| blob | b6f1dac00ad6bdb7445c917b56ab337cf67f019e |
1 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
3 PR ipa/60306
5 Revert:
6 2013-12-14 Jan Hubicka <jh@suse.cz>
7 PR middle-end/58477
8 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
10 2014-03-02 Jon Beniston <jon@beniston.com>
12 PR bootstrap/48230
13 PR bootstrap/50927
14 PR bootstrap/52466
15 PR target/46898
16 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
17 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
18 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
19 (simple_return, *simple_return): New patterns
20 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
21 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
23 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
25 * dwarf2out.c (gen_subprogram_die): Tidy.
27 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
29 PR target/60071
30 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
31 (*mov_t_msb_neg_negc): ... this new insn.
33 2014-02-28 Jason Merrill <jason@redhat.com>
35 PR c++/58678
36 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
37 function.
39 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
41 PR c++/60314
42 * dwarf2out.c (decltype_auto_die): New static.
43 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
44 (gen_type_die_with_usage): Handle 'decltype(auto)'.
45 (is_cxx_auto): Likewise.
47 2014-02-28 Ian Bolton <ian.bolton@arm.com>
49 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
50 we are not using general regs only.
52 2014-02-28 Richard Biener <rguenther@suse.de>
54 PR target/60280
55 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
56 previous fix and only allow to remove trivial pre-headers
57 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
58 (remove_forwarder_block): Properly update the latch of a loop.
60 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
62 PR debug/59992
63 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
64 (cselib_preserved_hash_table): New.
65 (preserve_constants_and_equivs): Move preserved vals to it.
66 (cselib_find_slot): Look it up first.
67 (cselib_init): Initialize it.
68 (cselib_finish): Release it.
69 (dump_cselib_table): Dump it.
71 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
73 PR debug/59992
74 * cselib.c (remove_useless_values): Skip to avoid quadratic
75 behavior if the condition moved from...
76 (cselib_process_insn): ... here holds.
78 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
80 PR debug/57232
81 * var-tracking.c (vt_initialize): Apply the same condition to
82 preserve the CFA base value.
84 2014-02-28 Joey Ye <joey.ye@arm.com>
86 PR target/PR60169
87 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
88 if reload in progress or completed.
90 2014-02-28 Tobias Burnus <burnus@net-b.de>
92 PR middle-end/60147
93 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
94 NAMELIST_DECL.
96 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
98 * doc/tm.texi.in (Condition Code Status): Update documention for
99 relative locations of cc0-setter and cc0-user.
101 2014-02-27 Jeff Law <law@redhat.com>
103 PR rtl-optimization/52714
104 * combine.c (try_combine): When splitting an unrecognized PARALLEL
105 into two independent simple sets, if I3 is a jump, ensure the
106 pattern we place into I3 is a (set (pc) ...).
108 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
109 Jeff Law <law@redhat.com>
111 PR rtl-optimization/49847
112 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
113 are in different blocks.
114 * doc/tm.texi (Condition Code Status): Update documention for
115 relative locations of cc0-setter and cc0-user.
117 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
119 PR target/59222
120 * lra.c (lra_emit_add): Check SUBREG too.
122 2014-02-27 Andreas Schwab <schwab@suse.de>
124 * config/m68k/m68k.c (m68k_option_override): Disable
125 -flive-range-shrinkage for classic m68k.
126 (m68k_override_options_after_change): Likewise.
128 2014-02-27 Marek Polacek <polacek@redhat.com>
130 PR middle-end/59223
131 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
132 -Wmaybe-uninitialized.
134 2014-02-27 Alan Modra <amodra@gmail.com>
136 PR target/57936
137 * reload1.c (emit_input_reload_insns): When reload_override_in,
138 set old to rl->in_reg when rl->in_reg is a subreg.
140 2014-02-26 Richard Biener <rguenther@suse.de>
142 PR bootstrap/60343
143 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
145 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
147 * common/config/i386/predicates.md (const1256_operand): Remove.
148 (const2356_operand): New.
149 (const_1_to_2_operand): Remove.
150 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
151 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
152 (*avx512pf_gatherpf<mode>sf): Ditto.
153 (avx512pf_gatherpf<mode>df): Ditto.
154 (*avx512pf_gatherpf<mode>df_mask): Ditto.
155 (*avx512pf_gatherpf<mode>df): Ditto.
156 (avx512pf_scatterpf<mode>sf): Ditto.
157 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
158 (*avx512pf_scatterpf<mode>sf): Ditto.
159 (avx512pf_scatterpf<mode>df): Ditto.
160 (*avx512pf_scatterpf<mode>df_mask): Ditto.
161 (*avx512pf_scatterpf<mode>df): Ditto.
162 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
164 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
166 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
167 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
168 (_mm512_mask_testn_epi64_mask): Move to ...
169 * config/i386/avx512cdintrin.h: Here.
170 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
171 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
172 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
173 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
174 TARGET_AVX512F from TARGET_AVX512CD.
176 2014-02-26 Richard Biener <rguenther@suse.de>
178 PR ipa/60327
179 * ipa.c (walk_polymorphic_call_targets): Properly guard
180 call to inline_update_overall_summary.
182 2014-02-26 Bin Cheng <bin.cheng@arm.com>
184 PR target/60280
185 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
186 and latches only if requested. Fix latch if it is removed.
187 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
188 LOOPS_HAVE_PREHEADERS.
190 2014-02-25 Andrew Pinski <apinski@cavium.com>
192 * builtins.c (expand_builtin_thread_pointer): Create a new target
193 when the target is NULL.
195 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
197 PR rtl-optimization/60317
198 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
199 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
200 * lra-assigns.c: Include params.h.
201 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
202 other reload pseudos considerations.
204 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
206 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
207 to use canonical form for nor<mode>3.
209 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
211 PR target/55426
212 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
213 conversions.
215 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
217 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
218 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
219 (ix86_handle_option): Handle OPT_mprefetchwt1.
220 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
221 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
222 PREFETCHWT1 CPUID.
223 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
224 OPTION_MASK_ISA_PREFETCHWT1.
225 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
226 (PTA_PREFETCHWT1): New.
227 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
228 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
229 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
230 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
231 (*prefetch_avx512pf_<mode>_: Change into ...
232 (*prefetch_prefetchwt1_<mode>: This.
233 * config/i386/i386.opt (mprefetchwt1): New.
234 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
235 (_mm_prefetch): Handle intent to write.
236 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
238 2014-02-25 Richard Biener <rguenther@suse.de>
240 PR middle-end/60291
241 * emit-rtl.c (mem_attrs_htab): Remove.
242 (mem_attrs_htab_hash): Likewise.
243 (mem_attrs_htab_eq): Likewise.
244 (set_mem_attrs): Always allocate new mem-attrs when something changed.
245 (init_emit_once): Do not allocate mem_attrs_htab.
247 2014-02-25 Richard Biener <rguenther@suse.de>
249 PR lto/60319
250 * lto-opts.c (lto_write_options): Output non-explicit conservative
251 -fwrapv, -fno-trapv and -fno-strict-overflow.
252 * lto-wrapper.c (merge_and_complain): Handle merging those options.
253 (run_gcc): And pass them through.
255 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
257 * sel-sched.c (calculate_new_fences): New parameter ptime.
258 Calculate it as a maximum over all fence cycles.
259 (sel_sched_region_2): Adjust the call to calculate_new_fences.
260 Print the final schedule timing when sched_verbose.
262 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
264 PR rtl-optimization/60292
265 * sel-sched.c (fill_vec_av_set): Do not reset target availability
266 bit fot the fence instruction.
268 2014-02-24 Alangi Derick <alangiderick@gmail.com>
270 * calls.h: Fix typo in comment.
272 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
274 * config/pa/pa.c (pa_output_move_double): Don't valididate when
275 adjusting offsetable addresses.
277 2014-02-24 Guozhi Wei <carrot@google.com>
279 * sparseset.h (sparseset_pop): Fix the wrong index.
281 2014-02-24 Walter Lee <walt@tilera.com>
283 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
284 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
285 triplet.
286 * common/config/tilegx/tilegx-common.c
287 (TARGET_DEFAULT_TARGET_FLAGS): Define.
288 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
289 (LINK_SPEC): Ditto.
290 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
291 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
292 (tilegx_gimplify_va_arg_expr): Handle big endian.
293 (tilegx_expand_unaligned_load): Ditto.
294 (tilegx_expand_unaligned_store): Ditto.
295 (TARGET_RETURN_IN_MSB): New.
296 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
297 (TARGET_ENDIAN_DEFAULT): New.
298 (TARGET_BIG_ENDIAN): Handle big endian.
299 (BYTES_BIG_ENDIAN): Ditto.
300 (WORDS_BIG_ENDIAN): Ditto.
301 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
302 (ENDIAN_SPEC): New.
303 (EXTRA_SPECS): New.
304 * config/tilegx/tilegx.md (extv): Handle big endian.
305 (extzv): Ditto.
306 (insn_st<n>): Ditto.
307 (insn_st<n>_add<bitsuffix>): Ditto.
308 (insn_stnt<n>): Ditto.
309 (insn_stnt<n>_add<bitsuffix>):Ditto.
310 (vec_interleave_highv8qi): Handle big endian.
311 (vec_interleave_highv8qi_be): New.
312 (vec_interleave_highv8qi_le): New.
313 (insn_v1int_h): Handle big endian.
314 (vec_interleave_lowv8qi): Handle big endian.
315 (vec_interleave_lowv8qi_be): New.
316 (vec_interleave_lowv8qi_le): New.
317 (insn_v1int_l): Handle big endian.
318 (vec_interleave_highv4hi): Handle big endian.
319 (vec_interleave_highv4hi_be): New.
320 (vec_interleave_highv4hi_le): New.
321 (insn_v2int_h): Handle big endian.
322 (vec_interleave_lowv4hi): Handle big endian.
323 (vec_interleave_lowv4hi_be): New.
324 (vec_interleave_lowv4hi_le): New.
325 (insn_v2int_l): Handle big endian.
326 (vec_interleave_highv2si): Handle big endian.
327 (vec_interleave_highv2si_be): New.
328 (vec_interleave_highv2si_le): New.
329 (insn_v4int_h): Handle big endian.
330 (vec_interleave_lowv2si): Handle big endian.
331 (vec_interleave_lowv2si_be): New.
332 (vec_interleave_lowv2si_le): New.
333 (insn_v4int_l): Handle big endian.
334 * config/tilegx/tilegx.opt (mbig-endian): New option.
335 (mlittle-endian): New option.
336 * doc/install.texi: Document tilegxbe-linux.
337 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
339 2014-02-24 Martin Jambor <mjambor@suse.cz>
341 PR ipa/60266
342 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
343 there are no parameter descriptors.
345 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
347 PR rtl-optimization/60268
348 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
349 initialization to ...
350 (sched_rgn_init): ... here.
351 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
353 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
355 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
356 names.
358 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
360 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
361 definition.
363 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
365 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
366 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
368 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
370 * config/microblaze/predicates.md: Add cmp_op predicate.
371 * config/microblaze/microblaze.md: Add branch_compare instruction
372 which uses cmp_op predicate and emits cmp insn before branch.
373 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
374 to microblaze_expand_conditional_branch and consolidate logic.
375 (microblaze_expand_conditional_branch): emit branch_compare
376 insn instead of handling cmp op separate from branch insn.
378 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
380 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
381 to permit subregs.
383 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
385 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
386 define_insn with define_expand and new define_insn
387 *altivec_lve<VI_char>x_internal.
388 (altivec_stve<VI_char>x): Replace define_insn with define_expand
389 and new define_insn *altivec_stve<VI_char>x_internal.
390 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
391 prototype.
392 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
393 lve*x built-ins.
394 (altivec_expand_stvex_be): New function.
396 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
398 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
399 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
400 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
401 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
403 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
405 PR target/60298
406 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
407 instead of emit_move_insn.
409 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
411 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
412 vspltw with vsldoi.
413 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
414 gen_altivec_vsumsws.
416 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
418 * config/rs6000/altivec.md (altivec_lvxl): Rename as
419 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
420 (altivec_lvxl_<mode>): New define_expand incorporating
421 -maltivec=be semantics where needed.
422 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
423 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
424 semantics where needed.
425 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
426 (altivec_stvx_<mode>): New define_expand incorporating
427 -maltivec=be semantics where needed.
428 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
429 VM2 iterator instead of V4SI.
430 (altivec_stvxl_<mode>): New define_expand incorporating
431 -maltivec=be semantics where needed.
432 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
433 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
434 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
435 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
436 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
437 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
438 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
439 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
440 ALTIVEC_BUILTIN_STVXL.
441 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
442 (altivec_expand_stvx_be): Likewise.
443 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
444 (altivec_expand_lvx_be): Likewise.
445 (altivec_expand_stvx_be): Likewise.
446 (altivec_expand_builtin): Add cases for
447 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
448 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
449 (altivec_init_builtins): Add definitions for
450 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
451 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
453 2014-02-21 Catherine Moore <clm@codesourcery.com>
455 * doc/invoke.texi (mvirt, mno-virt): Document.
456 * config/mips/mips.opt (mvirt): New option.
457 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
459 2014-02-21 Richard Biener <rguenther@suse.de>
461 PR tree-optimization/60276
462 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
463 (STMT_VINFO_MIN_NEG_DIST): New macro.
464 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
465 STMT_VINFO_MIN_NEG_DIST.
466 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
467 made for negative dependence distances still hold.
469 2014-02-21 Richard Biener <rguenther@suse.de>
471 PR middle-end/60291
472 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
473 DECL_INITIAL for globals not in the current function context.
475 2014-02-21 Jakub Jelinek <jakub@redhat.com>
477 PR tree-optimization/56490
478 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
479 * tree-ssa-uninit.c: Include params.h.
480 (compute_control_dep_chain): Add num_calls argument, return false
481 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
482 num_calls to recursive call.
483 (find_predicates): Change dep_chain into normal array,
484 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
485 variable and adjust compute_control_dep_chain caller.
486 (find_def_preds): Likewise.
488 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
490 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
491 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
493 2014-02-21 Nick Clifton <nickc@redhat.com>
495 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
496 (pushhi1): Likewise.
497 (popqi1): Add mode to pre_dec.
498 (pophi1): Likewise.
500 2014-02-21 Jakub Jelinek <jakub@redhat.com>
502 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
503 mode for mask of V8SFmode permutation.
505 2014-02-20 Richard Henderson <rth@redhat.com>
507 PR c++/60272
508 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
509 a new pseudo for OLDVAL.
511 2014-02-20 Jakub Jelinek <jakub@redhat.com>
513 PR target/57896
514 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
515 gen_reg_rtx if d->testing_p.
516 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
517 if d->testing_p and we will certainly return true.
518 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
519 if d->testing_p.
521 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
523 * emit-rtl.c (gen_reg_rtx): Assert that
524 crtl->emit.regno_pointer_align_length is non-zero.
526 2014-02-20 Richard Henderson <rth@redhat.com>
528 PR c++/60272
529 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
530 on failure the store back into EXPECT.
532 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
533 Sandra Loosemore <sandra@codesourcery.com>
535 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
536 * config/nios2/nios2.c (nios2_function_profiler): Add
537 -fPIC (flag_pic == 2) support.
538 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
539 (nios2_large_offset_p): New function.
540 (nios2_unspec_reloc_p): Move up position, update to use
541 nios2_large_offset_p.
542 (nios2_unspec_address): Remove function.
543 (nios2_unspec_offset): New function.
544 (nios2_large_got_address): New function.
545 (nios2_got_address): Add large offset support.
546 (nios2_legitimize_tls_address): Update usage of removed and new
547 functions.
548 (nios2_symbol_binds_local_p): New function.
549 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
550 (nios2_legitimize_address): Update to use nios2_large_offset_p.
551 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
552 (nios2_print_operand): Merge H/L processing, add hiadj/lo
553 processing for (const (unspec ...)).
554 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
556 2014-02-20 Richard Biener <rguenther@suse.de>
558 * tree-cfg.c (replace_uses_by): Mark altered BBs before
559 doing the substitution.
560 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
562 2014-02-20 Martin Jambor <mjambor@suse.cz>
564 PR ipa/55260
565 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
566 info when checking whether lattices are bottom.
568 2014-02-20 Richard Biener <rguenther@suse.de>
570 PR middle-end/60221
571 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
572 regions at -O0.
574 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
576 PR ipa/58555
577 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
578 parameter specifying the scaling.
579 (inline_call): Update.
580 (want_inline_recursively): Guard division by zero.
581 (recursive_inlining): Update.
582 * ipa-inline.h (clone_inlined_nodes): Update.
584 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
586 PR target/60204
587 * config/i386/i386.c (classify_argument): Pass structures of size
588 64 bytes or less in register.
590 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
591 Kirill Yukhin <kirill.yukhin@intel.com>
593 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
594 (_mm_rcp28_round_ss): Ditto.
595 (_mm_rsqrt28_round_sd): Ditto.
596 (_mm_rsqrt28_round_ss): Ditto.
597 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
598 (_mm_rcp14_round_ss): Ditto.
599 (_mm_rsqrt14_round_sd): Ditto.
600 (_mm_rsqrt14_round_ss): Ditto.
601 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
602 the first input operand, get rid of match_dup.
603 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
604 attribute to sse.
605 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
606 Ditto.
607 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
608 operand as the first input operand, set type attribute.
609 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
610 Set type attribute.
611 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
612 operand as the first input operand, set type attribute.
614 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
616 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
617 bit of zero.
619 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
621 PR target/60207
622 * config/i386/i386.c (construct_container): Remove TFmode check
623 for X86_64_INTEGER_CLASS.
625 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
627 PR target/59794
628 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
629 only when -Wpsabi is enabled.
631 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
633 PR target/59799
634 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
635 passing arrays in registers are the same as for structs, so remove the
636 special case for them.
638 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
640 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
641 destination type, extract only the valid bits if the source type is not
642 integral and has a different mode.
644 2014-02-19 Richard Biener <rguenther@suse.de>
646 PR ipa/60243
647 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
648 for all calls.
650 2014-02-19 Richard Biener <rguenther@suse.de>
652 PR ipa/60243
653 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
654 (ipa_modify_call_arguments): Emit an argument load explicitely and
655 preserve virtual SSA form there and for the replacement call.
656 Do not update SSA form nor free dominance info.
658 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
660 * ipa.c (function_and_variable_visibility): Also clear WEAK
661 flag when disolving COMDAT_GROUP.
663 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
665 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
666 * ipa-prop.c (ipa_set_jf_known_type): Return early when
667 not devirtualizing.
668 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
669 do more sanity checks.
670 (detect_type_change): Return true when giving up early.
671 (compute_complex_assign_jump_func): Fix type parameter of
672 ipa_set_ancestor_jf.
673 (compute_complex_ancestor_jump_func): Likewise.
674 (update_jump_functions_after_inlining): Fix updating of
675 ancestor function.
676 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
678 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
680 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
681 inline clones when edge disappears.
683 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
685 PR target/60203
686 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
687 Split 64-bit moves into 2 patterns. Do not allow the use of
688 direct move for TDmode in little endian, since the decimal value
689 has little endian bytes within a word, but the 64-bit pieces are
690 ordered in a big endian fashion, and normal subreg's of TDmode are
691 not allowed.
692 (mov<mode>_64bit_dm): Likewise.
693 (movtd_64bit_nodm): Likewise.
695 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
697 PR tree-optimization/60174
698 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
699 statement of an SSA_NAME that occurs in an abnormal PHI node.
701 2014-02-18 Jakub Jelinek <jakub@redhat.com>
703 PR sanitizer/60142
704 * final.c (SEEN_BB): Remove.
705 (SEEN_NOTE, SEEN_EMITTED): Renumber.
706 (final_scan_insn): Don't force_source_line on second
707 NOTE_INSN_BASIC_BLOCK.
709 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
711 PR target/60205
712 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
713 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
714 (type_natural_mode): Warn ABI change when %zmm register is not
715 available for AVX512F vector value passing.
717 2014-02-18 Kai Tietz <ktietz@redhat.com>
719 PR target/60193
720 * config/i386/i386.c (ix86_expand_prologue): Use value in
721 rax register as displacement when restoring %r10 or %rax.
722 Fix wrong offset when restoring both registers.
724 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
726 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
727 assertion with conditional return.
729 2014-02-18 Jakub Jelinek <jakub@redhat.com>
730 Uros Bizjak <ubizjak@gmail.com>
732 PR driver/60233
733 * config/i386/driver-i386.c (host_detect_local_cpu): If
734 YMM state is not saved by the OS, also clear has_f16c. Move
735 CPUID 0x80000001 handling before YMM state saving checking.
737 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
739 PR rtl-optimization/58960
740 * haifa-sched.c (alloc_global_sched_pressure_data): New,
741 factored out from ...
742 (sched_init): ... here.
743 (free_global_sched_pressure_data): New, factored out from ...
744 (sched_finish): ... here.
745 * sched-int.h (free_global_sched_pressure_data): Declare.
746 * sched-rgn.c (nr_regions_initial): New static global.
747 (haifa_find_rgns): Initialize it.
748 (schedule_region): Disable sched-pressure for the newly
749 generated regions.
751 2014-02-17 Richard Biener <rguenther@suse.de>
753 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
754 release SSA defs of pattern stmts.
756 2014-02-17 Richard Biener <rguenther@suse.de>
758 * tree-inline.c (expand_call_inline): Release the virtual
759 operand defined by the call we are about to inline.
761 2014-02-17 Richard Biener <rguenther@suse.de>
763 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
765 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
766 Ilya Tocar <ilya.tocar@intel.com>
768 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
769 arguments order in builtin.
770 (_mm512_permutexvar_epi64): Ditto.
771 (_mm512_mask_permutexvar_epi64): Ditto
772 (_mm512_maskz_permutexvar_epi32): Ditto
773 (_mm512_permutexvar_epi32): Ditto
774 (_mm512_mask_permutexvar_epi32): Ditto
776 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
778 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
779 (p8_vmrgow): Likewise.
781 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
783 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
784 endian targets.
786 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
788 PR target/60203
789 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
790 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
791 into 64-bit and 32-bit moves. On 64-bit moves, add support for
792 using direct move instructions on ISA 2.07. Also adjust
793 instruction length for 64-bit.
794 (mov<mode>_64bit, TFmode/TDmode): Likewise.
795 (mov<mode>_32bit, TFmode/TDmode): Likewise.
797 2014-02-15 Alan Modra <amodra@gmail.com>
799 PR target/58675
800 PR target/57935
801 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
802 find_replacement on parts of insn rtl that might be reloaded.
804 2014-02-15 Richard Biener <rguenther@suse.de>
806 PR tree-optimization/60183
807 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
808 (tree_ssa_phiprop): Calculate and free post-dominators.
810 2014-02-14 Jeff Law <law@redhat.com>
812 PR rtl-optimization/60131
813 * ree.c (get_extended_src_reg): New function.
814 (combine_reaching_defs): Use it rather than assuming location of REG.
815 (find_and_remove_re): Verify first operand of extension is
816 a REG before adding the insns to the copy list.
818 2014-02-14 Roland McGrath <mcgrathr@google.com>
820 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
821 * configure: Regenerated.
822 * config.in: Regenerated.
823 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
824 instead of ASM_SHORT.
826 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
827 Richard Earnshaw <rearnsha@arm.com>
829 PR rtl-optimization/59535
830 * lra-constraints.c (process_alt_operands): Encourage alternative
831 when unassigned pseudo class is superset of the alternative class.
832 (inherit_reload_reg): Don't inherit when optimizing for code size.
833 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
834 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
835 modes not less than 4 for Thumb1.
837 2014-02-14 Kyle McMartin <kyle@redhat.com>
839 PR pch/60010
840 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
842 2014-02-14 Richard Biener <rguenther@suse.de>
844 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
845 (get_frame_arg): Drop the assert with langhook types_compatible_p.
846 Do not strip INDIRECT_REFs.
848 2014-02-14 Richard Biener <rguenther@suse.de>
850 PR lto/60179
851 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
852 DECL_FUNCTION_SPECIFIC_TARGET.
853 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
854 * tree-streamer-out.c (pack_ts_target_option): Remove.
855 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
856 (write_ts_function_decl_tree_pointers): Do not stream
857 DECL_FUNCTION_SPECIFIC_TARGET.
858 * tree-streamer-in.c (unpack_ts_target_option): Remove.
859 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
860 (lto_input_ts_function_decl_tree_pointers): Do not stream
861 DECL_FUNCTION_SPECIFIC_TARGET.
863 2014-02-14 Jakub Jelinek <jakub@redhat.com>
865 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
866 (get_initial_def_for_induction, vectorizable_induction): Ignore
867 debug stmts when looking for exit_phi.
868 (vectorizable_live_operation): Fix up condition.
870 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
872 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
873 nreverse() because it changes the content of original tree list.
875 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
877 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
878 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
880 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
882 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
883 GNU coding standards.
885 2014-02-13 Jakub Jelinek <jakub@redhat.com>
887 PR debug/60152
888 * dwarf2out.c (gen_subprogram_die): Don't call
889 add_calling_convention_attribute if subr_die is old_die.
891 2014-02-13 Sharad Singhai <singhai@google.com>
893 * doc/optinfo.texi: Fix order of nodes.
895 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
897 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
898 operands[2], not operands[3].
900 2014-02-13 Richard Biener <rguenther@suse.de>
902 PR bootstrap/59878
903 * doc/install.texi (ISL): Update recommended version to 0.12.2,
904 mention the possibility of an in-tree build.
905 (CLooG): Update recommended version to 0.18.1, mention the
906 possibility of an in-tree build and clarify that the ISL
907 bundled with CLooG does not work.
909 2014-02-13 Jakub Jelinek <jakub@redhat.com>
911 PR target/43546
912 * expr.c (compress_float_constant): If x is a hard register,
913 extend into a pseudo and then move to x.
915 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
917 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
918 caused by bad second argument to warning_at() with -mhotpatch and
919 nested functions (e.g. with gfortran).
921 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
923 * opts.c (option_name): Remove "enabled by default" rider.
925 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
927 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
929 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
930 Uros Bizjak <ubizjak@gmail.com>
932 PR target/60151
933 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
934 * configure: Regenerated.
936 2014-02-12 Richard Biener <rguenther@suse.de>
938 * vec.c (vec_prefix::calculate_allocation): Move as
939 inline variant to vec.h.
940 (vec_prefix::calculate_allocation_1): New out-of-line version.
941 * vec.h (vec_prefix::calculate_allocation_1): Declare.
942 (vec_prefix::m_has_auto_buf): Rename to ...
943 (vec_prefix::m_using_auto_storage): ... this.
944 (vec_prefix::calculate_allocation): Inline the easy cases
945 and dispatch to calculate_allocation_1 which doesn't need the
946 prefix address.
947 (va_heap::reserve): Use gcc_checking_assert.
948 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
949 m_using_auto_storage.
950 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
951 member and adjust.
952 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
953 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
954 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
956 2014-02-12 Richard Biener <rguenther@suse.de>
958 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
959 when we found a dependence.
961 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
963 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
964 common code...
965 (maybe_fold_stmt): ... into this new function.
966 * omp-low.c (lower_omp): Update comment.
968 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
969 last use.
971 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
972 dereference.
974 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
976 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
977 identifiers in comments.
978 (cortexa53_extra_costs): Likewise.
979 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
980 (cortexa7_extra_costs): Likewise.
981 (cortexa12_extra_costs): Likewise.
982 (cortexa15_extra_costs): Likewise.
983 (v7m_extra_costs): Likewise.
985 2014-02-12 Richard Biener <rguenther@suse.de>
987 PR middle-end/60092
988 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
989 of posix_memalign being successful.
990 (lower_stmt): Restrict lowering of posix_memalign to when
991 -ftree-bit-ccp is enabled.
993 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
995 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
996 arg_loc.
997 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
999 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1001 PR rtl-optimization/60116
1002 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1003 other_insn once the combination has been validated.
1005 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1007 PR lto/59468
1008 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1009 and wrapper.
1010 * ipa-devirt.c: Include demangle.h
1011 (odr_violation_reported): New static variable.
1012 (add_type_duplicate): Update odr_violations.
1013 (maybe_record_node): Add completep parameter; update it.
1014 (record_target_from_binfo): Add COMPLETEP parameter;
1015 update it as needed.
1016 (possible_polymorphic_call_targets_1): Likewise.
1017 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1018 rename FINAL to COMPLETE.
1019 (record_targets_from_bases): Sanity check we found the binfo;
1020 fix COMPLETEP updating.
1021 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1022 parameter, fix computing of COMPLETEP.
1023 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1024 at LTO time do demangling.
1025 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1026 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1027 parameter.
1028 (gimple_get_virt_method_for_binfo): Likewise.
1029 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1030 gimple_get_virt_method_for_vtable): Update prototypes.
1032 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1034 PR target/49008
1035 * genautomata.c (add_presence_absence): Fix typo with
1036 {final_}presence_list.
1038 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1040 PR target/60137
1041 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1042 for VSX/Altivec vectors that land in GPR registers.
1044 2014-02-11 Richard Henderson <rth@redhat.com>
1045 Jakub Jelinek <jakub@redhat.com>
1047 PR debug/59776
1048 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1049 around drhs if type conversion to lacc->type is not useless.
1051 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1053 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1054 tuning struct.
1055 (cortex-a57.cortex-a53): Likewise.
1056 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1058 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1060 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1061 arm_restrict_it.
1063 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1065 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1066 add_options_for_arm_vfp3.
1068 2014-02-11 Jeff Law <law@redhat.com>
1070 PR middle-end/54041
1071 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1072 object with an undesirable mode.
1074 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1076 PR libgomp/60107
1077 * config/i386/sol2-9.h: New file.
1078 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1079 *-*-solaris2.9*): Use it.
1081 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1083 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1084 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1086 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1088 * config/microblaze/microblaze.c: Extend mcpu version format
1090 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1092 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1094 2014-02-10 Richard Henderson <rth@redhat.com>
1096 PR target/59927
1097 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1098 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1099 ms-abi vs -mno-accumulate-outgoing-args.
1100 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1101 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1102 respect to ms-abi.
1104 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1106 PR middle-end/60080
1107 * cfgexpand.c (expand_asm_operands): Attach source location to
1108 ASM_INPUT rtx objects.
1109 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1111 2014-02-10 Nick Clifton <nickc@redhat.com>
1113 * config/mn10300/mn10300.c (popcount): New function.
1114 (mn10300_expand_prologue): Include saved registers in stack usage
1115 count.
1117 2014-02-10 Jeff Law <law@redhat.com>
1119 PR middle-end/52306
1120 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1121 when changing the SET_DEST of a prior insn to avoid an input reload.
1123 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1125 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1126 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1127 -mcall-openbsd, or -mcall-linux.
1128 (CC1_ENDIAN_BIG_SPEC): Remove.
1129 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1130 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1131 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1132 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1133 and %cc1_endian_default.
1134 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1136 2014-02-10 Richard Biener <rguenther@suse.de>
1138 PR tree-optimization/60115
1139 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1140 MEM_REF handling. Properly verify that the accesses are not
1141 out of the objects bound.
1143 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1145 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1146 coretex to cortex.
1148 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1150 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1151 proper constants and fix formatting.
1152 (possible_polymorphic_call_targets): Fix formatting.
1154 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1155 Ilya Tocar <ilya.tocar@intel.com>
1157 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1158 (_mm512_loadu_epi32): Renamed into...
1159 (_mm512_loadu_si512): This.
1160 (_mm512_storeu_epi32): Renamed into...
1161 (_mm512_storeu_si512): This.
1162 (_mm512_maskz_ceil_ps): Removed.
1163 (_mm512_maskz_ceil_pd): Ditto.
1164 (_mm512_maskz_floor_ps): Ditto.
1165 (_mm512_maskz_floor_pd): Ditto.
1166 (_mm512_floor_round_ps): Ditto.
1167 (_mm512_floor_round_pd): Ditto.
1168 (_mm512_ceil_round_ps): Ditto.
1169 (_mm512_ceil_round_pd): Ditto.
1170 (_mm512_mask_floor_round_ps): Ditto.
1171 (_mm512_mask_floor_round_pd): Ditto.
1172 (_mm512_mask_ceil_round_ps): Ditto.
1173 (_mm512_mask_ceil_round_pd): Ditto.
1174 (_mm512_maskz_floor_round_ps): Ditto.
1175 (_mm512_maskz_floor_round_pd): Ditto.
1176 (_mm512_maskz_ceil_round_ps): Ditto.
1177 (_mm512_maskz_ceil_round_pd): Ditto.
1178 (_mm512_expand_pd): Ditto.
1179 (_mm512_expand_ps): Ditto.
1180 * config/i386/i386.c (ix86_builtins): Remove
1181 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1182 (bdesc_args): Ditto.
1183 * config/i386/predicates.md (const1256_operand): New.
1184 (const_1_to_2_operand): Ditto.
1185 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1186 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1187 (*avx512pf_gatherpf<mode>sf): Ditto.
1188 (avx512pf_gatherpf<mode>df): Ditto.
1189 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1190 (*avx512pf_gatherpf<mode>df): Ditto.
1191 (avx512pf_scatterpf<mode>sf): Ditto.
1192 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1193 (*avx512pf_scatterpf<mode>sf): Ditto.
1194 (avx512pf_scatterpf<mode>df): Ditto.
1195 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1196 (*avx512pf_scatterpf<mode>df): Ditto.
1197 (avx512f_expand<mode>): Removed.
1198 (<shift_insn><mode>3<mask_name>): Change predicate type.
1200 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1202 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1203 not at the end of datarefs vector use ordered_remove to avoid
1204 reordering datarefs vector.
1206 PR c/59984
1207 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1208 mark local addressable non-static vars as GOVD_PRIVATE
1209 instead of GOVD_LOCAL.
1210 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1211 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1212 than copying them.
1214 PR middle-end/60092
1215 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1216 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1217 assume_aligned or alloc_align attributes.
1218 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1219 arguments. Handle also assume_aligned and alloc_align attributes.
1220 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1221 calls to functions with assume_aligned or alloc_align attributes.
1222 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1224 2014-02-08 Terry Guo <terry.guo@arm.com>
1226 * doc/invoke.texi: Document ARM -march=armv7e-m.
1228 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1230 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1231 flag on __cilkrts_rethrow builtin.
1233 PR ipa/60026
1234 * ipa-cp.c (determine_versionability): Fail at -O0
1235 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1236 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1238 Revert:
1239 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1241 PR ipa/60026
1242 * tree-inline.c (copy_forbidden): Fail for
1243 __attribute__((optimize (0))) functions.
1245 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1247 * varpool.c: Include pointer-set.h.
1248 (varpool_remove_unreferenced_decls): Variables in other partitions
1249 will not be output; be however careful to not lose information
1250 about partitioning.
1252 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1254 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1255 lookup in the vtable constructor.
1257 2014-02-07 Jeff Law <law@redhat.com>
1259 PR target/40977
1260 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1261 define_insn_and_split.
1263 * ipa-inline.c (inline_small_functions): Fix typos.
1265 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1267 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1268 (s390_can_use_return_insn): Declare.
1269 * config/s390/s390.h (EPILOGUE_USES): Define.
1270 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1271 instructions.
1272 (s390_chunkify_start): Handle return JUMP_LABELs.
1273 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1274 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1275 (s390_can_use_return_insn): New functions.
1276 (s390_fix_long_loop_prediction): Handle conditional returns.
1277 (TARGET_SET_UP_BY_PROLOGUE): Define.
1278 * config/s390/s390.md (ANY_RETURN): New code iterator.
1279 (*creturn, *csimple_return, return, simple_return): New patterns.
1281 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1283 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1284 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1285 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1286 REG_CFA_RESTORE list when deciding not to restore a register.
1288 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1290 * config/s390/s390.c: Include tree-pass.h and context.h.
1291 (s390_early_mach): New function, split out from...
1292 (s390_emit_prologue): ...here.
1293 (pass_data_s390_early_mach): New pass structure.
1294 (pass_s390_early_mach): New class.
1295 (s390_option_override): Create and register early_mach pass.
1296 Move to end of file.
1298 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1300 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1301 to match for the exit block.
1303 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1305 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1306 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1307 Reject misaligned operands.
1309 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1311 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1313 2014-02-07 Richard Biener <rguenther@suse.de>
1315 PR middle-end/60092
1316 * gimple-low.c (lower_builtin_posix_memalign): New function.
1317 (lower_stmt): Call it to lower posix_memalign in a way
1318 to make alignment info accessible.
1320 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1322 PR c++/60082
1323 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1324 __builtin_setjmp_receiver.
1326 2014-02-07 Richard Biener <rguenther@suse.de>
1328 PR middle-end/60092
1329 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1330 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1331 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1332 Handle BUILT_IN_POSIX_MEMALIGN.
1333 (find_func_clobbers): Likewise.
1334 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1335 (call_may_clobber_ref_p_1): Likewise.
1337 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1339 PR ipa/59918
1340 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1341 sanity check.
1343 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1345 PR ipa/59469
1346 * lto-cgraph.c (lto_output_node): Use
1347 symtab_get_symbol_partitioning_class.
1348 (lto_output_varpool_node): likewise.
1349 (symtab_get_symbol_partitioning_class): Move here from
1350 lto/lto-partition.c
1351 * cgraph.h (symbol_partitioning_class): Likewise.
1352 (symtab_get_symbol_partitioning_class): Declare.
1354 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1356 * ggc.h (ggc_internal_cleared_alloc): New macro.
1357 * vec.h (vec_safe_copy): Handle memory stats.
1358 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1359 * target-globals.c (save_target_globals): Likewise.
1361 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1363 PR target/60077
1364 * expr.c (emit_move_resolve_push): Export; be bit more selective
1365 on when to clear alias set.
1366 * expr.h (emit_move_resolve_push): Declare.
1367 * function.h (struct function): Add tail_call_marked.
1368 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1369 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1370 * config/i386/i386.md (TImode move expander): De not call
1371 ix86_expand_push.
1372 (FP push expanders): Preserve memory attributes.
1373 * config/i386/sse.md (push<mode>1): Remove.
1374 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1375 (ix86_expand_push): Remove.
1376 * config/i386/mmx.md (push<mode>1): Remove.
1378 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1380 PR rtl-optimization/60030
1381 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1382 lopart with paradoxical subreg before shifting it up by hprec.
1384 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1386 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1387 Remove extra newline at end of file.
1388 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1389 (arm_issue_rate): Handle cortexa57.
1390 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1391 (cortex-a57.cortex-a53): Likewise.
1393 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1395 PR target/59575
1396 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1397 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1398 bitmask.
1399 (arm_expand_prologue): Adjust all callers.
1400 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1401 info, registers also at the lowest numbered registers side. Use
1402 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1403 XEXP.
1405 PR debug/59992
1406 * var-tracking.c (adjust_mems): Before adding a SET to
1407 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1409 2014-02-06 Alan Modra <amodra@gmail.com>
1411 PR target/60032
1412 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1413 change SDmode to DDmode when lra_in_progress.
1415 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1417 PR middle-end/59150
1418 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1419 free_data_ref on the dr first, and before goto again also set dr
1420 to the next dr. For simd_lane_access, free old datarefs[i] before
1421 overwriting it. For get_vectype_for_scalar_type failure, don't
1422 free_data_ref if simd_lane_access.
1424 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1426 PR target/60062
1427 * tree.h (opts_for_fn): New inline function.
1428 (opt_for_fn): Define.
1429 * config/i386/i386.c (ix86_function_regparm): Use
1430 opt_for_fn (decl, optimize) instead of optimize.
1432 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1434 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1435 for SYMBOL_REF in large memory model.
1437 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1439 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1440 and crypto support.
1441 (cortex-a57): Likewise.
1442 (cortex-a57.cortex-a53): Likewise.
1444 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1445 Kugan Vivekanandarajah <kuganv@linaro.org>
1447 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1448 unaligned_access.
1449 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1451 2014-02-06 Richard Biener <rguenther@suse.de>
1453 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1454 set_loop_copy and initialize_original_copy_tables.
1456 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1458 * config/aarch64/aarch64-simd.md
1459 (aarch64_ashr_simddi): Change QI to SI.
1461 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1462 Jakub Jelinek <jakub@redhat.com>
1464 PR middle-end/60013
1465 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1466 of the dataflow.
1468 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1470 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1471 CODE_FOR_altivec_vpku[hw]um to
1472 CODE_FOR_altivec_vpku[hw]um_direct.
1473 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1474 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1475 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1476 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1478 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1480 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1481 generation for -maltivec=be.
1482 (altivec_vsumsws): Simplify redundant test.
1484 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1486 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1487 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1488 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1489 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1490 gen_altivec_vpkuwum.
1491 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1492 BYTES_BIG_ENDIAN.
1493 (altivec_vpks<VI_char>ss): Likewise.
1494 (altivec_vpks<VI_char>us): Likewise.
1495 (altivec_vpku<VI_char>us): Likewise.
1496 (altivec_vpku<VI_char>um): Likewise.
1497 (altivec_vpku<VI_char>um_direct): New (copy of
1498 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1499 internal use).
1500 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1501 target is little endian and -maltivec=be is not specified.
1502 (*altivec_vupkhs<VU_char>_direct): New (copy of
1503 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1504 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1505 target is little endian and -maltivec=be is not specified.
1506 (*altivec_vupkls<VU_char>_direct): New (copy of
1507 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1508 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1509 little endian and -maltivec=be is not specified.
1510 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1511 little endian and -maltivec=be is not specified.
1513 2014-02-05 Richard Henderson <rth@redhat.com>
1515 PR debug/52727
1516 * combine-stack-adj.c: Revert r206943.
1517 * sched-int.h (struct deps_desc): Add last_args_size.
1518 * sched-deps.c (init_deps): Initialize it.
1519 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1520 contain REG_ARGS_SIZE notes.
1522 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1524 * lto-cgraph.c (asm_nodes_output): Make global.
1525 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1526 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1527 (driver_handle_option): Handle OPT_fwpa.
1529 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1531 PR ipa/59947
1532 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1533 a comment typo and formatting issue. If odr_hash hasn't been
1534 created, return vNULL and set *completep to false.
1536 PR middle-end/57499
1537 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1538 bb with no successors.
1540 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1542 PR target/59718
1543 * doc/invoke.texi (-march): Clarify documentation for ARM.
1544 (-mtune): Likewise.
1545 (-mcpu): Likewise.
1547 2014-02-05 Richard Biener <rguenther@suse.de>
1549 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1550 when not vectorizing because of too many alias checks.
1551 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1552 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1554 2014-02-05 Nick Clifton <nickc@redhat.com>
1556 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1557 accept extended registers in any mode when compiling for the MN10300.
1559 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1561 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1562 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1563 sanitization attributes.
1564 (can_inline_edge_p): Likewise.
1565 (sanitize_attrs_match_for_inline_p): New function.
1567 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1569 * ipa-prop.c (detect_type_change): Shor circuit testing of
1570 type changes on THIS pointer.
1572 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1574 PR target/59777
1575 * config/pa/pa.c (legitimize_tls_address): Return original address
1576 if not passed a SYMBOL_REF rtx.
1577 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1578 addresses.
1579 (pa_emit_move_sequence): Simplify TLS source operands.
1580 (pa_legitimate_constant_p): Reject all TLS constants.
1581 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1582 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1584 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1586 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1587 groups when we know they are controlled by LTO.
1588 * varasm.c (default_binds_local_p_1): If object is in other partition,
1589 it will be resolved locally.
1591 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1593 * config/host-linux.c (linux_gt_pch_use_address): Don't
1594 use SSIZE_MAX because it is not always defined.
1596 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1598 PR bootstrap/59913
1599 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1600 threshold for pseudo splitting.
1601 (update_ebb_live_info): Process call argument hard registers and
1602 hard registers from insn definition too.
1603 (max_small_class_regs_num): New constant.
1604 (inherit_in_ebb): Update live hard regs through EBBs. Update
1605 reloads_num only for small register classes. Don't split for
1606 outputs of jumps.
1608 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1610 PR ipa/60058
1611 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1612 is non-null.
1614 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1616 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1617 visibility is safe.
1619 2014-02-04 Marek Polacek <polacek@redhat.com>
1621 * gdbinit.in (pel): Define.
1623 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1625 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1626 behavior.
1628 2014-02-04 Richard Biener <rguenther@suse.de>
1630 PR lto/59723
1631 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1632 in function context local.
1633 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1634 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1635 similar to LTO_imported_decl_ref.
1637 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1639 PR tree-optimization/60002
1640 * cgraphclones.c (build_function_decl_skip_args): Clear
1641 DECL_LANG_SPECIFIC.
1643 PR tree-optimization/60023
1644 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1645 false to gsi_replace.
1646 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1647 has been in some EH region and vec_stmt could throw, add
1648 vec_stmt into the same EH region.
1649 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1650 has no lhs, ignore it.
1651 * internal-fn.c (expand_MASK_LOAD): Likewise.
1653 PR ipa/60026
1654 * tree-inline.c (copy_forbidden): Fail for
1655 __attribute__((optimize (0))) functions.
1657 PR other/58712
1658 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1659 is set, copy one less argument.
1660 (expand_simd_clones): Don't subtract clone_info->inbranch
1661 from simd_clone_struct_alloc argument.
1663 PR rtl-optimization/57915
1664 * recog.c (simplify_while_replacing): If all unary/binary/relational
1665 operation arguments are constant, attempt to simplify those.
1667 PR middle-end/59261
1668 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1669 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1671 2014-02-04 Richard Biener <rguenther@suse.de>
1673 PR tree-optimization/60012
1674 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1675 TBAA disambiguation to all DDRs.
1677 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1679 PR target/59788
1680 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1681 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1683 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1685 PR ipa/59882
1686 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1688 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1690 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1691 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1693 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1695 PR ipa/59831
1696 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1697 to figure out targets of polymorphic calls with known decl.
1698 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1699 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1700 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1701 (get_polymorphic_call_info): ... here.
1702 (get_polymorphic_call_info_from_invariant): New function.
1704 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1706 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1707 lookup via vtable pointer; check for type consistency
1708 and turn inconsitent facts into UNREACHABLE.
1709 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1710 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1711 type inconsistent querries; return UNREACHABLE instead.
1713 2014-02-03 Richard Henderson <rth@twiddle.net>
1715 PR tree-opt/59924
1716 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1717 already processed this node.
1718 (normalize_one_pred_1): Pass along mark_set.
1719 (normalize_one_pred): Create and destroy a pointer_set_t.
1720 (normalize_one_pred_chain): Likewise.
1722 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1724 PR gcov-profile/58602
1725 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1727 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1729 PR ipa/59831
1730 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1731 -fno-devirtualize; try to devirtualize by the knowledge of
1732 virtual table pointer given by aggregate propagation.
1733 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1734 (ipa_print_node_jump_functions): Dump also offset that
1735 is relevant for polymorphic calls.
1736 (determine_known_aggregate_parts): Add arg_type parameter; use it
1737 instead of determining the type from pointer type.
1738 (ipa_compute_jump_functions_for_edge): Update call of
1739 determine_known_aggregate_parts.
1740 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1741 (gimple_get_virt_method_for_binfo): ... here; simplify using
1742 vtable_pointer_value_to_vtable.
1743 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1744 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1745 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1746 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1747 POINTER_PLUS_EXPR.
1748 (vtable_pointer_value_to_binfo): ... here.
1749 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1751 2014-02-03 Teresa Johnson <tejohnson@google.com>
1753 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1754 redef of outer loop index variable.
1756 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1758 PR c++/53017
1759 PR c++/59211
1760 * doc/extend.texi (Function Attributes): Typo.
1762 2014-02-03 Cong Hou <congh@google.com>
1764 PR tree-optimization/60000
1765 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1766 if the vectorized statement is a store. A store statement can only
1767 appear at the end of pattern statements.
1769 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1771 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1772 (ix86_option_override_internal): Default long double to 64-bit for
1773 32-bit Bionic and to 128-bit for 64-bit Bionic.
1775 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1776 TARGET_LONG_DOUBLE_128 is true.
1777 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1779 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1780 (mlong-double-64): Negate -mlong-double-128.
1781 (mlong-double-128): New option.
1783 * config/i386/i386-c.c (ix86_target_macros): Define
1784 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1786 * doc/invoke.texi: Document -mlong-double-128.
1788 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1790 PR rtl-optimization/60024
1791 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1793 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1795 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1797 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1799 PR rtl-optimization/57662
1800 * sel-sched.c (code_motion_path_driver): Do not mark already not
1801 existing blocks in the visiting bitmap.
1803 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1805 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1806 on the insn being emitted.
1808 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1809 Will Deacon <will.deacon@arm.com>
1811 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1813 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1815 * config/arm/arm-tables.opt: Regenerate.
1817 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1819 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1820 for vector types other than V16QImode.
1821 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1822 define_expand, and call altivec_expand_vec_perm_le when producing
1823 code with little endian element order.
1824 (*altivec_vperm_<mode>_internal): New insn having previous
1825 behavior of altivec_vperm_<mode>.
1826 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1827 altivec_expand_vec_perm_le when producing code with little endian
1828 element order.
1829 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1830 behavior of altivec_vperm_<mode>_uns.
1832 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1834 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1835 (altivec_vsumsws): Add handling for -maltivec=be with a little
1836 endian target.
1837 (altivec_vsumsws_direct): New.
1838 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1839 gen_altivec_vsumsws.
1841 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1843 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1844 vtable_pointer_value_to_binfo): New functions.
1845 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1846 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1848 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1850 * config/nios2/nios2.md (load_got_register): Initialize GOT
1851 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1852 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1854 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1856 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1857 preserverd by passthrough, do not propagate the type.
1859 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1861 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1862 (mips_atomic_assign_expand_fenv): New function.
1863 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1865 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1867 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1868 (__builtin_mips_set_fcsr): Likewise.
1869 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1870 MIPS_USI_FTYPE_VOID.
1871 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1872 (mips16_expand_set_fcsr): Likewise.
1873 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1874 (mips16_set_fcsr_stub): Likewise.
1875 (mips16_get_fcsr_one_only_stub): New class.
1876 (mips16_set_fcsr_one_only_stub): Likewise.
1877 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1878 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1879 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1880 (hard_float): New availability predicate.
1881 (mips_builtins): Add get_fcsr and set_fcsr.
1882 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1883 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1884 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1885 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1886 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1887 patterns.
1889 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1891 * config/mips/mips.c (mips_one_only_stub): New class.
1892 (mips_need_mips16_rdhwr_p): Replace with...
1893 (mips16_rdhwr_stub): ...this new variable.
1894 (mips16_stub_call_address): New function.
1895 (mips16_rdhwr_one_only_stub): New class.
1896 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1897 (mips_output_mips16_rdhwr): Delete.
1898 (mips_finish_stub): New function.
1899 (mips_code_end): Use it to handle rdhwr stubs.
1901 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1903 PR target/60017
1904 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1905 when calculating size of integer atomic types.
1907 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1909 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1911 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1913 PR tree-optimization/60003
1914 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1915 * profile.c (branch_prob): Use gimple_call_builtin_p
1916 to check for BUILT_IN_SETJMP_RECEIVER.
1917 * tree-inline.c (copy_bb): Call notice_special_calls.
1919 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1921 PR bootstrap/59985
1922 * lra-constraints.c (process_alt_operands): Update reload_sum only
1923 on the first pass.
1925 2014-01-31 Richard Henderson <rth@redhat.com>
1927 PR middle-end/60004
1928 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1929 until after else_eh is processed.
1931 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1933 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1934 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1935 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1936 in smmintrin.h, remove them.
1937 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1938 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1939 * config/i386/i386.md (ROUND_SAE): Fix value.
1940 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1941 (const48_operand): New.
1942 * config/i386/subst.md (round), (round_expand): Use
1943 const_4_or_8_to_11_operand.
1944 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1946 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1948 * config/i386/constraints.md (Yk): Swap meaning with k.
1949 * config/i386/i386.md (movhi_internal): Change Yk to k.
1950 (movqi_internal): Ditto.
1951 (*k<logic><mode>): Ditto.
1952 (*andhi_1): Ditto.
1953 (*andqi_1): Ditto.
1954 (kandn<mode>): Ditto.
1955 (*<code>hi_1): Ditto.
1956 (*<code>qi_1): Ditto.
1957 (kxnor<mode>): Ditto.
1958 (kortestzhi): Ditto.
1959 (kortestchi): Ditto.
1960 (kunpckhi): Ditto.
1961 (*one_cmplhi2_1): Ditto.
1962 (*one_cmplqi2_1): Ditto.
1963 * config/i386/sse.md (): Change k to Yk.
1964 (avx512f_load<mode>_mask): Ditto.
1965 (avx512f_blendm<mode>): Ditto.
1966 (avx512f_store<mode>_mask): Ditto.
1967 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1968 (avx512f_storedqu<mode>_mask): Ditto.
1969 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1970 Ditto.
1971 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1972 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1973 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1974 (avx512f_maskcmp<mode>3): Ditto.
1975 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1976 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1977 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1978 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1979 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1980 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1981 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1982 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1983 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1984 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1985 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1986 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1987 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1988 (vec_extract_lo_<mode>_maskm): Ditto.
1989 (vec_extract_hi_<mode>_maskm): Ditto.
1990 (avx512f_vternlog<mode>_mask): Ditto.
1991 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1992 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1993 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1994 (avx512f_<code>v8div16qi2_mask): Ditto.
1995 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1996 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1997 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1998 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1999 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2000 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2001 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2002 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2003 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2004 (avx512cd_maskb_vec_dupv8di): Ditto.
2005 (avx512cd_maskw_vec_dupv16si): Ditto.
2006 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2007 (avx512f_vpermi2var<mode>3_mask): Ditto.
2008 (avx512f_vpermi2var<mode>3_mask): Ditto.
2009 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2010 (*avx512f_gathersi<mode>): Ditto.
2011 (*avx512f_gathersi<mode>_2): Ditto.
2012 (*avx512f_gatherdi<mode>): Ditto.
2013 (*avx512f_gatherdi<mode>_2): Ditto.
2014 (*avx512f_scattersi<mode>): Ditto.
2015 (*avx512f_scatterdi<mode>): Ditto.
2016 (avx512f_compress<mode>_mask): Ditto.
2017 (avx512f_compressstore<mode>_mask): Ditto.
2018 (avx512f_expand<mode>_mask): Ditto.
2019 * config/i386/subst.md (mask): Change k to Yk.
2020 (mask_scalar_merge): Ditto.
2021 (sd): Ditto.
2023 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2025 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2027 2014-01-31 Richard Biener <rguenther@suse.de>
2029 PR middle-end/59990
2030 * builtins.c (fold_builtin_memory_op): Make sure to not
2031 use a floating-point mode or a boolean or enumeral type for
2032 the copy operation.
2034 2014-01-30 DJ Delorie <dj@redhat.com>
2036 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2037 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2038 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2039 whenever main() has an epilogue.
2041 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2043 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2044 unused variable "field".
2045 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2046 (vsx_mergeh_<mode>): Likewise.
2047 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2048 (altivec_vmrghh): Likewise.
2049 (altivec_vmrghw): Likewise.
2050 (altivec_vmrglb): Likewise.
2051 (altivec_vmrglh): Likewise.
2052 (altivec_vmrglw): Likewise.
2053 (altivec_vspltb): Add missing uses.
2054 (altivec_vsplth): Likewise.
2055 (altivec_vspltw): Likewise.
2056 (altivec_vspltsf): Likewise.
2058 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2060 PR target/59923
2061 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2062 frame related instructions.
2064 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2066 PR rtl-optimization/59959
2067 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2068 any reload of register whose subreg is invalid.
2070 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2072 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2073 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2074 Add missing return type - void.
2076 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2078 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2079 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2080 remove element index adjustment for endian (now handled in vsx.md
2081 and altivec.md).
2082 (altivec_expand_vec_perm_const): Use
2083 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2084 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2085 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2086 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2087 define_expand and a new define_insn *altivec_vspltb_internal;
2088 adjust for -maltivec=be on a little endian target.
2089 (altivec_vspltb_direct): New.
2090 (altivec_vsplth): Divide into a define_expand and a new
2091 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2092 little endian target.
2093 (altivec_vsplth_direct): New.
2094 (altivec_vspltw): Divide into a define_expand and a new
2095 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2096 little endian target.
2097 (altivec_vspltw_direct): New.
2098 (altivec_vspltsf): Divide into a define_expand and a new
2099 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2100 a little endian target.
2102 2014-01-30 Richard Biener <rguenther@suse.de>
2104 PR tree-optimization/59993
2105 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2106 can propagate form the earlier stmt and avoid the transform
2107 when the intermediate result is needed.
2109 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2111 * README.Portability: Fix typo.
2113 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2115 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2116 comparison_operator with ordered_comparison_operator.
2118 2014-01-30 Nick Clifton <nickc@redhat.com>
2120 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2121 Rename to mn10300_store_multiple_regs.
2122 * config/mn10300/mn10300.c: Likewise.
2123 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2124 store_multiple_regs.
2125 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2126 Call mn10300_store_multiple_regs.
2128 2014-01-30 Nick Clifton <nickc@redhat.com>
2129 DJ Delorie <dj@redhat.com>
2131 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2132 %fp 2 to keep registers after it properly word-aligned.
2133 (rl78_alloc_physical_registers_umul): Handle the case where both
2134 input operands are the same.
2136 2014-01-30 Richard Biener <rguenther@suse.de>
2138 PR tree-optimization/59903
2139 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2140 check properly.
2142 2014-01-30 Jason Merrill <jason@redhat.com>
2144 PR c++/59633
2145 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2147 PR c++/59645
2148 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2150 2014-01-30 Richard Biener <rguenther@suse.de>
2152 PR tree-optimization/59951
2153 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2155 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2157 PR target/59784
2158 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2159 SFmode to DFmode case.
2161 2014-01-29 DJ Delorie <dj@redhat.com>
2163 * config/msp430/msp430.opt (-minrt): New.
2164 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2165 if -minrt given.
2166 (ENDFILE_SPEC): Likewise.
2168 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2170 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2171 (estimate_function_body_sizes): Use it.
2173 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2175 PR c++/58561
2176 * dwarf2out.c (is_cxx_auto): New.
2177 (is_base_type): Use it.
2178 (gen_type_die_with_usage): Likewise.
2180 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2182 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2183 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2184 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2185 -maltivec=be with LE targets.
2186 (vsx_mergeh_<mode>): Likewise.
2187 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2188 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2189 (altivec_vmrghb): Replace with define_expand and new
2190 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2191 (altivec_vmrghb_direct): New define_insn.
2192 (altivec_vmrghh): Replace with define_expand and new
2193 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2194 (altivec_vmrghh_direct): New define_insn.
2195 (altivec_vmrghw): Replace with define_expand and new
2196 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2197 (altivec_vmrghw_direct): New define_insn.
2198 (*altivec_vmrghsf): Adjust for endianness.
2199 (altivec_vmrglb): Replace with define_expand and new
2200 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2201 (altivec_vmrglb_direct): New define_insn.
2202 (altivec_vmrglh): Replace with define_expand and new
2203 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2204 (altivec_vmrglh_direct): New define_insn.
2205 (altivec_vmrglw): Replace with define_expand and new
2206 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2207 (altivec_vmrglw_direct): New define_insn.
2208 (*altivec_vmrglsf): Adjust for endianness.
2209 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2210 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2211 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2212 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2213 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2214 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2215 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2216 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2218 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2220 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2221 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2222 whitespace.
2224 2014-01-29 Richard Biener <rguenther@suse.de>
2226 PR tree-optimization/58742
2227 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2228 associate_pointerplus_align.
2229 (associate_pointerplus_diff): New function.
2230 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2231 and associate_pointerplus_diff.
2233 2014-01-29 Richard Biener <rguenther@suse.de>
2235 * lto-streamer.h (LTO_major_version): Bump to 3.
2236 (LTO_minor_version): Reset to 0.
2238 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2240 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2241 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2242 (arm_file_start): Generate correct asm header for armv7ve.
2243 * config/arm/bpabi.h: Add multilib support for armv7ve.
2244 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2245 and cortex-a15 to armv7ve.
2246 * config/arm/t-aprofile: Add multilib support for armv7ve.
2247 * doc/invoke.texi: Document -march=armv7ve.
2249 2014-01-29 Richard Biener <rguenther@suse.de>
2251 PR tree-optimization/58742
2252 * tree-ssa-forwprop.c (associate_plusminus): Return true
2253 if we changed sth, defer EH cleanup to ...
2254 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2255 (simplify_mult): New function.
2257 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2259 PR middle-end/59917
2260 PR tree-optimization/59920
2261 * tree.c (build_common_builtin_nodes): Remove
2262 __builtin_setjmp_dispatcher initialization.
2263 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2264 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2265 instead of gsi_after_labels + manually skipping debug stmts.
2266 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2267 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2268 * tree-inline.c (copy_edges_for_bb): Remove
2269 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2270 argument. Ignore computed_goto_p stmts. Don't call
2271 make_abnormal_goto_edges. If a call might need abnormal edges
2272 for non-local gotos, see if it already has an edge to
2273 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2274 with true argument, don't do anything then, otherwise add
2275 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2276 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2277 caller.
2278 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2279 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2280 (lower_stmt): Don't set data->calls_builtin_setjmp.
2281 (lower_builtin_setjmp): Adjust comment.
2282 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2283 * tree-cfg.c (found_computed_goto): Remove.
2284 (factor_computed_gotos): Remove.
2285 (make_goto_expr_edges): Return bool, true for computed gotos.
2286 Don't call make_abnormal_goto_edges.
2287 (build_gimple_cfg): Don't set found_computed_goto, don't call
2288 factor_computed_gotos.
2289 (computed_goto_p): No longer static.
2290 (make_blocks): Don't set found_computed_goto.
2291 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2292 (make_edges): If make_goto_expr_edges returns true, push bb
2293 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2294 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2295 vector. Record mapping between bbs and OpenMP regions if there
2296 are any, adjust make_gimple_omp_edges caller. Call
2297 handle_abnormal_edges.
2298 (make_abnormal_goto_edges): Remove.
2299 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2300 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2301 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2302 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2303 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2304 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2305 filling *region also set *region_idx to (*region)->entry->index.
2307 PR other/58712
2308 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2309 For REGs set ORIGINAL_REGNO.
2311 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2313 * doc/md.texi: Mention that a target shouldn't implement
2314 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2315 than hi/lo pair.
2317 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2319 PR tree-optimization/59594
2320 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2321 a copy of the datarefs vector rather than the vector itself.
2323 2014-01-28 Jason Merrill <jason@redhat.com>
2325 PR c++/53756
2326 * dwarf2out.c (auto_die): New static.
2327 (gen_type_die_with_usage): Handle C++1y 'auto'.
2328 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2329 on definition.
2331 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2333 PR target/59672
2334 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2335 (SPEC_X32): Likewise.
2336 (SPEC_64): Likewise.
2337 * config/i386/i386.c (ix86_option_override_internal): Turn off
2338 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2339 for TARGET_16BIT.
2340 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2341 * config/i386/i386.h (TARGET_16BIT): New macro.
2342 (TARGET_16BIT_P): Likewise.
2343 * config/i386/i386.opt: Add m16.
2344 * doc/invoke.texi: Document -m16.
2346 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2348 PR preprocessor/59935
2349 * input.c (location_get_source_line): Bail out on when line number
2350 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2352 2014-01-28 Richard Biener <rguenther@suse.de>
2354 PR tree-optimization/58742
2355 * tree-ssa-forwprop.c (associate_plusminus): Handle
2356 pointer subtraction of the form (T)(P + A) - (T)P.
2358 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2360 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2361 at const_int_cost.
2363 2014-01-28 Richard Biener <rguenther@suse.de>
2365 Revert
2366 2014-01-28 Richard Biener <rguenther@suse.de>
2368 PR rtl-optimization/45364
2369 PR rtl-optimization/59890
2370 * var-tracking.c (local_get_addr_clear_given_value): Handle
2371 already cleared slot.
2372 (val_reset): Handle not allocated local_get_addr_cache.
2373 (vt_find_locations): Use post-order on the inverted CFG.
2375 2014-01-28 Richard Biener <rguenther@suse.de>
2377 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2379 2014-01-28 Richard Biener <rguenther@suse.de>
2381 PR rtl-optimization/45364
2382 PR rtl-optimization/59890
2383 * var-tracking.c (local_get_addr_clear_given_value): Handle
2384 already cleared slot.
2385 (val_reset): Handle not allocated local_get_addr_cache.
2386 (vt_find_locations): Use post-order on the inverted CFG.
2388 2014-01-28 Alan Modra <amodra@gmail.com>
2390 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2391 * configure.ac <recursive call for build != host>: Define
2392 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2393 and LD_FOR_BUILD too.
2394 * configure: Regenerate.
2396 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2398 * config/i386/i386.c (get_builtin_code_for_version): Separate
2399 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2400 Broadwell from Haswell.
2402 2014-01-27 Steve Ellcey <sellcey@mips.com>
2404 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2405 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2406 * config/mips/mips.c (mips_option_override): Change setting
2407 of TARGET_DSP.
2408 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2409 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2410 Change from Mask to Var.
2412 2014-01-27 Jeff Law <law@redhat.com>
2414 * ipa-inline.c (inline_small_functions): Fix typo.
2416 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2418 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2419 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2420 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2421 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2422 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2423 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2424 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2425 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2426 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2427 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2428 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2429 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2430 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2431 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2432 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2433 (_mm512_storeu_epi64): Ditto.
2434 (_mm512_cmpge_epi32_mask): Ditto.
2435 (_mm512_cmpge_epu32_mask): Ditto.
2436 (_mm512_cmpge_epi64_mask): Ditto.
2437 (_mm512_cmpge_epu64_mask): Ditto.
2438 (_mm512_cmple_epi32_mask): Ditto.
2439 (_mm512_cmple_epu32_mask): Ditto.
2440 (_mm512_cmple_epi64_mask): Ditto.
2441 (_mm512_cmple_epu64_mask): Ditto.
2442 (_mm512_cmplt_epi32_mask): Ditto.
2443 (_mm512_cmplt_epu32_mask): Ditto.
2444 (_mm512_cmplt_epi64_mask): Ditto.
2445 (_mm512_cmplt_epu64_mask): Ditto.
2446 (_mm512_cmpneq_epi32_mask): Ditto.
2447 (_mm512_cmpneq_epu32_mask): Ditto.
2448 (_mm512_cmpneq_epi64_mask): Ditto.
2449 (_mm512_cmpneq_epu64_mask): Ditto.
2450 (_mm512_expand_pd): Ditto.
2451 (_mm512_expand_ps): Ditto.
2452 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2453 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2454 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2455 * config/i386/i386.c (ix86_builtins): Add
2456 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2457 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2458 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2459 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2460 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2461 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2462 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2463 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2464 IX86_BUILTIN_PMOVUSQW512_MEM.
2465 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2466 __builtin_ia32_pmovsqd512mem_mask,
2467 __builtin_ia32_pmovqd512mem_mask,
2468 __builtin_ia32_pmovusqw512mem_mask,
2469 __builtin_ia32_pmovsqw512mem_mask,
2470 __builtin_ia32_pmovqw512mem_mask,
2471 __builtin_ia32_pmovusdw512mem_mask,
2472 __builtin_ia32_pmovsdw512mem_mask,
2473 __builtin_ia32_pmovdw512mem_mask,
2474 __builtin_ia32_pmovqb512mem_mask,
2475 __builtin_ia32_pmovusqb512mem_mask,
2476 __builtin_ia32_pmovsqb512mem_mask,
2477 __builtin_ia32_pmovusdb512mem_mask,
2478 __builtin_ia32_pmovsdb512mem_mask,
2479 __builtin_ia32_pmovdb512mem_mask.
2480 (bdesc_args): Add __builtin_ia32_expanddf512,
2481 __builtin_ia32_expandsf512.
2482 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2483 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2484 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2485 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2486 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2487 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2488 (avx512f_<code>v8div16qi2_mask_store): This.
2489 (avx512f_expand<mode>): New.
2491 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2493 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2494 New.
2495 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2496 (_mm512_prefetch_i32scatter_pd): Ditto.
2497 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2498 (_mm512_prefetch_i64scatter_pd): Ditto.
2499 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2500 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2501 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2502 (_mm512_prefetch_i32scatter_ps): Ditto.
2503 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2504 (_mm512_prefetch_i64scatter_ps): Ditto.
2505 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2506 * config/i386/i386-builtin-types.def: Define
2507 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2508 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2509 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2510 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2511 IX86_BUILTIN_SCATTERPFQPD.
2512 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2513 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2514 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2515 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2516 __builtin_ia32_scatterpfqps.
2517 (ix86_expand_builtin): Expand new built-ins.
2518 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2519 fix memory access data type.
2520 (*avx512pf_gatherpf<mode>_mask): Ditto.
2521 (*avx512pf_gatherpf<mode>): Ditto.
2522 (avx512pf_scatterpf<mode>): Ditto.
2523 (*avx512pf_scatterpf<mode>_mask): Ditto.
2524 (*avx512pf_scatterpf<mode>): Ditto.
2525 (GATHER_SCATTER_SF_MEM_MODE): New.
2526 (avx512pf_gatherpf<mode>df): Ditto.
2527 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2528 (*avx512pf_scatterpf<mode>df): Ditto.
2530 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2532 PR bootstrap/59934
2533 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2534 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2535 reached.
2537 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2539 * common/config/arm/arm-common.c
2540 (arm_rewrite_mcpu): Handle multiple names.
2541 * config/arm/arm.h
2542 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2544 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2546 * gimple-builder.h (create_gimple_tmp): Delete.
2548 2014-01-27 Christian Bruel <christian.bruel@st.com>
2550 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2551 words comparisons.
2553 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2555 * config/pa/pa.md (call): Generate indirect long calls to non-local
2556 functions when outputing 32-bit code.
2557 (call_value): Likewise except for special call to buggy powf function.
2559 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2560 portable runtime and PIC indirect calls.
2561 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2562 and PIC call sequences. Use ldo instead of blr to set return register
2563 in PIC call sequence.
2565 2014-01-25 Walter Lee <walt@tilera.com>
2567 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2568 avoid clobbering a live register.
2570 2014-01-25 Walter Lee <walt@tilera.com>
2572 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2573 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2574 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2575 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2577 2014-01-25 Walter Lee <walt@tilera.com>
2579 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2580 arguments on even registers.
2581 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2582 STACK_BOUNDARY.
2583 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2584 (BIGGEST_ALIGNMENT): Ditto.
2585 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2587 2014-01-25 Walter Lee <walt@tilera.com>
2589 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2590 insns before bundling.
2591 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2593 2014-01-25 Walter Lee <walt@tilera.com>
2595 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2596 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2597 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2599 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2601 * config/mips/constraints.md (kl): Delete.
2602 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2603 define expands, using...
2604 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2605 instructions for MIPS16.
2606 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2607 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2609 2014-01-25 Walter Lee <walt@tilera.com>
2611 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2612 (clzdi2): Ditto.
2613 (ffsdi2): Ditto.
2615 2014-01-25 Walter Lee <walt@tilera.com>
2617 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2618 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2620 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2622 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2623 Handle XOR.
2625 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2627 * print-rtl.c (in_call_function_usage): New var.
2628 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2629 EXPR_LIST mode as mode and not as reg note name.
2631 PR middle-end/59561
2632 * cfgloopmanip.c (copy_loop_info): If
2633 loop->warned_aggressive_loop_optimizations, make sure
2634 the flag is set in target loop too.
2636 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2638 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2639 flag_cilkplus.
2640 * builtins.def: Likewise.
2641 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2642 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2643 * ira.c (ira_setup_eliminable_regset): Likewise.
2644 * omp-low.c (gate_expand_omp): Likewise.
2645 (execute_lower_omp): Likewise.
2646 (diagnose_sb_0): Likewise.
2647 (gate_diagnose_omp_blocks): Likewise.
2648 (simd_clone_clauses_extract): Likewise.
2649 (gate): Likewise.
2651 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2653 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2654 correction for little endian...
2655 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2656 here.
2658 2014-01-24 Jeff Law <law@redhat.com>
2660 PR tree-optimization/59919
2661 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2662 for non-returning calls.
2664 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2666 * common/config/aarch64/aarch64-common.c
2667 (aarch64_rewrite_mcpu): Handle multiple names.
2668 * config/aarch64/aarch64.h
2669 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2671 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2673 * input.c (add_file_to_cache_tab): Handle the case where fopen
2674 returns NULL.
2676 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2678 PR target/59929
2679 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2680 from push operand if code of push isn't PRE_DEC.
2682 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2684 PR target/59909
2685 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2686 -mquad-memory-atomic. Update -mquad-memory documentation to say
2687 it is only used for non-atomic loads/stores.
2689 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2690 -mquad-memory or -mquad-memory-atomic switches.
2692 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2693 -mquad-memory-atomic to ISA 2.07 support.
2695 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2696 to separate support of normal quad word memory operations (ldq, stq)
2697 from the atomic quad word memory operations.
2699 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2700 support to separate non-atomic quad word operations from atomic
2701 quad word operations. Disable non-atomic quad word operations in
2702 little endian mode so that we don't have to swap words after the
2703 load and before the store.
2704 (quad_load_store_p): Add comment about atomic quad word support.
2705 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2706 options printed with -mdebug=reg.
2708 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2709 -mquad-memory-atomic as the test for whether we have quad word
2710 atomic instructions.
2711 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2712 or -mp8-vector are used, allow byte/half-word atomic operations.
2714 * config/rs6000/sync.md (load_lockedti): Insure that the address
2715 is a proper indexed or indirect address for the lqarx instruction.
2716 On little endian systems, swap the hi/lo registers after the lqarx
2717 instruction.
2718 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2719 insure the address is valid for the lqarx instruction.
2720 (store_conditionalti): Insure that the address is a proper indexed
2721 or indirect address for the stqcrx. instruction. On little endian
2722 systems, swap the hi/lo registers before doing the stqcrx.
2723 instruction.
2724 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2725 insure the address is valid for the stqcrx. instruction.
2727 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2728 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2729 type of quad memory support is available.
2731 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2733 PR regression/59915
2734 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2735 there is a danger of looping.
2737 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2740 force flag_ira_loop_pressure if set via command line.
2742 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2744 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2745 (ashr_simd): New builtin handling DI mode.
2746 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2747 (aarch64_sshr_simddi): New match pattern.
2748 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2749 (vshrd_n_s64): Likewise.
2750 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2752 2014-01-23 Nick Clifton <nickc@redhat.com>
2754 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2755 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2756 favour of mcu specific scripts.
2757 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2758 430x multilibs.
2760 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2761 Alex Velenko <Alex.Velenko@arm.com>
2763 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2764 (vaddv_s16): Likewise.
2765 (vaddv_s32): Likewise.
2766 (vaddv_u8): Likewise.
2767 (vaddv_u16): Likewise.
2768 (vaddv_u32): Likewise.
2769 (vaddvq_s8): Likewise.
2770 (vaddvq_s16): Likewise.
2771 (vaddvq_s32): Likewise.
2772 (vaddvq_s64): Likewise.
2773 (vaddvq_u8): Likewise.
2774 (vaddvq_u16): Likewise.
2775 (vaddvq_u32): Likewise.
2776 (vaddvq_u64): Likewise.
2777 (vaddv_f32): Likewise.
2778 (vaddvq_f32): Likewise.
2779 (vaddvq_f64): Likewise.
2780 (vmaxv_f32): Likewise.
2781 (vmaxv_s8): Likewise.
2782 (vmaxv_s16): Likewise.
2783 (vmaxv_s32): Likewise.
2784 (vmaxv_u8): Likewise.
2785 (vmaxv_u16): Likewise.
2786 (vmaxv_u32): Likewise.
2787 (vmaxvq_f32): Likewise.
2788 (vmaxvq_f64): Likewise.
2789 (vmaxvq_s8): Likewise.
2790 (vmaxvq_s16): Likewise.
2791 (vmaxvq_s32): Likewise.
2792 (vmaxvq_u8): Likewise.
2793 (vmaxvq_u16): Likewise.
2794 (vmaxvq_u32): Likewise.
2795 (vmaxnmv_f32): Likewise.
2796 (vmaxnmvq_f32): Likewise.
2797 (vmaxnmvq_f64): Likewise.
2798 (vminv_f32): Likewise.
2799 (vminv_s8): Likewise.
2800 (vminv_s16): Likewise.
2801 (vminv_s32): Likewise.
2802 (vminv_u8): Likewise.
2803 (vminv_u16): Likewise.
2804 (vminv_u32): Likewise.
2805 (vminvq_f32): Likewise.
2806 (vminvq_f64): Likewise.
2807 (vminvq_s8): Likewise.
2808 (vminvq_s16): Likewise.
2809 (vminvq_s32): Likewise.
2810 (vminvq_u8): Likewise.
2811 (vminvq_u16): Likewise.
2812 (vminvq_u32): Likewise.
2813 (vminnmv_f32): Likewise.
2814 (vminnmvq_f32): Likewise.
2815 (vminnmvq_f64): Likewise.
2817 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2819 * config/aarch64/aarch64-simd.md
2820 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2821 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2822 (*aarch64_mul3_elt<mode>): Likewise.
2823 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2824 (*aarch64_mul3_elt_to_64v2df): Likewise.
2825 (*aarch64_mla_elt<mode>): Likewise.
2826 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2827 (*aarch64_mls_elt<mode>): Likewise.
2828 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2829 (*aarch64_fma4_elt<mode>): Likewise.
2830 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2831 (*aarch64_fma4_elt_to_64v2df): Likewise.
2832 (*aarch64_fnma4_elt<mode>): Likewise.
2833 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2834 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2835 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2836 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2837 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2838 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2839 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2840 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2841 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2843 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2845 * config/aarch64/aarch64-simd.md
2846 (aarch64_be_checked_get_lane<mode>): New define_expand.
2847 * config/aarch64/aarch64-simd-builtins.def
2848 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2849 New builtin definition.
2850 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2851 Use new safe be builtin.
2853 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2855 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2856 New define_insn.
2857 (aarch64_be_st1<mode>): Likewise.
2858 (aarch_ld1<VALL:mode>): Define_expand modified.
2859 (aarch_st1<VALL:mode>): Likewise.
2860 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2861 (UNSPEC_ST1): Likewise.
2863 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2865 * config/microblaze/microblaze.md: Add trap insn and attribute
2867 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2869 PR preprocessor/58580
2870 * input.h (location_get_source_line): Take an additional line_size
2871 parameter.
2872 (void diagnostics_file_cache_fini): Declare new function.
2873 * input.c (struct fcache): New type.
2874 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2875 New static constants.
2876 (diagnostic_file_cache_init, total_lines_num)
2877 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2878 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2879 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2880 (get_next_line, read_next_line, goto_next_line, read_line_num):
2881 New static function definitions.
2882 (diagnostic_file_cache_fini): New function.
2883 (location_get_source_line): Take an additional output line_len
2884 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2885 read_line_num.
2886 * diagnostic.c (diagnostic_finish): Call
2887 diagnostic_file_cache_fini.
2888 (adjust_line): Take an additional input parameter for the length
2889 of the line, rather than calculating it with strlen.
2890 (diagnostic_show_locus): Adjust the use of
2891 location_get_source_line and adjust_line with respect to their new
2892 signature. While displaying a line now, do not stop at the first
2893 null byte. Rather, display the zero byte as a space and keep
2894 going until we reach the size of the line.
2895 * Makefile.in: Add vec.o to OBJS-libcommon
2897 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2898 Ilya Tocar <ilya.tocar@intel.com>
2900 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2901 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2902 (__builtin_ia32_kmov16): Ditto.
2903 * config/i386/i386.md (UNSPEC_KMOV): New.
2904 (kmovw): Ditto.
2906 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2908 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2909 (_mm512_storeu_si512): Ditto.
2911 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2913 PR target/52125
2914 * rtl.h (get_referenced_operands): Declare.
2915 * recog.c (get_referenced_operands): New function.
2916 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2917 operands have been referenced when recording LO_SUM references.
2919 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2921 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2923 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2925 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2926 Enable for generic and recent AMD targets.
2928 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2930 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2931 ARG_SIZE note when adjustment was eliminated.
2933 2014-01-22 Jeff Law <law@redhat.com>
2935 PR tree-optimization/59597
2936 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2937 in file. Accept new argument REGISTERING and use it to modify
2938 dump output appropriately.
2939 (register_jump_thread): Corresponding changes.
2940 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2941 thread paths involving joiner blocks. Add code to dump cancelled
2942 jump threading paths.
2944 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2946 PR rtl-optimization/59477
2947 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2948 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2950 2014-01-22 Tom Tromey <tromey@redhat.com>
2952 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2953 PARAMS.
2954 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2956 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2958 PR rtl-optimization/59896
2959 * lra-constraints.c (process_alt_operands): Check unused note for
2960 matched operands of insn with no output reloads.
2962 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2964 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2965 (mips_move_from_gpr_cost): Likewise.
2967 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2969 PR rtl-optimization/59858
2970 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2971 ira_class_hard_regs_num.
2972 (process_alt_operands): Increase reject for dying matched operand.
2974 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2976 PR target/59003
2977 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2978 smaller than size, perform several stores or loads and stores
2979 at dst + count - size to store or copy all of size bytes, rather
2980 than just last modesize bytes.
2982 2014-01-20 DJ Delorie <dj@redhat.com>
2984 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2985 that CLOBBERs are REGs before propogating their values.
2987 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2989 PR middle-end/59789
2990 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2991 (cgraph_inline_failed_type): New function.
2992 * cgraph.h (DEFCIFCODE): Add type.
2993 (cgraph_inline_failed_type_t): New enum.
2994 (cgraph_inline_failed_type): New prototype.
2995 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2996 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2997 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2998 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2999 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3000 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3001 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3002 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3003 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3004 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3005 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3006 OPTIMIZATION_MISMATCH.
3007 * tree-inline.c (expand_call_inline): Emit errors during
3008 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3010 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3012 PR target/59685
3013 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3014 mode attribute in insn output.
3016 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3018 * output.h (output_constant): Delete.
3019 * varasm.c (output_constant): Make private.
3021 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3023 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3025 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3027 PR middle-end/59860
3028 * tree.h (fold_builtin_strcat): New prototype.
3029 * builtins.c (fold_builtin_strcat): No longer static. Add len
3030 argument, if non-NULL, don't call c_strlen. Optimize
3031 directly into __builtin_memcpy instead of __builtin_strcpy.
3032 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3033 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3035 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3037 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3038 for SImode_address_operand operands, having only a REG argument.
3040 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3042 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3043 loader name using mbig-endian.
3044 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3046 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3048 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3049 (-mtune): Likewise.
3050 (-mcpu): Likewise.
3052 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3054 * config/aarch64/aarch64-protos.h
3055 (aarch64_cannot_change_mode_class_ptr): Declare.
3056 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3057 aarch64_cannot_change_mode_class_ptr): New.
3058 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3059 backend hook aarch64_cannot_change_mode_class.
3061 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3063 * common/config/aarch64/aarch64-common.c
3064 (aarch64_handle_option): Don't handle any option order logic here.
3065 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3066 selected_cpu, warn on architecture version mismatch.
3067 (aarch64_override_options): Fix parsing order for option strings.
3069 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3070 Iain Sandoe <iain@codesourcery.com>
3072 PR bootstrap/59496
3073 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3074 warning. Amend comment to reflect current functionality.
3076 2014-01-20 Richard Biener <rguenther@suse.de>
3078 PR middle-end/59860
3079 * builtins.c (fold_builtin_strcat): Remove case better handled
3080 by tree-ssa-strlen.c.
3082 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3084 * config/aarch64/aarch64.opt
3085 (mcpu, march, mtune): Make case-insensitive.
3087 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3089 PR target/59880
3090 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3091 if operands[1] is a REG or ZERO_EXTEND of a REG.
3093 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3095 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3097 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3099 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3100 long non-pic millicode calls.
3102 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3104 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3106 2014-01-19 Kito Cheng <kito@0xlab.org>
3108 * builtins.c (expand_movstr): Check movstr expand done or fail.
3110 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3111 H.J. Lu <hongjiu.lu@intel.com>
3113 PR target/59379
3114 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3115 to DImode for zero-extended addresses.
3117 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3119 PR rtl-optimization/57763
3120 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3121 on the new indirect jump_insn and increment LABEL_NUSES (label).
3123 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3125 PR bootstrap/59580
3126 PR bootstrap/59583
3127 * config.gcc (x86_archs): New variable.
3128 (x86_64_archs): Likewise.
3129 (x86_cpus): Likewise.
3130 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3131 --with-arch/--with-cpu= options.
3132 Support --with-arch=/--with-cpu={nehalem,westmere,
3133 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3135 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3137 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3138 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3140 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3142 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3144 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3146 PR target/58944
3147 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3148 clear cpp_get_options (parse_in)->warn_unused_macros for
3149 ix86_target_macros_internal with cpp_define.
3151 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3153 * jump.c (delete_related_insns): Keep (use (insn))s.
3154 * reorg.c (redundant_insn): Check for barriers too.
3156 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3158 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3160 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3162 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3163 call to $$dyncall when TARGET_LONG_CALLS is true.
3165 2014-01-17 Jeff Law <law@redhat.com>
3167 * ree.c (combine_set_extension): Temporarily disable test for
3168 changing number of hard registers.
3170 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3172 PR middle-end/58125
3173 * ipa-inline-analysis.c (inline_free_summary):
3174 Do not free summary of aliases.
3176 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3178 PR middle-end/59706
3179 * gimplify.c (gimplify_expr): Use create_tmp_var
3180 instead of create_tmp_var_raw. If cond doesn't have
3181 integral type, don't add the IFN_ANNOTATE builtin at all.
3183 2014-01-17 Martin Jambor <mjambor@suse.cz>
3185 PR ipa/59736
3186 * ipa-cp.c (prev_edge_clone): New variable.
3187 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3188 Also resize prev_edge_clone vector.
3189 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3190 (ipcp_edge_removal_hook): New function.
3191 (ipcp_driver): Register ipcp_edge_removal_hook.
3193 2014-01-17 Andrew Pinski <apinski@cavium.com>
3194 Steve Ellcey <sellcey@mips.com>
3196 PR target/59462
3197 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3198 of operator mode.
3200 2014-01-17 Jeff Law <law@redhat.com>
3202 PR middle-end/57904
3203 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3204 so that pass_ccp runs first.
3206 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3208 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3209 (ix86_adjust_cost): Use !TARGET_XXX.
3210 (do_reorder_for_imul): Likewise.
3211 (swap_top_of_ready_list): Likewise.
3212 (ix86_sched_reorder): Likewise.
3214 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3216 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3217 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3218 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3219 (intel_memset): New. Duplicate slm_memset.
3220 (intel_cost): New. Duplicate slm_cost.
3221 (m_INTEL): New macro.
3222 (processor_target_table): Add "intel".
3223 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3224 with PROCESSOR_INTEL for "intel".
3225 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3226 PROCESSOR_SILVERMONT.
3227 (ix86_issue_rate): Likewise.
3228 (ix86_adjust_cost): Likewise.
3229 (ia32_multipass_dfa_lookahead): Likewise.
3230 (swap_top_of_ready_list): Likewise.
3231 (ix86_sched_reorder): Likewise.
3232 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3233 instead of TARGET_OPT_AGU.
3234 * config/i386/i386.h (TARGET_INTEL): New.
3235 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3236 (processor_type): Add PROCESSOR_INTEL.
3237 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3238 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3240 2014-01-17 Marek Polacek <polacek@redhat.com>
3242 PR c/58346
3243 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3244 size is zero.
3246 2014-01-17 Richard Biener <rguenther@suse.de>
3248 PR tree-optimization/46590
3249 * opts.c (default_options_table): Add entries for
3250 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3251 all enabled at -O1 but not for -Og.
3252 * common.opt (fbranch-count-reg): Remove Init(1).
3253 (fmove-loop-invariants): Likewise.
3254 (ftree-pta): Likewise.
3256 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3258 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3259 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3260 decls to at least the GCC 4.8 used alignments.
3262 PR fortran/59440
3263 * tree-nested.c (convert_nonlocal_reference_stmt,
3264 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3265 of GIMPLE_BIND stmts, adjust associated decls.
3267 2014-01-17 Richard Biener <rguenther@suse.de>
3269 PR tree-optimization/46590
3270 * vec.h (vec<>::bseach): New member function implementing
3271 binary search according to C89 bsearch.
3272 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3273 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3274 bitmap pointer again. Make accesses_in_loop a flat array.
3275 (mem_ref_obstack): New global.
3276 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3277 (mark_ref_stored): Likewise.
3278 (ref_indep_loop_p_2): Likewise.
3279 (set_ref_stored_in_loop): New helper function.
3280 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3281 (memref_free): Adjust.
3282 (record_mem_ref_loc): Simplify.
3283 (gather_mem_refs_stmt): Adjust.
3284 (sort_locs_in_loop_postorder_cmp): New function.
3285 (analyze_memory_references): Sort accesses_in_loop after
3286 loop postorder number.
3287 (find_ref_loc_in_loop_cmp): New function.
3288 (for_all_locs_in_loop): Find relevant cluster of locs in
3289 accesses_in_loop and iterate without recursion.
3290 (execute_sm): Avoid uninit warning.
3291 (struct ref_always_accessed): Simplify.
3292 (ref_always_accessed::operator ()): Likewise.
3293 (ref_always_accessed_p): Likewise.
3294 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3295 loop postorder numbers here.
3296 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3297 numbers.
3299 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3301 PR c++/57945
3302 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3303 on decls for which assemble_alias has been called.
3305 2014-01-17 Nick Clifton <nickc@redhat.com>
3307 * config/msp430/msp430.opt: (mcpu): New option.
3308 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3309 (msp430_option_override): Parse target_cpu. If the MCU name
3310 matches a generic string, clear target_mcu.
3311 (msp430_attr): Allow numeric interrupt values up to 63.
3312 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3313 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3314 option.
3315 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3316 Add mcpu matches.
3317 * config/msp430/msp430.md (popm): Use %J rather than %I.
3318 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3319 (addhi_cy_i): Use immediate_operand for operand 2.
3320 * doc/invoke.texi: Document -mcpu option.
3322 2014-01-17 Richard Biener <rguenther@suse.de>
3324 PR rtl-optimization/38518
3325 * df.h (df_analyze_loop): Declare.
3326 * df-core.c: Include cfgloop.h.
3327 (df_analyze_1): Split out main part of df_analyze.
3328 (df_analyze): Adjust.
3329 (loop_inverted_post_order_compute): New function.
3330 (loop_post_order_compute): Likewise.
3331 (df_analyze_loop): New function avoiding whole-function
3332 postorder computes.
3333 * loop-invariant.c (find_defs): Use df_analyze_loop.
3334 (find_invariants): Adjust.
3335 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3337 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3339 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3340 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3342 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3344 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3345 traversal when removing references.
3347 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3349 PR ipa/59775
3350 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3352 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3354 PR middle-end/56791
3355 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3356 pushing a reload for an autoinc when we had previously reloaded an
3357 inner part of the address.
3359 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3361 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3362 field.
3363 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3364 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3365 when not giving up or versioning for alias only because of
3366 loop->safelen.
3367 (vect_analyze_data_ref_dependences): Set to true.
3368 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3369 is a GIMPLE_PHI.
3370 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3371 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3372 to the condition.
3374 PR middle-end/58344
3375 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3377 PR target/59839
3378 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3379 operand 0 predicate for gathers, use a new pseudo as subtarget.
3381 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3383 PR middle-end/59609
3384 * lra-constraints.c (process_alt_operands): Add printing debug info.
3385 Check absence of input/output reloads for matched operands too.
3387 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3389 PR rtl-optimization/59835
3390 * ira.c (ira_init_register_move_cost): Increase cost for
3391 impossible modes.
3393 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3395 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3397 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3399 PR target/59780
3400 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3401 non-register objects. Use gen_(high/low)part more consistently.
3402 Fix assertions.
3404 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3406 PR target/59844
3407 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3408 endian support, remove tests for WORDS_BIG_ENDIAN.
3409 (p8_mfvsrd_3_<mode>): Likewise.
3410 (reload_gpr_from_vsx<mode>): Likewise.
3411 (reload_gpr_from_vsxsf): Likewise.
3412 (p8_mfvsrd_4_disf): Likewise.
3414 2014-01-16 Richard Biener <rguenther@suse.de>
3416 PR rtl-optimization/46590
3417 * lcm.c (compute_antinout_edge): Use postorder iteration.
3418 (compute_laterin): Use inverted postorder iteration.
3420 2014-01-16 Nick Clifton <nickc@redhat.com>
3422 PR middle-end/28865
3423 * varasm.c (output_constant): Return the number of bytes actually
3424 emitted.
3425 (output_constructor_array_range): Update the field size with the
3426 number of bytes emitted by output_constant.
3427 (output_constructor_regular_field): Likewise. Also do not
3428 complain if the total number of bytes emitted is now greater
3429 than the expected fieldpos.
3430 * output.h (output_constant): Update prototype and descriptive comment.
3432 2014-01-16 Marek Polacek <polacek@redhat.com>
3434 PR middle-end/59827
3435 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3436 it is error_mark_node.
3438 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3440 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3441 VALID_AVX256_REG_OR_OI_MODE.
3443 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3445 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3446 current procedure should be profiled.
3448 2014-01-15 Andrew Pinski <apinski@cavium.com>
3450 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3451 of moving from/to the STACK_REG register class.
3453 2014-01-15 Richard Henderson <rth@redhat.com>
3455 PR debug/54694
3456 * reginfo.c (global_regs_decl): Globalize.
3457 * rtl.h (global_regs_decl): Declare.
3458 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3459 reserved via global_regs.
3461 2014-01-15 Teresa Johnson <tejohnson@google.com>
3463 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3465 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3467 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3468 and vmulosh rather than call gen_vec_widen_smult_*.
3469 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3470 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3471 (vec_widen_smult_even_v16qi): Likewise.
3472 (vec_widen_umult_even_v8hi): Likewise.
3473 (vec_widen_smult_even_v8hi): Likewise.
3474 (vec_widen_umult_odd_v16qi): Likewise.
3475 (vec_widen_smult_odd_v16qi): Likewise.
3476 (vec_widen_umult_odd_v8hi): Likewise.
3477 (vec_widen_smult_odd_v8hi): Likewise.
3478 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3479 vmuloub rather than call gen_vec_widen_umult_*.
3480 (vec_widen_umult_lo_v16qi): Likewise.
3481 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3482 vmulosb rather than call gen_vec_widen_smult_*.
3483 (vec_widen_smult_lo_v16qi): Likewise.
3484 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3485 rather than call gen_vec_widen_umult_*.
3486 (vec_widen_umult_lo_v8hi): Likewise.
3487 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3488 rather than call gen_vec_widen_smult_*.
3489 (vec_widen_smult_lo_v8hi): Likewise.
3491 2014-01-15 Jeff Law <law@redhat.com>
3493 PR tree-optimization/59747
3494 * ree.c (find_and_remove_re): Properly handle case where a second
3495 eliminated extension requires widening a copy created for elimination
3496 of a prior extension.
3497 (combine_set_extension): Ensure that the number of hard regs needed
3498 for a destination register does not change when we widen it.
3500 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3502 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3503 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3504 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3505 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3506 (avr-*-rtems*): Likewise.
3507 (bfin*-rtems*): Likewise.
3508 (moxie-*-rtems*): Likewise.
3509 (h8300-*-rtems*): Likewise.
3510 (i[34567]86-*-rtems*): Likewise.
3511 (lm32-*-rtems*): Likewise.
3512 (m32r-*-rtems*): Likewise.
3513 (m68k-*-rtems*): Likewise.
3514 (microblaze*-*-rtems*): Likewise.
3515 (mips*-*-rtems*): Likewise.
3516 (powerpc-*-rtems*): Likewise.
3517 (sh-*-rtems*): Likewise.
3518 (sparc-*-rtems*): Likewise.
3519 (sparc64-*-rtems*): Likewise.
3520 (v850-*-rtems*): Likewise.
3521 (m32c-*-rtems*): Likewise.
3523 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3525 PR rtl-optimization/59511
3526 * ira.c (ira_init_register_move_cost): Use memory costs for some
3527 cases of register move cost calculations.
3528 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3529 instead of BB frequency.
3530 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3531 * lra-assigns.c (find_hard_regno_for): Ditto.
3533 2014-01-15 Richard Biener <rguenther@suse.de>
3535 PR tree-optimization/59822
3536 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3537 (vectorizable_load): Use it to hoist defs of uses of invariant
3538 loads out of the loop.
3540 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3541 Kugan Vivekanandarajah <kuganv@linaro.org>
3543 PR target/59695
3544 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3545 truncation.
3547 2014-01-15 Richard Biener <rguenther@suse.de>
3549 PR rtl-optimization/59802
3550 * lcm.c (compute_available): Use inverted postorder to seed
3551 the initial worklist.
3553 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3555 PR target/59803
3556 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3557 ADDR_REGS for invalid symrefs in non-PIC code.
3559 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3561 PR other/58712
3562 * builtins.c (determine_block_size): Initialize *probable_max_size
3563 even if len_rtx is CONST_INT.
3565 2014-01-14 Andrew Pinski <apinski@cavium.com>
3567 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3568 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3569 (cortexa53_tunings): Likewise.
3570 (aarch64_sched_issue_rate): New function.
3571 (TARGET_SCHED_ISSUE_RATE): Define.
3573 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3575 * ira-costs.c (find_costs_and_classes): Add missed
3576 ira_init_register_move_cost_if_necessary.
3578 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3580 PR target/59787
3581 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3583 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3585 PR target/59794
3586 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3587 to indicate if type is used for function return value. Warn ABI
3588 change if the vector mode isn't available for function return value.
3589 (ix86_function_arg_advance): Pass false to type_natural_mode.
3590 (ix86_function_arg): Likewise.
3591 (ix86_gimplify_va_arg): Likewise.
3592 (function_arg_32): Don't warn ABI change.
3593 (ix86_function_value): Pass true to type_natural_mode.
3594 (ix86_return_in_memory): Likewise.
3595 (ix86_struct_value_rtx): Removed.
3596 (TARGET_STRUCT_VALUE_RTX): Likewise.
3598 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3600 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3601 converting a conditional jump into a conditional return.
3603 2014-01-14 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/58921
3606 PR tree-optimization/59006
3607 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3608 hoisting invariant stmts.
3609 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3610 invariant loads on the preheader edge if possible.
3612 2014-01-14 Joey Ye <joey.ye@arm.com>
3614 * doc/plugin.texi (Building GCC plugins): Update to C++.
3616 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3618 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3619 (_mm_rcp28_round_ss): Ditto.
3620 (_mm_rsqrt28_round_sd): Ditto.
3621 (_mm_rsqrt28_round_ss): Ditto.
3622 (_mm_rcp28_sd): Ditto.
3623 (_mm_rcp28_ss): Ditto.
3624 (_mm_rsqrt28_sd): Ditto.
3625 (_mm_rsqrt28_ss): Ditto.
3626 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3627 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3628 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3629 (IX86_BUILTIN_RCP28SD): Ditto.
3630 (IX86_BUILTIN_RCP28SS): Ditto.
3631 (IX86_BUILTIN_RSQRT28SD): Ditto.
3632 (IX86_BUILTIN_RSQRT28SS): Ditto.
3633 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3634 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3635 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3636 (ix86_expand_special_args_builtin): Expand new FTYPE.
3637 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3638 (srcp14<mode>): Make insn unary.
3639 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3640 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3641 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3642 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3643 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3644 Fix rounding: make it SAE only.
3645 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3646 Ditto.
3647 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3648 Ditto.
3649 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3650 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3651 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3652 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3653 (round_saeonly_mask_scalar_operand4): Ditto.
3654 (round_saeonly_mask_scalar_op3): Ditto.
3655 (round_saeonly_mask_scalar_op4): Ditto.
3657 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3659 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3660 Implement -maltivec=be for vec_insert and vec_extract.
3662 2014-01-10 DJ Delorie <dj@redhat.com>
3664 * config/msp430/msp430.md (call_internal): Don't allow memory
3665 references with SP as the base register.
3666 (call_value_internal): Likewise.
3667 * config/msp430/constraints.md (Yc): New. For memory references
3668 that don't use SP as a base register.
3670 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3671 "an integer without a # prefix"
3672 * config/msp430/msp430.md (epilogue_helper): Use it.
3674 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3676 PR target/59617
3677 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3678 AVX512F gather builtins.
3679 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3680 on gather decls with INTEGER_TYPE masktype.
3681 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3682 directly into the builtin rather than hoisting it before loop.
3684 PR tree-optimization/59387
3685 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3686 (scev_const_prop): If folded_casts and type has undefined overflow,
3687 use force_gimple_operand instead of force_gimple_operand_gsi and
3688 for each added stmt if it is assign with
3689 arith_code_with_undefined_signed_overflow, call
3690 rewrite_to_defined_overflow.
3691 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3692 gimple-fold.h instead.
3693 (arith_code_with_undefined_signed_overflow,
3694 rewrite_to_defined_overflow): Moved to ...
3695 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3696 rewrite_to_defined_overflow): ... here. No longer static.
3697 Include gimplify-me.h.
3698 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3699 rewrite_to_defined_overflow): New prototypes.
3701 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3703 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3705 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3707 * builtins.c (get_object_alignment_2): Minor tweak.
3708 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3710 2014-01-13 Christian Bruel <christian.bruel@st.com>
3712 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3713 optimized non constant lengths.
3715 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3717 PR libgomp/59194
3718 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3719 load as __atomic_load_N if possible.
3721 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3723 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3724 target parameter.
3725 (rs6000_expand_builtin): Adjust call.
3727 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3729 PR target/58115
3730 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3731 * config/rs6000/rs6000.c: Include target-globals.h.
3732 (rs6000_set_current_function): Instead of doing target_reinit
3733 unconditionally, use save_target_globals_default_opts and
3734 restore_target_globals.
3736 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3737 FPSCR.
3738 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3739 (rs6000_expand_builtin): Handle mffs and mtfsf.
3740 (rs6000_init_builtins): Define mffs and mtfsf.
3741 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3742 (rs6000_mffs): New pattern.
3743 (rs6000_mtfsf): New pattern.
3745 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3747 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3748 Start narrowing with START. Apply candidate-use pair
3749 and check overall cost in narrowing.
3750 (iv_ca_prune): Pass new argument.
3752 2014-01-10 Jeff Law <law@redhat.com>
3754 PR middle-end/59743
3755 * ree.c (combine_reaching_defs): Ensure the defining statement
3756 occurs before the extension when optimizing extensions with
3757 different source and destination hard registers.
3759 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3761 PR ipa/58585
3762 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3763 vtables into the type inheritance graph.
3765 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3767 PR rtl-optimization/59754
3768 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3769 modes in the REGNO != REGNO case.
3771 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3773 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3775 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3777 PR tree-optimization/59745
3778 * tree-predcom.c (tree_predictive_commoning_loop): Call
3779 free_affine_expand_cache if giving up because components is NULL.
3781 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3782 GC in payload of target_globals struct instead of allocating them on
3783 the heap and the larger structs separately using GC.
3784 * target-globals.h (struct target_globals): Make regs, hard_regs,
3785 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3786 of GTY((skip)) and change type to void *.
3787 (reset_target_globals): Cast loads from those fields to corresponding
3788 types.
3790 2014-01-10 Steve Ellcey <sellcey@mips.com>
3792 PR plugins/59335
3793 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3794 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3795 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3797 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3799 PR target/59744
3800 * aarch64-modes.def (CC_Zmode): New flags mode.
3801 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3802 represents an equality.
3803 (aarch64_get_condition_code): Handle CC_Zmode.
3804 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3806 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3808 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3809 extraction in good case.
3811 2014-01-10 Richard Biener <rguenther@suse.de>
3813 PR tree-optimization/59374
3814 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3815 checking after SLP discovery. Mark stmts not participating
3816 in any SLP instance properly.
3818 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3820 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3821 when handling a SET rtx.
3823 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3825 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3826 (cortex-a57): Likewise.
3827 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3829 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3831 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3832 non-iwmmxt builtins.
3834 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3836 PR ipa/58252
3837 PR ipa/59226
3838 * ipa-devirt.c record_target_from_binfo): Take as argument
3839 stack of binfos and lookup matching one for virtual inheritance.
3840 (possible_polymorphic_call_targets_1): Update.
3842 2014-01-10 Huacai Chen <chenhc@lemote.com>
3844 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3845 kernel strings for Loongson-2E/2F/3A.
3847 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3849 PR middle-end/59670
3850 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3851 is_gimple_call before calling gimple_call_internal_p.
3853 2014-01-09 Steve Ellcey <sellcey@mips.com>
3855 * Makefile.in (TREE_FLOW_H): Remove.
3856 (TREE_SSA_H): Add file names from tree-flow.h.
3857 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3858 * tree.h: Remove tree-flow.h reference.
3859 * hash-table.h: Remove tree-flow.h reference.
3860 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3861 reference with tree-ssa-loop.h.
3863 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3865 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3866 default element-order behavior for -maltivec.
3867 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3868 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3869 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3870 when targeting big endian, at least for now.
3871 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3873 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3875 PR middle-end/47735
3876 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3877 var satisfies use_register_for_decl, just take into account type
3878 alignment, rather than decl alignment.
3880 PR tree-optimization/59622
3881 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3882 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3883 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3884 Don't devirtualize for inplace at all. For targets.length () == 1,
3885 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3887 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3889 * config/i386/i386.md (cpu): Remove the unused btver1.
3891 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3893 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3895 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3897 PR target/58115
3898 * tree-core.h (struct target_globals): New forward declaration.
3899 (struct tree_target_option): Add globals field.
3900 * tree.h (TREE_TARGET_GLOBALS): Define.
3901 (prepare_target_option_nodes_for_pch): New prototype.
3902 * target-globals.h (struct target_globals): Define even if
3903 !SWITCHABLE_TARGET.
3904 * tree.c (prepare_target_option_node_for_pch,
3905 prepare_target_option_nodes_for_pch): New functions.
3906 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3907 * config/i386/i386.c: Include target-globals.h.
3908 (ix86_set_current_function): Instead of doing target_reinit
3909 unconditionally, use save_target_globals_default_opts and
3910 restore_target_globals.
3912 2014-01-09 Richard Biener <rguenther@suse.de>
3914 PR tree-optimization/59715
3915 * tree-cfg.h (split_critical_edges): Declare.
3916 * tree-cfg.c (split_critical_edges): Export.
3917 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3919 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3921 * cfgexpand.c (expand_stack_vars): Optionally disable
3922 asan stack protection.
3923 (expand_used_vars): Likewise.
3924 (partition_stack_vars): Likewise.
3925 * asan.c (asan_emit_stack_protection): Optionally disable
3926 after return stack usage.
3927 (instrument_derefs): Optionally disable memory access instrumentation.
3928 (instrument_builtin_call): Likewise.
3929 (instrument_strlen_call): Likewise.
3930 (asan_protect_global): Optionally disable global variables protection.
3931 * doc/invoke.texi: Added doc for new options.
3932 * params.def: Added new options.
3933 * params.h: Likewise.
3935 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3937 PR rtl-optimization/59724
3938 * ifcvt.c (cond_exec_process_if_block): Don't call
3939 flow_find_head_matching_sequence with 0 longest_match.
3940 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3941 non-active insns if !stop_after.
3942 (try_head_merge_bb): Revert 2014-01-07 changes.
3944 2014-01-08 Jeff Law <law@redhat.com>
3946 * ree.c (get_sub_rtx): New function, extracted from...
3947 (merge_def_and_ext): Here.
3948 (combine_reaching_defs): Use get_sub_rtx.
3950 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3952 * cgraph.h (varpool_variable_node): Do not choke on null node.
3954 2014-01-08 Catherine Moore <clm@codesourcery.com>
3956 * config/mips/mips.md (simple_return): Attempt to use JRC
3957 for microMIPS.
3958 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3960 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3962 PR rtl-optimization/59137
3963 * reorg.c (steal_delay_list_from_target): Call update_block for
3964 elided insns.
3965 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3967 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3969 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3970 two duplicate entries.
3972 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3974 Revert:
3975 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3977 * config/mips/mips.c (mips_truncated_op_cost): New function.
3978 (mips_rtx_costs): Adjust test for BADDU.
3979 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3981 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3983 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3984 (*baddu_si): ...this new pattern.
3986 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3988 PR ipa/59722
3989 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3991 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3993 PR middle-end/57748
3994 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3995 inner_reference_p.
3996 (expand_expr, expand_normal): Adjust.
3997 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3998 inner_reference_p. Use inner_reference_p to expand inner references.
3999 (store_expr): Adjust.
4000 * cfgexpand.c (expand_call_stmt): Adjust.
4002 2014-01-08 Rong Xu <xur@google.com>
4004 * gcov-io.c (gcov_var): Move from gcov-io.h.
4005 (gcov_position): Ditto.
4006 (gcov_is_error): Ditto.
4007 (gcov_rewrite): Ditto.
4008 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4009 only part to libgcc/libgcov.h.
4011 2014-01-08 Marek Polacek <polacek@redhat.com>
4013 PR middle-end/59669
4014 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4016 2014-01-08 Marek Polacek <polacek@redhat.com>
4018 PR sanitizer/59667
4019 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4021 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4023 PR rtl-optimization/59649
4024 * stor-layout.c (get_mode_bounds): For BImode return
4025 0 and STORE_FLAG_VALUE.
4027 2014-01-08 Richard Biener <rguenther@suse.de>
4029 PR middle-end/59630
4030 * gimple.h (is_gimple_builtin_call): Remove.
4031 (gimple_builtin_call_types_compatible_p): New.
4032 (gimple_call_builtin_p): New overload.
4033 * gimple.c (is_gimple_builtin_call): Remove.
4034 (validate_call): Rename to ...
4035 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4036 check return types.
4037 (validate_type): New static function.
4038 (gimple_call_builtin_p): New overload and adjust.
4039 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4040 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4041 (gimple_fold_stmt_to_constant_1): Likewise.
4042 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4044 2014-01-08 Richard Biener <rguenther@suse.de>
4046 PR middle-end/59471
4047 * gimplify.c (gimplify_expr): Gimplify register-register type
4048 VIEW_CONVERT_EXPRs to separate stmts.
4050 2014-01-07 Jeff Law <law@redhat.com>
4052 PR middle-end/53623
4053 * ree.c (combine_set_extension): Handle case where source
4054 and destination registers in an extension insn are different.
4055 (combine_reaching_defs): Allow source and destination registers
4056 in extension to be different under limited circumstances.
4057 (add_removable_extension): Remove restriction that the
4058 source and destination registers in the extension are the same.
4059 (find_and_remove_re): Emit a copy from the extension's
4060 destination to its source after the defining insn if
4061 the source and destination registers are different.
4063 PR middle-end/59285
4064 * ifcvt.c (merge_if_block): If we are merging a block with more than
4065 one successor with a block with no successors, remove any BARRIER
4066 after the second block.
4068 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4070 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4072 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4074 PR target/59652
4075 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4076 for 14-bit register offsets when INT14_OK_STRICT is false.
4078 2014-01-07 Roland Stigge <stigge@antcom.de>
4079 Michael Meissner <meissner@linux.vnet.ibm.com>
4081 PR 57386/target
4082 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4083 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4085 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4087 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4088 -mcpu.
4090 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4092 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4093 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4094 rtx is const0_rtx or not.
4096 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4098 PR target/58115
4099 * target-globals.c (save_target_globals): Remove this_fn_optab
4100 handling.
4101 * toplev.c: Include optabs.h.
4102 (target_reinit): Temporarily restore the global options if another
4103 set of options are in force.
4105 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4107 PR rtl-optimization/58668
4108 * cfgcleanup.c (flow_find_cross_jump): Don't count
4109 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4110 to determine what is counted.
4111 (flow_find_head_matching_sequence): Use active_insn_p to determine
4112 what is counted.
4113 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4114 counting change.
4115 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4116 determine what is counted.
4118 PR tree-optimization/59643
4119 * tree-predcom.c (split_data_refs_to_components): If one dr is
4120 read and one write, determine_offset fails and the write isn't
4121 in the bad component, just put the read into the bad component.
4123 2014-01-07 Mike Stump <mikestump@comcast.net>
4124 Jakub Jelinek <jakub@redhat.com>
4126 PR pch/59436
4127 * tree-core.h (struct tree_optimization_option): Change optabs
4128 type from unsigned char * to void *.
4129 * optabs.c (init_tree_optimization_optabs): Adjust
4130 TREE_OPTIMIZATION_OPTABS initialization.
4132 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4134 PR target/59644
4135 * config/i386/i386.h (struct machine_function): Add
4136 no_drap_save_restore field.
4137 * config/i386/i386.c (ix86_save_reg): Use
4138 !cfun->machine->no_drap_save_restore instead of
4139 crtl->stack_realign_needed.
4140 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4141 this function clears frame_pointer_needed. Set
4142 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4143 and DRAP reg is needed.
4145 2014-01-06 Marek Polacek <polacek@redhat.com>
4147 PR c/57773
4148 * doc/implement-c.texi: Mention that other integer types are
4149 permitted as bit-field types in strictly conforming mode.
4151 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4153 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4154 is newly allocated.
4156 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4158 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4160 2014-01-06 Martin Jambor <mjambor@suse.cz>
4162 PR ipa/59008
4163 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4164 to int.
4165 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4167 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4169 PR debug/59350
4170 PR debug/59510
4171 * var-tracking.c (add_stores): Preserve the value of the source even if
4172 we don't record the store.
4174 2014-01-06 Terry Guo <terry.guo@arm.com>
4176 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4178 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4180 PR bootstrap/59541
4181 * config/darwin.c (darwin_function_section): Adjust return values to
4182 correspond to optimisation changes made in r206070.
4184 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4186 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4187 from prefetch_block tune setting.
4188 (nocona_cost): Correct size of prefetch block to 64.
4190 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4192 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4193 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4194 used to save the static chain register in the computation of the offset
4195 from which the FP registers need to be restored.
4197 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4199 PR tree-optimization/59519
4200 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4201 ICE if get_current_def (current_new_name) is already non-NULL, as long
4202 as it is a phi result of some other phi in *new_exit_bb that has
4203 the same argument.
4205 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4206 or vmovdqu* for misaligned_operand.
4207 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4208 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4209 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4210 aligned_mem for AVX512F masked aligned load and store builtins and for
4211 non-temporal moves.
4213 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4215 PR tree-optimization/59651
4216 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4217 Address range for negative step should be added by TYPE_SIZE_UNIT.
4219 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4221 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4222 overlapping registers also for registers other than the stack pointer.
4224 2014-01-03 Marek Polacek <polacek@redhat.com>
4226 PR other/59661
4227 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4228 __builtin_FILE.
4230 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4232 PR target/59625
4233 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4234 asm goto as jump.
4236 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4237 (push splitter): Use <P:MODE_SIZE> instead of
4238 GET_MODE_SIZE (<P:MODE>mode).
4239 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4240 (mov -1, reg peephole2): Likewise.
4241 * config/i386/sse.md (*mov<mode>_internal,
4242 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4243 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4244 *<code><mode>3, *andnot<mode>3<mask_name>,
4245 <mask_codefor><code><mode>3<mask_name>): Likewise.
4246 * config/i386/subst.md (mask_mode512bit_condition,
4247 sd_mask_mode512bit_condition): Likewise.
4249 2014-01-02 Xinliang David Li <davidxl@google.com>
4251 PR tree-optimization/59303
4252 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4253 (dump_predicates): Better output format.
4254 (pred_equal_p): New function.
4255 (is_neq_relop_p): Ditto.
4256 (is_neq_zero_form_p): Ditto.
4257 (pred_expr_equal_p): Ditto.
4258 (pred_neg_p): Ditto.
4259 (simplify_pred): Ditto.
4260 (simplify_preds_2): Ditto.
4261 (simplify_preds_3): Ditto.
4262 (simplify_preds_4): Ditto.
4263 (simplify_preds): Ditto.
4264 (push_pred): Ditto.
4265 (push_to_worklist): Ditto.
4266 (get_pred_info_from_cmp): Ditto.
4267 (is_degenerated_phi): Ditto.
4268 (normalize_one_pred_1): Ditto.
4269 (normalize_one_pred): Ditto.
4270 (normalize_one_pred_chain): Ditto.
4271 (normalize_preds): Ditto.
4272 (normalize_cond_1): Remove function.
4273 (normalize_cond): Ditto.
4274 (is_gcond_subset_of): Ditto.
4275 (is_subset_of_any): Ditto.
4276 (is_or_set_subset_of): Ditto.
4277 (is_and_set_subset_of): Ditto.
4278 (is_norm_cond_subset_of): Ditto.
4279 (pred_chain_length_cmp): Ditto.
4280 (convert_control_dep_chain_into_preds): Type change.
4281 (find_predicates): Ditto.
4282 (find_def_preds): Ditto.
4283 (destroy_predicates_vecs): Ditto.
4284 (find_matching_predicates_in_rest_chains): Ditto.
4285 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4286 (is_pred_expr_subset): Ditto.
4287 (is_pred_chain_subset_of): Ditto.
4288 (is_included_in): Ditto.
4289 (is_superset_of): Ditto.
4291 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4293 Update copyright years.
4295 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4297 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4298 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4299 config/arc/arc.md, config/arc/arc.opt,
4300 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4301 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4302 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4303 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4304 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4305 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4306 vtable-verify.c, vtable-verify.h: Use the standard form for the
4307 copyright notice.
4309 2014-01-02 Tobias Burnus <burnus@net-b.de>
4311 * gcc.c (process_command): Update copyright notice dates.
4312 * gcov-dump.c: Ditto.
4313 * gcov.c: Ditto.
4314 * doc/cpp.texi: Bump @copying's copyright year.
4315 * doc/cppinternals.texi: Ditto.
4316 * doc/gcc.texi: Ditto.
4317 * doc/gccint.texi: Ditto.
4318 * doc/gcov.texi: Ditto.
4319 * doc/install.texi: Ditto.
4320 * doc/invoke.texi: Ditto.
4322 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4324 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4326 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4328 * config/i386/sse.md (*mov<mode>_internal): Guard
4329 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4331 PR rtl-optimization/59647
4332 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4333 new_rtx into UNSIGNED_FLOAT rtxes.
4334 \f
4335 Copyright (C) 2014 Free Software Foundation, Inc.
4337 Copying and distribution of this file, with or without modification,
4338 are permitted in any medium without royalty provided the copyright
4339 notice and this notice are preserved.