyze_loop.
3587 (find_invariants): Adjust.
3588 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3590 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3592 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3593 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3595 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3597 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3598 traversal when removing references.
3600 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3603 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3605 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3608 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3609 pushing a reload for an autoinc when we had previously reloaded an
3610 inner part of the address.
3612 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3614 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3616 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3617 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3618 when not giving up or versioning for alias only because of
3620 (vect_analyze_data_ref_dependences): Set to true.
3621 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3623 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3624 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3628 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3631 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3632 operand 0 predicate for gathers, use a new pseudo as subtarget.
3634 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3637 * lra-constraints.c (process_alt_operands): Add printing debug info.
3638 Check absence of input/output reloads for matched operands too.
3640 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3642 PR rtl-optimization/59835
3643 * ira.c (ira_init_register_move_cost): Increase cost for
3646 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3648 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3650 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3653 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3654 non-register objects. Use gen_(high/low)part more consistently.
3657 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3660 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3661 endian support, remove tests for WORDS_BIG_ENDIAN.
3662 (p8_mfvsrd_3_<mode>): Likewise.
3663 (reload_gpr_from_vsx<mode>): Likewise.
3664 (reload_gpr_from_vsxsf): Likewise.
3665 (p8_mfvsrd_4_disf): Likewise.
3667 2014-01-16 Richard Biener <rguenther@suse.de>
3669 PR rtl-optimization/46590
3670 * lcm.c (compute_antinout_edge): Use postorder iteration.
3671 (compute_laterin): Use inverted postorder iteration.
3673 2014-01-16 Nick Clifton <nickc@redhat.com>
3676 * varasm.c (output_constant): Return the number of bytes actually
3678 (output_constructor_array_range): Update the field size with the
3679 number of bytes emitted by output_constant.
3680 (output_constructor_regular_field): Likewise. Also do not
3681 complain if the total number of bytes emitted is now greater
3682 than the expected fieldpos.
3683 * output.h (output_constant): Update prototype and descriptive comment.
3685 2014-01-16 Marek Polacek <polacek@redhat.com>
3688 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3689 it is error_mark_node.
3691 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3693 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3694 VALID_AVX256_REG_OR_OI_MODE.
3696 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3698 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3699 current procedure should be profiled.
3701 2014-01-15 Andrew Pinski <apinski@cavium.com>
3703 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3704 of moving from/to the STACK_REG register class.
3706 2014-01-15 Richard Henderson <rth@redhat.com>
3709 * reginfo.c (global_regs_decl): Globalize.
3710 * rtl.h (global_regs_decl): Declare.
3711 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3712 reserved via global_regs.
3714 2014-01-15 Teresa Johnson <tejohnson@google.com>
3716 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3718 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3720 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3721 and vmulosh rather than call gen_vec_widen_smult_*.
3722 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3723 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3724 (vec_widen_smult_even_v16qi): Likewise.
3725 (vec_widen_umult_even_v8hi): Likewise.
3726 (vec_widen_smult_even_v8hi): Likewise.
3727 (vec_widen_umult_odd_v16qi): Likewise.
3728 (vec_widen_smult_odd_v16qi): Likewise.
3729 (vec_widen_umult_odd_v8hi): Likewise.
3730 (vec_widen_smult_odd_v8hi): Likewise.
3731 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3732 vmuloub rather than call gen_vec_widen_umult_*.
3733 (vec_widen_umult_lo_v16qi): Likewise.
3734 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3735 vmulosb rather than call gen_vec_widen_smult_*.
3736 (vec_widen_smult_lo_v16qi): Likewise.
3737 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3738 rather than call gen_vec_widen_umult_*.
3739 (vec_widen_umult_lo_v8hi): Likewise.
3740 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3741 rather than call gen_vec_widen_smult_*.
3742 (vec_widen_smult_lo_v8hi): Likewise.
3744 2014-01-15 Jeff Law <law@redhat.com>
3746 PR tree-optimization/59747
3747 * ree.c (find_and_remove_re): Properly handle case where a second
3748 eliminated extension requires widening a copy created for elimination
3749 of a prior extension.
3750 (combine_set_extension): Ensure that the number of hard regs needed
3751 for a destination register does not change when we widen it.
3753 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3755 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3756 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3757 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3758 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3759 (avr-*-rtems*): Likewise.
3760 (bfin*-rtems*): Likewise.
3761 (moxie-*-rtems*): Likewise.
3762 (h8300-*-rtems*): Likewise.
3763 (i[34567]86-*-rtems*): Likewise.
3764 (lm32-*-rtems*): Likewise.
3765 (m32r-*-rtems*): Likewise.
3766 (m68k-*-rtems*): Likewise.
3767 (microblaze*-*-rtems*): Likewise.
3768 (mips*-*-rtems*): Likewise.
3769 (powerpc-*-rtems*): Likewise.
3770 (sh-*-rtems*): Likewise.
3771 (sparc-*-rtems*): Likewise.
3772 (sparc64-*-rtems*): Likewise.
3773 (v850-*-rtems*): Likewise.
3774 (m32c-*-rtems*): Likewise.
3776 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3778 PR rtl-optimization/59511
3779 * ira.c (ira_init_register_move_cost): Use memory costs for some
3780 cases of register move cost calculations.
3781 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3782 instead of BB frequency.
3783 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3784 * lra-assigns.c (find_hard_regno_for): Ditto.
3786 2014-01-15 Richard Biener <rguenther@suse.de>
3788 PR tree-optimization/59822
3789 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3790 (vectorizable_load): Use it to hoist defs of uses of invariant
3791 loads out of the loop.
3793 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3794 Kugan Vivekanandarajah <kuganv@linaro.org>
3797 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3800 2014-01-15 Richard Biener <rguenther@suse.de>
3802 PR rtl-optimization/59802
3803 * lcm.c (compute_available): Use inverted postorder to seed
3804 the initial worklist.
3806 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3809 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3810 ADDR_REGS for invalid symrefs in non-PIC code.
3812 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3815 * builtins.c (determine_block_size): Initialize *probable_max_size
3816 even if len_rtx is CONST_INT.
3818 2014-01-14 Andrew Pinski <apinski@cavium.com>
3820 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3821 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3822 (cortexa53_tunings): Likewise.
3823 (aarch64_sched_issue_rate): New function.
3824 (TARGET_SCHED_ISSUE_RATE): Define.
3826 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3828 * ira-costs.c (find_costs_and_classes): Add missed
3829 ira_init_register_move_cost_if_necessary.
3831 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3834 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3836 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3839 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3840 to indicate if type is used for function return value. Warn ABI
3841 change if the vector mode isn't available for function return value.
3842 (ix86_function_arg_advance): Pass false to type_natural_mode.
3843 (ix86_function_arg): Likewise.
3844 (ix86_gimplify_va_arg): Likewise.
3845 (function_arg_32): Don't warn ABI change.
3846 (ix86_function_value): Pass true to type_natural_mode.
3847 (ix86_return_in_memory): Likewise.
3848 (ix86_struct_value_rtx): Removed.
3849 (TARGET_STRUCT_VALUE_RTX): Likewise.
3851 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3853 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3854 converting a conditional jump into a conditional return.
3856 2014-01-14 Richard Biener <rguenther@suse.de>
3858 PR tree-optimization/58921
3859 PR tree-optimization/59006
3860 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3861 hoisting invariant stmts.
3862 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3863 invariant loads on the preheader edge if possible.
3865 2014-01-14 Joey Ye <joey.ye@arm.com>
3867 * doc/plugin.texi (Building GCC plugins): Update to C++.
3869 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3871 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3872 (_mm_rcp28_round_ss): Ditto.
3873 (_mm_rsqrt28_round_sd): Ditto.
3874 (_mm_rsqrt28_round_ss): Ditto.
3875 (_mm_rcp28_sd): Ditto.
3876 (_mm_rcp28_ss): Ditto.
3877 (_mm_rsqrt28_sd): Ditto.
3878 (_mm_rsqrt28_ss): Ditto.
3879 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3880 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3881 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3882 (IX86_BUILTIN_RCP28SD): Ditto.
3883 (IX86_BUILTIN_RCP28SS): Ditto.
3884 (IX86_BUILTIN_RSQRT28SD): Ditto.
3885 (IX86_BUILTIN_RSQRT28SS): Ditto.
3886 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3887 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3888 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3889 (ix86_expand_special_args_builtin): Expand new FTYPE.
3890 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3891 (srcp14<mode>): Make insn unary.
3892 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3893 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3894 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3895 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3896 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3897 Fix rounding: make it SAE only.
3898 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3900 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3902 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3903 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3904 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3905 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3906 (round_saeonly_mask_scalar_operand4): Ditto.
3907 (round_saeonly_mask_scalar_op3): Ditto.
3908 (round_saeonly_mask_scalar_op4): Ditto.
3910 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3912 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3913 Implement -maltivec=be for vec_insert and vec_extract.
3915 2014-01-10 DJ Delorie <dj@redhat.com>
3917 * config/msp430/msp430.md (call_internal): Don't allow memory
3918 references with SP as the base register.
3919 (call_value_internal): Likewise.
3920 * config/msp430/constraints.md (Yc): New. For memory references
3921 that don't use SP as a base register.
3923 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3924 "an integer without a # prefix"
3925 * config/msp430/msp430.md (epilogue_helper): Use it.
3927 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3930 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3931 AVX512F gather builtins.
3932 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3933 on gather decls with INTEGER_TYPE masktype.
3934 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3935 directly into the builtin rather than hoisting it before loop.
3937 PR tree-optimization/59387
3938 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3939 (scev_const_prop): If folded_casts and type has undefined overflow,
3940 use force_gimple_operand instead of force_gimple_operand_gsi and
3941 for each added stmt if it is assign with
3942 arith_code_with_undefined_signed_overflow, call
3943 rewrite_to_defined_overflow.
3944 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3945 gimple-fold.h instead.
3946 (arith_code_with_undefined_signed_overflow,
3947 rewrite_to_defined_overflow): Moved to ...
3948 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3949 rewrite_to_defined_overflow): ... here. No longer static.
3950 Include gimplify-me.h.
3951 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3952 rewrite_to_defined_overflow): New prototypes.
3954 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3956 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3958 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3960 * builtins.c (get_object_alignment_2): Minor tweak.
3961 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3963 2014-01-13 Christian Bruel <christian.bruel@st.com>
3965 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3966 optimized non constant lengths.
3968 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3971 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3972 load as __atomic_load_N if possible.
3974 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3976 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3978 (rs6000_expand_builtin): Adjust call.
3980 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3983 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3984 * config/rs6000/rs6000.c: Include target-globals.h.
3985 (rs6000_set_current_function): Instead of doing target_reinit
3986 unconditionally, use save_target_globals_default_opts and
3987 restore_target_globals.
3989 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3991 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3992 (rs6000_expand_builtin): Handle mffs and mtfsf.
3993 (rs6000_init_builtins): Define mffs and mtfsf.
3994 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3995 (rs6000_mffs): New pattern.
3996 (rs6000_mtfsf): New pattern.
3998 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4000 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4001 Start narrowing with START. Apply candidate-use pair
4002 and check overall cost in narrowing.
4003 (iv_ca_prune): Pass new argument.
4005 2014-01-10 Jeff Law <law@redhat.com>
4008 * ree.c (combine_reaching_defs): Ensure the defining statement
4009 occurs before the extension when optimizing extensions with
4010 different source and destination hard registers.
4012 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4015 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4016 vtables into the type inheritance graph.
4018 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4020 PR rtl-optimization/59754
4021 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4022 modes in the REGNO != REGNO case.
4024 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4026 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4028 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4030 PR tree-optimization/59745
4031 * tree-predcom.c (tree_predictive_commoning_loop): Call
4032 free_affine_expand_cache if giving up because components is NULL.
4034 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4035 GC in payload of target_globals struct instead of allocating them on
4036 the heap and the larger structs separately using GC.
4037 * target-globals.h (struct target_globals): Make regs, hard_regs,
4038 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4039 of GTY((skip)) and change type to void *.
4040 (reset_target_globals): Cast loads from those fields to corresponding
4043 2014-01-10 Steve Ellcey <sellcey@mips.com>
4046 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4047 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4048 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4050 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4053 * aarch64-modes.def (CC_Zmode): New flags mode.
4054 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4055 represents an equality.
4056 (aarch64_get_condition_code): Handle CC_Zmode.
4057 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4059 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4061 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4062 extraction in good case.
4064 2014-01-10 Richard Biener <rguenther@suse.de>
4066 PR tree-optimization/59374
4067 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4068 checking after SLP discovery. Mark stmts not participating
4069 in any SLP instance properly.
4071 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4073 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4074 when handling a SET rtx.
4076 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4078 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4079 (cortex-a57): Likewise.
4080 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4082 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4084 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4085 non-iwmmxt builtins.
4087 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4091 * ipa-devirt.c record_target_from_binfo): Take as argument
4092 stack of binfos and lookup matching one for virtual inheritance.
4093 (possible_polymorphic_call_targets_1): Update.
4095 2014-01-10 Huacai Chen <chenhc@lemote.com>
4097 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4098 kernel strings for Loongson-2E/2F/3A.
4100 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4103 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4104 is_gimple_call before calling gimple_call_internal_p.
4106 2014-01-09 Steve Ellcey <sellcey@mips.com>
4108 * Makefile.in (TREE_FLOW_H): Remove.
4109 (TREE_SSA_H): Add file names from tree-flow.h.
4110 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4111 * tree.h: Remove tree-flow.h reference.
4112 * hash-table.h: Remove tree-flow.h reference.
4113 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4114 reference with tree-ssa-loop.h.
4116 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4118 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4119 default element-order behavior for -maltivec.
4120 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4122 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4123 when targeting big endian, at least for now.
4124 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4126 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4129 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4130 var satisfies use_register_for_decl, just take into account type
4131 alignment, rather than decl alignment.
4133 PR tree-optimization/59622
4134 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4135 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4136 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4137 Don't devirtualize for inplace at all. For targets.length () == 1,
4138 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4140 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4142 * config/i386/i386.md (cpu): Remove the unused btver1.
4144 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4146 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4148 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4151 * tree-core.h (struct target_globals): New forward declaration.
4152 (struct tree_target_option): Add globals field.
4153 * tree.h (TREE_TARGET_GLOBALS): Define.
4154 (prepare_target_option_nodes_for_pch): New prototype.
4155 * target-globals.h (struct target_globals): Define even if
4157 * tree.c (prepare_target_option_node_for_pch,
4158 prepare_target_option_nodes_for_pch): New functions.
4159 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4160 * config/i386/i386.c: Include target-globals.h.
4161 (ix86_set_current_function): Instead of doing target_reinit
4162 unconditionally, use save_target_globals_default_opts and
4163 restore_target_globals.
4165 2014-01-09 Richard Biener <rguenther@suse.de>
4167 PR tree-optimization/59715
4168 * tree-cfg.h (split_critical_edges): Declare.
4169 * tree-cfg.c (split_critical_edges): Export.
4170 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4172 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4174 * cfgexpand.c (expand_stack_vars): Optionally disable
4175 asan stack protection.
4176 (expand_used_vars): Likewise.
4177 (partition_stack_vars): Likewise.
4178 * asan.c (asan_emit_stack_protection): Optionally disable
4179 after return stack usage.
4180 (instrument_derefs): Optionally disable memory access instrumentation.
4181 (instrument_builtin_call): Likewise.
4182 (instrument_strlen_call): Likewise.
4183 (asan_protect_global): Optionally disable global variables protection.
4184 * doc/invoke.texi: Added doc for new options.
4185 * params.def: Added new options.
4186 * params.h: Likewise.
4188 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4190 PR rtl-optimization/59724
4191 * ifcvt.c (cond_exec_process_if_block): Don't call
4192 flow_find_head_matching_sequence with 0 longest_match.
4193 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4194 non-active insns if !stop_after.
4195 (try_head_merge_bb): Revert 2014-01-07 changes.
4197 2014-01-08 Jeff Law <law@redhat.com>
4199 * ree.c (get_sub_rtx): New function, extracted from...
4200 (merge_def_and_ext): Here.
4201 (combine_reaching_defs): Use get_sub_rtx.
4203 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4205 * cgraph.h (varpool_variable_node): Do not choke on null node.
4207 2014-01-08 Catherine Moore <clm@codesourcery.com>
4209 * config/mips/mips.md (simple_return): Attempt to use JRC
4211 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4213 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4215 PR rtl-optimization/59137
4216 * reorg.c (steal_delay_list_from_target): Call update_block for
4218 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4220 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4222 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4223 two duplicate entries.
4225 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4228 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4230 * config/mips/mips.c (mips_truncated_op_cost): New function.
4231 (mips_rtx_costs): Adjust test for BADDU.
4232 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4234 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4236 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4237 (*baddu_si): ...this new pattern.
4239 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4242 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4244 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4247 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4249 (expand_expr, expand_normal): Adjust.
4250 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4251 inner_reference_p. Use inner_reference_p to expand inner references.
4252 (store_expr): Adjust.
4253 * cfgexpand.c (expand_call_stmt): Adjust.
4255 2014-01-08 Rong Xu <xur@google.com>
4257 * gcov-io.c (gcov_var): Move from gcov-io.h.
4258 (gcov_position): Ditto.
4259 (gcov_is_error): Ditto.
4260 (gcov_rewrite): Ditto.
4261 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4262 only part to libgcc/libgcov.h.
4264 2014-01-08 Marek Polacek <polacek@redhat.com>
4267 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4269 2014-01-08 Marek Polacek <polacek@redhat.com>
4272 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4274 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4276 PR rtl-optimization/59649
4277 * stor-layout.c (get_mode_bounds): For BImode return
4278 0 and STORE_FLAG_VALUE.
4280 2014-01-08 Richard Biener <rguenther@suse.de>
4283 * gimple.h (is_gimple_builtin_call): Remove.
4284 (gimple_builtin_call_types_compatible_p): New.
4285 (gimple_call_builtin_p): New overload.
4286 * gimple.c (is_gimple_builtin_call): Remove.
4287 (validate_call): Rename to ...
4288 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4290 (validate_type): New static function.
4291 (gimple_call_builtin_p): New overload and adjust.
4292 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4293 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4294 (gimple_fold_stmt_to_constant_1): Likewise.
4295 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4297 2014-01-08 Richard Biener <rguenther@suse.de>
4300 * gimplify.c (gimplify_expr): Gimplify register-register type
4301 VIEW_CONVERT_EXPRs to separate stmts.
4303 2014-01-07 Jeff Law <law@redhat.com>
4306 * ree.c (combine_set_extension): Handle case where source
4307 and destination registers in an extension insn are different.
4308 (combine_reaching_defs): Allow source and destination registers
4309 in extension to be different under limited circumstances.
4310 (add_removable_extension): Remove restriction that the
4311 source and destination registers in the extension are the same.
4312 (find_and_remove_re): Emit a copy from the extension's
4313 destination to its source after the defining insn if
4314 the source and destination registers are different.
4317 * ifcvt.c (merge_if_block): If we are merging a block with more than
4318 one successor with a block with no successors, remove any BARRIER
4319 after the second block.
4321 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4323 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4325 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4328 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4329 for 14-bit register offsets when INT14_OK_STRICT is false.
4331 2014-01-07 Roland Stigge <stigge@antcom.de>
4332 Michael Meissner <meissner@linux.vnet.ibm.com>
4335 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4336 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4338 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4340 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4343 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4345 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4346 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4347 rtx is const0_rtx or not.
4349 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4352 * target-globals.c (save_target_globals): Remove this_fn_optab
4354 * toplev.c: Include optabs.h.
4355 (target_reinit): Temporarily restore the global options if another
4356 set of options are in force.
4358 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4360 PR rtl-optimization/58668
4361 * cfgcleanup.c (flow_find_cross_jump): Don't count
4362 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4363 to determine what is counted.
4364 (flow_find_head_matching_sequence): Use active_insn_p to determine
4366 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4368 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4369 determine what is counted.
4371 PR tree-optimization/59643
4372 * tree-predcom.c (split_data_refs_to_components): If one dr is
4373 read and one write, determine_offset fails and the write isn't
4374 in the bad component, just put the read into the bad component.
4376 2014-01-07 Mike Stump <mikestump@comcast.net>
4377 Jakub Jelinek <jakub@redhat.com>
4380 * tree-core.h (struct tree_optimization_option): Change optabs
4381 type from unsigned char * to void *.
4382 * optabs.c (init_tree_optimization_optabs): Adjust
4383 TREE_OPTIMIZATION_OPTABS initialization.
4385 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4388 * config/i386/i386.h (struct machine_function): Add
4389 no_drap_save_restore field.
4390 * config/i386/i386.c (ix86_save_reg): Use
4391 !cfun->machine->no_drap_save_restore instead of
4392 crtl->stack_realign_needed.
4393 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4394 this function clears frame_pointer_needed. Set
4395 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4396 and DRAP reg is needed.
4398 2014-01-06 Marek Polacek <polacek@redhat.com>
4401 * doc/implement-c.texi: Mention that other integer types are
4402 permitted as bit-field types in strictly conforming mode.
4404 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4406 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4409 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4411 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4413 2014-01-06 Martin Jambor <mjambor@suse.cz>
4416 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4418 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4420 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4424 * var-tracking.c (add_stores): Preserve the value of the source even if
4425 we don't record the store.
4427 2014-01-06 Terry Guo <terry.guo@arm.com>
4429 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4431 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4434 * config/darwin.c (darwin_function_section): Adjust return values to
4435 correspond to optimisation changes made in r206070.
4437 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4439 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4440 from prefetch_block tune setting.
4441 (nocona_cost): Correct size of prefetch block to 64.
4443 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4445 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4446 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4447 used to save the static chain register in the computation of the offset
4448 from which the FP registers need to be restored.
4450 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4452 PR tree-optimization/59519
4453 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4454 ICE if get_current_def (current_new_name) is already non-NULL, as long
4455 as it is a phi result of some other phi in *new_exit_bb that has
4458 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4459 or vmovdqu* for misaligned_operand.
4460 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4461 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4462 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4463 aligned_mem for AVX512F masked aligned load and store builtins and for
4466 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4468 PR tree-optimization/59651
4469 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4470 Address range for negative step should be added by TYPE_SIZE_UNIT.
4472 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4474 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4475 overlapping registers also for registers other than the stack pointer.
4477 2014-01-03 Marek Polacek <polacek@redhat.com>
4480 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4483 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4486 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4489 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4490 (push splitter): Use <P:MODE_SIZE> instead of
4491 GET_MODE_SIZE (<P:MODE>mode).
4492 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4493 (mov -1, reg peephole2): Likewise.
4494 * config/i386/sse.md (*mov<mode>_internal,
4495 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4496 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4497 *<code><mode>3, *andnot<mode>3<mask_name>,
4498 <mask_codefor><code><mode>3<mask_name>): Likewise.
4499 * config/i386/subst.md (mask_mode512bit_condition,
4500 sd_mask_mode512bit_condition): Likewise.
4502 2014-01-02 Xinliang David Li <davidxl@google.com>
4504 PR tree-optimization/59303
4505 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4506 (dump_predicates): Better output format.
4507 (pred_equal_p): New function.
4508 (is_neq_relop_p): Ditto.
4509 (is_neq_zero_form_p): Ditto.
4510 (pred_expr_equal_p): Ditto.
4511 (pred_neg_p): Ditto.
4512 (simplify_pred): Ditto.
4513 (simplify_preds_2): Ditto.
4514 (simplify_preds_3): Ditto.
4515 (simplify_preds_4): Ditto.
4516 (simplify_preds): Ditto.
4518 (push_to_worklist): Ditto.
4519 (get_pred_info_from_cmp): Ditto.
4520 (is_degenerated_phi): Ditto.
4521 (normalize_one_pred_1): Ditto.
4522 (normalize_one_pred): Ditto.
4523 (normalize_one_pred_chain): Ditto.
4524 (normalize_preds): Ditto.
4525 (normalize_cond_1): Remove function.
4526 (normalize_cond): Ditto.
4527 (is_gcond_subset_of): Ditto.
4528 (is_subset_of_any): Ditto.
4529 (is_or_set_subset_of): Ditto.
4530 (is_and_set_subset_of): Ditto.
4531 (is_norm_cond_subset_of): Ditto.
4532 (pred_chain_length_cmp): Ditto.
4533 (convert_control_dep_chain_into_preds): Type change.
4534 (find_predicates): Ditto.
4535 (find_def_preds): Ditto.
4536 (destroy_predicates_vecs): Ditto.
4537 (find_matching_predicates_in_rest_chains): Ditto.
4538 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4539 (is_pred_expr_subset): Ditto.
4540 (is_pred_chain_subset_of): Ditto.
4541 (is_included_in): Ditto.
4542 (is_superset_of): Ditto.
4544 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4546 Update copyright years.
4548 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4550 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4551 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4552 config/arc/arc.md, config/arc/arc.opt,
4553 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4554 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4555 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4556 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4557 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4558 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4559 vtable-verify.c, vtable-verify.h: Use the standard form for the
4562 2014-01-02 Tobias Burnus <burnus@net-b.de>
4564 * gcc.c (process_command): Update copyright notice dates.
4565 * gcov-dump.c: Ditto.
4567 * doc/cpp.texi: Bump @copying's copyright year.
4568 * doc/cppinternals.texi: Ditto.
4569 * doc/gcc.texi: Ditto.
4570 * doc/gccint.texi: Ditto.
4571 * doc/gcov.texi: Ditto.
4572 * doc/install.texi: Ditto.
4573 * doc/invoke.texi: Ditto.
4575 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4577 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4579 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4581 * config/i386/sse.md (*mov<mode>_internal): Guard
4582 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4584 PR rtl-optimization/59647
4585 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4586 new_rtx into UNSIGNED_FLOAT rtxes.
4588 Copyright (C) 2014 Free Software Foundation, Inc.
4590 Copying and distribution of this file, with or without modification,
4591 are permitted in any medium without royalty provided the copyright
4592 notice and this notice are preserved.