1 2014-02-28 Richard Biener <rguenther@suse.de>
4 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
5 previous fix and only allow to remove trivial pre-headers
6 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
7 (remove_forwarder_block): Properly update the latch of a loop.
9 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
12 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
13 (cselib_preserved_hash_table): New.
14 (preserve_constants_and_equivs): Move preserved vals to it.
15 (cselib_find_slot): Look it up first.
16 (cselib_init): Initialize it.
17 (cselib_finish): Release it.
18 (dump_cselib_table): Dump it.
20 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
23 * cselib.c (remove_useless_values): Skip to avoid quadratic
24 behavior if the condition moved from...
25 (cselib_process_insn): ... here holds.
27 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
30 * var-tracking.c (vt_initialize): Apply the same condition to
31 preserve the CFA base value.
33 2014-02-28 Joey Ye <joey.ye@arm.com>
36 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
37 if reload in progress or completed.
39 2014-02-28 Tobias Burnus <burnus@net-b.de>
42 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
45 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
47 * doc/tm.texi.in (Condition Code Status): Update documention for
48 relative locations of cc0-setter and cc0-user.
50 2014-02-27 Jeff Law <law@redhat.com>
52 PR rtl-optimization/52714
53 * combine.c (try_combine): When splitting an unrecognized PARALLEL
54 into two independent simple sets, if I3 is a jump, ensure the
55 pattern we place into I3 is a (set (pc) ...).
57 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
58 Jeff Law <law@redhat.com>
60 PR rtl-optimization/49847
61 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
62 are in different blocks.
63 * doc/tm.texi (Condition Code Status): Update documention for
64 relative locations of cc0-setter and cc0-user.
66 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
69 * lra.c (lra_emit_add): Check SUBREG too.
71 2014-02-27 Andreas Schwab <schwab@suse.de>
73 * config/m68k/m68k.c (m68k_option_override): Disable
74 -flive-range-shrinkage for classic m68k.
75 (m68k_override_options_after_change): Likewise.
77 2014-02-27 Marek Polacek <polacek@redhat.com>
80 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
81 -Wmaybe-uninitialized.
83 2014-02-27 Alan Modra <amodra@gmail.com>
86 * reload1.c (emit_input_reload_insns): When reload_override_in,
87 set old to rl->in_reg when rl->in_reg is a subreg.
89 2014-02-26 Richard Biener <rguenther@suse.de>
92 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
94 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
96 * common/config/i386/predicates.md (const1256_operand): Remove.
97 (const2356_operand): New.
98 (const_1_to_2_operand): Remove.
99 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
100 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
101 (*avx512pf_gatherpf<mode>sf): Ditto.
102 (avx512pf_gatherpf<mode>df): Ditto.
103 (*avx512pf_gatherpf<mode>df_mask): Ditto.
104 (*avx512pf_gatherpf<mode>df): Ditto.
105 (avx512pf_scatterpf<mode>sf): Ditto.
106 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
107 (*avx512pf_scatterpf<mode>sf): Ditto.
108 (avx512pf_scatterpf<mode>df): Ditto.
109 (*avx512pf_scatterpf<mode>df_mask): Ditto.
110 (*avx512pf_scatterpf<mode>df): Ditto.
111 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
113 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
115 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
116 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
117 (_mm512_mask_testn_epi64_mask): Move to ...
118 * config/i386/avx512cdintrin.h: Here.
119 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
120 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
121 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
122 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
123 TARGET_AVX512F from TARGET_AVX512CD.
125 2014-02-26 Richard Biener <rguenther@suse.de>
128 * ipa.c (walk_polymorphic_call_targets): Properly guard
129 call to inline_update_overall_summary.
131 2014-02-26 Bin Cheng <bin.cheng@arm.com>
134 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
135 and latches only if requested. Fix latch if it is removed.
136 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
137 LOOPS_HAVE_PREHEADERS.
139 2014-02-25 Andrew Pinski <apinski@cavium.com>
141 * builtins.c (expand_builtin_thread_pointer): Create a new target
142 when the target is NULL.
144 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
146 PR rtl-optimization/60317
147 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
148 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
149 * lra-assigns.c: Include params.h.
150 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
151 other reload pseudos considerations.
153 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
155 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
156 to use canonical form for nor<mode>3.
158 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
161 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
164 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
166 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
167 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
168 (ix86_handle_option): Handle OPT_mprefetchwt1.
169 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
170 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
172 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
173 OPTION_MASK_ISA_PREFETCHWT1.
174 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
175 (PTA_PREFETCHWT1): New.
176 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
177 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
178 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
179 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
180 (*prefetch_avx512pf_<mode>_: Change into ...
181 (*prefetch_prefetchwt1_<mode>: This.
182 * config/i386/i386.opt (mprefetchwt1): New.
183 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
184 (_mm_prefetch): Handle intent to write.
185 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
187 2014-02-25 Richard Biener <rguenther@suse.de>
190 * emit-rtl.c (mem_attrs_htab): Remove.
191 (mem_attrs_htab_hash): Likewise.
192 (mem_attrs_htab_eq): Likewise.
193 (set_mem_attrs): Always allocate new mem-attrs when something changed.
194 (init_emit_once): Do not allocate mem_attrs_htab.
196 2014-02-25 Richard Biener <rguenther@suse.de>
199 * lto-opts.c (lto_write_options): Output non-explicit conservative
200 -fwrapv, -fno-trapv and -fno-strict-overflow.
201 * lto-wrapper.c (merge_and_complain): Handle merging those options.
202 (run_gcc): And pass them through.
204 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
206 * sel-sched.c (calculate_new_fences): New parameter ptime.
207 Calculate it as a maximum over all fence cycles.
208 (sel_sched_region_2): Adjust the call to calculate_new_fences.
209 Print the final schedule timing when sched_verbose.
211 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
213 PR rtl-optimization/60292
214 * sel-sched.c (fill_vec_av_set): Do not reset target availability
215 bit fot the fence instruction.
217 2014-02-24 Alangi Derick <alangiderick@gmail.com>
219 * calls.h: Fix typo in comment.
221 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
223 * config/pa/pa.c (pa_output_move_double): Don't valididate when
224 adjusting offsetable addresses.
226 2014-02-24 Guozhi Wei <carrot@google.com>
228 * sparseset.h (sparseset_pop): Fix the wrong index.
230 2014-02-24 Walter Lee <walt@tilera.com>
232 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
233 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
235 * common/config/tilegx/tilegx-common.c
236 (TARGET_DEFAULT_TARGET_FLAGS): Define.
237 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
239 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
240 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
241 (tilegx_gimplify_va_arg_expr): Handle big endian.
242 (tilegx_expand_unaligned_load): Ditto.
243 (tilegx_expand_unaligned_store): Ditto.
244 (TARGET_RETURN_IN_MSB): New.
245 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
246 (TARGET_ENDIAN_DEFAULT): New.
247 (TARGET_BIG_ENDIAN): Handle big endian.
248 (BYTES_BIG_ENDIAN): Ditto.
249 (WORDS_BIG_ENDIAN): Ditto.
250 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
253 * config/tilegx/tilegx.md (extv): Handle big endian.
256 (insn_st<n>_add<bitsuffix>): Ditto.
257 (insn_stnt<n>): Ditto.
258 (insn_stnt<n>_add<bitsuffix>):Ditto.
259 (vec_interleave_highv8qi): Handle big endian.
260 (vec_interleave_highv8qi_be): New.
261 (vec_interleave_highv8qi_le): New.
262 (insn_v1int_h): Handle big endian.
263 (vec_interleave_lowv8qi): Handle big endian.
264 (vec_interleave_lowv8qi_be): New.
265 (vec_interleave_lowv8qi_le): New.
266 (insn_v1int_l): Handle big endian.
267 (vec_interleave_highv4hi): Handle big endian.
268 (vec_interleave_highv4hi_be): New.
269 (vec_interleave_highv4hi_le): New.
270 (insn_v2int_h): Handle big endian.
271 (vec_interleave_lowv4hi): Handle big endian.
272 (vec_interleave_lowv4hi_be): New.
273 (vec_interleave_lowv4hi_le): New.
274 (insn_v2int_l): Handle big endian.
275 (vec_interleave_highv2si): Handle big endian.
276 (vec_interleave_highv2si_be): New.
277 (vec_interleave_highv2si_le): New.
278 (insn_v4int_h): Handle big endian.
279 (vec_interleave_lowv2si): Handle big endian.
280 (vec_interleave_lowv2si_be): New.
281 (vec_interleave_lowv2si_le): New.
282 (insn_v4int_l): Handle big endian.
283 * config/tilegx/tilegx.opt (mbig-endian): New option.
284 (mlittle-endian): New option.
285 * doc/install.texi: Document tilegxbe-linux.
286 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
288 2014-02-24 Martin Jambor <mjambor@suse.cz>
291 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
292 there are no parameter descriptors.
294 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
296 PR rtl-optimization/60268
297 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
298 initialization to ...
299 (sched_rgn_init): ... here.
300 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
302 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
304 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
307 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
309 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
312 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
314 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
315 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
317 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
319 * config/microblaze/predicates.md: Add cmp_op predicate.
320 * config/microblaze/microblaze.md: Add branch_compare instruction
321 which uses cmp_op predicate and emits cmp insn before branch.
322 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
323 to microblaze_expand_conditional_branch and consolidate logic.
324 (microblaze_expand_conditional_branch): emit branch_compare
325 insn instead of handling cmp op separate from branch insn.
327 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
329 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
332 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
334 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
335 define_insn with define_expand and new define_insn
336 *altivec_lve<VI_char>x_internal.
337 (altivec_stve<VI_char>x): Replace define_insn with define_expand
338 and new define_insn *altivec_stve<VI_char>x_internal.
339 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
341 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
343 (altivec_expand_stvex_be): New function.
345 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
347 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
348 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
349 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
350 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
352 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
355 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
356 instead of emit_move_insn.
358 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
360 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
362 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
365 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
367 * config/rs6000/altivec.md (altivec_lvxl): Rename as
368 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
369 (altivec_lvxl_<mode>): New define_expand incorporating
370 -maltivec=be semantics where needed.
371 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
372 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
373 semantics where needed.
374 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
375 (altivec_stvx_<mode>): New define_expand incorporating
376 -maltivec=be semantics where needed.
377 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
378 VM2 iterator instead of V4SI.
379 (altivec_stvxl_<mode>): New define_expand incorporating
380 -maltivec=be semantics where needed.
381 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
382 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
383 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
384 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
385 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
386 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
387 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
388 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
389 ALTIVEC_BUILTIN_STVXL.
390 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
391 (altivec_expand_stvx_be): Likewise.
392 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
393 (altivec_expand_lvx_be): Likewise.
394 (altivec_expand_stvx_be): Likewise.
395 (altivec_expand_builtin): Add cases for
396 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
397 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
398 (altivec_init_builtins): Add definitions for
399 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
400 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
402 2014-02-21 Catherine Moore <clm@codesourcery.com>
404 * doc/invoke.texi (mvirt, mno-virt): Document.
405 * config/mips/mips.opt (mvirt): New option.
406 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
408 2014-02-21 Richard Biener <rguenther@suse.de>
410 PR tree-optimization/60276
411 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
412 (STMT_VINFO_MIN_NEG_DIST): New macro.
413 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
414 STMT_VINFO_MIN_NEG_DIST.
415 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
416 made for negative dependence distances still hold.
418 2014-02-21 Richard Biener <rguenther@suse.de>
421 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
422 DECL_INITIAL for globals not in the current function context.
424 2014-02-21 Jakub Jelinek <jakub@redhat.com>
426 PR tree-optimization/56490
427 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
428 * tree-ssa-uninit.c: Include params.h.
429 (compute_control_dep_chain): Add num_calls argument, return false
430 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
431 num_calls to recursive call.
432 (find_predicates): Change dep_chain into normal array,
433 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
434 variable and adjust compute_control_dep_chain caller.
435 (find_def_preds): Likewise.
437 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
439 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
440 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
442 2014-02-21 Nick Clifton <nickc@redhat.com>
444 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
446 (popqi1): Add mode to pre_dec.
449 2014-02-21 Jakub Jelinek <jakub@redhat.com>
451 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
452 mode for mask of V8SFmode permutation.
454 2014-02-20 Richard Henderson <rth@redhat.com>
457 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
458 a new pseudo for OLDVAL.
460 2014-02-20 Jakub Jelinek <jakub@redhat.com>
463 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
464 gen_reg_rtx if d->testing_p.
465 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
466 if d->testing_p and we will certainly return true.
467 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
470 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
472 * emit-rtl.c (gen_reg_rtx): Assert that
473 crtl->emit.regno_pointer_align_length is non-zero.
475 2014-02-20 Richard Henderson <rth@redhat.com>
478 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
479 on failure the store back into EXPECT.
481 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
482 Sandra Loosemore <sandra@codesourcery.com>
484 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
485 * config/nios2/nios2.c (nios2_function_profiler): Add
486 -fPIC (flag_pic == 2) support.
487 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
488 (nios2_large_offset_p): New function.
489 (nios2_unspec_reloc_p): Move up position, update to use
490 nios2_large_offset_p.
491 (nios2_unspec_address): Remove function.
492 (nios2_unspec_offset): New function.
493 (nios2_large_got_address): New function.
494 (nios2_got_address): Add large offset support.
495 (nios2_legitimize_tls_address): Update usage of removed and new
497 (nios2_symbol_binds_local_p): New function.
498 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
499 (nios2_legitimize_address): Update to use nios2_large_offset_p.
500 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
501 (nios2_print_operand): Merge H/L processing, add hiadj/lo
502 processing for (const (unspec ...)).
503 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
505 2014-02-20 Richard Biener <rguenther@suse.de>
507 * tree-cfg.c (replace_uses_by): Mark altered BBs before
508 doing the substitution.
509 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
511 2014-02-20 Martin Jambor <mjambor@suse.cz>
514 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
515 info when checking whether lattices are bottom.
517 2014-02-20 Richard Biener <rguenther@suse.de>
520 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
523 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
526 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
527 parameter specifying the scaling.
528 (inline_call): Update.
529 (want_inline_recursively): Guard division by zero.
530 (recursive_inlining): Update.
531 * ipa-inline.h (clone_inlined_nodes): Update.
533 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
536 * config/i386/i386.c (classify_argument): Pass structures of size
537 64 bytes or less in register.
539 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
540 Kirill Yukhin <kirill.yukhin@intel.com>
542 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
543 (_mm_rcp28_round_ss): Ditto.
544 (_mm_rsqrt28_round_sd): Ditto.
545 (_mm_rsqrt28_round_ss): Ditto.
546 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
547 (_mm_rcp14_round_ss): Ditto.
548 (_mm_rsqrt14_round_sd): Ditto.
549 (_mm_rsqrt14_round_ss): Ditto.
550 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
551 the first input operand, get rid of match_dup.
552 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
554 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
556 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
557 operand as the first input operand, set type attribute.
558 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
560 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
561 operand as the first input operand, set type attribute.
563 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
565 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
568 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
571 * config/i386/i386.c (construct_container): Remove TFmode check
572 for X86_64_INTEGER_CLASS.
574 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
577 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
578 only when -Wpsabi is enabled.
580 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
583 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
584 passing arrays in registers are the same as for structs, so remove the
585 special case for them.
587 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
589 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
590 destination type, extract only the valid bits if the source type is not
591 integral and has a different mode.
593 2014-02-19 Richard Biener <rguenther@suse.de>
596 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
599 2014-02-19 Richard Biener <rguenther@suse.de>
602 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
603 (ipa_modify_call_arguments): Emit an argument load explicitely and
604 preserve virtual SSA form there and for the replacement call.
605 Do not update SSA form nor free dominance info.
607 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
609 * ipa.c (function_and_variable_visibility): Also clear WEAK
610 flag when disolving COMDAT_GROUP.
612 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
614 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
615 * ipa-prop.c (ipa_set_jf_known_type): Return early when
617 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
618 do more sanity checks.
619 (detect_type_change): Return true when giving up early.
620 (compute_complex_assign_jump_func): Fix type parameter of
622 (compute_complex_ancestor_jump_func): Likewise.
623 (update_jump_functions_after_inlining): Fix updating of
625 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
627 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
629 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
630 inline clones when edge disappears.
632 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
635 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
636 Split 64-bit moves into 2 patterns. Do not allow the use of
637 direct move for TDmode in little endian, since the decimal value
638 has little endian bytes within a word, but the 64-bit pieces are
639 ordered in a big endian fashion, and normal subreg's of TDmode are
641 (mov<mode>_64bit_dm): Likewise.
642 (movtd_64bit_nodm): Likewise.
644 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
646 PR tree-optimization/60174
647 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
648 statement of an SSA_NAME that occurs in an abnormal PHI node.
650 2014-02-18 Jakub Jelinek <jakub@redhat.com>
653 * final.c (SEEN_BB): Remove.
654 (SEEN_NOTE, SEEN_EMITTED): Renumber.
655 (final_scan_insn): Don't force_source_line on second
656 NOTE_INSN_BASIC_BLOCK.
658 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
661 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
662 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
663 (type_natural_mode): Warn ABI change when %zmm register is not
664 available for AVX512F vector value passing.
666 2014-02-18 Kai Tietz <ktietz@redhat.com>
669 * config/i386/i386.c (ix86_expand_prologue): Use value in
670 rax register as displacement when restoring %r10 or %rax.
671 Fix wrong offset when restoring both registers.
673 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
675 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
676 assertion with conditional return.
678 2014-02-18 Jakub Jelinek <jakub@redhat.com>
679 Uros Bizjak <ubizjak@gmail.com>
682 * config/i386/driver-i386.c (host_detect_local_cpu): If
683 YMM state is not saved by the OS, also clear has_f16c. Move
684 CPUID 0x80000001 handling before YMM state saving checking.
686 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
688 PR rtl-optimization/58960
689 * haifa-sched.c (alloc_global_sched_pressure_data): New,
690 factored out from ...
691 (sched_init): ... here.
692 (free_global_sched_pressure_data): New, factored out from ...
693 (sched_finish): ... here.
694 * sched-int.h (free_global_sched_pressure_data): Declare.
695 * sched-rgn.c (nr_regions_initial): New static global.
696 (haifa_find_rgns): Initialize it.
697 (schedule_region): Disable sched-pressure for the newly
700 2014-02-17 Richard Biener <rguenther@suse.de>
702 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
703 release SSA defs of pattern stmts.
705 2014-02-17 Richard Biener <rguenther@suse.de>
707 * tree-inline.c (expand_call_inline): Release the virtual
708 operand defined by the call we are about to inline.
710 2014-02-17 Richard Biener <rguenther@suse.de>
712 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
714 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
715 Ilya Tocar <ilya.tocar@intel.com>
717 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
718 arguments order in builtin.
719 (_mm512_permutexvar_epi64): Ditto.
720 (_mm512_mask_permutexvar_epi64): Ditto
721 (_mm512_maskz_permutexvar_epi32): Ditto
722 (_mm512_permutexvar_epi32): Ditto
723 (_mm512_mask_permutexvar_epi32): Ditto
725 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
727 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
728 (p8_vmrgow): Likewise.
730 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
732 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
735 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
738 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
739 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
740 into 64-bit and 32-bit moves. On 64-bit moves, add support for
741 using direct move instructions on ISA 2.07. Also adjust
742 instruction length for 64-bit.
743 (mov<mode>_64bit, TFmode/TDmode): Likewise.
744 (mov<mode>_32bit, TFmode/TDmode): Likewise.
746 2014-02-15 Alan Modra <amodra@gmail.com>
750 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
751 find_replacement on parts of insn rtl that might be reloaded.
753 2014-02-15 Richard Biener <rguenther@suse.de>
755 PR tree-optimization/60183
756 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
757 (tree_ssa_phiprop): Calculate and free post-dominators.
759 2014-02-14 Jeff Law <law@redhat.com>
761 PR rtl-optimization/60131
762 * ree.c (get_extended_src_reg): New function.
763 (combine_reaching_defs): Use it rather than assuming location of REG.
764 (find_and_remove_re): Verify first operand of extension is
765 a REG before adding the insns to the copy list.
767 2014-02-14 Roland McGrath <mcgrathr@google.com>
769 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
770 * configure: Regenerated.
771 * config.in: Regenerated.
772 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
773 instead of ASM_SHORT.
775 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
776 Richard Earnshaw <rearnsha@arm.com>
778 PR rtl-optimization/59535
779 * lra-constraints.c (process_alt_operands): Encourage alternative
780 when unassigned pseudo class is superset of the alternative class.
781 (inherit_reload_reg): Don't inherit when optimizing for code size.
782 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
783 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
784 modes not less than 4 for Thumb1.
786 2014-02-14 Kyle McMartin <kyle@redhat.com>
789 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
791 2014-02-14 Richard Biener <rguenther@suse.de>
793 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
794 (get_frame_arg): Drop the assert with langhook types_compatible_p.
795 Do not strip INDIRECT_REFs.
797 2014-02-14 Richard Biener <rguenther@suse.de>
800 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
801 DECL_FUNCTION_SPECIFIC_TARGET.
802 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
803 * tree-streamer-out.c (pack_ts_target_option): Remove.
804 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
805 (write_ts_function_decl_tree_pointers): Do not stream
806 DECL_FUNCTION_SPECIFIC_TARGET.
807 * tree-streamer-in.c (unpack_ts_target_option): Remove.
808 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
809 (lto_input_ts_function_decl_tree_pointers): Do not stream
810 DECL_FUNCTION_SPECIFIC_TARGET.
812 2014-02-14 Jakub Jelinek <jakub@redhat.com>
814 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
815 (get_initial_def_for_induction, vectorizable_induction): Ignore
816 debug stmts when looking for exit_phi.
817 (vectorizable_live_operation): Fix up condition.
819 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
821 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
822 nreverse() because it changes the content of original tree list.
824 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
826 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
827 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
829 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
831 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
832 GNU coding standards.
834 2014-02-13 Jakub Jelinek <jakub@redhat.com>
837 * dwarf2out.c (gen_subprogram_die): Don't call
838 add_calling_convention_attribute if subr_die is old_die.
840 2014-02-13 Sharad Singhai <singhai@google.com>
842 * doc/optinfo.texi: Fix order of nodes.
844 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
846 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
847 operands[2], not operands[3].
849 2014-02-13 Richard Biener <rguenther@suse.de>
852 * doc/install.texi (ISL): Update recommended version to 0.12.2,
853 mention the possibility of an in-tree build.
854 (CLooG): Update recommended version to 0.18.1, mention the
855 possibility of an in-tree build and clarify that the ISL
856 bundled with CLooG does not work.
858 2014-02-13 Jakub Jelinek <jakub@redhat.com>
861 * expr.c (compress_float_constant): If x is a hard register,
862 extend into a pseudo and then move to x.
864 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
866 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
867 caused by bad second argument to warning_at() with -mhotpatch and
868 nested functions (e.g. with gfortran).
870 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
872 * opts.c (option_name): Remove "enabled by default" rider.
874 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
876 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
878 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
879 Uros Bizjak <ubizjak@gmail.com>
882 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
883 * configure: Regenerated.
885 2014-02-12 Richard Biener <rguenther@suse.de>
887 * vec.c (vec_prefix::calculate_allocation): Move as
888 inline variant to vec.h.
889 (vec_prefix::calculate_allocation_1): New out-of-line version.
890 * vec.h (vec_prefix::calculate_allocation_1): Declare.
891 (vec_prefix::m_has_auto_buf): Rename to ...
892 (vec_prefix::m_using_auto_storage): ... this.
893 (vec_prefix::calculate_allocation): Inline the easy cases
894 and dispatch to calculate_allocation_1 which doesn't need the
896 (va_heap::reserve): Use gcc_checking_assert.
897 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
898 m_using_auto_storage.
899 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
901 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
902 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
903 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
905 2014-02-12 Richard Biener <rguenther@suse.de>
907 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
908 when we found a dependence.
910 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
912 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
914 (maybe_fold_stmt): ... into this new function.
915 * omp-low.c (lower_omp): Update comment.
917 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
920 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
923 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
925 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
926 identifiers in comments.
927 (cortexa53_extra_costs): Likewise.
928 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
929 (cortexa7_extra_costs): Likewise.
930 (cortexa12_extra_costs): Likewise.
931 (cortexa15_extra_costs): Likewise.
932 (v7m_extra_costs): Likewise.
934 2014-02-12 Richard Biener <rguenther@suse.de>
937 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
938 of posix_memalign being successful.
939 (lower_stmt): Restrict lowering of posix_memalign to when
940 -ftree-bit-ccp is enabled.
942 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
944 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
946 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
948 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
950 PR rtl-optimization/60116
951 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
952 other_insn once the combination has been validated.
954 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
957 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
959 * ipa-devirt.c: Include demangle.h
960 (odr_violation_reported): New static variable.
961 (add_type_duplicate): Update odr_violations.
962 (maybe_record_node): Add completep parameter; update it.
963 (record_target_from_binfo): Add COMPLETEP parameter;
965 (possible_polymorphic_call_targets_1): Likewise.
966 (struct polymorphic_call_target_d): Add nonconstruction_targets;
967 rename FINAL to COMPLETE.
968 (record_targets_from_bases): Sanity check we found the binfo;
969 fix COMPLETEP updating.
970 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
971 parameter, fix computing of COMPLETEP.
972 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
973 at LTO time do demangling.
974 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
975 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
977 (gimple_get_virt_method_for_binfo): Likewise.
978 * gimple-fold.h (gimple_get_virt_method_for_binfo,
979 gimple_get_virt_method_for_vtable): Update prototypes.
981 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
984 * genautomata.c (add_presence_absence): Fix typo with
985 {final_}presence_list.
987 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
990 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
991 for VSX/Altivec vectors that land in GPR registers.
993 2014-02-11 Richard Henderson <rth@redhat.com>
994 Jakub Jelinek <jakub@redhat.com>
997 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
998 around drhs if type conversion to lacc->type is not useless.
1000 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1002 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1004 (cortex-a57.cortex-a53): Likewise.
1005 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1007 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1009 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1012 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1014 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1015 add_options_for_arm_vfp3.
1017 2014-02-11 Jeff Law <law@redhat.com>
1020 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1021 object with an undesirable mode.
1023 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1026 * config/i386/sol2-9.h: New file.
1027 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1028 *-*-solaris2.9*): Use it.
1030 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1032 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1033 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1035 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1037 * config/microblaze/microblaze.c: Extend mcpu version format
1039 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1041 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1043 2014-02-10 Richard Henderson <rth@redhat.com>
1046 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1047 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1048 ms-abi vs -mno-accumulate-outgoing-args.
1049 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1050 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1053 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1056 * cfgexpand.c (expand_asm_operands): Attach source location to
1057 ASM_INPUT rtx objects.
1058 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1060 2014-02-10 Nick Clifton <nickc@redhat.com>
1062 * config/mn10300/mn10300.c (popcount): New function.
1063 (mn10300_expand_prologue): Include saved registers in stack usage
1066 2014-02-10 Jeff Law <law@redhat.com>
1069 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1070 when changing the SET_DEST of a prior insn to avoid an input reload.
1072 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1074 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1075 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1076 -mcall-openbsd, or -mcall-linux.
1077 (CC1_ENDIAN_BIG_SPEC): Remove.
1078 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1079 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1080 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1081 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1082 and %cc1_endian_default.
1083 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1085 2014-02-10 Richard Biener <rguenther@suse.de>
1087 PR tree-optimization/60115
1088 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1089 MEM_REF handling. Properly verify that the accesses are not
1090 out of the objects bound.
1092 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1094 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1097 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1099 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1100 proper constants and fix formatting.
1101 (possible_polymorphic_call_targets): Fix formatting.
1103 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1104 Ilya Tocar <ilya.tocar@intel.com>
1106 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1107 (_mm512_loadu_epi32): Renamed into...
1108 (_mm512_loadu_si512): This.
1109 (_mm512_storeu_epi32): Renamed into...
1110 (_mm512_storeu_si512): This.
1111 (_mm512_maskz_ceil_ps): Removed.
1112 (_mm512_maskz_ceil_pd): Ditto.
1113 (_mm512_maskz_floor_ps): Ditto.
1114 (_mm512_maskz_floor_pd): Ditto.
1115 (_mm512_floor_round_ps): Ditto.
1116 (_mm512_floor_round_pd): Ditto.
1117 (_mm512_ceil_round_ps): Ditto.
1118 (_mm512_ceil_round_pd): Ditto.
1119 (_mm512_mask_floor_round_ps): Ditto.
1120 (_mm512_mask_floor_round_pd): Ditto.
1121 (_mm512_mask_ceil_round_ps): Ditto.
1122 (_mm512_mask_ceil_round_pd): Ditto.
1123 (_mm512_maskz_floor_round_ps): Ditto.
1124 (_mm512_maskz_floor_round_pd): Ditto.
1125 (_mm512_maskz_ceil_round_ps): Ditto.
1126 (_mm512_maskz_ceil_round_pd): Ditto.
1127 (_mm512_expand_pd): Ditto.
1128 (_mm512_expand_ps): Ditto.
1129 * config/i386/i386.c (ix86_builtins): Remove
1130 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1131 (bdesc_args): Ditto.
1132 * config/i386/predicates.md (const1256_operand): New.
1133 (const_1_to_2_operand): Ditto.
1134 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1135 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1136 (*avx512pf_gatherpf<mode>sf): Ditto.
1137 (avx512pf_gatherpf<mode>df): Ditto.
1138 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1139 (*avx512pf_gatherpf<mode>df): Ditto.
1140 (avx512pf_scatterpf<mode>sf): Ditto.
1141 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1142 (*avx512pf_scatterpf<mode>sf): Ditto.
1143 (avx512pf_scatterpf<mode>df): Ditto.
1144 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1145 (*avx512pf_scatterpf<mode>df): Ditto.
1146 (avx512f_expand<mode>): Removed.
1147 (<shift_insn><mode>3<mask_name>): Change predicate type.
1149 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1151 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1152 not at the end of datarefs vector use ordered_remove to avoid
1153 reordering datarefs vector.
1156 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1157 mark local addressable non-static vars as GOVD_PRIVATE
1158 instead of GOVD_LOCAL.
1159 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1160 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1164 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1165 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1166 assume_aligned or alloc_align attributes.
1167 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1168 arguments. Handle also assume_aligned and alloc_align attributes.
1169 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1170 calls to functions with assume_aligned or alloc_align attributes.
1171 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1173 2014-02-08 Terry Guo <terry.guo@arm.com>
1175 * doc/invoke.texi: Document ARM -march=armv7e-m.
1177 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1179 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1180 flag on __cilkrts_rethrow builtin.
1183 * ipa-cp.c (determine_versionability): Fail at -O0
1184 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1185 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1188 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1191 * tree-inline.c (copy_forbidden): Fail for
1192 __attribute__((optimize (0))) functions.
1194 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1196 * varpool.c: Include pointer-set.h.
1197 (varpool_remove_unreferenced_decls): Variables in other partitions
1198 will not be output; be however careful to not lose information
1201 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1203 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1204 lookup in the vtable constructor.
1206 2014-02-07 Jeff Law <law@redhat.com>
1209 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1210 define_insn_and_split.
1212 * ipa-inline.c (inline_small_functions): Fix typos.
1214 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1216 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1217 (s390_can_use_return_insn): Declare.
1218 * config/s390/s390.h (EPILOGUE_USES): Define.
1219 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1221 (s390_chunkify_start): Handle return JUMP_LABELs.
1222 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1223 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1224 (s390_can_use_return_insn): New functions.
1225 (s390_fix_long_loop_prediction): Handle conditional returns.
1226 (TARGET_SET_UP_BY_PROLOGUE): Define.
1227 * config/s390/s390.md (ANY_RETURN): New code iterator.
1228 (*creturn, *csimple_return, return, simple_return): New patterns.
1230 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1232 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1233 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1234 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1235 REG_CFA_RESTORE list when deciding not to restore a register.
1237 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1239 * config/s390/s390.c: Include tree-pass.h and context.h.
1240 (s390_early_mach): New function, split out from...
1241 (s390_emit_prologue): ...here.
1242 (pass_data_s390_early_mach): New pass structure.
1243 (pass_s390_early_mach): New class.
1244 (s390_option_override): Create and register early_mach pass.
1245 Move to end of file.
1247 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1249 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1250 to match for the exit block.
1252 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1254 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1255 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1256 Reject misaligned operands.
1258 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1260 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1262 2014-02-07 Richard Biener <rguenther@suse.de>
1265 * gimple-low.c (lower_builtin_posix_memalign): New function.
1266 (lower_stmt): Call it to lower posix_memalign in a way
1267 to make alignment info accessible.
1269 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1272 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1273 __builtin_setjmp_receiver.
1275 2014-02-07 Richard Biener <rguenther@suse.de>
1278 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1279 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1280 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1281 Handle BUILT_IN_POSIX_MEMALIGN.
1282 (find_func_clobbers): Likewise.
1283 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1284 (call_may_clobber_ref_p_1): Likewise.
1286 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1289 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1292 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1295 * lto-cgraph.c (lto_output_node): Use
1296 symtab_get_symbol_partitioning_class.
1297 (lto_output_varpool_node): likewise.
1298 (symtab_get_symbol_partitioning_class): Move here from
1300 * cgraph.h (symbol_partitioning_class): Likewise.
1301 (symtab_get_symbol_partitioning_class): Declare.
1303 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1305 * ggc.h (ggc_internal_cleared_alloc): New macro.
1306 * vec.h (vec_safe_copy): Handle memory stats.
1307 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1308 * target-globals.c (save_target_globals): Likewise.
1310 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1313 * expr.c (emit_move_resolve_push): Export; be bit more selective
1314 on when to clear alias set.
1315 * expr.h (emit_move_resolve_push): Declare.
1316 * function.h (struct function): Add tail_call_marked.
1317 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1318 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1319 * config/i386/i386.md (TImode move expander): De not call
1321 (FP push expanders): Preserve memory attributes.
1322 * config/i386/sse.md (push<mode>1): Remove.
1323 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1324 (ix86_expand_push): Remove.
1325 * config/i386/mmx.md (push<mode>1): Remove.
1327 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1329 PR rtl-optimization/60030
1330 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1331 lopart with paradoxical subreg before shifting it up by hprec.
1333 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1335 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1336 Remove extra newline at end of file.
1337 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1338 (arm_issue_rate): Handle cortexa57.
1339 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1340 (cortex-a57.cortex-a53): Likewise.
1342 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1345 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1346 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1348 (arm_expand_prologue): Adjust all callers.
1349 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1350 info, registers also at the lowest numbered registers side. Use
1351 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1355 * var-tracking.c (adjust_mems): Before adding a SET to
1356 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1358 2014-02-06 Alan Modra <amodra@gmail.com>
1361 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1362 change SDmode to DDmode when lra_in_progress.
1364 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1367 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1368 free_data_ref on the dr first, and before goto again also set dr
1369 to the next dr. For simd_lane_access, free old datarefs[i] before
1370 overwriting it. For get_vectype_for_scalar_type failure, don't
1371 free_data_ref if simd_lane_access.
1373 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1376 * tree.h (opts_for_fn): New inline function.
1377 (opt_for_fn): Define.
1378 * config/i386/i386.c (ix86_function_regparm): Use
1379 opt_for_fn (decl, optimize) instead of optimize.
1381 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1383 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1384 for SYMBOL_REF in large memory model.
1386 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1388 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1390 (cortex-a57): Likewise.
1391 (cortex-a57.cortex-a53): Likewise.
1393 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1394 Kugan Vivekanandarajah <kuganv@linaro.org>
1396 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1398 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1400 2014-02-06 Richard Biener <rguenther@suse.de>
1402 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1403 set_loop_copy and initialize_original_copy_tables.
1405 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1407 * config/aarch64/aarch64-simd.md
1408 (aarch64_ashr_simddi): Change QI to SI.
1410 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1411 Jakub Jelinek <jakub@redhat.com>
1414 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1417 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1419 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1420 CODE_FOR_altivec_vpku[hw]um to
1421 CODE_FOR_altivec_vpku[hw]um_direct.
1422 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1423 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1424 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1425 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1427 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1429 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1430 generation for -maltivec=be.
1431 (altivec_vsumsws): Simplify redundant test.
1433 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1435 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1436 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1437 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1438 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1439 gen_altivec_vpkuwum.
1440 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1442 (altivec_vpks<VI_char>ss): Likewise.
1443 (altivec_vpks<VI_char>us): Likewise.
1444 (altivec_vpku<VI_char>us): Likewise.
1445 (altivec_vpku<VI_char>um): Likewise.
1446 (altivec_vpku<VI_char>um_direct): New (copy of
1447 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1449 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1450 target is little endian and -maltivec=be is not specified.
1451 (*altivec_vupkhs<VU_char>_direct): New (copy of
1452 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1453 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1454 target is little endian and -maltivec=be is not specified.
1455 (*altivec_vupkls<VU_char>_direct): New (copy of
1456 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1457 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1458 little endian and -maltivec=be is not specified.
1459 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1460 little endian and -maltivec=be is not specified.
1462 2014-02-05 Richard Henderson <rth@redhat.com>
1465 * combine-stack-adj.c: Revert r206943.
1466 * sched-int.h (struct deps_desc): Add last_args_size.
1467 * sched-deps.c (init_deps): Initialize it.
1468 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1469 contain REG_ARGS_SIZE notes.
1471 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1473 * lto-cgraph.c (asm_nodes_output): Make global.
1474 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1475 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1476 (driver_handle_option): Handle OPT_fwpa.
1478 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1481 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1482 a comment typo and formatting issue. If odr_hash hasn't been
1483 created, return vNULL and set *completep to false.
1486 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1487 bb with no successors.
1489 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1492 * doc/invoke.texi (-march): Clarify documentation for ARM.
1496 2014-02-05 Richard Biener <rguenther@suse.de>
1498 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1499 when not vectorizing because of too many alias checks.
1500 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1501 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1503 2014-02-05 Nick Clifton <nickc@redhat.com>
1505 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1506 accept extended registers in any mode when compiling for the MN10300.
1508 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1510 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1511 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1512 sanitization attributes.
1513 (can_inline_edge_p): Likewise.
1514 (sanitize_attrs_match_for_inline_p): New function.
1516 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1518 * ipa-prop.c (detect_type_change): Shor circuit testing of
1519 type changes on THIS pointer.
1521 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1524 * config/pa/pa.c (legitimize_tls_address): Return original address
1525 if not passed a SYMBOL_REF rtx.
1526 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1528 (pa_emit_move_sequence): Simplify TLS source operands.
1529 (pa_legitimate_constant_p): Reject all TLS constants.
1530 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1531 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1533 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1535 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1536 groups when we know they are controlled by LTO.
1537 * varasm.c (default_binds_local_p_1): If object is in other partition,
1538 it will be resolved locally.
1540 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1542 * config/host-linux.c (linux_gt_pch_use_address): Don't
1543 use SSIZE_MAX because it is not always defined.
1545 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1548 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1549 threshold for pseudo splitting.
1550 (update_ebb_live_info): Process call argument hard registers and
1551 hard registers from insn definition too.
1552 (max_small_class_regs_num): New constant.
1553 (inherit_in_ebb): Update live hard regs through EBBs. Update
1554 reloads_num only for small register classes. Don't split for
1557 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1560 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1563 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1565 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1568 2014-02-04 Marek Polacek <polacek@redhat.com>
1570 * gdbinit.in (pel): Define.
1572 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1574 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1577 2014-02-04 Richard Biener <rguenther@suse.de>
1580 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1581 in function context local.
1582 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1583 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1584 similar to LTO_imported_decl_ref.
1586 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1588 PR tree-optimization/60002
1589 * cgraphclones.c (build_function_decl_skip_args): Clear
1592 PR tree-optimization/60023
1593 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1594 false to gsi_replace.
1595 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1596 has been in some EH region and vec_stmt could throw, add
1597 vec_stmt into the same EH region.
1598 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1599 has no lhs, ignore it.
1600 * internal-fn.c (expand_MASK_LOAD): Likewise.
1603 * tree-inline.c (copy_forbidden): Fail for
1604 __attribute__((optimize (0))) functions.
1607 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1608 is set, copy one less argument.
1609 (expand_simd_clones): Don't subtract clone_info->inbranch
1610 from simd_clone_struct_alloc argument.
1612 PR rtl-optimization/57915
1613 * recog.c (simplify_while_replacing): If all unary/binary/relational
1614 operation arguments are constant, attempt to simplify those.
1617 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1618 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1620 2014-02-04 Richard Biener <rguenther@suse.de>
1622 PR tree-optimization/60012
1623 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1624 TBAA disambiguation to all DDRs.
1626 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1629 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1630 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1632 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1635 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1637 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1639 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1640 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1642 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1645 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1646 to figure out targets of polymorphic calls with known decl.
1647 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1648 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1649 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1650 (get_polymorphic_call_info): ... here.
1651 (get_polymorphic_call_info_from_invariant): New function.
1653 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1655 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1656 lookup via vtable pointer; check for type consistency
1657 and turn inconsitent facts into UNREACHABLE.
1658 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1659 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1660 type inconsistent querries; return UNREACHABLE instead.
1662 2014-02-03 Richard Henderson <rth@twiddle.net>
1665 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1666 already processed this node.
1667 (normalize_one_pred_1): Pass along mark_set.
1668 (normalize_one_pred): Create and destroy a pointer_set_t.
1669 (normalize_one_pred_chain): Likewise.
1671 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1673 PR gcov-profile/58602
1674 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1676 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1679 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1680 -fno-devirtualize; try to devirtualize by the knowledge of
1681 virtual table pointer given by aggregate propagation.
1682 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1683 (ipa_print_node_jump_functions): Dump also offset that
1684 is relevant for polymorphic calls.
1685 (determine_known_aggregate_parts): Add arg_type parameter; use it
1686 instead of determining the type from pointer type.
1687 (ipa_compute_jump_functions_for_edge): Update call of
1688 determine_known_aggregate_parts.
1689 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1690 (gimple_get_virt_method_for_binfo): ... here; simplify using
1691 vtable_pointer_value_to_vtable.
1692 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1693 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1694 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1695 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1697 (vtable_pointer_value_to_binfo): ... here.
1698 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1700 2014-02-03 Teresa Johnson <tejohnson@google.com>
1702 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1703 redef of outer loop index variable.
1705 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1709 * doc/extend.texi (Function Attributes): Typo.
1711 2014-02-03 Cong Hou <congh@google.com>
1713 PR tree-optimization/60000
1714 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1715 if the vectorized statement is a store. A store statement can only
1716 appear at the end of pattern statements.
1718 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1720 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1721 (ix86_option_override_internal): Default long double to 64-bit for
1722 32-bit Bionic and to 128-bit for 64-bit Bionic.
1724 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1725 TARGET_LONG_DOUBLE_128 is true.
1726 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1728 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1729 (mlong-double-64): Negate -mlong-double-128.
1730 (mlong-double-128): New option.
1732 * config/i386/i386-c.c (ix86_target_macros): Define
1733 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1735 * doc/invoke.texi: Document -mlong-double-128.
1737 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1739 PR rtl-optimization/60024
1740 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1742 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1744 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1746 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1748 PR rtl-optimization/57662
1749 * sel-sched.c (code_motion_path_driver): Do not mark already not
1750 existing blocks in the visiting bitmap.
1752 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1754 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1755 on the insn being emitted.
1757 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1758 Will Deacon <will.deacon@arm.com>
1760 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1762 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1764 * config/arm/arm-tables.opt: Regenerate.
1766 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1768 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1769 for vector types other than V16QImode.
1770 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1771 define_expand, and call altivec_expand_vec_perm_le when producing
1772 code with little endian element order.
1773 (*altivec_vperm_<mode>_internal): New insn having previous
1774 behavior of altivec_vperm_<mode>.
1775 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1776 altivec_expand_vec_perm_le when producing code with little endian
1778 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1779 behavior of altivec_vperm_<mode>_uns.
1781 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1783 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1784 (altivec_vsumsws): Add handling for -maltivec=be with a little
1786 (altivec_vsumsws_direct): New.
1787 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1788 gen_altivec_vsumsws.
1790 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1792 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1793 vtable_pointer_value_to_binfo): New functions.
1794 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1795 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1797 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1799 * config/nios2/nios2.md (load_got_register): Initialize GOT
1800 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1801 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1803 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1805 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1806 preserverd by passthrough, do not propagate the type.
1808 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1810 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1811 (mips_atomic_assign_expand_fenv): New function.
1812 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1814 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1816 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1817 (__builtin_mips_set_fcsr): Likewise.
1818 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1819 MIPS_USI_FTYPE_VOID.
1820 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1821 (mips16_expand_set_fcsr): Likewise.
1822 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1823 (mips16_set_fcsr_stub): Likewise.
1824 (mips16_get_fcsr_one_only_stub): New class.
1825 (mips16_set_fcsr_one_only_stub): Likewise.
1826 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1827 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1828 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1829 (hard_float): New availability predicate.
1830 (mips_builtins): Add get_fcsr and set_fcsr.
1831 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1832 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1833 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1834 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1835 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1838 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1840 * config/mips/mips.c (mips_one_only_stub): New class.
1841 (mips_need_mips16_rdhwr_p): Replace with...
1842 (mips16_rdhwr_stub): ...this new variable.
1843 (mips16_stub_call_address): New function.
1844 (mips16_rdhwr_one_only_stub): New class.
1845 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1846 (mips_output_mips16_rdhwr): Delete.
1847 (mips_finish_stub): New function.
1848 (mips_code_end): Use it to handle rdhwr stubs.
1850 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1853 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1854 when calculating size of integer atomic types.
1856 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1858 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1860 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1862 PR tree-optimization/60003
1863 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1864 * profile.c (branch_prob): Use gimple_call_builtin_p
1865 to check for BUILT_IN_SETJMP_RECEIVER.
1866 * tree-inline.c (copy_bb): Call notice_special_calls.
1868 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1871 * lra-constraints.c (process_alt_operands): Update reload_sum only
1874 2014-01-31 Richard Henderson <rth@redhat.com>
1877 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1878 until after else_eh is processed.
1880 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1882 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1883 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1884 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1885 in smmintrin.h, remove them.
1886 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1887 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1888 * config/i386/i386.md (ROUND_SAE): Fix value.
1889 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1890 (const48_operand): New.
1891 * config/i386/subst.md (round), (round_expand): Use
1892 const_4_or_8_to_11_operand.
1893 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1895 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1897 * config/i386/constraints.md (Yk): Swap meaning with k.
1898 * config/i386/i386.md (movhi_internal): Change Yk to k.
1899 (movqi_internal): Ditto.
1900 (*k<logic><mode>): Ditto.
1903 (kandn<mode>): Ditto.
1904 (*<code>hi_1): Ditto.
1905 (*<code>qi_1): Ditto.
1906 (kxnor<mode>): Ditto.
1907 (kortestzhi): Ditto.
1908 (kortestchi): Ditto.
1910 (*one_cmplhi2_1): Ditto.
1911 (*one_cmplqi2_1): Ditto.
1912 * config/i386/sse.md (): Change k to Yk.
1913 (avx512f_load<mode>_mask): Ditto.
1914 (avx512f_blendm<mode>): Ditto.
1915 (avx512f_store<mode>_mask): Ditto.
1916 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1917 (avx512f_storedqu<mode>_mask): Ditto.
1918 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1920 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1921 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1922 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1923 (avx512f_maskcmp<mode>3): Ditto.
1924 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1925 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1926 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1927 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1928 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1929 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1930 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1931 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1932 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1933 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1934 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1935 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1936 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1937 (vec_extract_lo_<mode>_maskm): Ditto.
1938 (vec_extract_hi_<mode>_maskm): Ditto.
1939 (avx512f_vternlog<mode>_mask): Ditto.
1940 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1941 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1942 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1943 (avx512f_<code>v8div16qi2_mask): Ditto.
1944 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1945 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1946 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1947 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1948 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1949 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1950 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1951 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1952 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1953 (avx512cd_maskb_vec_dupv8di): Ditto.
1954 (avx512cd_maskw_vec_dupv16si): Ditto.
1955 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1956 (avx512f_vpermi2var<mode>3_mask): Ditto.
1957 (avx512f_vpermi2var<mode>3_mask): Ditto.
1958 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1959 (*avx512f_gathersi<mode>): Ditto.
1960 (*avx512f_gathersi<mode>_2): Ditto.
1961 (*avx512f_gatherdi<mode>): Ditto.
1962 (*avx512f_gatherdi<mode>_2): Ditto.
1963 (*avx512f_scattersi<mode>): Ditto.
1964 (*avx512f_scatterdi<mode>): Ditto.
1965 (avx512f_compress<mode>_mask): Ditto.
1966 (avx512f_compressstore<mode>_mask): Ditto.
1967 (avx512f_expand<mode>_mask): Ditto.
1968 * config/i386/subst.md (mask): Change k to Yk.
1969 (mask_scalar_merge): Ditto.
1972 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1974 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1976 2014-01-31 Richard Biener <rguenther@suse.de>
1979 * builtins.c (fold_builtin_memory_op): Make sure to not
1980 use a floating-point mode or a boolean or enumeral type for
1983 2014-01-30 DJ Delorie <dj@redhat.com>
1985 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1986 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1987 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1988 whenever main() has an epilogue.
1990 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1992 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1993 unused variable "field".
1994 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1995 (vsx_mergeh_<mode>): Likewise.
1996 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1997 (altivec_vmrghh): Likewise.
1998 (altivec_vmrghw): Likewise.
1999 (altivec_vmrglb): Likewise.
2000 (altivec_vmrglh): Likewise.
2001 (altivec_vmrglw): Likewise.
2002 (altivec_vspltb): Add missing uses.
2003 (altivec_vsplth): Likewise.
2004 (altivec_vspltw): Likewise.
2005 (altivec_vspltsf): Likewise.
2007 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2010 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2011 frame related instructions.
2013 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2015 PR rtl-optimization/59959
2016 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2017 any reload of register whose subreg is invalid.
2019 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2021 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2022 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2023 Add missing return type - void.
2025 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2027 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2028 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2029 remove element index adjustment for endian (now handled in vsx.md
2031 (altivec_expand_vec_perm_const): Use
2032 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2033 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2034 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2035 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2036 define_expand and a new define_insn *altivec_vspltb_internal;
2037 adjust for -maltivec=be on a little endian target.
2038 (altivec_vspltb_direct): New.
2039 (altivec_vsplth): Divide into a define_expand and a new
2040 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2041 little endian target.
2042 (altivec_vsplth_direct): New.
2043 (altivec_vspltw): Divide into a define_expand and a new
2044 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2045 little endian target.
2046 (altivec_vspltw_direct): New.
2047 (altivec_vspltsf): Divide into a define_expand and a new
2048 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2049 a little endian target.
2051 2014-01-30 Richard Biener <rguenther@suse.de>
2053 PR tree-optimization/59993
2054 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2055 can propagate form the earlier stmt and avoid the transform
2056 when the intermediate result is needed.
2058 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2060 * README.Portability: Fix typo.
2062 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2064 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2065 comparison_operator with ordered_comparison_operator.
2067 2014-01-30 Nick Clifton <nickc@redhat.com>
2069 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2070 Rename to mn10300_store_multiple_regs.
2071 * config/mn10300/mn10300.c: Likewise.
2072 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2073 store_multiple_regs.
2074 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2075 Call mn10300_store_multiple_regs.
2077 2014-01-30 Nick Clifton <nickc@redhat.com>
2078 DJ Delorie <dj@redhat.com>
2080 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2081 %fp 2 to keep registers after it properly word-aligned.
2082 (rl78_alloc_physical_registers_umul): Handle the case where both
2083 input operands are the same.
2085 2014-01-30 Richard Biener <rguenther@suse.de>
2087 PR tree-optimization/59903
2088 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2091 2014-01-30 Jason Merrill <jason@redhat.com>
2094 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2097 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2099 2014-01-30 Richard Biener <rguenther@suse.de>
2101 PR tree-optimization/59951
2102 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2104 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2107 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2108 SFmode to DFmode case.
2110 2014-01-29 DJ Delorie <dj@redhat.com>
2112 * config/msp430/msp430.opt (-minrt): New.
2113 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2115 (ENDFILE_SPEC): Likewise.
2117 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2119 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2120 (estimate_function_body_sizes): Use it.
2122 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2125 * dwarf2out.c (is_cxx_auto): New.
2126 (is_base_type): Use it.
2127 (gen_type_die_with_usage): Likewise.
2129 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2131 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2132 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2133 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2134 -maltivec=be with LE targets.
2135 (vsx_mergeh_<mode>): Likewise.
2136 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2137 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2138 (altivec_vmrghb): Replace with define_expand and new
2139 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2140 (altivec_vmrghb_direct): New define_insn.
2141 (altivec_vmrghh): Replace with define_expand and new
2142 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2143 (altivec_vmrghh_direct): New define_insn.
2144 (altivec_vmrghw): Replace with define_expand and new
2145 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2146 (altivec_vmrghw_direct): New define_insn.
2147 (*altivec_vmrghsf): Adjust for endianness.
2148 (altivec_vmrglb): Replace with define_expand and new
2149 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2150 (altivec_vmrglb_direct): New define_insn.
2151 (altivec_vmrglh): Replace with define_expand and new
2152 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2153 (altivec_vmrglh_direct): New define_insn.
2154 (altivec_vmrglw): Replace with define_expand and new
2155 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2156 (altivec_vmrglw_direct): New define_insn.
2157 (*altivec_vmrglsf): Adjust for endianness.
2158 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2159 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2160 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2161 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2162 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2163 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2164 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2165 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2167 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2169 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2170 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2173 2014-01-29 Richard Biener <rguenther@suse.de>
2175 PR tree-optimization/58742
2176 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2177 associate_pointerplus_align.
2178 (associate_pointerplus_diff): New function.
2179 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2180 and associate_pointerplus_diff.
2182 2014-01-29 Richard Biener <rguenther@suse.de>
2184 * lto-streamer.h (LTO_major_version): Bump to 3.
2185 (LTO_minor_version): Reset to 0.
2187 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2189 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2190 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2191 (arm_file_start): Generate correct asm header for armv7ve.
2192 * config/arm/bpabi.h: Add multilib support for armv7ve.
2193 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2194 and cortex-a15 to armv7ve.
2195 * config/arm/t-aprofile: Add multilib support for armv7ve.
2196 * doc/invoke.texi: Document -march=armv7ve.
2198 2014-01-29 Richard Biener <rguenther@suse.de>
2200 PR tree-optimization/58742
2201 * tree-ssa-forwprop.c (associate_plusminus): Return true
2202 if we changed sth, defer EH cleanup to ...
2203 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2204 (simplify_mult): New function.
2206 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2209 PR tree-optimization/59920
2210 * tree.c (build_common_builtin_nodes): Remove
2211 __builtin_setjmp_dispatcher initialization.
2212 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2213 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2214 instead of gsi_after_labels + manually skipping debug stmts.
2215 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2216 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2217 * tree-inline.c (copy_edges_for_bb): Remove
2218 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2219 argument. Ignore computed_goto_p stmts. Don't call
2220 make_abnormal_goto_edges. If a call might need abnormal edges
2221 for non-local gotos, see if it already has an edge to
2222 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2223 with true argument, don't do anything then, otherwise add
2224 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2225 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2227 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2228 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2229 (lower_stmt): Don't set data->calls_builtin_setjmp.
2230 (lower_builtin_setjmp): Adjust comment.
2231 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2232 * tree-cfg.c (found_computed_goto): Remove.
2233 (factor_computed_gotos): Remove.
2234 (make_goto_expr_edges): Return bool, true for computed gotos.
2235 Don't call make_abnormal_goto_edges.
2236 (build_gimple_cfg): Don't set found_computed_goto, don't call
2237 factor_computed_gotos.
2238 (computed_goto_p): No longer static.
2239 (make_blocks): Don't set found_computed_goto.
2240 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2241 (make_edges): If make_goto_expr_edges returns true, push bb
2242 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2243 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2244 vector. Record mapping between bbs and OpenMP regions if there
2245 are any, adjust make_gimple_omp_edges caller. Call
2246 handle_abnormal_edges.
2247 (make_abnormal_goto_edges): Remove.
2248 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2249 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2250 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2251 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2252 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2253 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2254 filling *region also set *region_idx to (*region)->entry->index.
2257 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2258 For REGs set ORIGINAL_REGNO.
2260 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2262 * doc/md.texi: Mention that a target shouldn't implement
2263 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2266 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2268 PR tree-optimization/59594
2269 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2270 a copy of the datarefs vector rather than the vector itself.
2272 2014-01-28 Jason Merrill <jason@redhat.com>
2275 * dwarf2out.c (auto_die): New static.
2276 (gen_type_die_with_usage): Handle C++1y 'auto'.
2277 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2280 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2283 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2284 (SPEC_X32): Likewise.
2285 (SPEC_64): Likewise.
2286 * config/i386/i386.c (ix86_option_override_internal): Turn off
2287 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2289 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2290 * config/i386/i386.h (TARGET_16BIT): New macro.
2291 (TARGET_16BIT_P): Likewise.
2292 * config/i386/i386.opt: Add m16.
2293 * doc/invoke.texi: Document -m16.
2295 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2297 PR preprocessor/59935
2298 * input.c (location_get_source_line): Bail out on when line number
2299 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2301 2014-01-28 Richard Biener <rguenther@suse.de>
2303 PR tree-optimization/58742
2304 * tree-ssa-forwprop.c (associate_plusminus): Handle
2305 pointer subtraction of the form (T)(P + A) - (T)P.
2307 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2309 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2312 2014-01-28 Richard Biener <rguenther@suse.de>
2315 2014-01-28 Richard Biener <rguenther@suse.de>
2317 PR rtl-optimization/45364
2318 PR rtl-optimization/59890
2319 * var-tracking.c (local_get_addr_clear_given_value): Handle
2320 already cleared slot.
2321 (val_reset): Handle not allocated local_get_addr_cache.
2322 (vt_find_locations): Use post-order on the inverted CFG.
2324 2014-01-28 Richard Biener <rguenther@suse.de>
2326 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2328 2014-01-28 Richard Biener <rguenther@suse.de>
2330 PR rtl-optimization/45364
2331 PR rtl-optimization/59890
2332 * var-tracking.c (local_get_addr_clear_given_value): Handle
2333 already cleared slot.
2334 (val_reset): Handle not allocated local_get_addr_cache.
2335 (vt_find_locations): Use post-order on the inverted CFG.
2337 2014-01-28 Alan Modra <amodra@gmail.com>
2339 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2340 * configure.ac <recursive call for build != host>: Define
2341 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2342 and LD_FOR_BUILD too.
2343 * configure: Regenerate.
2345 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2347 * config/i386/i386.c (get_builtin_code_for_version): Separate
2348 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2349 Broadwell from Haswell.
2351 2014-01-27 Steve Ellcey <sellcey@mips.com>
2353 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2354 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2355 * config/mips/mips.c (mips_option_override): Change setting
2357 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2358 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2359 Change from Mask to Var.
2361 2014-01-27 Jeff Law <law@redhat.com>
2363 * ipa-inline.c (inline_small_functions): Fix typo.
2365 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2367 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2368 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2369 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2370 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2371 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2372 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2373 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2374 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2375 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2376 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2377 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2378 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2379 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2380 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2381 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2382 (_mm512_storeu_epi64): Ditto.
2383 (_mm512_cmpge_epi32_mask): Ditto.
2384 (_mm512_cmpge_epu32_mask): Ditto.
2385 (_mm512_cmpge_epi64_mask): Ditto.
2386 (_mm512_cmpge_epu64_mask): Ditto.
2387 (_mm512_cmple_epi32_mask): Ditto.
2388 (_mm512_cmple_epu32_mask): Ditto.
2389 (_mm512_cmple_epi64_mask): Ditto.
2390 (_mm512_cmple_epu64_mask): Ditto.
2391 (_mm512_cmplt_epi32_mask): Ditto.
2392 (_mm512_cmplt_epu32_mask): Ditto.
2393 (_mm512_cmplt_epi64_mask): Ditto.
2394 (_mm512_cmplt_epu64_mask): Ditto.
2395 (_mm512_cmpneq_epi32_mask): Ditto.
2396 (_mm512_cmpneq_epu32_mask): Ditto.
2397 (_mm512_cmpneq_epi64_mask): Ditto.
2398 (_mm512_cmpneq_epu64_mask): Ditto.
2399 (_mm512_expand_pd): Ditto.
2400 (_mm512_expand_ps): Ditto.
2401 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2402 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2403 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2404 * config/i386/i386.c (ix86_builtins): Add
2405 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2406 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2407 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2408 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2409 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2410 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2411 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2412 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2413 IX86_BUILTIN_PMOVUSQW512_MEM.
2414 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2415 __builtin_ia32_pmovsqd512mem_mask,
2416 __builtin_ia32_pmovqd512mem_mask,
2417 __builtin_ia32_pmovusqw512mem_mask,
2418 __builtin_ia32_pmovsqw512mem_mask,
2419 __builtin_ia32_pmovqw512mem_mask,
2420 __builtin_ia32_pmovusdw512mem_mask,
2421 __builtin_ia32_pmovsdw512mem_mask,
2422 __builtin_ia32_pmovdw512mem_mask,
2423 __builtin_ia32_pmovqb512mem_mask,
2424 __builtin_ia32_pmovusqb512mem_mask,
2425 __builtin_ia32_pmovsqb512mem_mask,
2426 __builtin_ia32_pmovusdb512mem_mask,
2427 __builtin_ia32_pmovsdb512mem_mask,
2428 __builtin_ia32_pmovdb512mem_mask.
2429 (bdesc_args): Add __builtin_ia32_expanddf512,
2430 __builtin_ia32_expandsf512.
2431 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2432 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2433 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2434 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2435 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2436 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2437 (avx512f_<code>v8div16qi2_mask_store): This.
2438 (avx512f_expand<mode>): New.
2440 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2442 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2444 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2445 (_mm512_prefetch_i32scatter_pd): Ditto.
2446 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2447 (_mm512_prefetch_i64scatter_pd): Ditto.
2448 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2449 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2450 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2451 (_mm512_prefetch_i32scatter_ps): Ditto.
2452 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2453 (_mm512_prefetch_i64scatter_ps): Ditto.
2454 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2455 * config/i386/i386-builtin-types.def: Define
2456 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2457 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2458 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2459 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2460 IX86_BUILTIN_SCATTERPFQPD.
2461 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2462 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2463 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2464 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2465 __builtin_ia32_scatterpfqps.
2466 (ix86_expand_builtin): Expand new built-ins.
2467 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2468 fix memory access data type.
2469 (*avx512pf_gatherpf<mode>_mask): Ditto.
2470 (*avx512pf_gatherpf<mode>): Ditto.
2471 (avx512pf_scatterpf<mode>): Ditto.
2472 (*avx512pf_scatterpf<mode>_mask): Ditto.
2473 (*avx512pf_scatterpf<mode>): Ditto.
2474 (GATHER_SCATTER_SF_MEM_MODE): New.
2475 (avx512pf_gatherpf<mode>df): Ditto.
2476 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2477 (*avx512pf_scatterpf<mode>df): Ditto.
2479 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2482 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2483 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2486 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2488 * common/config/arm/arm-common.c
2489 (arm_rewrite_mcpu): Handle multiple names.
2491 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2493 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2495 * gimple-builder.h (create_gimple_tmp): Delete.
2497 2014-01-27 Christian Bruel <christian.bruel@st.com>
2499 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2502 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2504 * config/pa/pa.md (call): Generate indirect long calls to non-local
2505 functions when outputing 32-bit code.
2506 (call_value): Likewise except for special call to buggy powf function.
2508 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2509 portable runtime and PIC indirect calls.
2510 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2511 and PIC call sequences. Use ldo instead of blr to set return register
2512 in PIC call sequence.
2514 2014-01-25 Walter Lee <walt@tilera.com>
2516 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2517 avoid clobbering a live register.
2519 2014-01-25 Walter Lee <walt@tilera.com>
2521 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2522 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2523 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2524 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2526 2014-01-25 Walter Lee <walt@tilera.com>
2528 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2529 arguments on even registers.
2530 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2532 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2533 (BIGGEST_ALIGNMENT): Ditto.
2534 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2536 2014-01-25 Walter Lee <walt@tilera.com>
2538 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2539 insns before bundling.
2540 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2542 2014-01-25 Walter Lee <walt@tilera.com>
2544 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2545 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2546 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2548 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2550 * config/mips/constraints.md (kl): Delete.
2551 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2552 define expands, using...
2553 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2554 instructions for MIPS16.
2555 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2556 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2558 2014-01-25 Walter Lee <walt@tilera.com>
2560 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2564 2014-01-25 Walter Lee <walt@tilera.com>
2566 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2567 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2569 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2571 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2574 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2576 * print-rtl.c (in_call_function_usage): New var.
2577 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2578 EXPR_LIST mode as mode and not as reg note name.
2581 * cfgloopmanip.c (copy_loop_info): If
2582 loop->warned_aggressive_loop_optimizations, make sure
2583 the flag is set in target loop too.
2585 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2587 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2589 * builtins.def: Likewise.
2590 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2591 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2592 * ira.c (ira_setup_eliminable_regset): Likewise.
2593 * omp-low.c (gate_expand_omp): Likewise.
2594 (execute_lower_omp): Likewise.
2595 (diagnose_sb_0): Likewise.
2596 (gate_diagnose_omp_blocks): Likewise.
2597 (simd_clone_clauses_extract): Likewise.
2600 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2602 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2603 correction for little endian...
2604 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2607 2014-01-24 Jeff Law <law@redhat.com>
2609 PR tree-optimization/59919
2610 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2611 for non-returning calls.
2613 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2615 * common/config/aarch64/aarch64-common.c
2616 (aarch64_rewrite_mcpu): Handle multiple names.
2617 * config/aarch64/aarch64.h
2618 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2620 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2622 * input.c (add_file_to_cache_tab): Handle the case where fopen
2625 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2628 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2629 from push operand if code of push isn't PRE_DEC.
2631 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2634 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2635 -mquad-memory-atomic. Update -mquad-memory documentation to say
2636 it is only used for non-atomic loads/stores.
2638 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2639 -mquad-memory or -mquad-memory-atomic switches.
2641 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2642 -mquad-memory-atomic to ISA 2.07 support.
2644 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2645 to separate support of normal quad word memory operations (ldq, stq)
2646 from the atomic quad word memory operations.
2648 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2649 support to separate non-atomic quad word operations from atomic
2650 quad word operations. Disable non-atomic quad word operations in
2651 little endian mode so that we don't have to swap words after the
2652 load and before the store.
2653 (quad_load_store_p): Add comment about atomic quad word support.
2654 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2655 options printed with -mdebug=reg.
2657 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2658 -mquad-memory-atomic as the test for whether we have quad word
2659 atomic instructions.
2660 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2661 or -mp8-vector are used, allow byte/half-word atomic operations.
2663 * config/rs6000/sync.md (load_lockedti): Insure that the address
2664 is a proper indexed or indirect address for the lqarx instruction.
2665 On little endian systems, swap the hi/lo registers after the lqarx
2667 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2668 insure the address is valid for the lqarx instruction.
2669 (store_conditionalti): Insure that the address is a proper indexed
2670 or indirect address for the stqcrx. instruction. On little endian
2671 systems, swap the hi/lo registers before doing the stqcrx.
2673 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2674 insure the address is valid for the stqcrx. instruction.
2676 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2677 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2678 type of quad memory support is available.
2680 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2683 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2684 there is a danger of looping.
2686 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2689 force flag_ira_loop_pressure if set via command line.
2691 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2693 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2694 (ashr_simd): New builtin handling DI mode.
2695 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2696 (aarch64_sshr_simddi): New match pattern.
2697 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2698 (vshrd_n_s64): Likewise.
2699 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2701 2014-01-23 Nick Clifton <nickc@redhat.com>
2703 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2704 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2705 favour of mcu specific scripts.
2706 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2709 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2710 Alex Velenko <Alex.Velenko@arm.com>
2712 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2713 (vaddv_s16): Likewise.
2714 (vaddv_s32): Likewise.
2715 (vaddv_u8): Likewise.
2716 (vaddv_u16): Likewise.
2717 (vaddv_u32): Likewise.
2718 (vaddvq_s8): Likewise.
2719 (vaddvq_s16): Likewise.
2720 (vaddvq_s32): Likewise.
2721 (vaddvq_s64): Likewise.
2722 (vaddvq_u8): Likewise.
2723 (vaddvq_u16): Likewise.
2724 (vaddvq_u32): Likewise.
2725 (vaddvq_u64): Likewise.
2726 (vaddv_f32): Likewise.
2727 (vaddvq_f32): Likewise.
2728 (vaddvq_f64): Likewise.
2729 (vmaxv_f32): Likewise.
2730 (vmaxv_s8): Likewise.
2731 (vmaxv_s16): Likewise.
2732 (vmaxv_s32): Likewise.
2733 (vmaxv_u8): Likewise.
2734 (vmaxv_u16): Likewise.
2735 (vmaxv_u32): Likewise.
2736 (vmaxvq_f32): Likewise.
2737 (vmaxvq_f64): Likewise.
2738 (vmaxvq_s8): Likewise.
2739 (vmaxvq_s16): Likewise.
2740 (vmaxvq_s32): Likewise.
2741 (vmaxvq_u8): Likewise.
2742 (vmaxvq_u16): Likewise.
2743 (vmaxvq_u32): Likewise.
2744 (vmaxnmv_f32): Likewise.
2745 (vmaxnmvq_f32): Likewise.
2746 (vmaxnmvq_f64): Likewise.
2747 (vminv_f32): Likewise.
2748 (vminv_s8): Likewise.
2749 (vminv_s16): Likewise.
2750 (vminv_s32): Likewise.
2751 (vminv_u8): Likewise.
2752 (vminv_u16): Likewise.
2753 (vminv_u32): Likewise.
2754 (vminvq_f32): Likewise.
2755 (vminvq_f64): Likewise.
2756 (vminvq_s8): Likewise.
2757 (vminvq_s16): Likewise.
2758 (vminvq_s32): Likewise.
2759 (vminvq_u8): Likewise.
2760 (vminvq_u16): Likewise.
2761 (vminvq_u32): Likewise.
2762 (vminnmv_f32): Likewise.
2763 (vminnmvq_f32): Likewise.
2764 (vminnmvq_f64): Likewise.
2766 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2768 * config/aarch64/aarch64-simd.md
2769 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2770 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2771 (*aarch64_mul3_elt<mode>): Likewise.
2772 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2773 (*aarch64_mul3_elt_to_64v2df): Likewise.
2774 (*aarch64_mla_elt<mode>): Likewise.
2775 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2776 (*aarch64_mls_elt<mode>): Likewise.
2777 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2778 (*aarch64_fma4_elt<mode>): Likewise.
2779 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2780 (*aarch64_fma4_elt_to_64v2df): Likewise.
2781 (*aarch64_fnma4_elt<mode>): Likewise.
2782 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2783 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2784 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2785 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2786 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2787 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2788 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2789 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2790 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2792 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2794 * config/aarch64/aarch64-simd.md
2795 (aarch64_be_checked_get_lane<mode>): New define_expand.
2796 * config/aarch64/aarch64-simd-builtins.def
2797 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2798 New builtin definition.
2799 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2800 Use new safe be builtin.
2802 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2804 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2806 (aarch64_be_st1<mode>): Likewise.
2807 (aarch_ld1<VALL:mode>): Define_expand modified.
2808 (aarch_st1<VALL:mode>): Likewise.
2809 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2810 (UNSPEC_ST1): Likewise.
2812 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2814 * config/microblaze/microblaze.md: Add trap insn and attribute
2816 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2818 PR preprocessor/58580
2819 * input.h (location_get_source_line): Take an additional line_size
2821 (void diagnostics_file_cache_fini): Declare new function.
2822 * input.c (struct fcache): New type.
2823 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2824 New static constants.
2825 (diagnostic_file_cache_init, total_lines_num)
2826 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2827 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2828 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2829 (get_next_line, read_next_line, goto_next_line, read_line_num):
2830 New static function definitions.
2831 (diagnostic_file_cache_fini): New function.
2832 (location_get_source_line): Take an additional output line_len
2833 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2835 * diagnostic.c (diagnostic_finish): Call
2836 diagnostic_file_cache_fini.
2837 (adjust_line): Take an additional input parameter for the length
2838 of the line, rather than calculating it with strlen.
2839 (diagnostic_show_locus): Adjust the use of
2840 location_get_source_line and adjust_line with respect to their new
2841 signature. While displaying a line now, do not stop at the first
2842 null byte. Rather, display the zero byte as a space and keep
2843 going until we reach the size of the line.
2844 * Makefile.in: Add vec.o to OBJS-libcommon
2846 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2847 Ilya Tocar <ilya.tocar@intel.com>
2849 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2850 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2851 (__builtin_ia32_kmov16): Ditto.
2852 * config/i386/i386.md (UNSPEC_KMOV): New.
2855 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2857 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2858 (_mm512_storeu_si512): Ditto.
2860 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2863 * rtl.h (get_referenced_operands): Declare.
2864 * recog.c (get_referenced_operands): New function.
2865 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2866 operands have been referenced when recording LO_SUM references.
2868 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2870 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2872 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2874 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2875 Enable for generic and recent AMD targets.
2877 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2879 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2880 ARG_SIZE note when adjustment was eliminated.
2882 2014-01-22 Jeff Law <law@redhat.com>
2884 PR tree-optimization/59597
2885 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2886 in file. Accept new argument REGISTERING and use it to modify
2887 dump output appropriately.
2888 (register_jump_thread): Corresponding changes.
2889 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2890 thread paths involving joiner blocks. Add code to dump cancelled
2891 jump threading paths.
2893 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2895 PR rtl-optimization/59477
2896 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2897 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2899 2014-01-22 Tom Tromey <tromey@redhat.com>
2901 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2903 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2905 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2907 PR rtl-optimization/59896
2908 * lra-constraints.c (process_alt_operands): Check unused note for
2909 matched operands of insn with no output reloads.
2911 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2913 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2914 (mips_move_from_gpr_cost): Likewise.
2916 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2918 PR rtl-optimization/59858
2919 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2920 ira_class_hard_regs_num.
2921 (process_alt_operands): Increase reject for dying matched operand.
2923 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2926 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2927 smaller than size, perform several stores or loads and stores
2928 at dst + count - size to store or copy all of size bytes, rather
2929 than just last modesize bytes.
2931 2014-01-20 DJ Delorie <dj@redhat.com>
2933 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2934 that CLOBBERs are REGs before propogating their values.
2936 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2939 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2940 (cgraph_inline_failed_type): New function.
2941 * cgraph.h (DEFCIFCODE): Add type.
2942 (cgraph_inline_failed_type_t): New enum.
2943 (cgraph_inline_failed_type): New prototype.
2944 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2945 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2946 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2947 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2948 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2949 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2950 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2951 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2952 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2953 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2954 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2955 OPTIMIZATION_MISMATCH.
2956 * tree-inline.c (expand_call_inline): Emit errors during
2957 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2959 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2962 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2963 mode attribute in insn output.
2965 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2967 * output.h (output_constant): Delete.
2968 * varasm.c (output_constant): Make private.
2970 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2972 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2974 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2977 * tree.h (fold_builtin_strcat): New prototype.
2978 * builtins.c (fold_builtin_strcat): No longer static. Add len
2979 argument, if non-NULL, don't call c_strlen. Optimize
2980 directly into __builtin_memcpy instead of __builtin_strcpy.
2981 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2982 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2984 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2986 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2987 for SImode_address_operand operands, having only a REG argument.
2989 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2991 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2992 loader name using mbig-endian.
2993 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2995 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2997 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3001 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3003 * config/aarch64/aarch64-protos.h
3004 (aarch64_cannot_change_mode_class_ptr): Declare.
3005 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3006 aarch64_cannot_change_mode_class_ptr): New.
3007 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3008 backend hook aarch64_cannot_change_mode_class.
3010 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3012 * common/config/aarch64/aarch64-common.c
3013 (aarch64_handle_option): Don't handle any option order logic here.
3014 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3015 selected_cpu, warn on architecture version mismatch.
3016 (aarch64_override_options): Fix parsing order for option strings.
3018 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3019 Iain Sandoe <iain@codesourcery.com>
3022 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3023 warning. Amend comment to reflect current functionality.
3025 2014-01-20 Richard Biener <rguenther@suse.de>
3028 * builtins.c (fold_builtin_strcat): Remove case better handled
3029 by tree-ssa-strlen.c.
3031 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3033 * config/aarch64/aarch64.opt
3034 (mcpu, march, mtune): Make case-insensitive.
3036 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3039 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3040 if operands[1] is a REG or ZERO_EXTEND of a REG.
3042 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3044 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3046 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3048 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3049 long non-pic millicode calls.
3051 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3053 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3055 2014-01-19 Kito Cheng <kito@0xlab.org>
3057 * builtins.c (expand_movstr): Check movstr expand done or fail.
3059 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3060 H.J. Lu <hongjiu.lu@intel.com>
3063 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3064 to DImode for zero-extended addresses.
3066 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3068 PR rtl-optimization/57763
3069 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3070 on the new indirect jump_insn and increment LABEL_NUSES (label).
3072 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3076 * config.gcc (x86_archs): New variable.
3077 (x86_64_archs): Likewise.
3078 (x86_cpus): Likewise.
3079 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3080 --with-arch/--with-cpu= options.
3081 Support --with-arch=/--with-cpu={nehalem,westmere,
3082 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3084 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3086 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3087 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3089 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3091 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3093 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3096 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3097 clear cpp_get_options (parse_in)->warn_unused_macros for
3098 ix86_target_macros_internal with cpp_define.
3100 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3102 * jump.c (delete_related_insns): Keep (use (insn))s.
3103 * reorg.c (redundant_insn): Check for barriers too.
3105 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3107 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3109 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3111 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3112 call to $$dyncall when TARGET_LONG_CALLS is true.
3114 2014-01-17 Jeff Law <law@redhat.com>
3116 * ree.c (combine_set_extension): Temporarily disable test for
3117 changing number of hard registers.
3119 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3122 * ipa-inline-analysis.c (inline_free_summary):
3123 Do not free summary of aliases.
3125 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3128 * gimplify.c (gimplify_expr): Use create_tmp_var
3129 instead of create_tmp_var_raw. If cond doesn't have
3130 integral type, don't add the IFN_ANNOTATE builtin at all.
3132 2014-01-17 Martin Jambor <mjambor@suse.cz>
3135 * ipa-cp.c (prev_edge_clone): New variable.
3136 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3137 Also resize prev_edge_clone vector.
3138 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3139 (ipcp_edge_removal_hook): New function.
3140 (ipcp_driver): Register ipcp_edge_removal_hook.
3142 2014-01-17 Andrew Pinski <apinski@cavium.com>
3143 Steve Ellcey <sellcey@mips.com>
3146 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3149 2014-01-17 Jeff Law <law@redhat.com>
3152 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3153 so that pass_ccp runs first.
3155 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3157 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3158 (ix86_adjust_cost): Use !TARGET_XXX.
3159 (do_reorder_for_imul): Likewise.
3160 (swap_top_of_ready_list): Likewise.
3161 (ix86_sched_reorder): Likewise.
3163 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3165 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3166 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3167 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3168 (intel_memset): New. Duplicate slm_memset.
3169 (intel_cost): New. Duplicate slm_cost.
3170 (m_INTEL): New macro.
3171 (processor_target_table): Add "intel".
3172 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3173 with PROCESSOR_INTEL for "intel".
3174 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3175 PROCESSOR_SILVERMONT.
3176 (ix86_issue_rate): Likewise.
3177 (ix86_adjust_cost): Likewise.
3178 (ia32_multipass_dfa_lookahead): Likewise.
3179 (swap_top_of_ready_list): Likewise.
3180 (ix86_sched_reorder): Likewise.
3181 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3182 instead of TARGET_OPT_AGU.
3183 * config/i386/i386.h (TARGET_INTEL): New.
3184 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3185 (processor_type): Add PROCESSOR_INTEL.
3186 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3187 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3189 2014-01-17 Marek Polacek <polacek@redhat.com>
3192 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3195 2014-01-17 Richard Biener <rguenther@suse.de>
3197 PR tree-optimization/46590
3198 * opts.c (default_options_table): Add entries for
3199 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3200 all enabled at -O1 but not for -Og.
3201 * common.opt (fbranch-count-reg): Remove Init(1).
3202 (fmove-loop-invariants): Likewise.
3203 (ftree-pta): Likewise.
3205 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3207 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3208 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3209 decls to at least the GCC 4.8 used alignments.
3212 * tree-nested.c (convert_nonlocal_reference_stmt,
3213 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3214 of GIMPLE_BIND stmts, adjust associated decls.
3216 2014-01-17 Richard Biener <rguenther@suse.de>
3218 PR tree-optimization/46590
3219 * vec.h (vec<>::bseach): New member function implementing
3220 binary search according to C89 bsearch.
3221 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3222 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3223 bitmap pointer again. Make accesses_in_loop a flat array.
3224 (mem_ref_obstack): New global.
3225 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3226 (mark_ref_stored): Likewise.
3227 (ref_indep_loop_p_2): Likewise.
3228 (set_ref_stored_in_loop): New helper function.
3229 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3230 (memref_free): Adjust.
3231 (record_mem_ref_loc): Simplify.
3232 (gather_mem_refs_stmt): Adjust.
3233 (sort_locs_in_loop_postorder_cmp): New function.
3234 (analyze_memory_references): Sort accesses_in_loop after
3235 loop postorder number.
3236 (find_ref_loc_in_loop_cmp): New function.
3237 (for_all_locs_in_loop): Find relevant cluster of locs in
3238 accesses_in_loop and iterate without recursion.
3239 (execute_sm): Avoid uninit warning.
3240 (struct ref_always_accessed): Simplify.
3241 (ref_always_accessed::operator ()): Likewise.
3242 (ref_always_accessed_p): Likewise.
3243 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3244 loop postorder numbers here.
3245 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3248 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3251 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3252 on decls for which assemble_alias has been called.
3254 2014-01-17 Nick Clifton <nickc@redhat.com>
3256 * config/msp430/msp430.opt: (mcpu): New option.
3257 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3258 (msp430_option_override): Parse target_cpu. If the MCU name
3259 matches a generic string, clear target_mcu.
3260 (msp430_attr): Allow numeric interrupt values up to 63.
3261 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3262 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3264 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3266 * config/msp430/msp430.md (popm): Use %J rather than %I.
3267 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3268 (addhi_cy_i): Use immediate_operand for operand 2.
3269 * doc/invoke.texi: Document -mcpu option.
3271 2014-01-17 Richard Biener <rguenther@suse.de>
3273 PR rtl-optimization/38518
3274 * df.h (df_analyze_loop): Declare.
3275 * df-core.c: Include cfgloop.h.
3276 (df_analyze_1): Split out main part of df_analyze.
3277 (df_analyze): Adjust.
3278 (loop_inverted_post_order_compute): New function.
3279 (loop_post_order_compute): Likewise.
3280 (df_analyze_loop): New function avoiding whole-function
3282 * loop-invariant.c (find_defs): Use df_analyze_loop.
3283 (find_invariants): Adjust.
3284 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3286 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3288 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3289 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3291 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3293 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3294 traversal when removing references.
3296 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3299 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3301 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3304 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3305 pushing a reload for an autoinc when we had previously reloaded an
3306 inner part of the address.
3308 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3310 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3312 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3313 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3314 when not giving up or versioning for alias only because of
3316 (vect_analyze_data_ref_dependences): Set to true.
3317 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3319 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3320 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3324 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3327 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3328 operand 0 predicate for gathers, use a new pseudo as subtarget.
3330 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3333 * lra-constraints.c (process_alt_operands): Add printing debug info.
3334 Check absence of input/output reloads for matched operands too.
3336 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3338 PR rtl-optimization/59835
3339 * ira.c (ira_init_register_move_cost): Increase cost for
3342 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3344 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3346 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3349 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3350 non-register objects. Use gen_(high/low)part more consistently.
3353 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3356 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3357 endian support, remove tests for WORDS_BIG_ENDIAN.
3358 (p8_mfvsrd_3_<mode>): Likewise.
3359 (reload_gpr_from_vsx<mode>): Likewise.
3360 (reload_gpr_from_vsxsf): Likewise.
3361 (p8_mfvsrd_4_disf): Likewise.
3363 2014-01-16 Richard Biener <rguenther@suse.de>
3365 PR rtl-optimization/46590
3366 * lcm.c (compute_antinout_edge): Use postorder iteration.
3367 (compute_laterin): Use inverted postorder iteration.
3369 2014-01-16 Nick Clifton <nickc@redhat.com>
3372 * varasm.c (output_constant): Return the number of bytes actually
3374 (output_constructor_array_range): Update the field size with the
3375 number of bytes emitted by output_constant.
3376 (output_constructor_regular_field): Likewise. Also do not
3377 complain if the total number of bytes emitted is now greater
3378 than the expected fieldpos.
3379 * output.h (output_constant): Update prototype and descriptive comment.
3381 2014-01-16 Marek Polacek <polacek@redhat.com>
3384 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3385 it is error_mark_node.
3387 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3389 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3390 VALID_AVX256_REG_OR_OI_MODE.
3392 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3394 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3395 current procedure should be profiled.
3397 2014-01-15 Andrew Pinski <apinski@cavium.com>
3399 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3400 of moving from/to the STACK_REG register class.
3402 2014-01-15 Richard Henderson <rth@redhat.com>
3405 * reginfo.c (global_regs_decl): Globalize.
3406 * rtl.h (global_regs_decl): Declare.
3407 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3408 reserved via global_regs.
3410 2014-01-15 Teresa Johnson <tejohnson@google.com>
3412 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3414 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3416 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3417 and vmulosh rather than call gen_vec_widen_smult_*.
3418 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3419 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3420 (vec_widen_smult_even_v16qi): Likewise.
3421 (vec_widen_umult_even_v8hi): Likewise.
3422 (vec_widen_smult_even_v8hi): Likewise.
3423 (vec_widen_umult_odd_v16qi): Likewise.
3424 (vec_widen_smult_odd_v16qi): Likewise.
3425 (vec_widen_umult_odd_v8hi): Likewise.
3426 (vec_widen_smult_odd_v8hi): Likewise.
3427 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3428 vmuloub rather than call gen_vec_widen_umult_*.
3429 (vec_widen_umult_lo_v16qi): Likewise.
3430 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3431 vmulosb rather than call gen_vec_widen_smult_*.
3432 (vec_widen_smult_lo_v16qi): Likewise.
3433 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3434 rather than call gen_vec_widen_umult_*.
3435 (vec_widen_umult_lo_v8hi): Likewise.
3436 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3437 rather than call gen_vec_widen_smult_*.
3438 (vec_widen_smult_lo_v8hi): Likewise.
3440 2014-01-15 Jeff Law <law@redhat.com>
3442 PR tree-optimization/59747
3443 * ree.c (find_and_remove_re): Properly handle case where a second
3444 eliminated extension requires widening a copy created for elimination
3445 of a prior extension.
3446 (combine_set_extension): Ensure that the number of hard regs needed
3447 for a destination register does not change when we widen it.
3449 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3451 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3452 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3453 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3454 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3455 (avr-*-rtems*): Likewise.
3456 (bfin*-rtems*): Likewise.
3457 (moxie-*-rtems*): Likewise.
3458 (h8300-*-rtems*): Likewise.
3459 (i[34567]86-*-rtems*): Likewise.
3460 (lm32-*-rtems*): Likewise.
3461 (m32r-*-rtems*): Likewise.
3462 (m68k-*-rtems*): Likewise.
3463 (microblaze*-*-rtems*): Likewise.
3464 (mips*-*-rtems*): Likewise.
3465 (powerpc-*-rtems*): Likewise.
3466 (sh-*-rtems*): Likewise.
3467 (sparc-*-rtems*): Likewise.
3468 (sparc64-*-rtems*): Likewise.
3469 (v850-*-rtems*): Likewise.
3470 (m32c-*-rtems*): Likewise.
3472 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3474 PR rtl-optimization/59511
3475 * ira.c (ira_init_register_move_cost): Use memory costs for some
3476 cases of register move cost calculations.
3477 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3478 instead of BB frequency.
3479 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3480 * lra-assigns.c (find_hard_regno_for): Ditto.
3482 2014-01-15 Richard Biener <rguenther@suse.de>
3484 PR tree-optimization/59822
3485 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3486 (vectorizable_load): Use it to hoist defs of uses of invariant
3487 loads out of the loop.
3489 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3490 Kugan Vivekanandarajah <kuganv@linaro.org>
3493 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3496 2014-01-15 Richard Biener <rguenther@suse.de>
3498 PR rtl-optimization/59802
3499 * lcm.c (compute_available): Use inverted postorder to seed
3500 the initial worklist.
3502 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3505 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3506 ADDR_REGS for invalid symrefs in non-PIC code.
3508 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3511 * builtins.c (determine_block_size): Initialize *probable_max_size
3512 even if len_rtx is CONST_INT.
3514 2014-01-14 Andrew Pinski <apinski@cavium.com>
3516 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3517 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3518 (cortexa53_tunings): Likewise.
3519 (aarch64_sched_issue_rate): New function.
3520 (TARGET_SCHED_ISSUE_RATE): Define.
3522 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3524 * ira-costs.c (find_costs_and_classes): Add missed
3525 ira_init_register_move_cost_if_necessary.
3527 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3530 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3532 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3535 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3536 to indicate if type is used for function return value. Warn ABI
3537 change if the vector mode isn't available for function return value.
3538 (ix86_function_arg_advance): Pass false to type_natural_mode.
3539 (ix86_function_arg): Likewise.
3540 (ix86_gimplify_va_arg): Likewise.
3541 (function_arg_32): Don't warn ABI change.
3542 (ix86_function_value): Pass true to type_natural_mode.
3543 (ix86_return_in_memory): Likewise.
3544 (ix86_struct_value_rtx): Removed.
3545 (TARGET_STRUCT_VALUE_RTX): Likewise.
3547 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3549 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3550 converting a conditional jump into a conditional return.
3552 2014-01-14 Richard Biener <rguenther@suse.de>
3554 PR tree-optimization/58921
3555 PR tree-optimization/59006
3556 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3557 hoisting invariant stmts.
3558 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3559 invariant loads on the preheader edge if possible.
3561 2014-01-14 Joey Ye <joey.ye@arm.com>
3563 * doc/plugin.texi (Building GCC plugins): Update to C++.
3565 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3567 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3568 (_mm_rcp28_round_ss): Ditto.
3569 (_mm_rsqrt28_round_sd): Ditto.
3570 (_mm_rsqrt28_round_ss): Ditto.
3571 (_mm_rcp28_sd): Ditto.
3572 (_mm_rcp28_ss): Ditto.
3573 (_mm_rsqrt28_sd): Ditto.
3574 (_mm_rsqrt28_ss): Ditto.
3575 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3576 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3577 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3578 (IX86_BUILTIN_RCP28SD): Ditto.
3579 (IX86_BUILTIN_RCP28SS): Ditto.
3580 (IX86_BUILTIN_RSQRT28SD): Ditto.
3581 (IX86_BUILTIN_RSQRT28SS): Ditto.
3582 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3583 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3584 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3585 (ix86_expand_special_args_builtin): Expand new FTYPE.
3586 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3587 (srcp14<mode>): Make insn unary.
3588 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3589 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3590 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3591 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3592 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3593 Fix rounding: make it SAE only.
3594 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3596 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3598 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3599 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3600 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3601 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3602 (round_saeonly_mask_scalar_operand4): Ditto.
3603 (round_saeonly_mask_scalar_op3): Ditto.
3604 (round_saeonly_mask_scalar_op4): Ditto.
3606 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3608 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3609 Implement -maltivec=be for vec_insert and vec_extract.
3611 2014-01-10 DJ Delorie <dj@redhat.com>
3613 * config/msp430/msp430.md (call_internal): Don't allow memory
3614 references with SP as the base register.
3615 (call_value_internal): Likewise.
3616 * config/msp430/constraints.md (Yc): New. For memory references
3617 that don't use SP as a base register.
3619 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3620 "an integer without a # prefix"
3621 * config/msp430/msp430.md (epilogue_helper): Use it.
3623 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3626 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3627 AVX512F gather builtins.
3628 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3629 on gather decls with INTEGER_TYPE masktype.
3630 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3631 directly into the builtin rather than hoisting it before loop.
3633 PR tree-optimization/59387
3634 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3635 (scev_const_prop): If folded_casts and type has undefined overflow,
3636 use force_gimple_operand instead of force_gimple_operand_gsi and
3637 for each added stmt if it is assign with
3638 arith_code_with_undefined_signed_overflow, call
3639 rewrite_to_defined_overflow.
3640 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3641 gimple-fold.h instead.
3642 (arith_code_with_undefined_signed_overflow,
3643 rewrite_to_defined_overflow): Moved to ...
3644 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3645 rewrite_to_defined_overflow): ... here. No longer static.
3646 Include gimplify-me.h.
3647 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3648 rewrite_to_defined_overflow): New prototypes.
3650 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3652 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3654 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3656 * builtins.c (get_object_alignment_2): Minor tweak.
3657 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3659 2014-01-13 Christian Bruel <christian.bruel@st.com>
3661 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3662 optimized non constant lengths.
3664 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3667 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3668 load as __atomic_load_N if possible.
3670 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3672 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3674 (rs6000_expand_builtin): Adjust call.
3676 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3679 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3680 * config/rs6000/rs6000.c: Include target-globals.h.
3681 (rs6000_set_current_function): Instead of doing target_reinit
3682 unconditionally, use save_target_globals_default_opts and
3683 restore_target_globals.
3685 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3687 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3688 (rs6000_expand_builtin): Handle mffs and mtfsf.
3689 (rs6000_init_builtins): Define mffs and mtfsf.
3690 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3691 (rs6000_mffs): New pattern.
3692 (rs6000_mtfsf): New pattern.
3694 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3696 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3697 Start narrowing with START. Apply candidate-use pair
3698 and check overall cost in narrowing.
3699 (iv_ca_prune): Pass new argument.
3701 2014-01-10 Jeff Law <law@redhat.com>
3704 * ree.c (combine_reaching_defs): Ensure the defining statement
3705 occurs before the extension when optimizing extensions with
3706 different source and destination hard registers.
3708 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3711 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3712 vtables into the type inheritance graph.
3714 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3716 PR rtl-optimization/59754
3717 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3718 modes in the REGNO != REGNO case.
3720 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3722 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3724 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3726 PR tree-optimization/59745
3727 * tree-predcom.c (tree_predictive_commoning_loop): Call
3728 free_affine_expand_cache if giving up because components is NULL.
3730 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3731 GC in payload of target_globals struct instead of allocating them on
3732 the heap and the larger structs separately using GC.
3733 * target-globals.h (struct target_globals): Make regs, hard_regs,
3734 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3735 of GTY((skip)) and change type to void *.
3736 (reset_target_globals): Cast loads from those fields to corresponding
3739 2014-01-10 Steve Ellcey <sellcey@mips.com>
3742 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3743 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3744 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3746 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3749 * aarch64-modes.def (CC_Zmode): New flags mode.
3750 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3751 represents an equality.
3752 (aarch64_get_condition_code): Handle CC_Zmode.
3753 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3755 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3757 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3758 extraction in good case.
3760 2014-01-10 Richard Biener <rguenther@suse.de>
3762 PR tree-optimization/59374
3763 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3764 checking after SLP discovery. Mark stmts not participating
3765 in any SLP instance properly.
3767 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3769 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3770 when handling a SET rtx.
3772 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3774 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3775 (cortex-a57): Likewise.
3776 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3778 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3780 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3781 non-iwmmxt builtins.
3783 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3787 * ipa-devirt.c record_target_from_binfo): Take as argument
3788 stack of binfos and lookup matching one for virtual inheritance.
3789 (possible_polymorphic_call_targets_1): Update.
3791 2014-01-10 Huacai Chen <chenhc@lemote.com>
3793 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3794 kernel strings for Loongson-2E/2F/3A.
3796 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3799 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3800 is_gimple_call before calling gimple_call_internal_p.
3802 2014-01-09 Steve Ellcey <sellcey@mips.com>
3804 * Makefile.in (TREE_FLOW_H): Remove.
3805 (TREE_SSA_H): Add file names from tree-flow.h.
3806 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3807 * tree.h: Remove tree-flow.h reference.
3808 * hash-table.h: Remove tree-flow.h reference.
3809 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3810 reference with tree-ssa-loop.h.
3812 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3814 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3815 default element-order behavior for -maltivec.
3816 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3818 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3819 when targeting big endian, at least for now.
3820 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3822 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3825 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3826 var satisfies use_register_for_decl, just take into account type
3827 alignment, rather than decl alignment.
3829 PR tree-optimization/59622
3830 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3831 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3832 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3833 Don't devirtualize for inplace at all. For targets.length () == 1,
3834 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3836 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3838 * config/i386/i386.md (cpu): Remove the unused btver1.
3840 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3842 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3844 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3847 * tree-core.h (struct target_globals): New forward declaration.
3848 (struct tree_target_option): Add globals field.
3849 * tree.h (TREE_TARGET_GLOBALS): Define.
3850 (prepare_target_option_nodes_for_pch): New prototype.
3851 * target-globals.h (struct target_globals): Define even if
3853 * tree.c (prepare_target_option_node_for_pch,
3854 prepare_target_option_nodes_for_pch): New functions.
3855 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3856 * config/i386/i386.c: Include target-globals.h.
3857 (ix86_set_current_function): Instead of doing target_reinit
3858 unconditionally, use save_target_globals_default_opts and
3859 restore_target_globals.
3861 2014-01-09 Richard Biener <rguenther@suse.de>
3863 PR tree-optimization/59715
3864 * tree-cfg.h (split_critical_edges): Declare.
3865 * tree-cfg.c (split_critical_edges): Export.
3866 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3868 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3870 * cfgexpand.c (expand_stack_vars): Optionally disable
3871 asan stack protection.
3872 (expand_used_vars): Likewise.
3873 (partition_stack_vars): Likewise.
3874 * asan.c (asan_emit_stack_protection): Optionally disable
3875 after return stack usage.
3876 (instrument_derefs): Optionally disable memory access instrumentation.
3877 (instrument_builtin_call): Likewise.
3878 (instrument_strlen_call): Likewise.
3879 (asan_protect_global): Optionally disable global variables protection.
3880 * doc/invoke.texi: Added doc for new options.
3881 * params.def: Added new options.
3882 * params.h: Likewise.
3884 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3886 PR rtl-optimization/59724
3887 * ifcvt.c (cond_exec_process_if_block): Don't call
3888 flow_find_head_matching_sequence with 0 longest_match.
3889 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3890 non-active insns if !stop_after.
3891 (try_head_merge_bb): Revert 2014-01-07 changes.
3893 2014-01-08 Jeff Law <law@redhat.com>
3895 * ree.c (get_sub_rtx): New function, extracted from...
3896 (merge_def_and_ext): Here.
3897 (combine_reaching_defs): Use get_sub_rtx.
3899 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3901 * cgraph.h (varpool_variable_node): Do not choke on null node.
3903 2014-01-08 Catherine Moore <clm@codesourcery.com>
3905 * config/mips/mips.md (simple_return): Attempt to use JRC
3907 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3909 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3911 PR rtl-optimization/59137
3912 * reorg.c (steal_delay_list_from_target): Call update_block for
3914 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3916 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3918 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3919 two duplicate entries.
3921 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3924 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3926 * config/mips/mips.c (mips_truncated_op_cost): New function.
3927 (mips_rtx_costs): Adjust test for BADDU.
3928 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3930 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3932 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3933 (*baddu_si): ...this new pattern.
3935 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3938 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3940 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3943 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3945 (expand_expr, expand_normal): Adjust.
3946 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3947 inner_reference_p. Use inner_reference_p to expand inner references.
3948 (store_expr): Adjust.
3949 * cfgexpand.c (expand_call_stmt): Adjust.
3951 2014-01-08 Rong Xu <xur@google.com>
3953 * gcov-io.c (gcov_var): Move from gcov-io.h.
3954 (gcov_position): Ditto.
3955 (gcov_is_error): Ditto.
3956 (gcov_rewrite): Ditto.
3957 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3958 only part to libgcc/libgcov.h.
3960 2014-01-08 Marek Polacek <polacek@redhat.com>
3963 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3965 2014-01-08 Marek Polacek <polacek@redhat.com>
3968 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3970 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3972 PR rtl-optimization/59649
3973 * stor-layout.c (get_mode_bounds): For BImode return
3974 0 and STORE_FLAG_VALUE.
3976 2014-01-08 Richard Biener <rguenther@suse.de>
3979 * gimple.h (is_gimple_builtin_call): Remove.
3980 (gimple_builtin_call_types_compatible_p): New.
3981 (gimple_call_builtin_p): New overload.
3982 * gimple.c (is_gimple_builtin_call): Remove.
3983 (validate_call): Rename to ...
3984 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3986 (validate_type): New static function.
3987 (gimple_call_builtin_p): New overload and adjust.
3988 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3989 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3990 (gimple_fold_stmt_to_constant_1): Likewise.
3991 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3993 2014-01-08 Richard Biener <rguenther@suse.de>
3996 * gimplify.c (gimplify_expr): Gimplify register-register type
3997 VIEW_CONVERT_EXPRs to separate stmts.
3999 2014-01-07 Jeff Law <law@redhat.com>
4002 * ree.c (combine_set_extension): Handle case where source
4003 and destination registers in an extension insn are different.
4004 (combine_reaching_defs): Allow source and destination registers
4005 in extension to be different under limited circumstances.
4006 (add_removable_extension): Remove restriction that the
4007 source and destination registers in the extension are the same.
4008 (find_and_remove_re): Emit a copy from the extension's
4009 destination to its source after the defining insn if
4010 the source and destination registers are different.
4013 * ifcvt.c (merge_if_block): If we are merging a block with more than
4014 one successor with a block with no successors, remove any BARRIER
4015 after the second block.
4017 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4019 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4021 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4024 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4025 for 14-bit register offsets when INT14_OK_STRICT is false.
4027 2014-01-07 Roland Stigge <stigge@antcom.de>
4028 Michael Meissner <meissner@linux.vnet.ibm.com>
4031 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4032 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4034 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4036 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4039 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4041 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4042 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4043 rtx is const0_rtx or not.
4045 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4048 * target-globals.c (save_target_globals): Remove this_fn_optab
4050 * toplev.c: Include optabs.h.
4051 (target_reinit): Temporarily restore the global options if another
4052 set of options are in force.
4054 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4056 PR rtl-optimization/58668
4057 * cfgcleanup.c (flow_find_cross_jump): Don't count
4058 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4059 to determine what is counted.
4060 (flow_find_head_matching_sequence): Use active_insn_p to determine
4062 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4064 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4065 determine what is counted.
4067 PR tree-optimization/59643
4068 * tree-predcom.c (split_data_refs_to_components): If one dr is
4069 read and one write, determine_offset fails and the write isn't
4070 in the bad component, just put the read into the bad component.
4072 2014-01-07 Mike Stump <mikestump@comcast.net>
4073 Jakub Jelinek <jakub@redhat.com>
4076 * tree-core.h (struct tree_optimization_option): Change optabs
4077 type from unsigned char * to void *.
4078 * optabs.c (init_tree_optimization_optabs): Adjust
4079 TREE_OPTIMIZATION_OPTABS initialization.
4081 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4084 * config/i386/i386.h (struct machine_function): Add
4085 no_drap_save_restore field.
4086 * config/i386/i386.c (ix86_save_reg): Use
4087 !cfun->machine->no_drap_save_restore instead of
4088 crtl->stack_realign_needed.
4089 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4090 this function clears frame_pointer_needed. Set
4091 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4092 and DRAP reg is needed.
4094 2014-01-06 Marek Polacek <polacek@redhat.com>
4097 * doc/implement-c.texi: Mention that other integer types are
4098 permitted as bit-field types in strictly conforming mode.
4100 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4102 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4105 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4107 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4109 2014-01-06 Martin Jambor <mjambor@suse.cz>
4112 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4114 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4116 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4120 * var-tracking.c (add_stores): Preserve the value of the source even if
4121 we don't record the store.
4123 2014-01-06 Terry Guo <terry.guo@arm.com>
4125 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4127 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4130 * config/darwin.c (darwin_function_section): Adjust return values to
4131 correspond to optimisation changes made in r206070.
4133 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4135 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4136 from prefetch_block tune setting.
4137 (nocona_cost): Correct size of prefetch block to 64.
4139 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4141 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4142 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4143 used to save the static chain register in the computation of the offset
4144 from which the FP registers need to be restored.
4146 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4148 PR tree-optimization/59519
4149 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4150 ICE if get_current_def (current_new_name) is already non-NULL, as long
4151 as it is a phi result of some other phi in *new_exit_bb that has
4154 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4155 or vmovdqu* for misaligned_operand.
4156 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4157 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4158 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4159 aligned_mem for AVX512F masked aligned load and store builtins and for
4162 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4164 PR tree-optimization/59651
4165 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4166 Address range for negative step should be added by TYPE_SIZE_UNIT.
4168 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4170 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4171 overlapping registers also for registers other than the stack pointer.
4173 2014-01-03 Marek Polacek <polacek@redhat.com>
4176 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4179 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4182 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4185 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4186 (push splitter): Use <P:MODE_SIZE> instead of
4187 GET_MODE_SIZE (<P:MODE>mode).
4188 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4189 (mov -1, reg peephole2): Likewise.
4190 * config/i386/sse.md (*mov<mode>_internal,
4191 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4192 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4193 *<code><mode>3, *andnot<mode>3<mask_name>,
4194 <mask_codefor><code><mode>3<mask_name>): Likewise.
4195 * config/i386/subst.md (mask_mode512bit_condition,
4196 sd_mask_mode512bit_condition): Likewise.
4198 2014-01-02 Xinliang David Li <davidxl@google.com>
4200 PR tree-optimization/59303
4201 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4202 (dump_predicates): Better output format.
4203 (pred_equal_p): New function.
4204 (is_neq_relop_p): Ditto.
4205 (is_neq_zero_form_p): Ditto.
4206 (pred_expr_equal_p): Ditto.
4207 (pred_neg_p): Ditto.
4208 (simplify_pred): Ditto.
4209 (simplify_preds_2): Ditto.
4210 (simplify_preds_3): Ditto.
4211 (simplify_preds_4): Ditto.
4212 (simplify_preds): Ditto.
4214 (push_to_worklist): Ditto.
4215 (get_pred_info_from_cmp): Ditto.
4216 (is_degenerated_phi): Ditto.
4217 (normalize_one_pred_1): Ditto.
4218 (normalize_one_pred): Ditto.
4219 (normalize_one_pred_chain): Ditto.
4220 (normalize_preds): Ditto.
4221 (normalize_cond_1): Remove function.
4222 (normalize_cond): Ditto.
4223 (is_gcond_subset_of): Ditto.
4224 (is_subset_of_any): Ditto.
4225 (is_or_set_subset_of): Ditto.
4226 (is_and_set_subset_of): Ditto.
4227 (is_norm_cond_subset_of): Ditto.
4228 (pred_chain_length_cmp): Ditto.
4229 (convert_control_dep_chain_into_preds): Type change.
4230 (find_predicates): Ditto.
4231 (find_def_preds): Ditto.
4232 (destroy_predicates_vecs): Ditto.
4233 (find_matching_predicates_in_rest_chains): Ditto.
4234 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4235 (is_pred_expr_subset): Ditto.
4236 (is_pred_chain_subset_of): Ditto.
4237 (is_included_in): Ditto.
4238 (is_superset_of): Ditto.
4240 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4242 Update copyright years.
4244 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4246 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4247 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4248 config/arc/arc.md, config/arc/arc.opt,
4249 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4250 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4251 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4252 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4253 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4254 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4255 vtable-verify.c, vtable-verify.h: Use the standard form for the
4258 2014-01-02 Tobias Burnus <burnus@net-b.de>
4260 * gcc.c (process_command): Update copyright notice dates.
4261 * gcov-dump.c: Ditto.
4263 * doc/cpp.texi: Bump @copying's copyright year.
4264 * doc/cppinternals.texi: Ditto.
4265 * doc/gcc.texi: Ditto.
4266 * doc/gccint.texi: Ditto.
4267 * doc/gcov.texi: Ditto.
4268 * doc/install.texi: Ditto.
4269 * doc/invoke.texi: Ditto.
4271 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4273 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4275 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4277 * config/i386/sse.md (*mov<mode>_internal): Guard
4278 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4280 PR rtl-optimization/59647
4281 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4282 new_rtx into UNSIGNED_FLOAT rtxes.
4284 Copyright (C) 2014 Free Software Foundation, Inc.
4286 Copying and distribution of this file, with or without modification,
4287 are permitted in any medium without royalty provided the copyright
4288 notice and this notice are preserved.