1 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
4 * ipa.c (function_and_variable_visibility): When dissolving comdat
5 group, also set all symbols to local.
7 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
12 2013-12-14 Jan Hubicka <jh@suse.cz>
14 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
16 2014-03-02 Jon Beniston <jon@beniston.com>
22 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
23 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
24 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
25 (simple_return, *simple_return): New patterns
26 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
27 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
29 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
31 * dwarf2out.c (gen_subprogram_die): Tidy.
33 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
36 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
37 (*mov_t_msb_neg_negc): ... this new insn.
39 2014-02-28 Jason Merrill <jason@redhat.com>
42 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
45 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
48 * dwarf2out.c (decltype_auto_die): New static.
49 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
50 (gen_type_die_with_usage): Handle 'decltype(auto)'.
51 (is_cxx_auto): Likewise.
53 2014-02-28 Ian Bolton <ian.bolton@arm.com>
55 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
56 we are not using general regs only.
58 2014-02-28 Richard Biener <rguenther@suse.de>
61 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
62 previous fix and only allow to remove trivial pre-headers
63 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
64 (remove_forwarder_block): Properly update the latch of a loop.
66 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
69 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
70 (cselib_preserved_hash_table): New.
71 (preserve_constants_and_equivs): Move preserved vals to it.
72 (cselib_find_slot): Look it up first.
73 (cselib_init): Initialize it.
74 (cselib_finish): Release it.
75 (dump_cselib_table): Dump it.
77 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
80 * cselib.c (remove_useless_values): Skip to avoid quadratic
81 behavior if the condition moved from...
82 (cselib_process_insn): ... here holds.
84 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
87 * var-tracking.c (vt_initialize): Apply the same condition to
88 preserve the CFA base value.
90 2014-02-28 Joey Ye <joey.ye@arm.com>
93 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
94 if reload in progress or completed.
96 2014-02-28 Tobias Burnus <burnus@net-b.de>
99 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
102 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
104 * doc/tm.texi.in (Condition Code Status): Update documention for
105 relative locations of cc0-setter and cc0-user.
107 2014-02-27 Jeff Law <law@redhat.com>
109 PR rtl-optimization/52714
110 * combine.c (try_combine): When splitting an unrecognized PARALLEL
111 into two independent simple sets, if I3 is a jump, ensure the
112 pattern we place into I3 is a (set (pc) ...).
114 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
115 Jeff Law <law@redhat.com>
117 PR rtl-optimization/49847
118 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
119 are in different blocks.
120 * doc/tm.texi (Condition Code Status): Update documention for
121 relative locations of cc0-setter and cc0-user.
123 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
126 * lra.c (lra_emit_add): Check SUBREG too.
128 2014-02-27 Andreas Schwab <schwab@suse.de>
130 * config/m68k/m68k.c (m68k_option_override): Disable
131 -flive-range-shrinkage for classic m68k.
132 (m68k_override_options_after_change): Likewise.
134 2014-02-27 Marek Polacek <polacek@redhat.com>
137 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
138 -Wmaybe-uninitialized.
140 2014-02-27 Alan Modra <amodra@gmail.com>
143 * reload1.c (emit_input_reload_insns): When reload_override_in,
144 set old to rl->in_reg when rl->in_reg is a subreg.
146 2014-02-26 Richard Biener <rguenther@suse.de>
149 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
151 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
153 * common/config/i386/predicates.md (const1256_operand): Remove.
154 (const2356_operand): New.
155 (const_1_to_2_operand): Remove.
156 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
157 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
158 (*avx512pf_gatherpf<mode>sf): Ditto.
159 (avx512pf_gatherpf<mode>df): Ditto.
160 (*avx512pf_gatherpf<mode>df_mask): Ditto.
161 (*avx512pf_gatherpf<mode>df): Ditto.
162 (avx512pf_scatterpf<mode>sf): Ditto.
163 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
164 (*avx512pf_scatterpf<mode>sf): Ditto.
165 (avx512pf_scatterpf<mode>df): Ditto.
166 (*avx512pf_scatterpf<mode>df_mask): Ditto.
167 (*avx512pf_scatterpf<mode>df): Ditto.
168 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
170 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
172 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
173 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
174 (_mm512_mask_testn_epi64_mask): Move to ...
175 * config/i386/avx512cdintrin.h: Here.
176 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
177 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
178 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
179 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
180 TARGET_AVX512F from TARGET_AVX512CD.
182 2014-02-26 Richard Biener <rguenther@suse.de>
185 * ipa.c (walk_polymorphic_call_targets): Properly guard
186 call to inline_update_overall_summary.
188 2014-02-26 Bin Cheng <bin.cheng@arm.com>
191 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
192 and latches only if requested. Fix latch if it is removed.
193 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
194 LOOPS_HAVE_PREHEADERS.
196 2014-02-25 Andrew Pinski <apinski@cavium.com>
198 * builtins.c (expand_builtin_thread_pointer): Create a new target
199 when the target is NULL.
201 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
203 PR rtl-optimization/60317
204 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
205 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
206 * lra-assigns.c: Include params.h.
207 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
208 other reload pseudos considerations.
210 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
212 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
213 to use canonical form for nor<mode>3.
215 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
218 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
221 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
223 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
224 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
225 (ix86_handle_option): Handle OPT_mprefetchwt1.
226 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
227 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
229 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
230 OPTION_MASK_ISA_PREFETCHWT1.
231 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
232 (PTA_PREFETCHWT1): New.
233 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
234 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
235 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
236 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
237 (*prefetch_avx512pf_<mode>_: Change into ...
238 (*prefetch_prefetchwt1_<mode>: This.
239 * config/i386/i386.opt (mprefetchwt1): New.
240 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
241 (_mm_prefetch): Handle intent to write.
242 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
244 2014-02-25 Richard Biener <rguenther@suse.de>
247 * emit-rtl.c (mem_attrs_htab): Remove.
248 (mem_attrs_htab_hash): Likewise.
249 (mem_attrs_htab_eq): Likewise.
250 (set_mem_attrs): Always allocate new mem-attrs when something changed.
251 (init_emit_once): Do not allocate mem_attrs_htab.
253 2014-02-25 Richard Biener <rguenther@suse.de>
256 * lto-opts.c (lto_write_options): Output non-explicit conservative
257 -fwrapv, -fno-trapv and -fno-strict-overflow.
258 * lto-wrapper.c (merge_and_complain): Handle merging those options.
259 (run_gcc): And pass them through.
261 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
263 * sel-sched.c (calculate_new_fences): New parameter ptime.
264 Calculate it as a maximum over all fence cycles.
265 (sel_sched_region_2): Adjust the call to calculate_new_fences.
266 Print the final schedule timing when sched_verbose.
268 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
270 PR rtl-optimization/60292
271 * sel-sched.c (fill_vec_av_set): Do not reset target availability
272 bit fot the fence instruction.
274 2014-02-24 Alangi Derick <alangiderick@gmail.com>
276 * calls.h: Fix typo in comment.
278 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
280 * config/pa/pa.c (pa_output_move_double): Don't valididate when
281 adjusting offsetable addresses.
283 2014-02-24 Guozhi Wei <carrot@google.com>
285 * sparseset.h (sparseset_pop): Fix the wrong index.
287 2014-02-24 Walter Lee <walt@tilera.com>
289 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
290 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
292 * common/config/tilegx/tilegx-common.c
293 (TARGET_DEFAULT_TARGET_FLAGS): Define.
294 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
296 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
297 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
298 (tilegx_gimplify_va_arg_expr): Handle big endian.
299 (tilegx_expand_unaligned_load): Ditto.
300 (tilegx_expand_unaligned_store): Ditto.
301 (TARGET_RETURN_IN_MSB): New.
302 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
303 (TARGET_ENDIAN_DEFAULT): New.
304 (TARGET_BIG_ENDIAN): Handle big endian.
305 (BYTES_BIG_ENDIAN): Ditto.
306 (WORDS_BIG_ENDIAN): Ditto.
307 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
310 * config/tilegx/tilegx.md (extv): Handle big endian.
313 (insn_st<n>_add<bitsuffix>): Ditto.
314 (insn_stnt<n>): Ditto.
315 (insn_stnt<n>_add<bitsuffix>):Ditto.
316 (vec_interleave_highv8qi): Handle big endian.
317 (vec_interleave_highv8qi_be): New.
318 (vec_interleave_highv8qi_le): New.
319 (insn_v1int_h): Handle big endian.
320 (vec_interleave_lowv8qi): Handle big endian.
321 (vec_interleave_lowv8qi_be): New.
322 (vec_interleave_lowv8qi_le): New.
323 (insn_v1int_l): Handle big endian.
324 (vec_interleave_highv4hi): Handle big endian.
325 (vec_interleave_highv4hi_be): New.
326 (vec_interleave_highv4hi_le): New.
327 (insn_v2int_h): Handle big endian.
328 (vec_interleave_lowv4hi): Handle big endian.
329 (vec_interleave_lowv4hi_be): New.
330 (vec_interleave_lowv4hi_le): New.
331 (insn_v2int_l): Handle big endian.
332 (vec_interleave_highv2si): Handle big endian.
333 (vec_interleave_highv2si_be): New.
334 (vec_interleave_highv2si_le): New.
335 (insn_v4int_h): Handle big endian.
336 (vec_interleave_lowv2si): Handle big endian.
337 (vec_interleave_lowv2si_be): New.
338 (vec_interleave_lowv2si_le): New.
339 (insn_v4int_l): Handle big endian.
340 * config/tilegx/tilegx.opt (mbig-endian): New option.
341 (mlittle-endian): New option.
342 * doc/install.texi: Document tilegxbe-linux.
343 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
345 2014-02-24 Martin Jambor <mjambor@suse.cz>
348 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
349 there are no parameter descriptors.
351 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
353 PR rtl-optimization/60268
354 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
355 initialization to ...
356 (sched_rgn_init): ... here.
357 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
359 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
361 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
364 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
366 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
369 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
371 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
372 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
374 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
376 * config/microblaze/predicates.md: Add cmp_op predicate.
377 * config/microblaze/microblaze.md: Add branch_compare instruction
378 which uses cmp_op predicate and emits cmp insn before branch.
379 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
380 to microblaze_expand_conditional_branch and consolidate logic.
381 (microblaze_expand_conditional_branch): emit branch_compare
382 insn instead of handling cmp op separate from branch insn.
384 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
386 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
389 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
391 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
392 define_insn with define_expand and new define_insn
393 *altivec_lve<VI_char>x_internal.
394 (altivec_stve<VI_char>x): Replace define_insn with define_expand
395 and new define_insn *altivec_stve<VI_char>x_internal.
396 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
398 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
400 (altivec_expand_stvex_be): New function.
402 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
404 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
405 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
406 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
407 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
409 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
412 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
413 instead of emit_move_insn.
415 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
417 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
419 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
422 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
424 * config/rs6000/altivec.md (altivec_lvxl): Rename as
425 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
426 (altivec_lvxl_<mode>): New define_expand incorporating
427 -maltivec=be semantics where needed.
428 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
429 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
430 semantics where needed.
431 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
432 (altivec_stvx_<mode>): New define_expand incorporating
433 -maltivec=be semantics where needed.
434 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
435 VM2 iterator instead of V4SI.
436 (altivec_stvxl_<mode>): New define_expand incorporating
437 -maltivec=be semantics where needed.
438 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
439 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
440 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
441 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
442 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
443 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
444 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
445 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
446 ALTIVEC_BUILTIN_STVXL.
447 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
448 (altivec_expand_stvx_be): Likewise.
449 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
450 (altivec_expand_lvx_be): Likewise.
451 (altivec_expand_stvx_be): Likewise.
452 (altivec_expand_builtin): Add cases for
453 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
454 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
455 (altivec_init_builtins): Add definitions for
456 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
457 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
459 2014-02-21 Catherine Moore <clm@codesourcery.com>
461 * doc/invoke.texi (mvirt, mno-virt): Document.
462 * config/mips/mips.opt (mvirt): New option.
463 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
465 2014-02-21 Richard Biener <rguenther@suse.de>
467 PR tree-optimization/60276
468 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
469 (STMT_VINFO_MIN_NEG_DIST): New macro.
470 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
471 STMT_VINFO_MIN_NEG_DIST.
472 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
473 made for negative dependence distances still hold.
475 2014-02-21 Richard Biener <rguenther@suse.de>
478 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
479 DECL_INITIAL for globals not in the current function context.
481 2014-02-21 Jakub Jelinek <jakub@redhat.com>
483 PR tree-optimization/56490
484 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
485 * tree-ssa-uninit.c: Include params.h.
486 (compute_control_dep_chain): Add num_calls argument, return false
487 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
488 num_calls to recursive call.
489 (find_predicates): Change dep_chain into normal array,
490 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
491 variable and adjust compute_control_dep_chain caller.
492 (find_def_preds): Likewise.
494 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
496 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
497 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
499 2014-02-21 Nick Clifton <nickc@redhat.com>
501 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
503 (popqi1): Add mode to pre_dec.
506 2014-02-21 Jakub Jelinek <jakub@redhat.com>
508 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
509 mode for mask of V8SFmode permutation.
511 2014-02-20 Richard Henderson <rth@redhat.com>
514 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
515 a new pseudo for OLDVAL.
517 2014-02-20 Jakub Jelinek <jakub@redhat.com>
520 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
521 gen_reg_rtx if d->testing_p.
522 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
523 if d->testing_p and we will certainly return true.
524 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
527 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
529 * emit-rtl.c (gen_reg_rtx): Assert that
530 crtl->emit.regno_pointer_align_length is non-zero.
532 2014-02-20 Richard Henderson <rth@redhat.com>
535 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
536 on failure the store back into EXPECT.
538 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
539 Sandra Loosemore <sandra@codesourcery.com>
541 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
542 * config/nios2/nios2.c (nios2_function_profiler): Add
543 -fPIC (flag_pic == 2) support.
544 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
545 (nios2_large_offset_p): New function.
546 (nios2_unspec_reloc_p): Move up position, update to use
547 nios2_large_offset_p.
548 (nios2_unspec_address): Remove function.
549 (nios2_unspec_offset): New function.
550 (nios2_large_got_address): New function.
551 (nios2_got_address): Add large offset support.
552 (nios2_legitimize_tls_address): Update usage of removed and new
554 (nios2_symbol_binds_local_p): New function.
555 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
556 (nios2_legitimize_address): Update to use nios2_large_offset_p.
557 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
558 (nios2_print_operand): Merge H/L processing, add hiadj/lo
559 processing for (const (unspec ...)).
560 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
562 2014-02-20 Richard Biener <rguenther@suse.de>
564 * tree-cfg.c (replace_uses_by): Mark altered BBs before
565 doing the substitution.
566 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
568 2014-02-20 Martin Jambor <mjambor@suse.cz>
571 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
572 info when checking whether lattices are bottom.
574 2014-02-20 Richard Biener <rguenther@suse.de>
577 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
580 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
583 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
584 parameter specifying the scaling.
585 (inline_call): Update.
586 (want_inline_recursively): Guard division by zero.
587 (recursive_inlining): Update.
588 * ipa-inline.h (clone_inlined_nodes): Update.
590 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
593 * config/i386/i386.c (classify_argument): Pass structures of size
594 64 bytes or less in register.
596 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
597 Kirill Yukhin <kirill.yukhin@intel.com>
599 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
600 (_mm_rcp28_round_ss): Ditto.
601 (_mm_rsqrt28_round_sd): Ditto.
602 (_mm_rsqrt28_round_ss): Ditto.
603 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
604 (_mm_rcp14_round_ss): Ditto.
605 (_mm_rsqrt14_round_sd): Ditto.
606 (_mm_rsqrt14_round_ss): Ditto.
607 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
608 the first input operand, get rid of match_dup.
609 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
611 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
613 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
614 operand as the first input operand, set type attribute.
615 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
617 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
618 operand as the first input operand, set type attribute.
620 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
622 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
625 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
628 * config/i386/i386.c (construct_container): Remove TFmode check
629 for X86_64_INTEGER_CLASS.
631 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
634 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
635 only when -Wpsabi is enabled.
637 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
640 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
641 passing arrays in registers are the same as for structs, so remove the
642 special case for them.
644 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
646 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
647 destination type, extract only the valid bits if the source type is not
648 integral and has a different mode.
650 2014-02-19 Richard Biener <rguenther@suse.de>
653 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
656 2014-02-19 Richard Biener <rguenther@suse.de>
659 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
660 (ipa_modify_call_arguments): Emit an argument load explicitely and
661 preserve virtual SSA form there and for the replacement call.
662 Do not update SSA form nor free dominance info.
664 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
666 * ipa.c (function_and_variable_visibility): Also clear WEAK
667 flag when disolving COMDAT_GROUP.
669 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
671 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
672 * ipa-prop.c (ipa_set_jf_known_type): Return early when
674 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
675 do more sanity checks.
676 (detect_type_change): Return true when giving up early.
677 (compute_complex_assign_jump_func): Fix type parameter of
679 (compute_complex_ancestor_jump_func): Likewise.
680 (update_jump_functions_after_inlining): Fix updating of
682 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
684 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
686 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
687 inline clones when edge disappears.
689 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
692 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
693 Split 64-bit moves into 2 patterns. Do not allow the use of
694 direct move for TDmode in little endian, since the decimal value
695 has little endian bytes within a word, but the 64-bit pieces are
696 ordered in a big endian fashion, and normal subreg's of TDmode are
698 (mov<mode>_64bit_dm): Likewise.
699 (movtd_64bit_nodm): Likewise.
701 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
703 PR tree-optimization/60174
704 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
705 statement of an SSA_NAME that occurs in an abnormal PHI node.
707 2014-02-18 Jakub Jelinek <jakub@redhat.com>
710 * final.c (SEEN_BB): Remove.
711 (SEEN_NOTE, SEEN_EMITTED): Renumber.
712 (final_scan_insn): Don't force_source_line on second
713 NOTE_INSN_BASIC_BLOCK.
715 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
718 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
719 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
720 (type_natural_mode): Warn ABI change when %zmm register is not
721 available for AVX512F vector value passing.
723 2014-02-18 Kai Tietz <ktietz@redhat.com>
726 * config/i386/i386.c (ix86_expand_prologue): Use value in
727 rax register as displacement when restoring %r10 or %rax.
728 Fix wrong offset when restoring both registers.
730 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
732 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
733 assertion with conditional return.
735 2014-02-18 Jakub Jelinek <jakub@redhat.com>
736 Uros Bizjak <ubizjak@gmail.com>
739 * config/i386/driver-i386.c (host_detect_local_cpu): If
740 YMM state is not saved by the OS, also clear has_f16c. Move
741 CPUID 0x80000001 handling before YMM state saving checking.
743 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
745 PR rtl-optimization/58960
746 * haifa-sched.c (alloc_global_sched_pressure_data): New,
747 factored out from ...
748 (sched_init): ... here.
749 (free_global_sched_pressure_data): New, factored out from ...
750 (sched_finish): ... here.
751 * sched-int.h (free_global_sched_pressure_data): Declare.
752 * sched-rgn.c (nr_regions_initial): New static global.
753 (haifa_find_rgns): Initialize it.
754 (schedule_region): Disable sched-pressure for the newly
757 2014-02-17 Richard Biener <rguenther@suse.de>
759 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
760 release SSA defs of pattern stmts.
762 2014-02-17 Richard Biener <rguenther@suse.de>
764 * tree-inline.c (expand_call_inline): Release the virtual
765 operand defined by the call we are about to inline.
767 2014-02-17 Richard Biener <rguenther@suse.de>
769 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
771 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
772 Ilya Tocar <ilya.tocar@intel.com>
774 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
775 arguments order in builtin.
776 (_mm512_permutexvar_epi64): Ditto.
777 (_mm512_mask_permutexvar_epi64): Ditto
778 (_mm512_maskz_permutexvar_epi32): Ditto
779 (_mm512_permutexvar_epi32): Ditto
780 (_mm512_mask_permutexvar_epi32): Ditto
782 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
784 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
785 (p8_vmrgow): Likewise.
787 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
789 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
792 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
795 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
796 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
797 into 64-bit and 32-bit moves. On 64-bit moves, add support for
798 using direct move instructions on ISA 2.07. Also adjust
799 instruction length for 64-bit.
800 (mov<mode>_64bit, TFmode/TDmode): Likewise.
801 (mov<mode>_32bit, TFmode/TDmode): Likewise.
803 2014-02-15 Alan Modra <amodra@gmail.com>
807 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
808 find_replacement on parts of insn rtl that might be reloaded.
810 2014-02-15 Richard Biener <rguenther@suse.de>
812 PR tree-optimization/60183
813 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
814 (tree_ssa_phiprop): Calculate and free post-dominators.
816 2014-02-14 Jeff Law <law@redhat.com>
818 PR rtl-optimization/60131
819 * ree.c (get_extended_src_reg): New function.
820 (combine_reaching_defs): Use it rather than assuming location of REG.
821 (find_and_remove_re): Verify first operand of extension is
822 a REG before adding the insns to the copy list.
824 2014-02-14 Roland McGrath <mcgrathr@google.com>
826 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
827 * configure: Regenerated.
828 * config.in: Regenerated.
829 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
830 instead of ASM_SHORT.
832 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
833 Richard Earnshaw <rearnsha@arm.com>
835 PR rtl-optimization/59535
836 * lra-constraints.c (process_alt_operands): Encourage alternative
837 when unassigned pseudo class is superset of the alternative class.
838 (inherit_reload_reg): Don't inherit when optimizing for code size.
839 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
840 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
841 modes not less than 4 for Thumb1.
843 2014-02-14 Kyle McMartin <kyle@redhat.com>
846 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
848 2014-02-14 Richard Biener <rguenther@suse.de>
850 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
851 (get_frame_arg): Drop the assert with langhook types_compatible_p.
852 Do not strip INDIRECT_REFs.
854 2014-02-14 Richard Biener <rguenther@suse.de>
857 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
858 DECL_FUNCTION_SPECIFIC_TARGET.
859 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
860 * tree-streamer-out.c (pack_ts_target_option): Remove.
861 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
862 (write_ts_function_decl_tree_pointers): Do not stream
863 DECL_FUNCTION_SPECIFIC_TARGET.
864 * tree-streamer-in.c (unpack_ts_target_option): Remove.
865 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
866 (lto_input_ts_function_decl_tree_pointers): Do not stream
867 DECL_FUNCTION_SPECIFIC_TARGET.
869 2014-02-14 Jakub Jelinek <jakub@redhat.com>
871 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
872 (get_initial_def_for_induction, vectorizable_induction): Ignore
873 debug stmts when looking for exit_phi.
874 (vectorizable_live_operation): Fix up condition.
876 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
878 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
879 nreverse() because it changes the content of original tree list.
881 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
883 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
884 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
886 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
888 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
889 GNU coding standards.
891 2014-02-13 Jakub Jelinek <jakub@redhat.com>
894 * dwarf2out.c (gen_subprogram_die): Don't call
895 add_calling_convention_attribute if subr_die is old_die.
897 2014-02-13 Sharad Singhai <singhai@google.com>
899 * doc/optinfo.texi: Fix order of nodes.
901 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
903 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
904 operands[2], not operands[3].
906 2014-02-13 Richard Biener <rguenther@suse.de>
909 * doc/install.texi (ISL): Update recommended version to 0.12.2,
910 mention the possibility of an in-tree build.
911 (CLooG): Update recommended version to 0.18.1, mention the
912 possibility of an in-tree build and clarify that the ISL
913 bundled with CLooG does not work.
915 2014-02-13 Jakub Jelinek <jakub@redhat.com>
918 * expr.c (compress_float_constant): If x is a hard register,
919 extend into a pseudo and then move to x.
921 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
923 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
924 caused by bad second argument to warning_at() with -mhotpatch and
925 nested functions (e.g. with gfortran).
927 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
929 * opts.c (option_name): Remove "enabled by default" rider.
931 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
933 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
935 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
936 Uros Bizjak <ubizjak@gmail.com>
939 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
940 * configure: Regenerated.
942 2014-02-12 Richard Biener <rguenther@suse.de>
944 * vec.c (vec_prefix::calculate_allocation): Move as
945 inline variant to vec.h.
946 (vec_prefix::calculate_allocation_1): New out-of-line version.
947 * vec.h (vec_prefix::calculate_allocation_1): Declare.
948 (vec_prefix::m_has_auto_buf): Rename to ...
949 (vec_prefix::m_using_auto_storage): ... this.
950 (vec_prefix::calculate_allocation): Inline the easy cases
951 and dispatch to calculate_allocation_1 which doesn't need the
953 (va_heap::reserve): Use gcc_checking_assert.
954 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
955 m_using_auto_storage.
956 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
958 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
959 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
960 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
962 2014-02-12 Richard Biener <rguenther@suse.de>
964 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
965 when we found a dependence.
967 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
969 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
971 (maybe_fold_stmt): ... into this new function.
972 * omp-low.c (lower_omp): Update comment.
974 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
977 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
980 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
982 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
983 identifiers in comments.
984 (cortexa53_extra_costs): Likewise.
985 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
986 (cortexa7_extra_costs): Likewise.
987 (cortexa12_extra_costs): Likewise.
988 (cortexa15_extra_costs): Likewise.
989 (v7m_extra_costs): Likewise.
991 2014-02-12 Richard Biener <rguenther@suse.de>
994 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
995 of posix_memalign being successful.
996 (lower_stmt): Restrict lowering of posix_memalign to when
997 -ftree-bit-ccp is enabled.
999 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1001 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1003 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1005 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1007 PR rtl-optimization/60116
1008 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1009 other_insn once the combination has been validated.
1011 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1014 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1016 * ipa-devirt.c: Include demangle.h
1017 (odr_violation_reported): New static variable.
1018 (add_type_duplicate): Update odr_violations.
1019 (maybe_record_node): Add completep parameter; update it.
1020 (record_target_from_binfo): Add COMPLETEP parameter;
1021 update it as needed.
1022 (possible_polymorphic_call_targets_1): Likewise.
1023 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1024 rename FINAL to COMPLETE.
1025 (record_targets_from_bases): Sanity check we found the binfo;
1026 fix COMPLETEP updating.
1027 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1028 parameter, fix computing of COMPLETEP.
1029 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1030 at LTO time do demangling.
1031 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1032 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1034 (gimple_get_virt_method_for_binfo): Likewise.
1035 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1036 gimple_get_virt_method_for_vtable): Update prototypes.
1038 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1041 * genautomata.c (add_presence_absence): Fix typo with
1042 {final_}presence_list.
1044 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1047 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1048 for VSX/Altivec vectors that land in GPR registers.
1050 2014-02-11 Richard Henderson <rth@redhat.com>
1051 Jakub Jelinek <jakub@redhat.com>
1054 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1055 around drhs if type conversion to lacc->type is not useless.
1057 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1059 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1061 (cortex-a57.cortex-a53): Likewise.
1062 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1064 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1066 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1069 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1071 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1072 add_options_for_arm_vfp3.
1074 2014-02-11 Jeff Law <law@redhat.com>
1077 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1078 object with an undesirable mode.
1080 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1083 * config/i386/sol2-9.h: New file.
1084 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1085 *-*-solaris2.9*): Use it.
1087 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1089 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1090 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1092 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1094 * config/microblaze/microblaze.c: Extend mcpu version format
1096 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1098 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1100 2014-02-10 Richard Henderson <rth@redhat.com>
1103 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1104 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1105 ms-abi vs -mno-accumulate-outgoing-args.
1106 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1107 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1110 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1113 * cfgexpand.c (expand_asm_operands): Attach source location to
1114 ASM_INPUT rtx objects.
1115 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1117 2014-02-10 Nick Clifton <nickc@redhat.com>
1119 * config/mn10300/mn10300.c (popcount): New function.
1120 (mn10300_expand_prologue): Include saved registers in stack usage
1123 2014-02-10 Jeff Law <law@redhat.com>
1126 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1127 when changing the SET_DEST of a prior insn to avoid an input reload.
1129 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1131 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1132 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1133 -mcall-openbsd, or -mcall-linux.
1134 (CC1_ENDIAN_BIG_SPEC): Remove.
1135 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1136 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1137 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1138 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1139 and %cc1_endian_default.
1140 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1142 2014-02-10 Richard Biener <rguenther@suse.de>
1144 PR tree-optimization/60115
1145 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1146 MEM_REF handling. Properly verify that the accesses are not
1147 out of the objects bound.
1149 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1151 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1154 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1156 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1157 proper constants and fix formatting.
1158 (possible_polymorphic_call_targets): Fix formatting.
1160 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1161 Ilya Tocar <ilya.tocar@intel.com>
1163 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1164 (_mm512_loadu_epi32): Renamed into...
1165 (_mm512_loadu_si512): This.
1166 (_mm512_storeu_epi32): Renamed into...
1167 (_mm512_storeu_si512): This.
1168 (_mm512_maskz_ceil_ps): Removed.
1169 (_mm512_maskz_ceil_pd): Ditto.
1170 (_mm512_maskz_floor_ps): Ditto.
1171 (_mm512_maskz_floor_pd): Ditto.
1172 (_mm512_floor_round_ps): Ditto.
1173 (_mm512_floor_round_pd): Ditto.
1174 (_mm512_ceil_round_ps): Ditto.
1175 (_mm512_ceil_round_pd): Ditto.
1176 (_mm512_mask_floor_round_ps): Ditto.
1177 (_mm512_mask_floor_round_pd): Ditto.
1178 (_mm512_mask_ceil_round_ps): Ditto.
1179 (_mm512_mask_ceil_round_pd): Ditto.
1180 (_mm512_maskz_floor_round_ps): Ditto.
1181 (_mm512_maskz_floor_round_pd): Ditto.
1182 (_mm512_maskz_ceil_round_ps): Ditto.
1183 (_mm512_maskz_ceil_round_pd): Ditto.
1184 (_mm512_expand_pd): Ditto.
1185 (_mm512_expand_ps): Ditto.
1186 * config/i386/i386.c (ix86_builtins): Remove
1187 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1188 (bdesc_args): Ditto.
1189 * config/i386/predicates.md (const1256_operand): New.
1190 (const_1_to_2_operand): Ditto.
1191 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1192 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1193 (*avx512pf_gatherpf<mode>sf): Ditto.
1194 (avx512pf_gatherpf<mode>df): Ditto.
1195 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1196 (*avx512pf_gatherpf<mode>df): Ditto.
1197 (avx512pf_scatterpf<mode>sf): Ditto.
1198 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1199 (*avx512pf_scatterpf<mode>sf): Ditto.
1200 (avx512pf_scatterpf<mode>df): Ditto.
1201 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1202 (*avx512pf_scatterpf<mode>df): Ditto.
1203 (avx512f_expand<mode>): Removed.
1204 (<shift_insn><mode>3<mask_name>): Change predicate type.
1206 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1208 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1209 not at the end of datarefs vector use ordered_remove to avoid
1210 reordering datarefs vector.
1213 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1214 mark local addressable non-static vars as GOVD_PRIVATE
1215 instead of GOVD_LOCAL.
1216 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1217 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1221 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1222 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1223 assume_aligned or alloc_align attributes.
1224 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1225 arguments. Handle also assume_aligned and alloc_align attributes.
1226 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1227 calls to functions with assume_aligned or alloc_align attributes.
1228 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1230 2014-02-08 Terry Guo <terry.guo@arm.com>
1232 * doc/invoke.texi: Document ARM -march=armv7e-m.
1234 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1236 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1237 flag on __cilkrts_rethrow builtin.
1240 * ipa-cp.c (determine_versionability): Fail at -O0
1241 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1242 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1245 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1248 * tree-inline.c (copy_forbidden): Fail for
1249 __attribute__((optimize (0))) functions.
1251 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1253 * varpool.c: Include pointer-set.h.
1254 (varpool_remove_unreferenced_decls): Variables in other partitions
1255 will not be output; be however careful to not lose information
1258 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1260 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1261 lookup in the vtable constructor.
1263 2014-02-07 Jeff Law <law@redhat.com>
1266 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1267 define_insn_and_split.
1269 * ipa-inline.c (inline_small_functions): Fix typos.
1271 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1273 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1274 (s390_can_use_return_insn): Declare.
1275 * config/s390/s390.h (EPILOGUE_USES): Define.
1276 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1278 (s390_chunkify_start): Handle return JUMP_LABELs.
1279 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1280 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1281 (s390_can_use_return_insn): New functions.
1282 (s390_fix_long_loop_prediction): Handle conditional returns.
1283 (TARGET_SET_UP_BY_PROLOGUE): Define.
1284 * config/s390/s390.md (ANY_RETURN): New code iterator.
1285 (*creturn, *csimple_return, return, simple_return): New patterns.
1287 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1289 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1290 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1291 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1292 REG_CFA_RESTORE list when deciding not to restore a register.
1294 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1296 * config/s390/s390.c: Include tree-pass.h and context.h.
1297 (s390_early_mach): New function, split out from...
1298 (s390_emit_prologue): ...here.
1299 (pass_data_s390_early_mach): New pass structure.
1300 (pass_s390_early_mach): New class.
1301 (s390_option_override): Create and register early_mach pass.
1302 Move to end of file.
1304 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1306 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1307 to match for the exit block.
1309 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1311 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1312 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1313 Reject misaligned operands.
1315 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1317 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1319 2014-02-07 Richard Biener <rguenther@suse.de>
1322 * gimple-low.c (lower_builtin_posix_memalign): New function.
1323 (lower_stmt): Call it to lower posix_memalign in a way
1324 to make alignment info accessible.
1326 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1329 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1330 __builtin_setjmp_receiver.
1332 2014-02-07 Richard Biener <rguenther@suse.de>
1335 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1336 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1337 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1338 Handle BUILT_IN_POSIX_MEMALIGN.
1339 (find_func_clobbers): Likewise.
1340 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1341 (call_may_clobber_ref_p_1): Likewise.
1343 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1346 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1349 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1352 * lto-cgraph.c (lto_output_node): Use
1353 symtab_get_symbol_partitioning_class.
1354 (lto_output_varpool_node): likewise.
1355 (symtab_get_symbol_partitioning_class): Move here from
1357 * cgraph.h (symbol_partitioning_class): Likewise.
1358 (symtab_get_symbol_partitioning_class): Declare.
1360 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1362 * ggc.h (ggc_internal_cleared_alloc): New macro.
1363 * vec.h (vec_safe_copy): Handle memory stats.
1364 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1365 * target-globals.c (save_target_globals): Likewise.
1367 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1370 * expr.c (emit_move_resolve_push): Export; be bit more selective
1371 on when to clear alias set.
1372 * expr.h (emit_move_resolve_push): Declare.
1373 * function.h (struct function): Add tail_call_marked.
1374 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1375 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1376 * config/i386/i386.md (TImode move expander): De not call
1378 (FP push expanders): Preserve memory attributes.
1379 * config/i386/sse.md (push<mode>1): Remove.
1380 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1381 (ix86_expand_push): Remove.
1382 * config/i386/mmx.md (push<mode>1): Remove.
1384 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1386 PR rtl-optimization/60030
1387 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1388 lopart with paradoxical subreg before shifting it up by hprec.
1390 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1392 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1393 Remove extra newline at end of file.
1394 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1395 (arm_issue_rate): Handle cortexa57.
1396 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1397 (cortex-a57.cortex-a53): Likewise.
1399 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1402 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1403 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1405 (arm_expand_prologue): Adjust all callers.
1406 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1407 info, registers also at the lowest numbered registers side. Use
1408 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1412 * var-tracking.c (adjust_mems): Before adding a SET to
1413 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1415 2014-02-06 Alan Modra <amodra@gmail.com>
1418 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1419 change SDmode to DDmode when lra_in_progress.
1421 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1424 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1425 free_data_ref on the dr first, and before goto again also set dr
1426 to the next dr. For simd_lane_access, free old datarefs[i] before
1427 overwriting it. For get_vectype_for_scalar_type failure, don't
1428 free_data_ref if simd_lane_access.
1430 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1433 * tree.h (opts_for_fn): New inline function.
1434 (opt_for_fn): Define.
1435 * config/i386/i386.c (ix86_function_regparm): Use
1436 opt_for_fn (decl, optimize) instead of optimize.
1438 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1440 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1441 for SYMBOL_REF in large memory model.
1443 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1445 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1447 (cortex-a57): Likewise.
1448 (cortex-a57.cortex-a53): Likewise.
1450 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1451 Kugan Vivekanandarajah <kuganv@linaro.org>
1453 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1455 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1457 2014-02-06 Richard Biener <rguenther@suse.de>
1459 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1460 set_loop_copy and initialize_original_copy_tables.
1462 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1464 * config/aarch64/aarch64-simd.md
1465 (aarch64_ashr_simddi): Change QI to SI.
1467 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1468 Jakub Jelinek <jakub@redhat.com>
1471 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1474 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1476 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1477 CODE_FOR_altivec_vpku[hw]um to
1478 CODE_FOR_altivec_vpku[hw]um_direct.
1479 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1480 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1481 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1482 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1484 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1486 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1487 generation for -maltivec=be.
1488 (altivec_vsumsws): Simplify redundant test.
1490 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1492 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1493 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1494 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1495 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1496 gen_altivec_vpkuwum.
1497 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1499 (altivec_vpks<VI_char>ss): Likewise.
1500 (altivec_vpks<VI_char>us): Likewise.
1501 (altivec_vpku<VI_char>us): Likewise.
1502 (altivec_vpku<VI_char>um): Likewise.
1503 (altivec_vpku<VI_char>um_direct): New (copy of
1504 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1506 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1507 target is little endian and -maltivec=be is not specified.
1508 (*altivec_vupkhs<VU_char>_direct): New (copy of
1509 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1510 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1511 target is little endian and -maltivec=be is not specified.
1512 (*altivec_vupkls<VU_char>_direct): New (copy of
1513 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1514 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1515 little endian and -maltivec=be is not specified.
1516 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1517 little endian and -maltivec=be is not specified.
1519 2014-02-05 Richard Henderson <rth@redhat.com>
1522 * combine-stack-adj.c: Revert r206943.
1523 * sched-int.h (struct deps_desc): Add last_args_size.
1524 * sched-deps.c (init_deps): Initialize it.
1525 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1526 contain REG_ARGS_SIZE notes.
1528 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1530 * lto-cgraph.c (asm_nodes_output): Make global.
1531 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1532 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1533 (driver_handle_option): Handle OPT_fwpa.
1535 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1538 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1539 a comment typo and formatting issue. If odr_hash hasn't been
1540 created, return vNULL and set *completep to false.
1543 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1544 bb with no successors.
1546 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1549 * doc/invoke.texi (-march): Clarify documentation for ARM.
1553 2014-02-05 Richard Biener <rguenther@suse.de>
1555 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1556 when not vectorizing because of too many alias checks.
1557 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1558 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1560 2014-02-05 Nick Clifton <nickc@redhat.com>
1562 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1563 accept extended registers in any mode when compiling for the MN10300.
1565 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1567 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1568 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1569 sanitization attributes.
1570 (can_inline_edge_p): Likewise.
1571 (sanitize_attrs_match_for_inline_p): New function.
1573 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1575 * ipa-prop.c (detect_type_change): Shor circuit testing of
1576 type changes on THIS pointer.
1578 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1581 * config/pa/pa.c (legitimize_tls_address): Return original address
1582 if not passed a SYMBOL_REF rtx.
1583 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1585 (pa_emit_move_sequence): Simplify TLS source operands.
1586 (pa_legitimate_constant_p): Reject all TLS constants.
1587 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1588 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1590 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1592 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1593 groups when we know they are controlled by LTO.
1594 * varasm.c (default_binds_local_p_1): If object is in other partition,
1595 it will be resolved locally.
1597 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1599 * config/host-linux.c (linux_gt_pch_use_address): Don't
1600 use SSIZE_MAX because it is not always defined.
1602 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1605 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1606 threshold for pseudo splitting.
1607 (update_ebb_live_info): Process call argument hard registers and
1608 hard registers from insn definition too.
1609 (max_small_class_regs_num): New constant.
1610 (inherit_in_ebb): Update live hard regs through EBBs. Update
1611 reloads_num only for small register classes. Don't split for
1614 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1617 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1620 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1622 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1625 2014-02-04 Marek Polacek <polacek@redhat.com>
1627 * gdbinit.in (pel): Define.
1629 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1631 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1634 2014-02-04 Richard Biener <rguenther@suse.de>
1637 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1638 in function context local.
1639 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1640 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1641 similar to LTO_imported_decl_ref.
1643 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1645 PR tree-optimization/60002
1646 * cgraphclones.c (build_function_decl_skip_args): Clear
1649 PR tree-optimization/60023
1650 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1651 false to gsi_replace.
1652 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1653 has been in some EH region and vec_stmt could throw, add
1654 vec_stmt into the same EH region.
1655 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1656 has no lhs, ignore it.
1657 * internal-fn.c (expand_MASK_LOAD): Likewise.
1660 * tree-inline.c (copy_forbidden): Fail for
1661 __attribute__((optimize (0))) functions.
1664 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1665 is set, copy one less argument.
1666 (expand_simd_clones): Don't subtract clone_info->inbranch
1667 from simd_clone_struct_alloc argument.
1669 PR rtl-optimization/57915
1670 * recog.c (simplify_while_replacing): If all unary/binary/relational
1671 operation arguments are constant, attempt to simplify those.
1674 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1675 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1677 2014-02-04 Richard Biener <rguenther@suse.de>
1679 PR tree-optimization/60012
1680 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1681 TBAA disambiguation to all DDRs.
1683 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1686 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1687 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1689 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1692 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1694 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1696 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1697 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1699 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1702 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1703 to figure out targets of polymorphic calls with known decl.
1704 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1705 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1706 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1707 (get_polymorphic_call_info): ... here.
1708 (get_polymorphic_call_info_from_invariant): New function.
1710 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1712 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1713 lookup via vtable pointer; check for type consistency
1714 and turn inconsitent facts into UNREACHABLE.
1715 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1716 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1717 type inconsistent querries; return UNREACHABLE instead.
1719 2014-02-03 Richard Henderson <rth@twiddle.net>
1722 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1723 already processed this node.
1724 (normalize_one_pred_1): Pass along mark_set.
1725 (normalize_one_pred): Create and destroy a pointer_set_t.
1726 (normalize_one_pred_chain): Likewise.
1728 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1730 PR gcov-profile/58602
1731 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1733 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1736 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1737 -fno-devirtualize; try to devirtualize by the knowledge of
1738 virtual table pointer given by aggregate propagation.
1739 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1740 (ipa_print_node_jump_functions): Dump also offset that
1741 is relevant for polymorphic calls.
1742 (determine_known_aggregate_parts): Add arg_type parameter; use it
1743 instead of determining the type from pointer type.
1744 (ipa_compute_jump_functions_for_edge): Update call of
1745 determine_known_aggregate_parts.
1746 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1747 (gimple_get_virt_method_for_binfo): ... here; simplify using
1748 vtable_pointer_value_to_vtable.
1749 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1750 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1751 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1752 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1754 (vtable_pointer_value_to_binfo): ... here.
1755 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1757 2014-02-03 Teresa Johnson <tejohnson@google.com>
1759 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1760 redef of outer loop index variable.
1762 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1766 * doc/extend.texi (Function Attributes): Typo.
1768 2014-02-03 Cong Hou <congh@google.com>
1770 PR tree-optimization/60000
1771 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1772 if the vectorized statement is a store. A store statement can only
1773 appear at the end of pattern statements.
1775 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1777 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1778 (ix86_option_override_internal): Default long double to 64-bit for
1779 32-bit Bionic and to 128-bit for 64-bit Bionic.
1781 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1782 TARGET_LONG_DOUBLE_128 is true.
1783 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1785 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1786 (mlong-double-64): Negate -mlong-double-128.
1787 (mlong-double-128): New option.
1789 * config/i386/i386-c.c (ix86_target_macros): Define
1790 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1792 * doc/invoke.texi: Document -mlong-double-128.
1794 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1796 PR rtl-optimization/60024
1797 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1799 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1801 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1803 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1805 PR rtl-optimization/57662
1806 * sel-sched.c (code_motion_path_driver): Do not mark already not
1807 existing blocks in the visiting bitmap.
1809 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1811 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1812 on the insn being emitted.
1814 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1815 Will Deacon <will.deacon@arm.com>
1817 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1819 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1821 * config/arm/arm-tables.opt: Regenerate.
1823 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1825 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1826 for vector types other than V16QImode.
1827 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1828 define_expand, and call altivec_expand_vec_perm_le when producing
1829 code with little endian element order.
1830 (*altivec_vperm_<mode>_internal): New insn having previous
1831 behavior of altivec_vperm_<mode>.
1832 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1833 altivec_expand_vec_perm_le when producing code with little endian
1835 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1836 behavior of altivec_vperm_<mode>_uns.
1838 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1840 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1841 (altivec_vsumsws): Add handling for -maltivec=be with a little
1843 (altivec_vsumsws_direct): New.
1844 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1845 gen_altivec_vsumsws.
1847 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1849 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1850 vtable_pointer_value_to_binfo): New functions.
1851 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1852 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1854 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1856 * config/nios2/nios2.md (load_got_register): Initialize GOT
1857 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1858 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1860 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1862 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1863 preserverd by passthrough, do not propagate the type.
1865 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1867 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1868 (mips_atomic_assign_expand_fenv): New function.
1869 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1871 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1873 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1874 (__builtin_mips_set_fcsr): Likewise.
1875 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1876 MIPS_USI_FTYPE_VOID.
1877 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1878 (mips16_expand_set_fcsr): Likewise.
1879 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1880 (mips16_set_fcsr_stub): Likewise.
1881 (mips16_get_fcsr_one_only_stub): New class.
1882 (mips16_set_fcsr_one_only_stub): Likewise.
1883 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1884 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1885 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1886 (hard_float): New availability predicate.
1887 (mips_builtins): Add get_fcsr and set_fcsr.
1888 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1889 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1890 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1891 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1892 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1895 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1897 * config/mips/mips.c (mips_one_only_stub): New class.
1898 (mips_need_mips16_rdhwr_p): Replace with...
1899 (mips16_rdhwr_stub): ...this new variable.
1900 (mips16_stub_call_address): New function.
1901 (mips16_rdhwr_one_only_stub): New class.
1902 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1903 (mips_output_mips16_rdhwr): Delete.
1904 (mips_finish_stub): New function.
1905 (mips_code_end): Use it to handle rdhwr stubs.
1907 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1910 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1911 when calculating size of integer atomic types.
1913 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1915 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1917 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1919 PR tree-optimization/60003
1920 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1921 * profile.c (branch_prob): Use gimple_call_builtin_p
1922 to check for BUILT_IN_SETJMP_RECEIVER.
1923 * tree-inline.c (copy_bb): Call notice_special_calls.
1925 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1928 * lra-constraints.c (process_alt_operands): Update reload_sum only
1931 2014-01-31 Richard Henderson <rth@redhat.com>
1934 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1935 until after else_eh is processed.
1937 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1939 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1940 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1941 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1942 in smmintrin.h, remove them.
1943 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1944 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1945 * config/i386/i386.md (ROUND_SAE): Fix value.
1946 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1947 (const48_operand): New.
1948 * config/i386/subst.md (round), (round_expand): Use
1949 const_4_or_8_to_11_operand.
1950 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1952 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1954 * config/i386/constraints.md (Yk): Swap meaning with k.
1955 * config/i386/i386.md (movhi_internal): Change Yk to k.
1956 (movqi_internal): Ditto.
1957 (*k<logic><mode>): Ditto.
1960 (kandn<mode>): Ditto.
1961 (*<code>hi_1): Ditto.
1962 (*<code>qi_1): Ditto.
1963 (kxnor<mode>): Ditto.
1964 (kortestzhi): Ditto.
1965 (kortestchi): Ditto.
1967 (*one_cmplhi2_1): Ditto.
1968 (*one_cmplqi2_1): Ditto.
1969 * config/i386/sse.md (): Change k to Yk.
1970 (avx512f_load<mode>_mask): Ditto.
1971 (avx512f_blendm<mode>): Ditto.
1972 (avx512f_store<mode>_mask): Ditto.
1973 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1974 (avx512f_storedqu<mode>_mask): Ditto.
1975 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1977 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1978 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1979 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1980 (avx512f_maskcmp<mode>3): Ditto.
1981 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1982 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1983 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1984 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1985 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1986 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1987 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1988 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1989 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1990 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1991 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1992 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1993 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1994 (vec_extract_lo_<mode>_maskm): Ditto.
1995 (vec_extract_hi_<mode>_maskm): Ditto.
1996 (avx512f_vternlog<mode>_mask): Ditto.
1997 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1998 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1999 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2000 (avx512f_<code>v8div16qi2_mask): Ditto.
2001 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2002 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2003 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2004 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2005 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2006 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2007 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2008 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2009 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2010 (avx512cd_maskb_vec_dupv8di): Ditto.
2011 (avx512cd_maskw_vec_dupv16si): Ditto.
2012 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2013 (avx512f_vpermi2var<mode>3_mask): Ditto.
2014 (avx512f_vpermi2var<mode>3_mask): Ditto.
2015 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2016 (*avx512f_gathersi<mode>): Ditto.
2017 (*avx512f_gathersi<mode>_2): Ditto.
2018 (*avx512f_gatherdi<mode>): Ditto.
2019 (*avx512f_gatherdi<mode>_2): Ditto.
2020 (*avx512f_scattersi<mode>): Ditto.
2021 (*avx512f_scatterdi<mode>): Ditto.
2022 (avx512f_compress<mode>_mask): Ditto.
2023 (avx512f_compressstore<mode>_mask): Ditto.
2024 (avx512f_expand<mode>_mask): Ditto.
2025 * config/i386/subst.md (mask): Change k to Yk.
2026 (mask_scalar_merge): Ditto.
2029 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2031 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2033 2014-01-31 Richard Biener <rguenther@suse.de>
2036 * builtins.c (fold_builtin_memory_op): Make sure to not
2037 use a floating-point mode or a boolean or enumeral type for
2040 2014-01-30 DJ Delorie <dj@redhat.com>
2042 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2043 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2044 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2045 whenever main() has an epilogue.
2047 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2049 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2050 unused variable "field".
2051 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2052 (vsx_mergeh_<mode>): Likewise.
2053 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2054 (altivec_vmrghh): Likewise.
2055 (altivec_vmrghw): Likewise.
2056 (altivec_vmrglb): Likewise.
2057 (altivec_vmrglh): Likewise.
2058 (altivec_vmrglw): Likewise.
2059 (altivec_vspltb): Add missing uses.
2060 (altivec_vsplth): Likewise.
2061 (altivec_vspltw): Likewise.
2062 (altivec_vspltsf): Likewise.
2064 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2067 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2068 frame related instructions.
2070 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2072 PR rtl-optimization/59959
2073 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2074 any reload of register whose subreg is invalid.
2076 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2078 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2079 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2080 Add missing return type - void.
2082 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2084 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2085 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2086 remove element index adjustment for endian (now handled in vsx.md
2088 (altivec_expand_vec_perm_const): Use
2089 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2090 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2091 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2092 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2093 define_expand and a new define_insn *altivec_vspltb_internal;
2094 adjust for -maltivec=be on a little endian target.
2095 (altivec_vspltb_direct): New.
2096 (altivec_vsplth): Divide into a define_expand and a new
2097 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2098 little endian target.
2099 (altivec_vsplth_direct): New.
2100 (altivec_vspltw): Divide into a define_expand and a new
2101 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2102 little endian target.
2103 (altivec_vspltw_direct): New.
2104 (altivec_vspltsf): Divide into a define_expand and a new
2105 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2106 a little endian target.
2108 2014-01-30 Richard Biener <rguenther@suse.de>
2110 PR tree-optimization/59993
2111 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2112 can propagate form the earlier stmt and avoid the transform
2113 when the intermediate result is needed.
2115 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2117 * README.Portability: Fix typo.
2119 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2121 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2122 comparison_operator with ordered_comparison_operator.
2124 2014-01-30 Nick Clifton <nickc@redhat.com>
2126 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2127 Rename to mn10300_store_multiple_regs.
2128 * config/mn10300/mn10300.c: Likewise.
2129 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2130 store_multiple_regs.
2131 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2132 Call mn10300_store_multiple_regs.
2134 2014-01-30 Nick Clifton <nickc@redhat.com>
2135 DJ Delorie <dj@redhat.com>
2137 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2138 %fp 2 to keep registers after it properly word-aligned.
2139 (rl78_alloc_physical_registers_umul): Handle the case where both
2140 input operands are the same.
2142 2014-01-30 Richard Biener <rguenther@suse.de>
2144 PR tree-optimization/59903
2145 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2148 2014-01-30 Jason Merrill <jason@redhat.com>
2151 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2154 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2156 2014-01-30 Richard Biener <rguenther@suse.de>
2158 PR tree-optimization/59951
2159 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2161 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2164 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2165 SFmode to DFmode case.
2167 2014-01-29 DJ Delorie <dj@redhat.com>
2169 * config/msp430/msp430.opt (-minrt): New.
2170 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2172 (ENDFILE_SPEC): Likewise.
2174 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2176 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2177 (estimate_function_body_sizes): Use it.
2179 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2182 * dwarf2out.c (is_cxx_auto): New.
2183 (is_base_type): Use it.
2184 (gen_type_die_with_usage): Likewise.
2186 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2188 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2189 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2190 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2191 -maltivec=be with LE targets.
2192 (vsx_mergeh_<mode>): Likewise.
2193 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2194 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2195 (altivec_vmrghb): Replace with define_expand and new
2196 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2197 (altivec_vmrghb_direct): New define_insn.
2198 (altivec_vmrghh): Replace with define_expand and new
2199 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2200 (altivec_vmrghh_direct): New define_insn.
2201 (altivec_vmrghw): Replace with define_expand and new
2202 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2203 (altivec_vmrghw_direct): New define_insn.
2204 (*altivec_vmrghsf): Adjust for endianness.
2205 (altivec_vmrglb): Replace with define_expand and new
2206 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2207 (altivec_vmrglb_direct): New define_insn.
2208 (altivec_vmrglh): Replace with define_expand and new
2209 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2210 (altivec_vmrglh_direct): New define_insn.
2211 (altivec_vmrglw): Replace with define_expand and new
2212 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2213 (altivec_vmrglw_direct): New define_insn.
2214 (*altivec_vmrglsf): Adjust for endianness.
2215 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2216 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2217 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2218 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2219 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2220 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2221 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2222 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2224 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2226 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2227 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2230 2014-01-29 Richard Biener <rguenther@suse.de>
2232 PR tree-optimization/58742
2233 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2234 associate_pointerplus_align.
2235 (associate_pointerplus_diff): New function.
2236 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2237 and associate_pointerplus_diff.
2239 2014-01-29 Richard Biener <rguenther@suse.de>
2241 * lto-streamer.h (LTO_major_version): Bump to 3.
2242 (LTO_minor_version): Reset to 0.
2244 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2246 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2247 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2248 (arm_file_start): Generate correct asm header for armv7ve.
2249 * config/arm/bpabi.h: Add multilib support for armv7ve.
2250 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2251 and cortex-a15 to armv7ve.
2252 * config/arm/t-aprofile: Add multilib support for armv7ve.
2253 * doc/invoke.texi: Document -march=armv7ve.
2255 2014-01-29 Richard Biener <rguenther@suse.de>
2257 PR tree-optimization/58742
2258 * tree-ssa-forwprop.c (associate_plusminus): Return true
2259 if we changed sth, defer EH cleanup to ...
2260 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2261 (simplify_mult): New function.
2263 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2266 PR tree-optimization/59920
2267 * tree.c (build_common_builtin_nodes): Remove
2268 __builtin_setjmp_dispatcher initialization.
2269 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2270 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2271 instead of gsi_after_labels + manually skipping debug stmts.
2272 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2273 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2274 * tree-inline.c (copy_edges_for_bb): Remove
2275 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2276 argument. Ignore computed_goto_p stmts. Don't call
2277 make_abnormal_goto_edges. If a call might need abnormal edges
2278 for non-local gotos, see if it already has an edge to
2279 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2280 with true argument, don't do anything then, otherwise add
2281 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2282 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2284 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2285 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2286 (lower_stmt): Don't set data->calls_builtin_setjmp.
2287 (lower_builtin_setjmp): Adjust comment.
2288 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2289 * tree-cfg.c (found_computed_goto): Remove.
2290 (factor_computed_gotos): Remove.
2291 (make_goto_expr_edges): Return bool, true for computed gotos.
2292 Don't call make_abnormal_goto_edges.
2293 (build_gimple_cfg): Don't set found_computed_goto, don't call
2294 factor_computed_gotos.
2295 (computed_goto_p): No longer static.
2296 (make_blocks): Don't set found_computed_goto.
2297 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2298 (make_edges): If make_goto_expr_edges returns true, push bb
2299 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2300 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2301 vector. Record mapping between bbs and OpenMP regions if there
2302 are any, adjust make_gimple_omp_edges caller. Call
2303 handle_abnormal_edges.
2304 (make_abnormal_goto_edges): Remove.
2305 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2306 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2307 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2308 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2309 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2310 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2311 filling *region also set *region_idx to (*region)->entry->index.
2314 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2315 For REGs set ORIGINAL_REGNO.
2317 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2319 * doc/md.texi: Mention that a target shouldn't implement
2320 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2323 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2325 PR tree-optimization/59594
2326 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2327 a copy of the datarefs vector rather than the vector itself.
2329 2014-01-28 Jason Merrill <jason@redhat.com>
2332 * dwarf2out.c (auto_die): New static.
2333 (gen_type_die_with_usage): Handle C++1y 'auto'.
2334 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2337 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2340 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2341 (SPEC_X32): Likewise.
2342 (SPEC_64): Likewise.
2343 * config/i386/i386.c (ix86_option_override_internal): Turn off
2344 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2346 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2347 * config/i386/i386.h (TARGET_16BIT): New macro.
2348 (TARGET_16BIT_P): Likewise.
2349 * config/i386/i386.opt: Add m16.
2350 * doc/invoke.texi: Document -m16.
2352 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2354 PR preprocessor/59935
2355 * input.c (location_get_source_line): Bail out on when line number
2356 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2358 2014-01-28 Richard Biener <rguenther@suse.de>
2360 PR tree-optimization/58742
2361 * tree-ssa-forwprop.c (associate_plusminus): Handle
2362 pointer subtraction of the form (T)(P + A) - (T)P.
2364 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2366 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2369 2014-01-28 Richard Biener <rguenther@suse.de>
2372 2014-01-28 Richard Biener <rguenther@suse.de>
2374 PR rtl-optimization/45364
2375 PR rtl-optimization/59890
2376 * var-tracking.c (local_get_addr_clear_given_value): Handle
2377 already cleared slot.
2378 (val_reset): Handle not allocated local_get_addr_cache.
2379 (vt_find_locations): Use post-order on the inverted CFG.
2381 2014-01-28 Richard Biener <rguenther@suse.de>
2383 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2385 2014-01-28 Richard Biener <rguenther@suse.de>
2387 PR rtl-optimization/45364
2388 PR rtl-optimization/59890
2389 * var-tracking.c (local_get_addr_clear_given_value): Handle
2390 already cleared slot.
2391 (val_reset): Handle not allocated local_get_addr_cache.
2392 (vt_find_locations): Use post-order on the inverted CFG.
2394 2014-01-28 Alan Modra <amodra@gmail.com>
2396 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2397 * configure.ac <recursive call for build != host>: Define
2398 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2399 and LD_FOR_BUILD too.
2400 * configure: Regenerate.
2402 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2404 * config/i386/i386.c (get_builtin_code_for_version): Separate
2405 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2406 Broadwell from Haswell.
2408 2014-01-27 Steve Ellcey <sellcey@mips.com>
2410 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2411 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2412 * config/mips/mips.c (mips_option_override): Change setting
2414 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2415 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2416 Change from Mask to Var.
2418 2014-01-27 Jeff Law <law@redhat.com>
2420 * ipa-inline.c (inline_small_functions): Fix typo.
2422 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2424 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2425 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2426 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2427 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2428 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2429 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2430 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2431 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2432 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2433 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2434 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2435 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2436 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2437 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2438 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2439 (_mm512_storeu_epi64): Ditto.
2440 (_mm512_cmpge_epi32_mask): Ditto.
2441 (_mm512_cmpge_epu32_mask): Ditto.
2442 (_mm512_cmpge_epi64_mask): Ditto.
2443 (_mm512_cmpge_epu64_mask): Ditto.
2444 (_mm512_cmple_epi32_mask): Ditto.
2445 (_mm512_cmple_epu32_mask): Ditto.
2446 (_mm512_cmple_epi64_mask): Ditto.
2447 (_mm512_cmple_epu64_mask): Ditto.
2448 (_mm512_cmplt_epi32_mask): Ditto.
2449 (_mm512_cmplt_epu32_mask): Ditto.
2450 (_mm512_cmplt_epi64_mask): Ditto.
2451 (_mm512_cmplt_epu64_mask): Ditto.
2452 (_mm512_cmpneq_epi32_mask): Ditto.
2453 (_mm512_cmpneq_epu32_mask): Ditto.
2454 (_mm512_cmpneq_epi64_mask): Ditto.
2455 (_mm512_cmpneq_epu64_mask): Ditto.
2456 (_mm512_expand_pd): Ditto.
2457 (_mm512_expand_ps): Ditto.
2458 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2459 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2460 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2461 * config/i386/i386.c (ix86_builtins): Add
2462 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2463 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2464 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2465 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2466 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2467 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2468 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2469 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2470 IX86_BUILTIN_PMOVUSQW512_MEM.
2471 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2472 __builtin_ia32_pmovsqd512mem_mask,
2473 __builtin_ia32_pmovqd512mem_mask,
2474 __builtin_ia32_pmovusqw512mem_mask,
2475 __builtin_ia32_pmovsqw512mem_mask,
2476 __builtin_ia32_pmovqw512mem_mask,
2477 __builtin_ia32_pmovusdw512mem_mask,
2478 __builtin_ia32_pmovsdw512mem_mask,
2479 __builtin_ia32_pmovdw512mem_mask,
2480 __builtin_ia32_pmovqb512mem_mask,
2481 __builtin_ia32_pmovusqb512mem_mask,
2482 __builtin_ia32_pmovsqb512mem_mask,
2483 __builtin_ia32_pmovusdb512mem_mask,
2484 __builtin_ia32_pmovsdb512mem_mask,
2485 __builtin_ia32_pmovdb512mem_mask.
2486 (bdesc_args): Add __builtin_ia32_expanddf512,
2487 __builtin_ia32_expandsf512.
2488 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2489 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2490 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2491 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2492 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2493 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2494 (avx512f_<code>v8div16qi2_mask_store): This.
2495 (avx512f_expand<mode>): New.
2497 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2499 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2501 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2502 (_mm512_prefetch_i32scatter_pd): Ditto.
2503 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2504 (_mm512_prefetch_i64scatter_pd): Ditto.
2505 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2506 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2507 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2508 (_mm512_prefetch_i32scatter_ps): Ditto.
2509 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2510 (_mm512_prefetch_i64scatter_ps): Ditto.
2511 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2512 * config/i386/i386-builtin-types.def: Define
2513 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2514 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2515 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2516 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2517 IX86_BUILTIN_SCATTERPFQPD.
2518 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2519 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2520 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2521 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2522 __builtin_ia32_scatterpfqps.
2523 (ix86_expand_builtin): Expand new built-ins.
2524 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2525 fix memory access data type.
2526 (*avx512pf_gatherpf<mode>_mask): Ditto.
2527 (*avx512pf_gatherpf<mode>): Ditto.
2528 (avx512pf_scatterpf<mode>): Ditto.
2529 (*avx512pf_scatterpf<mode>_mask): Ditto.
2530 (*avx512pf_scatterpf<mode>): Ditto.
2531 (GATHER_SCATTER_SF_MEM_MODE): New.
2532 (avx512pf_gatherpf<mode>df): Ditto.
2533 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2534 (*avx512pf_scatterpf<mode>df): Ditto.
2536 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2539 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2540 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2543 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2545 * common/config/arm/arm-common.c
2546 (arm_rewrite_mcpu): Handle multiple names.
2548 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2550 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2552 * gimple-builder.h (create_gimple_tmp): Delete.
2554 2014-01-27 Christian Bruel <christian.bruel@st.com>
2556 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2559 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2561 * config/pa/pa.md (call): Generate indirect long calls to non-local
2562 functions when outputing 32-bit code.
2563 (call_value): Likewise except for special call to buggy powf function.
2565 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2566 portable runtime and PIC indirect calls.
2567 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2568 and PIC call sequences. Use ldo instead of blr to set return register
2569 in PIC call sequence.
2571 2014-01-25 Walter Lee <walt@tilera.com>
2573 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2574 avoid clobbering a live register.
2576 2014-01-25 Walter Lee <walt@tilera.com>
2578 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2579 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2580 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2581 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2583 2014-01-25 Walter Lee <walt@tilera.com>
2585 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2586 arguments on even registers.
2587 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2589 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2590 (BIGGEST_ALIGNMENT): Ditto.
2591 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2593 2014-01-25 Walter Lee <walt@tilera.com>
2595 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2596 insns before bundling.
2597 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2599 2014-01-25 Walter Lee <walt@tilera.com>
2601 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2602 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2603 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2605 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2607 * config/mips/constraints.md (kl): Delete.
2608 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2609 define expands, using...
2610 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2611 instructions for MIPS16.
2612 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2613 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2615 2014-01-25 Walter Lee <walt@tilera.com>
2617 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2621 2014-01-25 Walter Lee <walt@tilera.com>
2623 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2624 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2626 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2628 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2631 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2633 * print-rtl.c (in_call_function_usage): New var.
2634 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2635 EXPR_LIST mode as mode and not as reg note name.
2638 * cfgloopmanip.c (copy_loop_info): If
2639 loop->warned_aggressive_loop_optimizations, make sure
2640 the flag is set in target loop too.
2642 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2644 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2646 * builtins.def: Likewise.
2647 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2648 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2649 * ira.c (ira_setup_eliminable_regset): Likewise.
2650 * omp-low.c (gate_expand_omp): Likewise.
2651 (execute_lower_omp): Likewise.
2652 (diagnose_sb_0): Likewise.
2653 (gate_diagnose_omp_blocks): Likewise.
2654 (simd_clone_clauses_extract): Likewise.
2657 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2659 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2660 correction for little endian...
2661 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2664 2014-01-24 Jeff Law <law@redhat.com>
2666 PR tree-optimization/59919
2667 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2668 for non-returning calls.
2670 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2672 * common/config/aarch64/aarch64-common.c
2673 (aarch64_rewrite_mcpu): Handle multiple names.
2674 * config/aarch64/aarch64.h
2675 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2677 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2679 * input.c (add_file_to_cache_tab): Handle the case where fopen
2682 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2685 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2686 from push operand if code of push isn't PRE_DEC.
2688 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2691 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2692 -mquad-memory-atomic. Update -mquad-memory documentation to say
2693 it is only used for non-atomic loads/stores.
2695 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2696 -mquad-memory or -mquad-memory-atomic switches.
2698 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2699 -mquad-memory-atomic to ISA 2.07 support.
2701 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2702 to separate support of normal quad word memory operations (ldq, stq)
2703 from the atomic quad word memory operations.
2705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2706 support to separate non-atomic quad word operations from atomic
2707 quad word operations. Disable non-atomic quad word operations in
2708 little endian mode so that we don't have to swap words after the
2709 load and before the store.
2710 (quad_load_store_p): Add comment about atomic quad word support.
2711 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2712 options printed with -mdebug=reg.
2714 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2715 -mquad-memory-atomic as the test for whether we have quad word
2716 atomic instructions.
2717 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2718 or -mp8-vector are used, allow byte/half-word atomic operations.
2720 * config/rs6000/sync.md (load_lockedti): Insure that the address
2721 is a proper indexed or indirect address for the lqarx instruction.
2722 On little endian systems, swap the hi/lo registers after the lqarx
2724 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2725 insure the address is valid for the lqarx instruction.
2726 (store_conditionalti): Insure that the address is a proper indexed
2727 or indirect address for the stqcrx. instruction. On little endian
2728 systems, swap the hi/lo registers before doing the stqcrx.
2730 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2731 insure the address is valid for the stqcrx. instruction.
2733 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2734 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2735 type of quad memory support is available.
2737 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2740 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2741 there is a danger of looping.
2743 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2745 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2746 force flag_ira_loop_pressure if set via command line.
2748 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2750 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2751 (ashr_simd): New builtin handling DI mode.
2752 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2753 (aarch64_sshr_simddi): New match pattern.
2754 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2755 (vshrd_n_s64): Likewise.
2756 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2758 2014-01-23 Nick Clifton <nickc@redhat.com>
2760 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2761 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2762 favour of mcu specific scripts.
2763 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2766 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2767 Alex Velenko <Alex.Velenko@arm.com>
2769 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2770 (vaddv_s16): Likewise.
2771 (vaddv_s32): Likewise.
2772 (vaddv_u8): Likewise.
2773 (vaddv_u16): Likewise.
2774 (vaddv_u32): Likewise.
2775 (vaddvq_s8): Likewise.
2776 (vaddvq_s16): Likewise.
2777 (vaddvq_s32): Likewise.
2778 (vaddvq_s64): Likewise.
2779 (vaddvq_u8): Likewise.
2780 (vaddvq_u16): Likewise.
2781 (vaddvq_u32): Likewise.
2782 (vaddvq_u64): Likewise.
2783 (vaddv_f32): Likewise.
2784 (vaddvq_f32): Likewise.
2785 (vaddvq_f64): Likewise.
2786 (vmaxv_f32): Likewise.
2787 (vmaxv_s8): Likewise.
2788 (vmaxv_s16): Likewise.
2789 (vmaxv_s32): Likewise.
2790 (vmaxv_u8): Likewise.
2791 (vmaxv_u16): Likewise.
2792 (vmaxv_u32): Likewise.
2793 (vmaxvq_f32): Likewise.
2794 (vmaxvq_f64): Likewise.
2795 (vmaxvq_s8): Likewise.
2796 (vmaxvq_s16): Likewise.
2797 (vmaxvq_s32): Likewise.
2798 (vmaxvq_u8): Likewise.
2799 (vmaxvq_u16): Likewise.
2800 (vmaxvq_u32): Likewise.
2801 (vmaxnmv_f32): Likewise.
2802 (vmaxnmvq_f32): Likewise.
2803 (vmaxnmvq_f64): Likewise.
2804 (vminv_f32): Likewise.
2805 (vminv_s8): Likewise.
2806 (vminv_s16): Likewise.
2807 (vminv_s32): Likewise.
2808 (vminv_u8): Likewise.
2809 (vminv_u16): Likewise.
2810 (vminv_u32): Likewise.
2811 (vminvq_f32): Likewise.
2812 (vminvq_f64): Likewise.
2813 (vminvq_s8): Likewise.
2814 (vminvq_s16): Likewise.
2815 (vminvq_s32): Likewise.
2816 (vminvq_u8): Likewise.
2817 (vminvq_u16): Likewise.
2818 (vminvq_u32): Likewise.
2819 (vminnmv_f32): Likewise.
2820 (vminnmvq_f32): Likewise.
2821 (vminnmvq_f64): Likewise.
2823 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2825 * config/aarch64/aarch64-simd.md
2826 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2827 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2828 (*aarch64_mul3_elt<mode>): Likewise.
2829 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2830 (*aarch64_mul3_elt_to_64v2df): Likewise.
2831 (*aarch64_mla_elt<mode>): Likewise.
2832 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2833 (*aarch64_mls_elt<mode>): Likewise.
2834 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2835 (*aarch64_fma4_elt<mode>): Likewise.
2836 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2837 (*aarch64_fma4_elt_to_64v2df): Likewise.
2838 (*aarch64_fnma4_elt<mode>): Likewise.
2839 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2840 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2841 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2842 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2843 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2844 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2845 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2846 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2847 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2849 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2851 * config/aarch64/aarch64-simd.md
2852 (aarch64_be_checked_get_lane<mode>): New define_expand.
2853 * config/aarch64/aarch64-simd-builtins.def
2854 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2855 New builtin definition.
2856 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2857 Use new safe be builtin.
2859 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2861 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2863 (aarch64_be_st1<mode>): Likewise.
2864 (aarch_ld1<VALL:mode>): Define_expand modified.
2865 (aarch_st1<VALL:mode>): Likewise.
2866 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2867 (UNSPEC_ST1): Likewise.
2869 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2871 * config/microblaze/microblaze.md: Add trap insn and attribute
2873 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2875 PR preprocessor/58580
2876 * input.h (location_get_source_line): Take an additional line_size
2878 (void diagnostics_file_cache_fini): Declare new function.
2879 * input.c (struct fcache): New type.
2880 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2881 New static constants.
2882 (diagnostic_file_cache_init, total_lines_num)
2883 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2884 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2885 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2886 (get_next_line, read_next_line, goto_next_line, read_line_num):
2887 New static function definitions.
2888 (diagnostic_file_cache_fini): New function.
2889 (location_get_source_line): Take an additional output line_len
2890 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2892 * diagnostic.c (diagnostic_finish): Call
2893 diagnostic_file_cache_fini.
2894 (adjust_line): Take an additional input parameter for the length
2895 of the line, rather than calculating it with strlen.
2896 (diagnostic_show_locus): Adjust the use of
2897 location_get_source_line and adjust_line with respect to their new
2898 signature. While displaying a line now, do not stop at the first
2899 null byte. Rather, display the zero byte as a space and keep
2900 going until we reach the size of the line.
2901 * Makefile.in: Add vec.o to OBJS-libcommon
2903 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2904 Ilya Tocar <ilya.tocar@intel.com>
2906 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2907 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2908 (__builtin_ia32_kmov16): Ditto.
2909 * config/i386/i386.md (UNSPEC_KMOV): New.
2912 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2914 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2915 (_mm512_storeu_si512): Ditto.
2917 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2920 * rtl.h (get_referenced_operands): Declare.
2921 * recog.c (get_referenced_operands): New function.
2922 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2923 operands have been referenced when recording LO_SUM references.
2925 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2927 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2929 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2931 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2932 Enable for generic and recent AMD targets.
2934 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2936 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2937 ARG_SIZE note when adjustment was eliminated.
2939 2014-01-22 Jeff Law <law@redhat.com>
2941 PR tree-optimization/59597
2942 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2943 in file. Accept new argument REGISTERING and use it to modify
2944 dump output appropriately.
2945 (register_jump_thread): Corresponding changes.
2946 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2947 thread paths involving joiner blocks. Add code to dump cancelled
2948 jump threading paths.
2950 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2952 PR rtl-optimization/59477
2953 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2954 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2956 2014-01-22 Tom Tromey <tromey@redhat.com>
2958 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2960 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2962 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2964 PR rtl-optimization/59896
2965 * lra-constraints.c (process_alt_operands): Check unused note for
2966 matched operands of insn with no output reloads.
2968 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2970 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2971 (mips_move_from_gpr_cost): Likewise.
2973 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2975 PR rtl-optimization/59858
2976 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2977 ira_class_hard_regs_num.
2978 (process_alt_operands): Increase reject for dying matched operand.
2980 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2983 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2984 smaller than size, perform several stores or loads and stores
2985 at dst + count - size to store or copy all of size bytes, rather
2986 than just last modesize bytes.
2988 2014-01-20 DJ Delorie <dj@redhat.com>
2990 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2991 that CLOBBERs are REGs before propogating their values.
2993 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2996 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2997 (cgraph_inline_failed_type): New function.
2998 * cgraph.h (DEFCIFCODE): Add type.
2999 (cgraph_inline_failed_type_t): New enum.
3000 (cgraph_inline_failed_type): New prototype.
3001 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3002 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3003 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3004 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3005 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3006 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3007 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3008 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3009 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3010 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3011 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3012 OPTIMIZATION_MISMATCH.
3013 * tree-inline.c (expand_call_inline): Emit errors during
3014 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3016 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3019 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3020 mode attribute in insn output.
3022 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3024 * output.h (output_constant): Delete.
3025 * varasm.c (output_constant): Make private.
3027 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3029 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3031 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3034 * tree.h (fold_builtin_strcat): New prototype.
3035 * builtins.c (fold_builtin_strcat): No longer static. Add len
3036 argument, if non-NULL, don't call c_strlen. Optimize
3037 directly into __builtin_memcpy instead of __builtin_strcpy.
3038 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3039 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3041 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3043 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3044 for SImode_address_operand operands, having only a REG argument.
3046 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3048 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3049 loader name using mbig-endian.
3050 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3052 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3054 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3058 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3060 * config/aarch64/aarch64-protos.h
3061 (aarch64_cannot_change_mode_class_ptr): Declare.
3062 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3063 aarch64_cannot_change_mode_class_ptr): New.
3064 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3065 backend hook aarch64_cannot_change_mode_class.
3067 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3069 * common/config/aarch64/aarch64-common.c
3070 (aarch64_handle_option): Don't handle any option order logic here.
3071 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3072 selected_cpu, warn on architecture version mismatch.
3073 (aarch64_override_options): Fix parsing order for option strings.
3075 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3076 Iain Sandoe <iain@codesourcery.com>
3079 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3080 warning. Amend comment to reflect current functionality.
3082 2014-01-20 Richard Biener <rguenther@suse.de>
3085 * builtins.c (fold_builtin_strcat): Remove case better handled
3086 by tree-ssa-strlen.c.
3088 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3090 * config/aarch64/aarch64.opt
3091 (mcpu, march, mtune): Make case-insensitive.
3093 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3096 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3097 if operands[1] is a REG or ZERO_EXTEND of a REG.
3099 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3101 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3103 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3105 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3106 long non-pic millicode calls.
3108 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3110 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3112 2014-01-19 Kito Cheng <kito@0xlab.org>
3114 * builtins.c (expand_movstr): Check movstr expand done or fail.
3116 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3117 H.J. Lu <hongjiu.lu@intel.com>
3120 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3121 to DImode for zero-extended addresses.
3123 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3125 PR rtl-optimization/57763
3126 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3127 on the new indirect jump_insn and increment LABEL_NUSES (label).
3129 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3133 * config.gcc (x86_archs): New variable.
3134 (x86_64_archs): Likewise.
3135 (x86_cpus): Likewise.
3136 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3137 --with-arch/--with-cpu= options.
3138 Support --with-arch=/--with-cpu={nehalem,westmere,
3139 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3141 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3143 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3144 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3146 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3148 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3150 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3153 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3154 clear cpp_get_options (parse_in)->warn_unused_macros for
3155 ix86_target_macros_internal with cpp_define.
3157 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3159 * jump.c (delete_related_insns): Keep (use (insn))s.
3160 * reorg.c (redundant_insn): Check for barriers too.
3162 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3164 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3166 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3168 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3169 call to $$dyncall when TARGET_LONG_CALLS is true.
3171 2014-01-17 Jeff Law <law@redhat.com>
3173 * ree.c (combine_set_extension): Temporarily disable test for
3174 changing number of hard registers.
3176 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3179 * ipa-inline-analysis.c (inline_free_summary):
3180 Do not free summary of aliases.
3182 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3185 * gimplify.c (gimplify_expr): Use create_tmp_var
3186 instead of create_tmp_var_raw. If cond doesn't have
3187 integral type, don't add the IFN_ANNOTATE builtin at all.
3189 2014-01-17 Martin Jambor <mjambor@suse.cz>
3192 * ipa-cp.c (prev_edge_clone): New variable.
3193 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3194 Also resize prev_edge_clone vector.
3195 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3196 (ipcp_edge_removal_hook): New function.
3197 (ipcp_driver): Register ipcp_edge_removal_hook.
3199 2014-01-17 Andrew Pinski <apinski@cavium.com>
3200 Steve Ellcey <sellcey@mips.com>
3203 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3206 2014-01-17 Jeff Law <law@redhat.com>
3209 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3210 so that pass_ccp runs first.
3212 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3214 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3215 (ix86_adjust_cost): Use !TARGET_XXX.
3216 (do_reorder_for_imul): Likewise.
3217 (swap_top_of_ready_list): Likewise.
3218 (ix86_sched_reorder): Likewise.
3220 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3222 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3223 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3224 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3225 (intel_memset): New. Duplicate slm_memset.
3226 (intel_cost): New. Duplicate slm_cost.
3227 (m_INTEL): New macro.
3228 (processor_target_table): Add "intel".
3229 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3230 with PROCESSOR_INTEL for "intel".
3231 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3232 PROCESSOR_SILVERMONT.
3233 (ix86_issue_rate): Likewise.
3234 (ix86_adjust_cost): Likewise.
3235 (ia32_multipass_dfa_lookahead): Likewise.
3236 (swap_top_of_ready_list): Likewise.
3237 (ix86_sched_reorder): Likewise.
3238 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3239 instead of TARGET_OPT_AGU.
3240 * config/i386/i386.h (TARGET_INTEL): New.
3241 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3242 (processor_type): Add PROCESSOR_INTEL.
3243 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3244 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3246 2014-01-17 Marek Polacek <polacek@redhat.com>
3249 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3252 2014-01-17 Richard Biener <rguenther@suse.de>
3254 PR tree-optimization/46590
3255 * opts.c (default_options_table): Add entries for
3256 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3257 all enabled at -O1 but not for -Og.
3258 * common.opt (fbranch-count-reg): Remove Init(1).
3259 (fmove-loop-invariants): Likewise.
3260 (ftree-pta): Likewise.
3262 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3264 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3265 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3266 decls to at least the GCC 4.8 used alignments.
3269 * tree-nested.c (convert_nonlocal_reference_stmt,
3270 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3271 of GIMPLE_BIND stmts, adjust associated decls.
3273 2014-01-17 Richard Biener <rguenther@suse.de>
3275 PR tree-optimization/46590
3276 * vec.h (vec<>::bseach): New member function implementing
3277 binary search according to C89 bsearch.
3278 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3279 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3280 bitmap pointer again. Make accesses_in_loop a flat array.
3281 (mem_ref_obstack): New global.
3282 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3283 (mark_ref_stored): Likewise.
3284 (ref_indep_loop_p_2): Likewise.
3285 (set_ref_stored_in_loop): New helper function.
3286 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3287 (memref_free): Adjust.
3288 (record_mem_ref_loc): Simplify.
3289 (gather_mem_refs_stmt): Adjust.
3290 (sort_locs_in_loop_postorder_cmp): New function.
3291 (analyze_memory_references): Sort accesses_in_loop after
3292 loop postorder number.
3293 (find_ref_loc_in_loop_cmp): New function.
3294 (for_all_locs_in_loop): Find relevant cluster of locs in
3295 accesses_in_loop and iterate without recursion.
3296 (execute_sm): Avoid uninit warning.
3297 (struct ref_always_accessed): Simplify.
3298 (ref_always_accessed::operator ()): Likewise.
3299 (ref_always_accessed_p): Likewise.
3300 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3301 loop postorder numbers here.
3302 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3305 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3308 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3309 on decls for which assemble_alias has been called.
3311 2014-01-17 Nick Clifton <nickc@redhat.com>
3313 * config/msp430/msp430.opt: (mcpu): New option.
3314 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3315 (msp430_option_override): Parse target_cpu. If the MCU name
3316 matches a generic string, clear target_mcu.
3317 (msp430_attr): Allow numeric interrupt values up to 63.
3318 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3319 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3321 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3323 * config/msp430/msp430.md (popm): Use %J rather than %I.
3324 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3325 (addhi_cy_i): Use immediate_operand for operand 2.
3326 * doc/invoke.texi: Document -mcpu option.
3328 2014-01-17 Richard Biener <rguenther@suse.de>
3330 PR rtl-optimization/38518
3331 * df.h (df_analyze_loop): Declare.
3332 * df-core.c: Include cfgloop.h.
3333 (df_analyze_1): Split out main part of df_analyze.
3334 (df_analyze): Adjust.
3335 (loop_inverted_post_order_compute): New function.
3336 (loop_post_order_compute): Likewise.
3337 (df_analyze_loop): New function avoiding whole-function
3339 * loop-invariant.c (find_defs): Use df_analyze_loop.
3340 (find_invariants): Adjust.
3341 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3343 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3345 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3346 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3348 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3350 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3351 traversal when removing references.
3353 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3356 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3358 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3361 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3362 pushing a reload for an autoinc when we had previously reloaded an
3363 inner part of the address.
3365 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3367 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3369 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3370 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3371 when not giving up or versioning for alias only because of
3373 (vect_analyze_data_ref_dependences): Set to true.
3374 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3376 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3377 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3381 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3384 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3385 operand 0 predicate for gathers, use a new pseudo as subtarget.
3387 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3390 * lra-constraints.c (process_alt_operands): Add printing debug info.
3391 Check absence of input/output reloads for matched operands too.
3393 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3395 PR rtl-optimization/59835
3396 * ira.c (ira_init_register_move_cost): Increase cost for
3399 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3401 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3403 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3406 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3407 non-register objects. Use gen_(high/low)part more consistently.
3410 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3413 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3414 endian support, remove tests for WORDS_BIG_ENDIAN.
3415 (p8_mfvsrd_3_<mode>): Likewise.
3416 (reload_gpr_from_vsx<mode>): Likewise.
3417 (reload_gpr_from_vsxsf): Likewise.
3418 (p8_mfvsrd_4_disf): Likewise.
3420 2014-01-16 Richard Biener <rguenther@suse.de>
3422 PR rtl-optimization/46590
3423 * lcm.c (compute_antinout_edge): Use postorder iteration.
3424 (compute_laterin): Use inverted postorder iteration.
3426 2014-01-16 Nick Clifton <nickc@redhat.com>
3429 * varasm.c (output_constant): Return the number of bytes actually
3431 (output_constructor_array_range): Update the field size with the
3432 number of bytes emitted by output_constant.
3433 (output_constructor_regular_field): Likewise. Also do not
3434 complain if the total number of bytes emitted is now greater
3435 than the expected fieldpos.
3436 * output.h (output_constant): Update prototype and descriptive comment.
3438 2014-01-16 Marek Polacek <polacek@redhat.com>
3441 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3442 it is error_mark_node.
3444 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3446 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3447 VALID_AVX256_REG_OR_OI_MODE.
3449 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3451 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3452 current procedure should be profiled.
3454 2014-01-15 Andrew Pinski <apinski@cavium.com>
3456 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3457 of moving from/to the STACK_REG register class.
3459 2014-01-15 Richard Henderson <rth@redhat.com>
3462 * reginfo.c (global_regs_decl): Globalize.
3463 * rtl.h (global_regs_decl): Declare.
3464 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3465 reserved via global_regs.
3467 2014-01-15 Teresa Johnson <tejohnson@google.com>
3469 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3471 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3473 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3474 and vmulosh rather than call gen_vec_widen_smult_*.
3475 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3476 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3477 (vec_widen_smult_even_v16qi): Likewise.
3478 (vec_widen_umult_even_v8hi): Likewise.
3479 (vec_widen_smult_even_v8hi): Likewise.
3480 (vec_widen_umult_odd_v16qi): Likewise.
3481 (vec_widen_smult_odd_v16qi): Likewise.
3482 (vec_widen_umult_odd_v8hi): Likewise.
3483 (vec_widen_smult_odd_v8hi): Likewise.
3484 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3485 vmuloub rather than call gen_vec_widen_umult_*.
3486 (vec_widen_umult_lo_v16qi): Likewise.
3487 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3488 vmulosb rather than call gen_vec_widen_smult_*.
3489 (vec_widen_smult_lo_v16qi): Likewise.
3490 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3491 rather than call gen_vec_widen_umult_*.
3492 (vec_widen_umult_lo_v8hi): Likewise.
3493 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3494 rather than call gen_vec_widen_smult_*.
3495 (vec_widen_smult_lo_v8hi): Likewise.
3497 2014-01-15 Jeff Law <law@redhat.com>
3499 PR tree-optimization/59747
3500 * ree.c (find_and_remove_re): Properly handle case where a second
3501 eliminated extension requires widening a copy created for elimination
3502 of a prior extension.
3503 (combine_set_extension): Ensure that the number of hard regs needed
3504 for a destination register does not change when we widen it.
3506 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3508 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3509 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3510 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3511 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3512 (avr-*-rtems*): Likewise.
3513 (bfin*-rtems*): Likewise.
3514 (moxie-*-rtems*): Likewise.
3515 (h8300-*-rtems*): Likewise.
3516 (i[34567]86-*-rtems*): Likewise.
3517 (lm32-*-rtems*): Likewise.
3518 (m32r-*-rtems*): Likewise.
3519 (m68k-*-rtems*): Likewise.
3520 (microblaze*-*-rtems*): Likewise.
3521 (mips*-*-rtems*): Likewise.
3522 (powerpc-*-rtems*): Likewise.
3523 (sh-*-rtems*): Likewise.
3524 (sparc-*-rtems*): Likewise.
3525 (sparc64-*-rtems*): Likewise.
3526 (v850-*-rtems*): Likewise.
3527 (m32c-*-rtems*): Likewise.
3529 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3531 PR rtl-optimization/59511
3532 * ira.c (ira_init_register_move_cost): Use memory costs for some
3533 cases of register move cost calculations.
3534 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3535 instead of BB frequency.
3536 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3537 * lra-assigns.c (find_hard_regno_for): Ditto.
3539 2014-01-15 Richard Biener <rguenther@suse.de>
3541 PR tree-optimization/59822
3542 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3543 (vectorizable_load): Use it to hoist defs of uses of invariant
3544 loads out of the loop.
3546 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3547 Kugan Vivekanandarajah <kuganv@linaro.org>
3550 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3553 2014-01-15 Richard Biener <rguenther@suse.de>
3555 PR rtl-optimization/59802
3556 * lcm.c (compute_available): Use inverted postorder to seed
3557 the initial worklist.
3559 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3562 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3563 ADDR_REGS for invalid symrefs in non-PIC code.
3565 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3568 * builtins.c (determine_block_size): Initialize *probable_max_size
3569 even if len_rtx is CONST_INT.
3571 2014-01-14 Andrew Pinski <apinski@cavium.com>
3573 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3574 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3575 (cortexa53_tunings): Likewise.
3576 (aarch64_sched_issue_rate): New function.
3577 (TARGET_SCHED_ISSUE_RATE): Define.
3579 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3581 * ira-costs.c (find_costs_and_classes): Add missed
3582 ira_init_register_move_cost_if_necessary.
3584 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3587 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3589 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3592 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3593 to indicate if type is used for function return value. Warn ABI
3594 change if the vector mode isn't available for function return value.
3595 (ix86_function_arg_advance): Pass false to type_natural_mode.
3596 (ix86_function_arg): Likewise.
3597 (ix86_gimplify_va_arg): Likewise.
3598 (function_arg_32): Don't warn ABI change.
3599 (ix86_function_value): Pass true to type_natural_mode.
3600 (ix86_return_in_memory): Likewise.
3601 (ix86_struct_value_rtx): Removed.
3602 (TARGET_STRUCT_VALUE_RTX): Likewise.
3604 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3606 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3607 converting a conditional jump into a conditional return.
3609 2014-01-14 Richard Biener <rguenther@suse.de>
3611 PR tree-optimization/58921
3612 PR tree-optimization/59006
3613 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3614 hoisting invariant stmts.
3615 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3616 invariant loads on the preheader edge if possible.
3618 2014-01-14 Joey Ye <joey.ye@arm.com>
3620 * doc/plugin.texi (Building GCC plugins): Update to C++.
3622 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3624 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3625 (_mm_rcp28_round_ss): Ditto.
3626 (_mm_rsqrt28_round_sd): Ditto.
3627 (_mm_rsqrt28_round_ss): Ditto.
3628 (_mm_rcp28_sd): Ditto.
3629 (_mm_rcp28_ss): Ditto.
3630 (_mm_rsqrt28_sd): Ditto.
3631 (_mm_rsqrt28_ss): Ditto.
3632 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3633 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3634 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3635 (IX86_BUILTIN_RCP28SD): Ditto.
3636 (IX86_BUILTIN_RCP28SS): Ditto.
3637 (IX86_BUILTIN_RSQRT28SD): Ditto.
3638 (IX86_BUILTIN_RSQRT28SS): Ditto.
3639 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3640 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3641 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3642 (ix86_expand_special_args_builtin): Expand new FTYPE.
3643 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3644 (srcp14<mode>): Make insn unary.
3645 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3646 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3647 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3648 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3649 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3650 Fix rounding: make it SAE only.
3651 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3653 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3655 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3656 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3657 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3658 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3659 (round_saeonly_mask_scalar_operand4): Ditto.
3660 (round_saeonly_mask_scalar_op3): Ditto.
3661 (round_saeonly_mask_scalar_op4): Ditto.
3663 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3665 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3666 Implement -maltivec=be for vec_insert and vec_extract.
3668 2014-01-10 DJ Delorie <dj@redhat.com>
3670 * config/msp430/msp430.md (call_internal): Don't allow memory
3671 references with SP as the base register.
3672 (call_value_internal): Likewise.
3673 * config/msp430/constraints.md (Yc): New. For memory references
3674 that don't use SP as a base register.
3676 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3677 "an integer without a # prefix"
3678 * config/msp430/msp430.md (epilogue_helper): Use it.
3680 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3683 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3684 AVX512F gather builtins.
3685 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3686 on gather decls with INTEGER_TYPE masktype.
3687 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3688 directly into the builtin rather than hoisting it before loop.
3690 PR tree-optimization/59387
3691 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3692 (scev_const_prop): If folded_casts and type has undefined overflow,
3693 use force_gimple_operand instead of force_gimple_operand_gsi and
3694 for each added stmt if it is assign with
3695 arith_code_with_undefined_signed_overflow, call
3696 rewrite_to_defined_overflow.
3697 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3698 gimple-fold.h instead.
3699 (arith_code_with_undefined_signed_overflow,
3700 rewrite_to_defined_overflow): Moved to ...
3701 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3702 rewrite_to_defined_overflow): ... here. No longer static.
3703 Include gimplify-me.h.
3704 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3705 rewrite_to_defined_overflow): New prototypes.
3707 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3709 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3711 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3713 * builtins.c (get_object_alignment_2): Minor tweak.
3714 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3716 2014-01-13 Christian Bruel <christian.bruel@st.com>
3718 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3719 optimized non constant lengths.
3721 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3724 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3725 load as __atomic_load_N if possible.
3727 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3729 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3731 (rs6000_expand_builtin): Adjust call.
3733 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3736 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3737 * config/rs6000/rs6000.c: Include target-globals.h.
3738 (rs6000_set_current_function): Instead of doing target_reinit
3739 unconditionally, use save_target_globals_default_opts and
3740 restore_target_globals.
3742 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3744 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3745 (rs6000_expand_builtin): Handle mffs and mtfsf.
3746 (rs6000_init_builtins): Define mffs and mtfsf.
3747 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3748 (rs6000_mffs): New pattern.
3749 (rs6000_mtfsf): New pattern.
3751 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3753 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3754 Start narrowing with START. Apply candidate-use pair
3755 and check overall cost in narrowing.
3756 (iv_ca_prune): Pass new argument.
3758 2014-01-10 Jeff Law <law@redhat.com>
3761 * ree.c (combine_reaching_defs): Ensure the defining statement
3762 occurs before the extension when optimizing extensions with
3763 different source and destination hard registers.
3765 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3768 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3769 vtables into the type inheritance graph.
3771 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3773 PR rtl-optimization/59754
3774 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3775 modes in the REGNO != REGNO case.
3777 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3779 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3781 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3783 PR tree-optimization/59745
3784 * tree-predcom.c (tree_predictive_commoning_loop): Call
3785 free_affine_expand_cache if giving up because components is NULL.
3787 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3788 GC in payload of target_globals struct instead of allocating them on
3789 the heap and the larger structs separately using GC.
3790 * target-globals.h (struct target_globals): Make regs, hard_regs,
3791 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3792 of GTY((skip)) and change type to void *.
3793 (reset_target_globals): Cast loads from those fields to corresponding
3796 2014-01-10 Steve Ellcey <sellcey@mips.com>
3799 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3800 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3801 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3803 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3806 * aarch64-modes.def (CC_Zmode): New flags mode.
3807 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3808 represents an equality.
3809 (aarch64_get_condition_code): Handle CC_Zmode.
3810 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3812 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3814 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3815 extraction in good case.
3817 2014-01-10 Richard Biener <rguenther@suse.de>
3819 PR tree-optimization/59374
3820 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3821 checking after SLP discovery. Mark stmts not participating
3822 in any SLP instance properly.
3824 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3826 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3827 when handling a SET rtx.
3829 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3831 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3832 (cortex-a57): Likewise.
3833 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3835 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3837 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3838 non-iwmmxt builtins.
3840 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3844 * ipa-devirt.c record_target_from_binfo): Take as argument
3845 stack of binfos and lookup matching one for virtual inheritance.
3846 (possible_polymorphic_call_targets_1): Update.
3848 2014-01-10 Huacai Chen <chenhc@lemote.com>
3850 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3851 kernel strings for Loongson-2E/2F/3A.
3853 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3856 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3857 is_gimple_call before calling gimple_call_internal_p.
3859 2014-01-09 Steve Ellcey <sellcey@mips.com>
3861 * Makefile.in (TREE_FLOW_H): Remove.
3862 (TREE_SSA_H): Add file names from tree-flow.h.
3863 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3864 * tree.h: Remove tree-flow.h reference.
3865 * hash-table.h: Remove tree-flow.h reference.
3866 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3867 reference with tree-ssa-loop.h.
3869 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3871 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3872 default element-order behavior for -maltivec.
3873 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3874 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3875 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3876 when targeting big endian, at least for now.
3877 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3879 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3882 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3883 var satisfies use_register_for_decl, just take into account type
3884 alignment, rather than decl alignment.
3886 PR tree-optimization/59622
3887 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3888 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3889 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3890 Don't devirtualize for inplace at all. For targets.length () == 1,
3891 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3893 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3895 * config/i386/i386.md (cpu): Remove the unused btver1.
3897 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3899 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3901 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3904 * tree-core.h (struct target_globals): New forward declaration.
3905 (struct tree_target_option): Add globals field.
3906 * tree.h (TREE_TARGET_GLOBALS): Define.
3907 (prepare_target_option_nodes_for_pch): New prototype.
3908 * target-globals.h (struct target_globals): Define even if
3910 * tree.c (prepare_target_option_node_for_pch,
3911 prepare_target_option_nodes_for_pch): New functions.
3912 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3913 * config/i386/i386.c: Include target-globals.h.
3914 (ix86_set_current_function): Instead of doing target_reinit
3915 unconditionally, use save_target_globals_default_opts and
3916 restore_target_globals.
3918 2014-01-09 Richard Biener <rguenther@suse.de>
3920 PR tree-optimization/59715
3921 * tree-cfg.h (split_critical_edges): Declare.
3922 * tree-cfg.c (split_critical_edges): Export.
3923 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3925 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3927 * cfgexpand.c (expand_stack_vars): Optionally disable
3928 asan stack protection.
3929 (expand_used_vars): Likewise.
3930 (partition_stack_vars): Likewise.
3931 * asan.c (asan_emit_stack_protection): Optionally disable
3932 after return stack usage.
3933 (instrument_derefs): Optionally disable memory access instrumentation.
3934 (instrument_builtin_call): Likewise.
3935 (instrument_strlen_call): Likewise.
3936 (asan_protect_global): Optionally disable global variables protection.
3937 * doc/invoke.texi: Added doc for new options.
3938 * params.def: Added new options.
3939 * params.h: Likewise.
3941 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3943 PR rtl-optimization/59724
3944 * ifcvt.c (cond_exec_process_if_block): Don't call
3945 flow_find_head_matching_sequence with 0 longest_match.
3946 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3947 non-active insns if !stop_after.
3948 (try_head_merge_bb): Revert 2014-01-07 changes.
3950 2014-01-08 Jeff Law <law@redhat.com>
3952 * ree.c (get_sub_rtx): New function, extracted from...
3953 (merge_def_and_ext): Here.
3954 (combine_reaching_defs): Use get_sub_rtx.
3956 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3958 * cgraph.h (varpool_variable_node): Do not choke on null node.
3960 2014-01-08 Catherine Moore <clm@codesourcery.com>
3962 * config/mips/mips.md (simple_return): Attempt to use JRC
3964 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3966 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3968 PR rtl-optimization/59137
3969 * reorg.c (steal_delay_list_from_target): Call update_block for
3971 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3973 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3975 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3976 two duplicate entries.
3978 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3981 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3983 * config/mips/mips.c (mips_truncated_op_cost): New function.
3984 (mips_rtx_costs): Adjust test for BADDU.
3985 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3987 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3989 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3990 (*baddu_si): ...this new pattern.
3992 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3995 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3997 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4000 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4002 (expand_expr, expand_normal): Adjust.
4003 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4004 inner_reference_p. Use inner_reference_p to expand inner references.
4005 (store_expr): Adjust.
4006 * cfgexpand.c (expand_call_stmt): Adjust.
4008 2014-01-08 Rong Xu <xur@google.com>
4010 * gcov-io.c (gcov_var): Move from gcov-io.h.
4011 (gcov_position): Ditto.
4012 (gcov_is_error): Ditto.
4013 (gcov_rewrite): Ditto.
4014 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4015 only part to libgcc/libgcov.h.
4017 2014-01-08 Marek Polacek <polacek@redhat.com>
4020 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4022 2014-01-08 Marek Polacek <polacek@redhat.com>
4025 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4027 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4029 PR rtl-optimization/59649
4030 * stor-layout.c (get_mode_bounds): For BImode return
4031 0 and STORE_FLAG_VALUE.
4033 2014-01-08 Richard Biener <rguenther@suse.de>
4036 * gimple.h (is_gimple_builtin_call): Remove.
4037 (gimple_builtin_call_types_compatible_p): New.
4038 (gimple_call_builtin_p): New overload.
4039 * gimple.c (is_gimple_builtin_call): Remove.
4040 (validate_call): Rename to ...
4041 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4043 (validate_type): New static function.
4044 (gimple_call_builtin_p): New overload and adjust.
4045 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4046 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4047 (gimple_fold_stmt_to_constant_1): Likewise.
4048 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4050 2014-01-08 Richard Biener <rguenther@suse.de>
4053 * gimplify.c (gimplify_expr): Gimplify register-register type
4054 VIEW_CONVERT_EXPRs to separate stmts.
4056 2014-01-07 Jeff Law <law@redhat.com>
4059 * ree.c (combine_set_extension): Handle case where source
4060 and destination registers in an extension insn are different.
4061 (combine_reaching_defs): Allow source and destination registers
4062 in extension to be different under limited circumstances.
4063 (add_removable_extension): Remove restriction that the
4064 source and destination registers in the extension are the same.
4065 (find_and_remove_re): Emit a copy from the extension's
4066 destination to its source after the defining insn if
4067 the source and destination registers are different.
4070 * ifcvt.c (merge_if_block): If we are merging a block with more than
4071 one successor with a block with no successors, remove any BARRIER
4072 after the second block.
4074 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4076 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4078 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4081 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4082 for 14-bit register offsets when INT14_OK_STRICT is false.
4084 2014-01-07 Roland Stigge <stigge@antcom.de>
4085 Michael Meissner <meissner@linux.vnet.ibm.com>
4088 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4089 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4091 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4093 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4096 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4098 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4099 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4100 rtx is const0_rtx or not.
4102 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4105 * target-globals.c (save_target_globals): Remove this_fn_optab
4107 * toplev.c: Include optabs.h.
4108 (target_reinit): Temporarily restore the global options if another
4109 set of options are in force.
4111 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4113 PR rtl-optimization/58668
4114 * cfgcleanup.c (flow_find_cross_jump): Don't count
4115 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4116 to determine what is counted.
4117 (flow_find_head_matching_sequence): Use active_insn_p to determine
4119 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4121 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4122 determine what is counted.
4124 PR tree-optimization/59643
4125 * tree-predcom.c (split_data_refs_to_components): If one dr is
4126 read and one write, determine_offset fails and the write isn't
4127 in the bad component, just put the read into the bad component.
4129 2014-01-07 Mike Stump <mikestump@comcast.net>
4130 Jakub Jelinek <jakub@redhat.com>
4133 * tree-core.h (struct tree_optimization_option): Change optabs
4134 type from unsigned char * to void *.
4135 * optabs.c (init_tree_optimization_optabs): Adjust
4136 TREE_OPTIMIZATION_OPTABS initialization.
4138 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4141 * config/i386/i386.h (struct machine_function): Add
4142 no_drap_save_restore field.
4143 * config/i386/i386.c (ix86_save_reg): Use
4144 !cfun->machine->no_drap_save_restore instead of
4145 crtl->stack_realign_needed.
4146 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4147 this function clears frame_pointer_needed. Set
4148 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4149 and DRAP reg is needed.
4151 2014-01-06 Marek Polacek <polacek@redhat.com>
4154 * doc/implement-c.texi: Mention that other integer types are
4155 permitted as bit-field types in strictly conforming mode.
4157 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4159 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4162 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4164 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4166 2014-01-06 Martin Jambor <mjambor@suse.cz>
4169 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4171 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4173 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4177 * var-tracking.c (add_stores): Preserve the value of the source even if
4178 we don't record the store.
4180 2014-01-06 Terry Guo <terry.guo@arm.com>
4182 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4184 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4187 * config/darwin.c (darwin_function_section): Adjust return values to
4188 correspond to optimisation changes made in r206070.
4190 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4192 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4193 from prefetch_block tune setting.
4194 (nocona_cost): Correct size of prefetch block to 64.
4196 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4198 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4199 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4200 used to save the static chain register in the computation of the offset
4201 from which the FP registers need to be restored.
4203 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4205 PR tree-optimization/59519
4206 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4207 ICE if get_current_def (current_new_name) is already non-NULL, as long
4208 as it is a phi result of some other phi in *new_exit_bb that has
4211 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4212 or vmovdqu* for misaligned_operand.
4213 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4214 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4215 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4216 aligned_mem for AVX512F masked aligned load and store builtins and for
4219 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4221 PR tree-optimization/59651
4222 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4223 Address range for negative step should be added by TYPE_SIZE_UNIT.
4225 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4227 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4228 overlapping registers also for registers other than the stack pointer.
4230 2014-01-03 Marek Polacek <polacek@redhat.com>
4233 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4236 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4239 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4242 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4243 (push splitter): Use <P:MODE_SIZE> instead of
4244 GET_MODE_SIZE (<P:MODE>mode).
4245 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4246 (mov -1, reg peephole2): Likewise.
4247 * config/i386/sse.md (*mov<mode>_internal,
4248 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4249 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4250 *<code><mode>3, *andnot<mode>3<mask_name>,
4251 <mask_codefor><code><mode>3<mask_name>): Likewise.
4252 * config/i386/subst.md (mask_mode512bit_condition,
4253 sd_mask_mode512bit_condition): Likewise.
4255 2014-01-02 Xinliang David Li <davidxl@google.com>
4257 PR tree-optimization/59303
4258 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4259 (dump_predicates): Better output format.
4260 (pred_equal_p): New function.
4261 (is_neq_relop_p): Ditto.
4262 (is_neq_zero_form_p): Ditto.
4263 (pred_expr_equal_p): Ditto.
4264 (pred_neg_p): Ditto.
4265 (simplify_pred): Ditto.
4266 (simplify_preds_2): Ditto.
4267 (simplify_preds_3): Ditto.
4268 (simplify_preds_4): Ditto.
4269 (simplify_preds): Ditto.
4271 (push_to_worklist): Ditto.
4272 (get_pred_info_from_cmp): Ditto.
4273 (is_degenerated_phi): Ditto.
4274 (normalize_one_pred_1): Ditto.
4275 (normalize_one_pred): Ditto.
4276 (normalize_one_pred_chain): Ditto.
4277 (normalize_preds): Ditto.
4278 (normalize_cond_1): Remove function.
4279 (normalize_cond): Ditto.
4280 (is_gcond_subset_of): Ditto.
4281 (is_subset_of_any): Ditto.
4282 (is_or_set_subset_of): Ditto.
4283 (is_and_set_subset_of): Ditto.
4284 (is_norm_cond_subset_of): Ditto.
4285 (pred_chain_length_cmp): Ditto.
4286 (convert_control_dep_chain_into_preds): Type change.
4287 (find_predicates): Ditto.
4288 (find_def_preds): Ditto.
4289 (destroy_predicates_vecs): Ditto.
4290 (find_matching_predicates_in_rest_chains): Ditto.
4291 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4292 (is_pred_expr_subset): Ditto.
4293 (is_pred_chain_subset_of): Ditto.
4294 (is_included_in): Ditto.
4295 (is_superset_of): Ditto.
4297 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4299 Update copyright years.
4301 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4303 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4304 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4305 config/arc/arc.md, config/arc/arc.opt,
4306 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4307 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4308 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4309 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4310 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4311 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4312 vtable-verify.c, vtable-verify.h: Use the standard form for the
4315 2014-01-02 Tobias Burnus <burnus@net-b.de>
4317 * gcc.c (process_command): Update copyright notice dates.
4318 * gcov-dump.c: Ditto.
4320 * doc/cpp.texi: Bump @copying's copyright year.
4321 * doc/cppinternals.texi: Ditto.
4322 * doc/gcc.texi: Ditto.
4323 * doc/gccint.texi: Ditto.
4324 * doc/gcov.texi: Ditto.
4325 * doc/install.texi: Ditto.
4326 * doc/invoke.texi: Ditto.
4328 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4330 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4332 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4334 * config/i386/sse.md (*mov<mode>_internal): Guard
4335 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4337 PR rtl-optimization/59647
4338 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4339 new_rtx into UNSIGNED_FLOAT rtxes.
4341 Copyright (C) 2014 Free Software Foundation, Inc.
4343 Copying and distribution of this file, with or without modification,
4344 are permitted in any medium without royalty provided the copyright
4345 notice and this notice are preserved.