1 2014-02-25 Richard Biener <rguenther@suse.de>
4 * emit-rtl.c (mem_attrs_htab): Remove.
5 (mem_attrs_htab_hash): Likewise.
6 (mem_attrs_htab_eq): Likewise.
7 (set_mem_attrs): Always allocate new mem-attrs when something
9 (init_emit_once): Do not allocate mem_attrs_htab.
11 2014-02-25 Richard Biener <rguenther@suse.de>
14 * lto-opts.c (lto_write_options): Output non-explicit conservative
15 -fwrapv, -fno-trapv and -fno-strict-overflow.
16 * lto-wrapper.c (merge_and_complain): Handle merging those options.
17 (run_gcc): And pass them through.
19 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
21 * sel-sched.c (calculate_new_fences): New parameter ptime.
22 Calculate it as a maximum over all fence cycles.
23 (sel_sched_region_2): Adjust the call to calculate_new_fences.
24 Print the final schedule timing when sched_verbose.
26 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
28 PR rtl-optimization/60292
29 * sel-sched.c (fill_vec_av_set): Do not reset target availability
30 bit fot the fence instruction.
32 2014-02-24 Alangi Derick <alangiderick@gmail.com>
34 * calls.h: Fix typo in comment.
36 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
38 * config/pa/pa.c (pa_output_move_double): Don't valididate when
39 adjusting offsetable addresses.
41 2014-02-24 Guozhi Wei <carrot@google.com>
43 * sparseset.h (sparseset_pop): Fix the wrong index.
45 2014-02-24 Walter Lee <walt@tilera.com>
47 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
48 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
50 * common/config/tilegx/tilegx-common.c
51 (TARGET_DEFAULT_TARGET_FLAGS): Define.
52 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
54 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
55 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
56 (tilegx_gimplify_va_arg_expr): Handle big endian.
57 (tilegx_expand_unaligned_load): Ditto.
58 (tilegx_expand_unaligned_store): Ditto.
59 (TARGET_RETURN_IN_MSB): New.
60 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
61 (TARGET_ENDIAN_DEFAULT): New.
62 (TARGET_BIG_ENDIAN): Handle big endian.
63 (BYTES_BIG_ENDIAN): Ditto.
64 (WORDS_BIG_ENDIAN): Ditto.
65 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
68 * config/tilegx/tilegx.md (extv): Handle big endian.
71 (insn_st<n>_add<bitsuffix>): Ditto.
72 (insn_stnt<n>): Ditto.
73 (insn_stnt<n>_add<bitsuffix>):Ditto.
74 (vec_interleave_highv8qi): Handle big endian.
75 (vec_interleave_highv8qi_be): New.
76 (vec_interleave_highv8qi_le): New.
77 (insn_v1int_h): Handle big endian.
78 (vec_interleave_lowv8qi): Handle big endian.
79 (vec_interleave_lowv8qi_be): New.
80 (vec_interleave_lowv8qi_le): New.
81 (insn_v1int_l): Handle big endian.
82 (vec_interleave_highv4hi): Handle big endian.
83 (vec_interleave_highv4hi_be): New.
84 (vec_interleave_highv4hi_le): New.
85 (insn_v2int_h): Handle big endian.
86 (vec_interleave_lowv4hi): Handle big endian.
87 (vec_interleave_lowv4hi_be): New.
88 (vec_interleave_lowv4hi_le): New.
89 (insn_v2int_l): Handle big endian.
90 (vec_interleave_highv2si): Handle big endian.
91 (vec_interleave_highv2si_be): New.
92 (vec_interleave_highv2si_le): New.
93 (insn_v4int_h): Handle big endian.
94 (vec_interleave_lowv2si): Handle big endian.
95 (vec_interleave_lowv2si_be): New.
96 (vec_interleave_lowv2si_le): New.
97 (insn_v4int_l): Handle big endian.
98 * config/tilegx/tilegx.opt (mbig-endian): New option.
99 (mlittle-endian): New option.
100 * doc/install.texi: Document tilegxbe-linux.
101 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
103 2014-02-24 Martin Jambor <mjambor@suse.cz>
106 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
107 there are no parameter descriptors.
109 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
111 PR rtl-optimization/60268
112 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
113 initialization to ...
114 (sched_rgn_init): ... here.
115 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
117 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
119 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
122 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
124 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
127 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
129 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk
130 and define TARGET_ASM_OUTPUT_MI_THUNK and
131 TARGET_ASM_CAN_OUTPUT_MI_THUNK.
133 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
135 * config/microblaze/predicates.md: Add cmp_op predicate.
136 * config/microblaze/microblaze.md: Add branch_compare instruction
137 which uses cmp_op predicate and emits cmp insn before branch.
138 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
139 to microblaze_expand_conditional_branch and consolidate logic.
140 (microblaze_expand_conditional_branch): emit branch_compare
141 insn instead of handling cmp op separate from branch insn.
143 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
145 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
148 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
150 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
151 define_insn with define_expand and new define_insn
152 *altivec_lve<VI_char>x_internal.
153 (altivec_stve<VI_char>x): Replace define_insn with define_expand
154 and new define_insn *altivec_stve<VI_char>x_internal.
155 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
157 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
159 (altivec_expand_stvex_be): New function.
161 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
163 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
165 to STACK_POINTER_REGNUM if !frame_pointer_needed.
166 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
167 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
169 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
172 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
173 instead of emit_move_insn.
175 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
177 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
179 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
182 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
184 * config/rs6000/altivec.md (altivec_lvxl): Rename as
185 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
187 (altivec_lvxl_<mode>): New define_expand incorporating
188 -maltivec=be semantics where needed.
189 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
190 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
191 semantics where needed.
192 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
193 (altivec_stvx_<mode>): New define_expand incorporating
194 -maltivec=be semantics where needed.
195 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
196 VM2 iterator instead of V4SI.
197 (altivec_stvxl_<mode>): New define_expand incorporating
198 -maltivec=be semantics where needed.
199 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
200 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
201 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
202 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
203 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
205 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
206 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
207 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
208 ALTIVEC_BUILTIN_STVXL.
209 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
211 (altivec_expand_stvx_be): Likewise.
212 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
213 (altivec_expand_lvx_be): Likewise.
214 (altivec_expand_stvx_be): Likewise.
215 (altivec_expand_builtin): Add cases for
216 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
217 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
218 (altivec_init_builtins): Add definitions for
219 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
220 __builtin_altivec_stvx_<mode>, and
221 __builtin_altivec_stvxl_<mode>.
223 2014-02-21 Catherine Moore <clm@codesourcery.com>
225 * doc/invoke.texi (mvirt, mno-virt): Document.
226 * config/mips/mips.opt (mvirt): New option.
227 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
229 2014-02-21 Richard Biener <rguenther@suse.de>
231 PR tree-optimization/60276
232 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
233 (STMT_VINFO_MIN_NEG_DIST): New macro.
234 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
235 STMT_VINFO_MIN_NEG_DIST.
236 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
237 made for negative dependence distances still hold.
239 2014-02-21 Richard Biener <rguenther@suse.de>
242 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
243 DECL_INITIAL for globals not in the current function context.
245 2014-02-21 Jakub Jelinek <jakub@redhat.com>
247 PR tree-optimization/56490
248 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
249 * tree-ssa-uninit.c: Include params.h.
250 (compute_control_dep_chain): Add num_calls argument, return false
251 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
252 num_calls to recursive call.
253 (find_predicates): Change dep_chain into normal array,
254 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
255 variable and adjust compute_control_dep_chain caller.
256 (find_def_preds): Likewise.
258 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
260 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
261 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
263 2014-02-21 Nick Clifton <nickc@redhat.com>
265 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
267 (popqi1): Add mode to pre_dec.
270 2014-02-21 Jakub Jelinek <jakub@redhat.com>
272 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
273 mode for mask of V8SFmode permutation.
275 2014-02-20 Richard Henderson <rth@redhat.com>
278 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
279 a new pseudo for OLDVAL.
281 2014-02-20 Jakub Jelinek <jakub@redhat.com>
284 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
285 gen_reg_rtx if d->testing_p.
286 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
287 if d->testing_p and we will certainly return true.
288 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
291 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
293 * emit-rtl.c (gen_reg_rtx): Assert that
294 crtl->emit.regno_pointer_align_length is non-zero.
296 2014-02-20 Richard Henderson <rth@redhat.com>
299 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
300 on failure the store back into EXPECT.
302 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
303 Sandra Loosemore <sandra@codesourcery.com>
305 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
306 * config/nios2/nios2.c (nios2_function_profiler):
307 Add -fPIC (flag_pic == 2) support.
308 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
309 (nios2_large_offset_p): New function.
310 (nios2_unspec_reloc_p): Move up position, update to use
311 nios2_large_offset_p.
312 (nios2_unspec_address): Remove function.
313 (nios2_unspec_offset): New function.
314 (nios2_large_got_address): New function.
315 (nios2_got_address): Add large offset support.
316 (nios2_legitimize_tls_address): Update usage of removed and new
318 (nios2_symbol_binds_local_p): New function.
319 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
320 (nios2_legitimize_address): Update to use nios2_large_offset_p.
321 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
322 (nios2_print_operand): Merge H/L processing, add hiadj/lo
323 processing for (const (unspec ...)).
324 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
326 2014-02-20 Richard Biener <rguenther@suse.de>
328 * tree-cfg.c (replace_uses_by): Mark altered BBs before
329 doing the substitution.
330 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
332 2014-02-20 Martin Jambor <mjambor@suse.cz>
335 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
336 info when checking whether lattices are bottom.
338 2014-02-20 Richard Biener <rguenther@suse.de>
341 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
344 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
347 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
348 parameter specifying the scaling.
349 (inline_call): Update.
350 (want_inline_recursively): Guard division by zero.
351 (recursive_inlining): Update.
352 * ipa-inline.h (clone_inlined_nodes): Update.
354 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
357 * config/i386/i386.c (classify_argument): Pass structures of size
358 64 bytes or less in register.
360 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
361 Kirill Yukhin <kirill.yukhin@intel.com>
363 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
364 (_mm_rcp28_round_ss): Ditto.
365 (_mm_rsqrt28_round_sd): Ditto.
366 (_mm_rsqrt28_round_ss): Ditto.
367 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
368 (_mm_rcp14_round_ss): Ditto.
369 (_mm_rsqrt14_round_sd): Ditto.
370 (_mm_rsqrt14_round_ss): Ditto.
371 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
372 the first input operand, get rid of match_dup.
373 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
375 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
377 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
378 operand as the first input operand, set type attribute.
379 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
381 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
382 operand as the first input operand, set type attribute.
384 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
386 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
389 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
392 * config/i386/i386.c (construct_container): Remove TFmode check
393 for X86_64_INTEGER_CLASS.
395 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
398 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
399 only when -Wpsabi is enabled.
401 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
404 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
405 passing arrays in registers are the same as for structs, so remove the
406 special case for them.
408 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
410 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
411 destination type, extract only the valid bits if the source type is not
412 integral and has a different mode.
414 2014-02-19 Richard Biener <rguenther@suse.de>
417 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
420 2014-02-19 Richard Biener <rguenther@suse.de>
423 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
424 (ipa_modify_call_arguments): Emit an argument load explicitely and
425 preserve virtual SSA form there and for the replacement call.
426 Do not update SSA form nor free dominance info.
428 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
430 * ipa.c (function_and_variable_visibility): Also clear WEAK
431 flag when disolving COMDAT_GROUP.
433 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
435 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
436 * ipa-prop.c (ipa_set_jf_known_type): Return early when
438 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
439 do more sanity checks.
440 (detect_type_change): Return true when giving up early.
441 (compute_complex_assign_jump_func): Fix type parameter of
443 (compute_complex_ancestor_jump_func): Likewise.
444 (update_jump_functions_after_inlining): Fix updating of
446 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
448 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
450 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
451 inline clones when edge disappears.
453 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
456 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
457 Split 64-bit moves into 2 patterns. Do not allow the use of
458 direct move for TDmode in little endian, since the decimal value
459 has little endian bytes within a word, but the 64-bit pieces are
460 ordered in a big endian fashion, and normal subreg's of TDmode are
462 (mov<mode>_64bit_dm): Likewise.
463 (movtd_64bit_nodm): Likewise.
465 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
467 PR tree-optimization/60174
468 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
469 statement of an SSA_NAME that occurs in an abnormal PHI node.
471 2014-02-18 Jakub Jelinek <jakub@redhat.com>
474 * final.c (SEEN_BB): Remove.
475 (SEEN_NOTE, SEEN_EMITTED): Renumber.
476 (final_scan_insn): Don't force_source_line on second
477 NOTE_INSN_BASIC_BLOCK.
479 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
482 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
483 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
484 (type_natural_mode): Warn ABI change when %zmm register is not
485 available for AVX512F vector value passing.
487 2014-02-18 Kai Tietz <ktietz@redhat.com>
490 * config/i386/i386.c (ix86_expand_prologue): Use value in
491 rax register as displacement when restoring %r10 or %rax.
492 Fix wrong offset when restoring both registers.
494 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
496 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
497 assertion with conditional return.
499 2014-02-18 Jakub Jelinek <jakub@redhat.com>
500 Uros Bizjak <ubizjak@gmail.com>
503 * config/i386/driver-i386.c (host_detect_local_cpu): If
504 YMM state is not saved by the OS, also clear has_f16c. Move
505 CPUID 0x80000001 handling before YMM state saving checking.
507 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
509 PR rtl-optimization/58960
510 * haifa-sched.c (alloc_global_sched_pressure_data): New,
511 factored out from ...
512 (sched_init): ... here.
513 (free_global_sched_pressure_data): New, factored out from ...
514 (sched_finish): ... here.
515 * sched-int.h (free_global_sched_pressure_data): Declare.
516 * sched-rgn.c (nr_regions_initial): New static global.
517 (haifa_find_rgns): Initialize it.
518 (schedule_region): Disable sched-pressure for the newly
521 2014-02-17 Richard Biener <rguenther@suse.de>
523 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
524 release SSA defs of pattern stmts.
526 2014-02-17 Richard Biener <rguenther@suse.de>
528 * tree-inline.c (expand_call_inline): Release the virtual
529 operand defined by the call we are about to inline.
531 2014-02-17 Richard Biener <rguenther@suse.de>
533 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
535 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
536 Ilya Tocar <ilya.tocar@intel.com>
538 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
539 arguments order in builtin.
540 (_mm512_permutexvar_epi64): Ditto.
541 (_mm512_mask_permutexvar_epi64): Ditto
542 (_mm512_maskz_permutexvar_epi32): Ditto
543 (_mm512_permutexvar_epi32): Ditto
544 (_mm512_mask_permutexvar_epi32): Ditto
546 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
548 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
549 (p8_vmrgow): Likewise.
551 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
553 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
556 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
559 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
560 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
561 into 64-bit and 32-bit moves. On 64-bit moves, add support for
562 using direct move instructions on ISA 2.07. Also adjust
563 instruction length for 64-bit.
564 (mov<mode>_64bit, TFmode/TDmode): Likewise.
565 (mov<mode>_32bit, TFmode/TDmode): Likewise.
567 2014-02-15 Alan Modra <amodra@gmail.com>
571 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
572 find_replacement on parts of insn rtl that might be reloaded.
574 2014-02-15 Richard Biener <rguenther@suse.de>
576 PR tree-optimization/60183
577 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
578 (tree_ssa_phiprop): Calculate and free post-dominators.
580 2014-02-14 Jeff Law <law@redhat.com>
582 PR rtl-optimization/60131
583 * ree.c (get_extended_src_reg): New function.
584 (combine_reaching_defs): Use it rather than assuming location of REG.
585 (find_and_remove_re): Verify first operand of extension is
586 a REG before adding the insns to the copy list.
588 2014-02-14 Roland McGrath <mcgrathr@google.com>
590 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
591 * configure: Regenerated.
592 * config.in: Regenerated.
593 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
594 instead of ASM_SHORT.
596 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
597 Richard Earnshaw <rearnsha@arm.com>
599 PR rtl-optimization/59535
600 * lra-constraints.c (process_alt_operands): Encourage alternative
601 when unassigned pseudo class is superset of the alternative class.
602 (inherit_reload_reg): Don't inherit when optimizing for code size.
603 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
604 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
605 modes not less than 4 for Thumb1.
607 2014-02-14 Kyle McMartin <kyle@redhat.com>
610 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
612 2014-02-14 Richard Biener <rguenther@suse.de>
614 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
615 (get_frame_arg): Drop the assert with langhook types_compatible_p.
616 Do not strip INDIRECT_REFs.
618 2014-02-14 Richard Biener <rguenther@suse.de>
621 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
622 DECL_FUNCTION_SPECIFIC_TARGET.
623 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
624 * tree-streamer-out.c (pack_ts_target_option): Remove.
625 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
626 (write_ts_function_decl_tree_pointers): Do not stream
627 DECL_FUNCTION_SPECIFIC_TARGET.
628 * tree-streamer-in.c (unpack_ts_target_option): Remove.
629 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
630 (lto_input_ts_function_decl_tree_pointers): Do not stream
631 DECL_FUNCTION_SPECIFIC_TARGET.
633 2014-02-14 Jakub Jelinek <jakub@redhat.com>
635 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
636 (get_initial_def_for_induction, vectorizable_induction): Ignore
637 debug stmts when looking for exit_phi.
638 (vectorizable_live_operation): Fix up condition.
640 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
642 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
643 nreverse() because it changes the content of original tree list.
645 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
647 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
648 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
650 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
652 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
653 GNU coding standards.
655 2014-02-13 Jakub Jelinek <jakub@redhat.com>
658 * dwarf2out.c (gen_subprogram_die): Don't call
659 add_calling_convention_attribute if subr_die is old_die.
661 2014-02-13 Sharad Singhai <singhai@google.com>
663 * doc/optinfo.texi: Fix order of nodes.
665 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
667 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
668 operands[2], not operands[3].
670 2014-02-13 Richard Biener <rguenther@suse.de>
673 * doc/install.texi (ISL): Update recommended version to 0.12.2,
674 mention the possibility of an in-tree build.
675 (CLooG): Update recommended version to 0.18.1, mention the
676 possibility of an in-tree build and clarify that the ISL
677 bundled with CLooG does not work.
679 2014-02-13 Jakub Jelinek <jakub@redhat.com>
682 * expr.c (compress_float_constant): If x is a hard register,
683 extend into a pseudo and then move to x.
685 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
687 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
688 caused by bad second argument to warning_at() with -mhotpatch and
689 nested functions (e.g. with gfortran).
691 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
693 * opts.c (option_name): Remove "enabled by default" rider.
695 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
697 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
699 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
700 Uros Bizjak <ubizjak@gmail.com>
703 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
704 * configure: Regenerated.
706 2014-02-12 Richard Biener <rguenther@suse.de>
708 * vec.c (vec_prefix::calculate_allocation): Move as
709 inline variant to vec.h.
710 (vec_prefix::calculate_allocation_1): New out-of-line version.
711 * vec.h (vec_prefix::calculate_allocation_1): Declare.
712 (vec_prefix::m_has_auto_buf): Rename to ...
713 (vec_prefix::m_using_auto_storage): ... this.
714 (vec_prefix::calculate_allocation): Inline the easy cases
715 and dispatch to calculate_allocation_1 which doesn't need the
717 (va_heap::reserve): Use gcc_checking_assert.
718 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
719 m_using_auto_storage.
720 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
722 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
723 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
724 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
726 2014-02-12 Richard Biener <rguenther@suse.de>
728 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
729 when we found a dependence.
731 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
733 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
735 (maybe_fold_stmt): ... into this new function.
736 * omp-low.c (lower_omp): Update comment.
738 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
741 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
744 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
746 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
747 identifiers in comments.
748 (cortexa53_extra_costs): Likewise.
749 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
750 (cortexa7_extra_costs): Likewise.
751 (cortexa12_extra_costs): Likewise.
752 (cortexa15_extra_costs): Likewise.
753 (v7m_extra_costs): Likewise.
755 2014-02-12 Richard Biener <rguenther@suse.de>
758 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
759 of posix_memalign being successful.
760 (lower_stmt): Restrict lowering of posix_memalign to when
761 -ftree-bit-ccp is enabled.
763 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
765 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
767 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
769 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
771 PR rtl-optimization/60116
772 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
773 other_insn once the combination has been validated.
775 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
778 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
780 * ipa-devirt.c: Include demangle.h
781 (odr_violation_reported): New static variable.
782 (add_type_duplicate): Update odr_violations.
783 (maybe_record_node): Add completep parameter; update it.
784 (record_target_from_binfo): Add COMPLETEP parameter;
786 (possible_polymorphic_call_targets_1): Likewise.
787 (struct polymorphic_call_target_d): Add nonconstruction_targets;
788 rename FINAL to COMPLETE.
789 (record_targets_from_bases): Sanity check we found the binfo;
790 fix COMPLETEP updating.
791 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
792 parameter, fix computing of COMPLETEP.
793 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
794 at LTO time do demangling.
795 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
796 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
798 (gimple_get_virt_method_for_binfo): Likewise.
799 * gimple-fold.h (gimple_get_virt_method_for_binfo,
800 gimple_get_virt_method_for_vtable): Update prototypes.
802 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
805 * genautomata.c (add_presence_absence): Fix typo with
806 {final_}presence_list.
808 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
811 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
812 for VSX/Altivec vectors that land in GPR registers.
814 2014-02-11 Richard Henderson <rth@redhat.com>
815 Jakub Jelinek <jakub@redhat.com>
818 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
819 around drhs if type conversion to lacc->type is not useless.
821 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
823 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
825 (cortex-a57.cortex-a53): Likewise.
826 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
828 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
830 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
833 2014-02-11 Renlin Li <Renlin.Li@arm.com>
835 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
836 add_options_for_arm_vfp3.
838 2014-02-11 Jeff Law <law@redhat.com>
841 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
842 object with an undesirable mode.
844 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
847 * config/i386/sol2-9.h: New file.
848 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
849 *-*-solaris2.9*): Use it.
851 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
853 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
854 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
856 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
858 * config/microblaze/microblaze.c: Extend mcpu version format
860 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
862 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
864 2014-02-10 Richard Henderson <rth@redhat.com>
867 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
868 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
869 ms-abi vs -mno-accumulate-outgoing-args.
870 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
871 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
874 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
877 * cfgexpand.c (expand_asm_operands): Attach source location to
878 ASM_INPUT rtx objects.
879 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
881 2014-02-10 Nick Clifton <nickc@redhat.com>
883 * config/mn10300/mn10300.c (popcount): New function.
884 (mn10300_expand_prologue): Include saved registers in stack usage
887 2014-02-10 Jeff Law <law@redhat.com>
890 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
891 when changing the SET_DEST of a prior insn to avoid an input reload.
893 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
895 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
896 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
897 -mcall-openbsd, or -mcall-linux.
898 (CC1_ENDIAN_BIG_SPEC): Remove.
899 (CC1_ENDIAN_LITTLE_SPEC): Remove.
900 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
901 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
902 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
903 and %cc1_endian_default.
904 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
906 2014-02-10 Richard Biener <rguenther@suse.de>
908 PR tree-optimization/60115
909 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
910 MEM_REF handling. Properly verify that the accesses are not
911 out of the objects bound.
913 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
915 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
918 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
920 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
921 proper constants and fix formatting.
922 (possible_polymorphic_call_targets): Fix formatting.
924 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
925 Ilya Tocar <ilya.tocar@intel.com>
927 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
928 (_mm512_loadu_epi32): Renamed into...
929 (_mm512_loadu_si512): This.
930 (_mm512_storeu_epi32): Renamed into...
931 (_mm512_storeu_si512): This.
932 (_mm512_maskz_ceil_ps): Removed.
933 (_mm512_maskz_ceil_pd): Ditto.
934 (_mm512_maskz_floor_ps): Ditto.
935 (_mm512_maskz_floor_pd): Ditto.
936 (_mm512_floor_round_ps): Ditto.
937 (_mm512_floor_round_pd): Ditto.
938 (_mm512_ceil_round_ps): Ditto.
939 (_mm512_ceil_round_pd): Ditto.
940 (_mm512_mask_floor_round_ps): Ditto.
941 (_mm512_mask_floor_round_pd): Ditto.
942 (_mm512_mask_ceil_round_ps): Ditto.
943 (_mm512_mask_ceil_round_pd): Ditto.
944 (_mm512_maskz_floor_round_ps): Ditto.
945 (_mm512_maskz_floor_round_pd): Ditto.
946 (_mm512_maskz_ceil_round_ps): Ditto.
947 (_mm512_maskz_ceil_round_pd): Ditto.
948 (_mm512_expand_pd): Ditto.
949 (_mm512_expand_ps): Ditto.
950 * config/i386/i386.c (ix86_builtins): Remove
951 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
953 * config/i386/predicates.md (const1256_operand): New.
954 (const_1_to_2_operand): Ditto.
955 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
956 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
957 (*avx512pf_gatherpf<mode>sf): Ditto.
958 (avx512pf_gatherpf<mode>df): Ditto.
959 (*avx512pf_gatherpf<mode>df_mask): Ditto.
960 (*avx512pf_gatherpf<mode>df): Ditto.
961 (avx512pf_scatterpf<mode>sf): Ditto.
962 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
963 (*avx512pf_scatterpf<mode>sf): Ditto.
964 (avx512pf_scatterpf<mode>df): Ditto.
965 (*avx512pf_scatterpf<mode>df_mask): Ditto.
966 (*avx512pf_scatterpf<mode>df): Ditto.
967 (avx512f_expand<mode>): Removed.
968 (<shift_insn><mode>3<mask_name>): Change predicate type.
970 2014-02-08 Jakub Jelinek <jakub@redhat.com>
972 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
973 not at the end of datarefs vector use ordered_remove to avoid
974 reordering datarefs vector.
977 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
978 mark local addressable non-static vars as GOVD_PRIVATE
979 instead of GOVD_LOCAL.
980 * omp-low.c (lower_omp_for): Move gimple_bind_vars
981 and BLOCK_VARS of gimple_bind_block to new_stmt rather
985 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
986 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
987 assume_aligned or alloc_align attributes.
988 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
989 arguments. Handle also assume_aligned and alloc_align attributes.
990 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
991 calls to functions with assume_aligned or alloc_align attributes.
992 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
994 2014-02-08 Terry Guo <terry.guo@arm.com>
996 * doc/invoke.texi: Document ARM -march=armv7e-m.
998 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1000 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1001 flag on __cilkrts_rethrow builtin.
1004 * ipa-cp.c (determine_versionability): Fail at -O0
1005 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1006 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1009 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1012 * tree-inline.c (copy_forbidden): Fail for
1013 __attribute__((optimize (0))) functions.
1015 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1017 * varpool.c: Include pointer-set.h.
1018 (varpool_remove_unreferenced_decls): Variables in other partitions
1019 will not be output; be however careful to not lose information
1022 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1024 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1025 lookup in the vtable constructor.
1027 2014-02-07 Jeff Law <law@redhat.com>
1030 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1031 define_insn_and_split.
1033 * ipa-inline.c (inline_small_functions): Fix typos.
1035 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1037 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1038 (s390_can_use_return_insn): Declare.
1039 * config/s390/s390.h (EPILOGUE_USES): Define.
1040 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1042 (s390_chunkify_start): Handle return JUMP_LABELs.
1043 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1044 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1045 (s390_can_use_return_insn): New functions.
1046 (s390_fix_long_loop_prediction): Handle conditional returns.
1047 (TARGET_SET_UP_BY_PROLOGUE): Define.
1048 * config/s390/s390.md (ANY_RETURN): New code iterator.
1049 (*creturn, *csimple_return, return, simple_return): New patterns.
1051 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1053 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1054 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1055 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1056 REG_CFA_RESTORE list when deciding not to restore a register.
1058 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1060 * config/s390/s390.c: Include tree-pass.h and context.h.
1061 (s390_early_mach): New function, split out from...
1062 (s390_emit_prologue): ...here.
1063 (pass_data_s390_early_mach): New pass structure.
1064 (pass_s390_early_mach): New class.
1065 (s390_option_override): Create and register early_mach pass.
1066 Move to end of file.
1068 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1070 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1071 to match for the exit block.
1073 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1075 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1076 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1077 Reject misaligned operands.
1079 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1081 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1083 2014-02-07 Richard Biener <rguenther@suse.de>
1086 * gimple-low.c (lower_builtin_posix_memalign): New function.
1087 (lower_stmt): Call it to lower posix_memalign in a way
1088 to make alignment info accessible.
1090 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1093 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1094 __builtin_setjmp_receiver.
1096 2014-02-07 Richard Biener <rguenther@suse.de>
1099 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1100 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1101 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1102 Handle BUILT_IN_POSIX_MEMALIGN.
1103 (find_func_clobbers): Likewise.
1104 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1105 (call_may_clobber_ref_p_1): Likewise.
1107 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1110 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1113 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1116 * lto-cgraph.c (lto_output_node): Use
1117 symtab_get_symbol_partitioning_class.
1118 (lto_output_varpool_node): likewise.
1119 (symtab_get_symbol_partitioning_class): Move here from
1121 * cgraph.h (symbol_partitioning_class): Likewise.
1122 (symtab_get_symbol_partitioning_class): Declare.
1124 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1126 * ggc.h (ggc_internal_cleared_alloc): New macro.
1127 * vec.h (vec_safe_copy): Handle memory stats.
1128 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1129 * target-globals.c (save_target_globals): Likewise.
1131 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1134 * expr.c (emit_move_resolve_push): Export; be bit more selective
1135 on when to clear alias set.
1136 * expr.h (emit_move_resolve_push): Declare.
1137 * function.h (struct function): Add tail_call_marked.
1138 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1139 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1140 * config/i386/i386.md (TImode move expander): De not call
1142 (FP push expanders): Preserve memory attributes.
1143 * config/i386/sse.md (push<mode>1): Remove.
1144 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1145 (ix86_expand_push): Remove.
1146 * config/i386/mmx.md (push<mode>1): Remove.
1148 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1150 PR rtl-optimization/60030
1151 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1152 lopart with paradoxical subreg before shifting it up by hprec.
1154 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1156 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1157 Remove extra newline at end of file.
1158 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1159 (arm_issue_rate): Handle cortexa57.
1160 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1161 (cortex-a57.cortex-a53): Likewise.
1163 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1166 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1167 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1169 (arm_expand_prologue): Adjust all callers.
1170 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1171 info, registers also at the lowest numbered registers side. Use
1172 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1176 * var-tracking.c (adjust_mems): Before adding a SET to
1177 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1179 2014-02-06 Alan Modra <amodra@gmail.com>
1182 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1183 change SDmode to DDmode when lra_in_progress.
1185 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1188 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1189 free_data_ref on the dr first, and before goto again also set dr
1190 to the next dr. For simd_lane_access, free old datarefs[i] before
1191 overwriting it. For get_vectype_for_scalar_type failure, don't
1192 free_data_ref if simd_lane_access.
1194 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1197 * tree.h (opts_for_fn): New inline function.
1198 (opt_for_fn): Define.
1199 * config/i386/i386.c (ix86_function_regparm): Use
1200 opt_for_fn (decl, optimize) instead of optimize.
1202 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1204 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1205 for SYMBOL_REF in large memory model.
1207 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1209 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1211 (cortex-a57): Likewise.
1212 (cortex-a57.cortex-a53): Likewise.
1214 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1215 Kugan Vivekanandarajah <kuganv@linaro.org>
1217 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1219 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1221 2014-02-06 Richard Biener <rguenther@suse.de>
1223 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1224 set_loop_copy and initialize_original_copy_tables.
1226 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1228 * config/aarch64/aarch64-simd.md
1229 (aarch64_ashr_simddi): Change QI to SI.
1231 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1232 Jakub Jelinek <jakub@redhat.com>
1235 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1238 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1240 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1241 CODE_FOR_altivec_vpku[hw]um to
1242 CODE_FOR_altivec_vpku[hw]um_direct.
1243 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1244 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1245 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1246 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1248 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1250 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1251 generation for -maltivec=be.
1252 (altivec_vsumsws): Simplify redundant test.
1254 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1256 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1257 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1258 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1259 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1260 gen_altivec_vpkuwum.
1261 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1263 (altivec_vpks<VI_char>ss): Likewise.
1264 (altivec_vpks<VI_char>us): Likewise.
1265 (altivec_vpku<VI_char>us): Likewise.
1266 (altivec_vpku<VI_char>um): Likewise.
1267 (altivec_vpku<VI_char>um_direct): New (copy of
1268 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1270 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1271 target is little endian and -maltivec=be is not specified.
1272 (*altivec_vupkhs<VU_char>_direct): New (copy of
1273 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1274 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1275 target is little endian and -maltivec=be is not specified.
1276 (*altivec_vupkls<VU_char>_direct): New (copy of
1277 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1278 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1279 little endian and -maltivec=be is not specified.
1280 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1281 little endian and -maltivec=be is not specified.
1283 2014-02-05 Richard Henderson <rth@redhat.com>
1286 * combine-stack-adj.c: Revert r206943.
1287 * sched-int.h (struct deps_desc): Add last_args_size.
1288 * sched-deps.c (init_deps): Initialize it.
1289 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1290 contain REG_ARGS_SIZE notes.
1292 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1294 * lto-cgraph.c (asm_nodes_output): Make global.
1295 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1296 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1297 (driver_handle_option): Handle OPT_fwpa.
1299 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1302 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1303 a comment typo and formatting issue. If odr_hash hasn't been
1304 created, return vNULL and set *completep to false.
1307 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1308 bb with no successors.
1310 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1313 * doc/invoke.texi (-march): Clarify documentation for ARM.
1317 2014-02-05 Richard Biener <rguenther@suse.de>
1319 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1320 when not vectorizing because of too many alias checks.
1321 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1322 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1324 2014-02-05 Nick Clifton <nickc@redhat.com>
1326 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1327 accept extended registers in any mode when compiling for the MN10300.
1329 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1331 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1332 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1333 sanitization attributes.
1334 (can_inline_edge_p): Likewise.
1335 (sanitize_attrs_match_for_inline_p): New function.
1337 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1339 * ipa-prop.c (detect_type_change): Shor circuit testing of
1340 type changes on THIS pointer.
1342 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1345 * config/pa/pa.c (legitimize_tls_address): Return original address
1346 if not passed a SYMBOL_REF rtx.
1347 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1349 (pa_emit_move_sequence): Simplify TLS source operands.
1350 (pa_legitimate_constant_p): Reject all TLS constants.
1351 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1352 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1354 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1356 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1357 groups when we know they are controlled by LTO.
1358 * varasm.c (default_binds_local_p_1): If object is in other partition,
1359 it will be resolved locally.
1361 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1363 * config/host-linux.c (linux_gt_pch_use_address): Don't
1364 use SSIZE_MAX because it is not always defined.
1366 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1369 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1370 threshold for pseudo splitting.
1371 (update_ebb_live_info): Process call argument hard registers and
1372 hard registers from insn definition too.
1373 (max_small_class_regs_num): New constant.
1374 (inherit_in_ebb): Update live hard regs through EBBs. Update
1375 reloads_num only for small register classes. Don't split for
1378 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1381 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1384 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1386 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1389 2014-02-04 Marek Polacek <polacek@redhat.com>
1391 * gdbinit.in (pel): Define.
1393 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1395 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1398 2014-02-04 Richard Biener <rguenther@suse.de>
1401 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1402 in function context local.
1403 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1404 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1405 similar to LTO_imported_decl_ref.
1407 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1409 PR tree-optimization/60002
1410 * cgraphclones.c (build_function_decl_skip_args): Clear
1413 PR tree-optimization/60023
1414 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1415 false to gsi_replace.
1416 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1417 has been in some EH region and vec_stmt could throw, add
1418 vec_stmt into the same EH region.
1419 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1420 has no lhs, ignore it.
1421 * internal-fn.c (expand_MASK_LOAD): Likewise.
1424 * tree-inline.c (copy_forbidden): Fail for
1425 __attribute__((optimize (0))) functions.
1428 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1429 is set, copy one less argument.
1430 (expand_simd_clones): Don't subtract clone_info->inbranch
1431 from simd_clone_struct_alloc argument.
1433 PR rtl-optimization/57915
1434 * recog.c (simplify_while_replacing): If all unary/binary/relational
1435 operation arguments are constant, attempt to simplify those.
1438 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1439 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1441 2014-02-04 Richard Biener <rguenther@suse.de>
1443 PR tree-optimization/60012
1444 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1445 TBAA disambiguation to all DDRs.
1447 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1450 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1451 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1453 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1456 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1458 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1460 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1461 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1463 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1466 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1467 to figure out targets of polymorphic calls with known decl.
1468 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1469 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1470 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1471 (get_polymorphic_call_info): ... here.
1472 (get_polymorphic_call_info_from_invariant): New function.
1474 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1476 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1477 lookup via vtable pointer; check for type consistency
1478 and turn inconsitent facts into UNREACHABLE.
1479 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1480 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1481 type inconsistent querries; return UNREACHABLE instead.
1483 2014-02-03 Richard Henderson <rth@twiddle.net>
1486 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1487 already processed this node.
1488 (normalize_one_pred_1): Pass along mark_set.
1489 (normalize_one_pred): Create and destroy a pointer_set_t.
1490 (normalize_one_pred_chain): Likewise.
1492 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1494 PR gcov-profile/58602
1495 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1497 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1500 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1501 -fno-devirtualize; try to devirtualize by the knowledge of
1502 virtual table pointer given by aggregate propagation.
1503 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1504 (ipa_print_node_jump_functions): Dump also offset that
1505 is relevant for polymorphic calls.
1506 (determine_known_aggregate_parts): Add arg_type parameter; use it
1507 instead of determining the type from pointer type.
1508 (ipa_compute_jump_functions_for_edge): Update call of
1509 determine_known_aggregate_parts.
1510 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1511 (gimple_get_virt_method_for_binfo): ... here; simplify using
1512 vtable_pointer_value_to_vtable.
1513 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1514 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1515 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1516 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1518 (vtable_pointer_value_to_binfo): ... here.
1519 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1521 2014-02-03 Teresa Johnson <tejohnson@google.com>
1523 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1524 redef of outer loop index variable.
1526 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1530 * doc/extend.texi (Function Attributes): Typo.
1532 2014-02-03 Cong Hou <congh@google.com>
1534 PR tree-optimization/60000
1535 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1536 if the vectorized statement is a store. A store statement can only
1537 appear at the end of pattern statements.
1539 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1541 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1542 (ix86_option_override_internal): Default long double to 64-bit for
1543 32-bit Bionic and to 128-bit for 64-bit Bionic.
1545 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1546 TARGET_LONG_DOUBLE_128 is true.
1547 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1549 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1550 (mlong-double-64): Negate -mlong-double-128.
1551 (mlong-double-128): New option.
1553 * config/i386/i386-c.c (ix86_target_macros): Define
1554 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1556 * doc/invoke.texi: Document -mlong-double-128.
1558 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1560 PR rtl-optimization/60024
1561 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1563 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1565 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1567 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1569 PR rtl-optimization/57662
1570 * sel-sched.c (code_motion_path_driver): Do not mark already not
1571 existing blocks in the visiting bitmap.
1573 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1575 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1576 on the insn being emitted.
1578 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1579 Will Deacon <will.deacon@arm.com>
1581 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1583 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1585 * config/arm/arm-tables.opt: Regenerate.
1587 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1589 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1590 for vector types other than V16QImode.
1591 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1592 define_expand, and call altivec_expand_vec_perm_le when producing
1593 code with little endian element order.
1594 (*altivec_vperm_<mode>_internal): New insn having previous
1595 behavior of altivec_vperm_<mode>.
1596 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1597 altivec_expand_vec_perm_le when producing code with little endian
1599 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1600 behavior of altivec_vperm_<mode>_uns.
1602 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1604 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1605 (altivec_vsumsws): Add handling for -maltivec=be with a little
1607 (altivec_vsumsws_direct): New.
1608 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1609 gen_altivec_vsumsws.
1611 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1613 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1614 vtable_pointer_value_to_binfo): New functions.
1615 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1616 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1618 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1620 * config/nios2/nios2.md (load_got_register): Initialize GOT
1621 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1622 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1624 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1626 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1627 preserverd by passthrough, do not propagate the type.
1629 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1631 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1632 (mips_atomic_assign_expand_fenv): New function.
1633 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1635 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1637 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1638 (__builtin_mips_set_fcsr): Likewise.
1639 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1640 MIPS_USI_FTYPE_VOID.
1641 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1642 (mips16_expand_set_fcsr): Likewise.
1643 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1644 (mips16_set_fcsr_stub): Likewise.
1645 (mips16_get_fcsr_one_only_stub): New class.
1646 (mips16_set_fcsr_one_only_stub): Likewise.
1647 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1648 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1649 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1650 (hard_float): New availability predicate.
1651 (mips_builtins): Add get_fcsr and set_fcsr.
1652 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1653 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1654 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1655 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1656 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1659 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1661 * config/mips/mips.c (mips_one_only_stub): New class.
1662 (mips_need_mips16_rdhwr_p): Replace with...
1663 (mips16_rdhwr_stub): ...this new variable.
1664 (mips16_stub_call_address): New function.
1665 (mips16_rdhwr_one_only_stub): New class.
1666 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1667 (mips_output_mips16_rdhwr): Delete.
1668 (mips_finish_stub): New function.
1669 (mips_code_end): Use it to handle rdhwr stubs.
1671 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1674 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1675 when calculating size of integer atomic types.
1677 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1679 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1681 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1683 PR tree-optimization/60003
1684 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1685 * profile.c (branch_prob): Use gimple_call_builtin_p
1686 to check for BUILT_IN_SETJMP_RECEIVER.
1687 * tree-inline.c (copy_bb): Call notice_special_calls.
1689 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1692 * lra-constraints.c (process_alt_operands): Update reload_sum only
1695 2014-01-31 Richard Henderson <rth@redhat.com>
1698 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1699 until after else_eh is processed.
1701 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1703 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1704 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1705 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1706 in smmintrin.h, remove them.
1707 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1708 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1709 * config/i386/i386.md (ROUND_SAE): Fix value.
1710 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1711 (const48_operand): New.
1712 * config/i386/subst.md (round), (round_expand): Use
1713 const_4_or_8_to_11_operand.
1714 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1716 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1718 * config/i386/constraints.md (Yk): Swap meaning with k.
1719 * config/i386/i386.md (movhi_internal): Change Yk to k.
1720 (movqi_internal): Ditto.
1721 (*k<logic><mode>): Ditto.
1724 (kandn<mode>): Ditto.
1725 (*<code>hi_1): Ditto.
1726 (*<code>qi_1): Ditto.
1727 (kxnor<mode>): Ditto.
1728 (kortestzhi): Ditto.
1729 (kortestchi): Ditto.
1731 (*one_cmplhi2_1): Ditto.
1732 (*one_cmplqi2_1): Ditto.
1733 * config/i386/sse.md (): Change k to Yk.
1734 (avx512f_load<mode>_mask): Ditto.
1735 (avx512f_blendm<mode>): Ditto.
1736 (avx512f_store<mode>_mask): Ditto.
1737 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1738 (avx512f_storedqu<mode>_mask): Ditto.
1739 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1741 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1742 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1743 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1744 (avx512f_maskcmp<mode>3): Ditto.
1745 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1746 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1747 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1748 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1749 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1750 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1751 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1752 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1753 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1754 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1755 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1756 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1757 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1758 (vec_extract_lo_<mode>_maskm): Ditto.
1759 (vec_extract_hi_<mode>_maskm): Ditto.
1760 (avx512f_vternlog<mode>_mask): Ditto.
1761 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1762 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1763 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1764 (avx512f_<code>v8div16qi2_mask): Ditto.
1765 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1766 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1767 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1768 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1769 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1770 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1771 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1772 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1773 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1774 (avx512cd_maskb_vec_dupv8di): Ditto.
1775 (avx512cd_maskw_vec_dupv16si): Ditto.
1776 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1777 (avx512f_vpermi2var<mode>3_mask): Ditto.
1778 (avx512f_vpermi2var<mode>3_mask): Ditto.
1779 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1780 (*avx512f_gathersi<mode>): Ditto.
1781 (*avx512f_gathersi<mode>_2): Ditto.
1782 (*avx512f_gatherdi<mode>): Ditto.
1783 (*avx512f_gatherdi<mode>_2): Ditto.
1784 (*avx512f_scattersi<mode>): Ditto.
1785 (*avx512f_scatterdi<mode>): Ditto.
1786 (avx512f_compress<mode>_mask): Ditto.
1787 (avx512f_compressstore<mode>_mask): Ditto.
1788 (avx512f_expand<mode>_mask): Ditto.
1789 * config/i386/subst.md (mask): Change k to Yk.
1790 (mask_scalar_merge): Ditto.
1793 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1795 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1797 2014-01-31 Richard Biener <rguenther@suse.de>
1800 * builtins.c (fold_builtin_memory_op): Make sure to not
1801 use a floating-point mode or a boolean or enumeral type for
1804 2014-01-30 DJ Delorie <dj@redhat.com>
1806 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1807 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1808 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1809 whenever main() has an epilogue.
1811 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1813 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1814 unused variable "field".
1815 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1816 (vsx_mergeh_<mode>): Likewise.
1817 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1818 (altivec_vmrghh): Likewise.
1819 (altivec_vmrghw): Likewise.
1820 (altivec_vmrglb): Likewise.
1821 (altivec_vmrglh): Likewise.
1822 (altivec_vmrglw): Likewise.
1823 (altivec_vspltb): Add missing uses.
1824 (altivec_vsplth): Likewise.
1825 (altivec_vspltw): Likewise.
1826 (altivec_vspltsf): Likewise.
1828 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1831 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1832 frame related instructions.
1834 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1836 PR rtl-optimization/59959
1837 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1838 any reload of register whose subreg is invalid.
1840 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1842 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1843 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1844 Add missing return type - void.
1846 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1848 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1849 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1850 remove element index adjustment for endian (now handled in vsx.md
1852 (altivec_expand_vec_perm_const): Use
1853 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1854 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1855 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1856 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1857 define_expand and a new define_insn *altivec_vspltb_internal;
1858 adjust for -maltivec=be on a little endian target.
1859 (altivec_vspltb_direct): New.
1860 (altivec_vsplth): Divide into a define_expand and a new
1861 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1862 little endian target.
1863 (altivec_vsplth_direct): New.
1864 (altivec_vspltw): Divide into a define_expand and a new
1865 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1866 little endian target.
1867 (altivec_vspltw_direct): New.
1868 (altivec_vspltsf): Divide into a define_expand and a new
1869 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1870 a little endian target.
1872 2014-01-30 Richard Biener <rguenther@suse.de>
1874 PR tree-optimization/59993
1875 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1876 can propagate form the earlier stmt and avoid the transform
1877 when the intermediate result is needed.
1879 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1881 * README.Portability: Fix typo.
1883 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1885 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1886 comparison_operator with ordered_comparison_operator.
1888 2014-01-30 Nick Clifton <nickc@redhat.com>
1890 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1891 Rename to mn10300_store_multiple_regs.
1892 * config/mn10300/mn10300.c: Likewise.
1893 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1894 store_multiple_regs.
1895 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1896 Call mn10300_store_multiple_regs.
1898 2014-01-30 Nick Clifton <nickc@redhat.com>
1899 DJ Delorie <dj@redhat.com>
1901 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1902 %fp 2 to keep registers after it properly word-aligned.
1903 (rl78_alloc_physical_registers_umul): Handle the case where both
1904 input operands are the same.
1906 2014-01-30 Richard Biener <rguenther@suse.de>
1908 PR tree-optimization/59903
1909 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1912 2014-01-30 Jason Merrill <jason@redhat.com>
1915 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1918 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1920 2014-01-30 Richard Biener <rguenther@suse.de>
1922 PR tree-optimization/59951
1923 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1925 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1928 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1929 SFmode to DFmode case.
1931 2014-01-29 DJ Delorie <dj@redhat.com>
1933 * config/msp430/msp430.opt (-minrt): New.
1934 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1936 (ENDFILE_SPEC): Likewise.
1938 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1940 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1941 (estimate_function_body_sizes): Use it.
1943 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1946 * dwarf2out.c (is_cxx_auto): New.
1947 (is_base_type): Use it.
1948 (gen_type_die_with_usage): Likewise.
1950 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1952 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1953 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1954 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1955 -maltivec=be with LE targets.
1956 (vsx_mergeh_<mode>): Likewise.
1957 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1958 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1959 (altivec_vmrghb): Replace with define_expand and new
1960 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1961 (altivec_vmrghb_direct): New define_insn.
1962 (altivec_vmrghh): Replace with define_expand and new
1963 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1964 (altivec_vmrghh_direct): New define_insn.
1965 (altivec_vmrghw): Replace with define_expand and new
1966 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1967 (altivec_vmrghw_direct): New define_insn.
1968 (*altivec_vmrghsf): Adjust for endianness.
1969 (altivec_vmrglb): Replace with define_expand and new
1970 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1971 (altivec_vmrglb_direct): New define_insn.
1972 (altivec_vmrglh): Replace with define_expand and new
1973 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1974 (altivec_vmrglh_direct): New define_insn.
1975 (altivec_vmrglw): Replace with define_expand and new
1976 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1977 (altivec_vmrglw_direct): New define_insn.
1978 (*altivec_vmrglsf): Adjust for endianness.
1979 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1980 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1981 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1982 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1983 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1984 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1985 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1986 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1988 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1990 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1991 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1994 2014-01-29 Richard Biener <rguenther@suse.de>
1996 PR tree-optimization/58742
1997 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1998 associate_pointerplus_align.
1999 (associate_pointerplus_diff): New function.
2000 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2001 and associate_pointerplus_diff.
2003 2014-01-29 Richard Biener <rguenther@suse.de>
2005 * lto-streamer.h (LTO_major_version): Bump to 3.
2006 (LTO_minor_version): Reset to 0.
2008 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2010 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2011 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2012 (arm_file_start): Generate correct asm header for armv7ve.
2013 * config/arm/bpabi.h: Add multilib support for armv7ve.
2014 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2015 and cortex-a15 to armv7ve.
2016 * config/arm/t-aprofile: Add multilib support for armv7ve.
2017 * doc/invoke.texi: Document -march=armv7ve.
2019 2014-01-29 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/58742
2022 * tree-ssa-forwprop.c (associate_plusminus): Return true
2023 if we changed sth, defer EH cleanup to ...
2024 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2025 (simplify_mult): New function.
2027 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2030 PR tree-optimization/59920
2031 * tree.c (build_common_builtin_nodes): Remove
2032 __builtin_setjmp_dispatcher initialization.
2033 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2034 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2035 instead of gsi_after_labels + manually skipping debug stmts.
2036 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2037 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2038 * tree-inline.c (copy_edges_for_bb): Remove
2039 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2040 argument. Ignore computed_goto_p stmts. Don't call
2041 make_abnormal_goto_edges. If a call might need abnormal edges
2042 for non-local gotos, see if it already has an edge to
2043 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2044 with true argument, don't do anything then, otherwise add
2045 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2046 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2048 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2049 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2050 (lower_stmt): Don't set data->calls_builtin_setjmp.
2051 (lower_builtin_setjmp): Adjust comment.
2052 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2053 * tree-cfg.c (found_computed_goto): Remove.
2054 (factor_computed_gotos): Remove.
2055 (make_goto_expr_edges): Return bool, true for computed gotos.
2056 Don't call make_abnormal_goto_edges.
2057 (build_gimple_cfg): Don't set found_computed_goto, don't call
2058 factor_computed_gotos.
2059 (computed_goto_p): No longer static.
2060 (make_blocks): Don't set found_computed_goto.
2061 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2062 (make_edges): If make_goto_expr_edges returns true, push bb
2063 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2064 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2065 vector. Record mapping between bbs and OpenMP regions if there
2066 are any, adjust make_gimple_omp_edges caller. Call
2067 handle_abnormal_edges.
2068 (make_abnormal_goto_edges): Remove.
2069 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2070 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2071 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2072 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2073 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2074 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2075 filling *region also set *region_idx to (*region)->entry->index.
2078 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2079 For REGs set ORIGINAL_REGNO.
2081 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2083 * doc/md.texi: Mention that a target shouldn't implement
2084 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2087 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2089 PR tree-optimization/59594
2090 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2091 a copy of the datarefs vector rather than the vector itself.
2093 2014-01-28 Jason Merrill <jason@redhat.com>
2096 * dwarf2out.c (auto_die): New static.
2097 (gen_type_die_with_usage): Handle C++1y 'auto'.
2098 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2101 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2104 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2105 (SPEC_X32): Likewise.
2106 (SPEC_64): Likewise.
2107 * config/i386/i386.c (ix86_option_override_internal): Turn off
2108 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2110 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2111 * config/i386/i386.h (TARGET_16BIT): New macro.
2112 (TARGET_16BIT_P): Likewise.
2113 * config/i386/i386.opt: Add m16.
2114 * doc/invoke.texi: Document -m16.
2116 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2118 PR preprocessor/59935
2119 * input.c (location_get_source_line): Bail out on when line number
2120 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2122 2014-01-28 Richard Biener <rguenther@suse.de>
2124 PR tree-optimization/58742
2125 * tree-ssa-forwprop.c (associate_plusminus): Handle
2126 pointer subtraction of the form (T)(P + A) - (T)P.
2128 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2130 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2133 2014-01-28 Richard Biener <rguenther@suse.de>
2136 2014-01-28 Richard Biener <rguenther@suse.de>
2138 PR rtl-optimization/45364
2139 PR rtl-optimization/59890
2140 * var-tracking.c (local_get_addr_clear_given_value): Handle
2141 already cleared slot.
2142 (val_reset): Handle not allocated local_get_addr_cache.
2143 (vt_find_locations): Use post-order on the inverted CFG.
2145 2014-01-28 Richard Biener <rguenther@suse.de>
2147 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2149 2014-01-28 Richard Biener <rguenther@suse.de>
2151 PR rtl-optimization/45364
2152 PR rtl-optimization/59890
2153 * var-tracking.c (local_get_addr_clear_given_value): Handle
2154 already cleared slot.
2155 (val_reset): Handle not allocated local_get_addr_cache.
2156 (vt_find_locations): Use post-order on the inverted CFG.
2158 2014-01-28 Alan Modra <amodra@gmail.com>
2160 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2161 * configure.ac <recursive call for build != host>: Define
2162 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2163 and LD_FOR_BUILD too.
2164 * configure: Regenerate.
2166 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2168 * config/i386/i386.c (get_builtin_code_for_version): Separate
2169 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2170 Broadwell from Haswell.
2172 2014-01-27 Steve Ellcey <sellcey@mips.com>
2174 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2175 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2176 * config/mips/mips.c (mips_option_override): Change setting
2178 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2179 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2180 Change from Mask to Var.
2182 2014-01-27 Jeff Law <law@redhat.com>
2184 * ipa-inline.c (inline_small_functions): Fix typo.
2186 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2188 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2189 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2190 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2191 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2192 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2193 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2194 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2195 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2196 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2197 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2198 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2199 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2200 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2201 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2202 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2203 (_mm512_storeu_epi64): Ditto.
2204 (_mm512_cmpge_epi32_mask): Ditto.
2205 (_mm512_cmpge_epu32_mask): Ditto.
2206 (_mm512_cmpge_epi64_mask): Ditto.
2207 (_mm512_cmpge_epu64_mask): Ditto.
2208 (_mm512_cmple_epi32_mask): Ditto.
2209 (_mm512_cmple_epu32_mask): Ditto.
2210 (_mm512_cmple_epi64_mask): Ditto.
2211 (_mm512_cmple_epu64_mask): Ditto.
2212 (_mm512_cmplt_epi32_mask): Ditto.
2213 (_mm512_cmplt_epu32_mask): Ditto.
2214 (_mm512_cmplt_epi64_mask): Ditto.
2215 (_mm512_cmplt_epu64_mask): Ditto.
2216 (_mm512_cmpneq_epi32_mask): Ditto.
2217 (_mm512_cmpneq_epu32_mask): Ditto.
2218 (_mm512_cmpneq_epi64_mask): Ditto.
2219 (_mm512_cmpneq_epu64_mask): Ditto.
2220 (_mm512_expand_pd): Ditto.
2221 (_mm512_expand_ps): Ditto.
2222 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2223 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2224 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2225 * config/i386/i386.c (ix86_builtins): Add
2226 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2227 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2228 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2229 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2230 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2231 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2232 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2233 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2234 IX86_BUILTIN_PMOVUSQW512_MEM.
2235 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2236 __builtin_ia32_pmovsqd512mem_mask,
2237 __builtin_ia32_pmovqd512mem_mask,
2238 __builtin_ia32_pmovusqw512mem_mask,
2239 __builtin_ia32_pmovsqw512mem_mask,
2240 __builtin_ia32_pmovqw512mem_mask,
2241 __builtin_ia32_pmovusdw512mem_mask,
2242 __builtin_ia32_pmovsdw512mem_mask,
2243 __builtin_ia32_pmovdw512mem_mask,
2244 __builtin_ia32_pmovqb512mem_mask,
2245 __builtin_ia32_pmovusqb512mem_mask,
2246 __builtin_ia32_pmovsqb512mem_mask,
2247 __builtin_ia32_pmovusdb512mem_mask,
2248 __builtin_ia32_pmovsdb512mem_mask,
2249 __builtin_ia32_pmovdb512mem_mask.
2250 (bdesc_args): Add __builtin_ia32_expanddf512,
2251 __builtin_ia32_expandsf512.
2252 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2253 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2254 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2255 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2256 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2257 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2258 (avx512f_<code>v8div16qi2_mask_store): This.
2259 (avx512f_expand<mode>): New.
2261 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2263 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2265 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2266 (_mm512_prefetch_i32scatter_pd): Ditto.
2267 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2268 (_mm512_prefetch_i64scatter_pd): Ditto.
2269 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2270 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2271 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2272 (_mm512_prefetch_i32scatter_ps): Ditto.
2273 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2274 (_mm512_prefetch_i64scatter_ps): Ditto.
2275 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2276 * config/i386/i386-builtin-types.def: Define
2277 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2278 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2279 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2280 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2281 IX86_BUILTIN_SCATTERPFQPD.
2282 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2283 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2284 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2285 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2286 __builtin_ia32_scatterpfqps.
2287 (ix86_expand_builtin): Expand new built-ins.
2288 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2289 fix memory access data type.
2290 (*avx512pf_gatherpf<mode>_mask): Ditto.
2291 (*avx512pf_gatherpf<mode>): Ditto.
2292 (avx512pf_scatterpf<mode>): Ditto.
2293 (*avx512pf_scatterpf<mode>_mask): Ditto.
2294 (*avx512pf_scatterpf<mode>): Ditto.
2295 (GATHER_SCATTER_SF_MEM_MODE): New.
2296 (avx512pf_gatherpf<mode>df): Ditto.
2297 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2298 (*avx512pf_scatterpf<mode>df): Ditto.
2300 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2303 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2304 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2307 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2309 * common/config/arm/arm-common.c
2310 (arm_rewrite_mcpu): Handle multiple names.
2312 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2314 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2316 * gimple-builder.h (create_gimple_tmp): Delete.
2318 2014-01-27 Christian Bruel <christian.bruel@st.com>
2320 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2323 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2325 * config/pa/pa.md (call): Generate indirect long calls to non-local
2326 functions when outputing 32-bit code.
2327 (call_value): Likewise except for special call to buggy powf function.
2329 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2330 portable runtime and PIC indirect calls.
2331 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2332 and PIC call sequences. Use ldo instead of blr to set return register
2333 in PIC call sequence.
2335 2014-01-25 Walter Lee <walt@tilera.com>
2337 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2338 avoid clobbering a live register.
2340 2014-01-25 Walter Lee <walt@tilera.com>
2342 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2343 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2344 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2345 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2347 2014-01-25 Walter Lee <walt@tilera.com>
2349 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2350 arguments on even registers.
2351 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2353 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2354 (BIGGEST_ALIGNMENT): Ditto.
2355 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2357 2014-01-25 Walter Lee <walt@tilera.com>
2359 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2360 insns before bundling.
2361 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2363 2014-01-25 Walter Lee <walt@tilera.com>
2365 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2366 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2367 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2369 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2371 * config/mips/constraints.md (kl): Delete.
2372 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2373 define expands, using...
2374 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2375 instructions for MIPS16.
2376 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2377 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2379 2014-01-25 Walter Lee <walt@tilera.com>
2381 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2385 2014-01-25 Walter Lee <walt@tilera.com>
2387 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2388 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2390 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2392 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2395 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2397 * print-rtl.c (in_call_function_usage): New var.
2398 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2399 EXPR_LIST mode as mode and not as reg note name.
2402 * cfgloopmanip.c (copy_loop_info): If
2403 loop->warned_aggressive_loop_optimizations, make sure
2404 the flag is set in target loop too.
2406 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2408 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2410 * builtins.def: Likewise.
2411 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2412 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2413 * ira.c (ira_setup_eliminable_regset): Likewise.
2414 * omp-low.c (gate_expand_omp): Likewise.
2415 (execute_lower_omp): Likewise.
2416 (diagnose_sb_0): Likewise.
2417 (gate_diagnose_omp_blocks): Likewise.
2418 (simd_clone_clauses_extract): Likewise.
2421 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2423 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2424 correction for little endian...
2425 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2428 2014-01-24 Jeff Law <law@redhat.com>
2430 PR tree-optimization/59919
2431 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2432 for non-returning calls.
2434 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2436 * common/config/aarch64/aarch64-common.c
2437 (aarch64_rewrite_mcpu): Handle multiple names.
2438 * config/aarch64/aarch64.h
2439 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2441 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2443 * input.c (add_file_to_cache_tab): Handle the case where fopen
2446 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2449 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2450 from push operand if code of push isn't PRE_DEC.
2452 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2455 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2456 -mquad-memory-atomic. Update -mquad-memory documentation to say
2457 it is only used for non-atomic loads/stores.
2459 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2460 -mquad-memory or -mquad-memory-atomic switches.
2462 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2463 -mquad-memory-atomic to ISA 2.07 support.
2465 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2466 to separate support of normal quad word memory operations (ldq, stq)
2467 from the atomic quad word memory operations.
2469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2470 support to separate non-atomic quad word operations from atomic
2471 quad word operations. Disable non-atomic quad word operations in
2472 little endian mode so that we don't have to swap words after the
2473 load and before the store.
2474 (quad_load_store_p): Add comment about atomic quad word support.
2475 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2476 options printed with -mdebug=reg.
2478 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2479 -mquad-memory-atomic as the test for whether we have quad word
2480 atomic instructions.
2481 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2482 or -mp8-vector are used, allow byte/half-word atomic operations.
2484 * config/rs6000/sync.md (load_lockedti): Insure that the address
2485 is a proper indexed or indirect address for the lqarx instruction.
2486 On little endian systems, swap the hi/lo registers after the lqarx
2488 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2489 insure the address is valid for the lqarx instruction.
2490 (store_conditionalti): Insure that the address is a proper indexed
2491 or indirect address for the stqcrx. instruction. On little endian
2492 systems, swap the hi/lo registers before doing the stqcrx.
2494 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2495 insure the address is valid for the stqcrx. instruction.
2497 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2498 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2499 type of quad memory support is available.
2501 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2504 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2505 there is a danger of looping.
2507 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2509 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2510 force flag_ira_loop_pressure if set via command line.
2512 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2514 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2515 (ashr_simd): New builtin handling DI mode.
2516 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2517 (aarch64_sshr_simddi): New match pattern.
2518 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2519 (vshrd_n_s64): Likewise.
2520 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2522 2014-01-23 Nick Clifton <nickc@redhat.com>
2524 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2525 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2526 favour of mcu specific scripts.
2527 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2530 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2531 Alex Velenko <Alex.Velenko@arm.com>
2533 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2534 (vaddv_s16): Likewise.
2535 (vaddv_s32): Likewise.
2536 (vaddv_u8): Likewise.
2537 (vaddv_u16): Likewise.
2538 (vaddv_u32): Likewise.
2539 (vaddvq_s8): Likewise.
2540 (vaddvq_s16): Likewise.
2541 (vaddvq_s32): Likewise.
2542 (vaddvq_s64): Likewise.
2543 (vaddvq_u8): Likewise.
2544 (vaddvq_u16): Likewise.
2545 (vaddvq_u32): Likewise.
2546 (vaddvq_u64): Likewise.
2547 (vaddv_f32): Likewise.
2548 (vaddvq_f32): Likewise.
2549 (vaddvq_f64): Likewise.
2550 (vmaxv_f32): Likewise.
2551 (vmaxv_s8): Likewise.
2552 (vmaxv_s16): Likewise.
2553 (vmaxv_s32): Likewise.
2554 (vmaxv_u8): Likewise.
2555 (vmaxv_u16): Likewise.
2556 (vmaxv_u32): Likewise.
2557 (vmaxvq_f32): Likewise.
2558 (vmaxvq_f64): Likewise.
2559 (vmaxvq_s8): Likewise.
2560 (vmaxvq_s16): Likewise.
2561 (vmaxvq_s32): Likewise.
2562 (vmaxvq_u8): Likewise.
2563 (vmaxvq_u16): Likewise.
2564 (vmaxvq_u32): Likewise.
2565 (vmaxnmv_f32): Likewise.
2566 (vmaxnmvq_f32): Likewise.
2567 (vmaxnmvq_f64): Likewise.
2568 (vminv_f32): Likewise.
2569 (vminv_s8): Likewise.
2570 (vminv_s16): Likewise.
2571 (vminv_s32): Likewise.
2572 (vminv_u8): Likewise.
2573 (vminv_u16): Likewise.
2574 (vminv_u32): Likewise.
2575 (vminvq_f32): Likewise.
2576 (vminvq_f64): Likewise.
2577 (vminvq_s8): Likewise.
2578 (vminvq_s16): Likewise.
2579 (vminvq_s32): Likewise.
2580 (vminvq_u8): Likewise.
2581 (vminvq_u16): Likewise.
2582 (vminvq_u32): Likewise.
2583 (vminnmv_f32): Likewise.
2584 (vminnmvq_f32): Likewise.
2585 (vminnmvq_f64): Likewise.
2587 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2589 * config/aarch64/aarch64-simd.md
2590 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2591 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2592 (*aarch64_mul3_elt<mode>): Likewise.
2593 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2594 (*aarch64_mul3_elt_to_64v2df): Likewise.
2595 (*aarch64_mla_elt<mode>): Likewise.
2596 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2597 (*aarch64_mls_elt<mode>): Likewise.
2598 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2599 (*aarch64_fma4_elt<mode>): Likewise.
2600 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2601 (*aarch64_fma4_elt_to_64v2df): Likewise.
2602 (*aarch64_fnma4_elt<mode>): Likewise.
2603 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2604 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2605 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2606 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2607 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2608 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2609 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2610 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2611 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2613 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2615 * config/aarch64/aarch64-simd.md
2616 (aarch64_be_checked_get_lane<mode>): New define_expand.
2617 * config/aarch64/aarch64-simd-builtins.def
2618 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2619 New builtin definition.
2620 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2621 Use new safe be builtin.
2623 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2625 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2627 (aarch64_be_st1<mode>): Likewise.
2628 (aarch_ld1<VALL:mode>): Define_expand modified.
2629 (aarch_st1<VALL:mode>): Likewise.
2630 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2631 (UNSPEC_ST1): Likewise.
2633 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2635 * config/microblaze/microblaze.md: Add trap insn and attribute
2637 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2639 PR preprocessor/58580
2640 * input.h (location_get_source_line): Take an additional line_size
2642 (void diagnostics_file_cache_fini): Declare new function.
2643 * input.c (struct fcache): New type.
2644 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2645 New static constants.
2646 (diagnostic_file_cache_init, total_lines_num)
2647 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2648 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2649 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2650 (get_next_line, read_next_line, goto_next_line, read_line_num):
2651 New static function definitions.
2652 (diagnostic_file_cache_fini): New function.
2653 (location_get_source_line): Take an additional output line_len
2654 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2656 * diagnostic.c (diagnostic_finish): Call
2657 diagnostic_file_cache_fini.
2658 (adjust_line): Take an additional input parameter for the length
2659 of the line, rather than calculating it with strlen.
2660 (diagnostic_show_locus): Adjust the use of
2661 location_get_source_line and adjust_line with respect to their new
2662 signature. While displaying a line now, do not stop at the first
2663 null byte. Rather, display the zero byte as a space and keep
2664 going until we reach the size of the line.
2665 * Makefile.in: Add vec.o to OBJS-libcommon
2667 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2668 Ilya Tocar <ilya.tocar@intel.com>
2670 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2671 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2672 (__builtin_ia32_kmov16): Ditto.
2673 * config/i386/i386.md (UNSPEC_KMOV): New.
2676 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2678 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2679 (_mm512_storeu_si512): Ditto.
2681 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2684 * rtl.h (get_referenced_operands): Declare.
2685 * recog.c (get_referenced_operands): New function.
2686 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2687 operands have been referenced when recording LO_SUM references.
2689 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2691 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2693 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2695 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2696 Enable for generic and recent AMD targets.
2698 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2700 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2701 ARG_SIZE note when adjustment was eliminated.
2703 2014-01-22 Jeff Law <law@redhat.com>
2705 PR tree-optimization/59597
2706 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2707 in file. Accept new argument REGISTERING and use it to modify
2708 dump output appropriately.
2709 (register_jump_thread): Corresponding changes.
2710 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2711 thread paths involving joiner blocks. Add code to dump cancelled
2712 jump threading paths.
2714 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2716 PR rtl-optimization/59477
2717 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2718 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2720 2014-01-22 Tom Tromey <tromey@redhat.com>
2722 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2724 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2726 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2728 PR rtl-optimization/59896
2729 * lra-constraints.c (process_alt_operands): Check unused note for
2730 matched operands of insn with no output reloads.
2732 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2734 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2735 (mips_move_from_gpr_cost): Likewise.
2737 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2739 PR rtl-optimization/59858
2740 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2741 ira_class_hard_regs_num.
2742 (process_alt_operands): Increase reject for dying matched operand.
2744 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2747 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2748 smaller than size, perform several stores or loads and stores
2749 at dst + count - size to store or copy all of size bytes, rather
2750 than just last modesize bytes.
2752 2014-01-20 DJ Delorie <dj@redhat.com>
2754 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2755 that CLOBBERs are REGs before propogating their values.
2757 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2760 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2761 (cgraph_inline_failed_type): New function.
2762 * cgraph.h (DEFCIFCODE): Add type.
2763 (cgraph_inline_failed_type_t): New enum.
2764 (cgraph_inline_failed_type): New prototype.
2765 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2766 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2767 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2768 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2769 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2770 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2771 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2772 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2773 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2774 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2775 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2776 OPTIMIZATION_MISMATCH.
2777 * tree-inline.c (expand_call_inline): Emit errors during
2778 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2780 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2783 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2784 mode attribute in insn output.
2786 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2788 * output.h (output_constant): Delete.
2789 * varasm.c (output_constant): Make private.
2791 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2793 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2795 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2798 * tree.h (fold_builtin_strcat): New prototype.
2799 * builtins.c (fold_builtin_strcat): No longer static. Add len
2800 argument, if non-NULL, don't call c_strlen. Optimize
2801 directly into __builtin_memcpy instead of __builtin_strcpy.
2802 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2803 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2805 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2807 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2808 for SImode_address_operand operands, having only a REG argument.
2810 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2812 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2813 loader name using mbig-endian.
2814 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2816 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2818 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2822 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2824 * config/aarch64/aarch64-protos.h
2825 (aarch64_cannot_change_mode_class_ptr): Declare.
2826 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2827 aarch64_cannot_change_mode_class_ptr): New.
2828 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2829 backend hook aarch64_cannot_change_mode_class.
2831 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2833 * common/config/aarch64/aarch64-common.c
2834 (aarch64_handle_option): Don't handle any option order logic here.
2835 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2836 selected_cpu, warn on architecture version mismatch.
2837 (aarch64_override_options): Fix parsing order for option strings.
2839 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2840 Iain Sandoe <iain@codesourcery.com>
2843 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2844 warning. Amend comment to reflect current functionality.
2846 2014-01-20 Richard Biener <rguenther@suse.de>
2849 * builtins.c (fold_builtin_strcat): Remove case better handled
2850 by tree-ssa-strlen.c.
2852 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2854 * config/aarch64/aarch64.opt
2855 (mcpu, march, mtune): Make case-insensitive.
2857 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2860 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2861 if operands[1] is a REG or ZERO_EXTEND of a REG.
2863 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2865 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2867 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2869 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2870 long non-pic millicode calls.
2872 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2874 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2876 2014-01-19 Kito Cheng <kito@0xlab.org>
2878 * builtins.c (expand_movstr): Check movstr expand done or fail.
2880 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2881 H.J. Lu <hongjiu.lu@intel.com>
2884 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2885 to DImode for zero-extended addresses.
2887 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2889 PR rtl-optimization/57763
2890 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2891 on the new indirect jump_insn and increment LABEL_NUSES (label).
2893 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2897 * config.gcc (x86_archs): New variable.
2898 (x86_64_archs): Likewise.
2899 (x86_cpus): Likewise.
2900 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2901 --with-arch/--with-cpu= options.
2902 Support --with-arch=/--with-cpu={nehalem,westmere,
2903 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2905 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2907 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2908 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2910 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2912 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2914 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2917 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2918 clear cpp_get_options (parse_in)->warn_unused_macros for
2919 ix86_target_macros_internal with cpp_define.
2921 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2923 * jump.c (delete_related_insns): Keep (use (insn))s.
2924 * reorg.c (redundant_insn): Check for barriers too.
2926 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2928 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2930 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2932 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2933 call to $$dyncall when TARGET_LONG_CALLS is true.
2935 2014-01-17 Jeff Law <law@redhat.com>
2937 * ree.c (combine_set_extension): Temporarily disable test for
2938 changing number of hard registers.
2940 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2943 * ipa-inline-analysis.c (inline_free_summary):
2944 Do not free summary of aliases.
2946 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2949 * gimplify.c (gimplify_expr): Use create_tmp_var
2950 instead of create_tmp_var_raw. If cond doesn't have
2951 integral type, don't add the IFN_ANNOTATE builtin at all.
2953 2014-01-17 Martin Jambor <mjambor@suse.cz>
2956 * ipa-cp.c (prev_edge_clone): New variable.
2957 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2958 Also resize prev_edge_clone vector.
2959 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2960 (ipcp_edge_removal_hook): New function.
2961 (ipcp_driver): Register ipcp_edge_removal_hook.
2963 2014-01-17 Andrew Pinski <apinski@cavium.com>
2964 Steve Ellcey <sellcey@mips.com>
2967 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2970 2014-01-17 Jeff Law <law@redhat.com>
2973 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2974 so that pass_ccp runs first.
2976 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2978 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2979 (ix86_adjust_cost): Use !TARGET_XXX.
2980 (do_reorder_for_imul): Likewise.
2981 (swap_top_of_ready_list): Likewise.
2982 (ix86_sched_reorder): Likewise.
2984 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2986 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2987 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2988 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2989 (intel_memset): New. Duplicate slm_memset.
2990 (intel_cost): New. Duplicate slm_cost.
2991 (m_INTEL): New macro.
2992 (processor_target_table): Add "intel".
2993 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2994 with PROCESSOR_INTEL for "intel".
2995 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2996 PROCESSOR_SILVERMONT.
2997 (ix86_issue_rate): Likewise.
2998 (ix86_adjust_cost): Likewise.
2999 (ia32_multipass_dfa_lookahead): Likewise.
3000 (swap_top_of_ready_list): Likewise.
3001 (ix86_sched_reorder): Likewise.
3002 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3003 instead of TARGET_OPT_AGU.
3004 * config/i386/i386.h (TARGET_INTEL): New.
3005 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3006 (processor_type): Add PROCESSOR_INTEL.
3007 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3008 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3010 2014-01-17 Marek Polacek <polacek@redhat.com>
3013 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3016 2014-01-17 Richard Biener <rguenther@suse.de>
3018 PR tree-optimization/46590
3019 * opts.c (default_options_table): Add entries for
3020 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3021 all enabled at -O1 but not for -Og.
3022 * common.opt (fbranch-count-reg): Remove Init(1).
3023 (fmove-loop-invariants): Likewise.
3024 (ftree-pta): Likewise.
3026 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3028 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3029 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3030 decls to at least the GCC 4.8 used alignments.
3033 * tree-nested.c (convert_nonlocal_reference_stmt,
3034 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3035 of GIMPLE_BIND stmts, adjust associated decls.
3037 2014-01-17 Richard Biener <rguenther@suse.de>
3039 PR tree-optimization/46590
3040 * vec.h (vec<>::bseach): New member function implementing
3041 binary search according to C89 bsearch.
3042 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3043 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3044 bitmap pointer again. Make accesses_in_loop a flat array.
3045 (mem_ref_obstack): New global.
3046 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3047 (mark_ref_stored): Likewise.
3048 (ref_indep_loop_p_2): Likewise.
3049 (set_ref_stored_in_loop): New helper function.
3050 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3051 (memref_free): Adjust.
3052 (record_mem_ref_loc): Simplify.
3053 (gather_mem_refs_stmt): Adjust.
3054 (sort_locs_in_loop_postorder_cmp): New function.
3055 (analyze_memory_references): Sort accesses_in_loop after
3056 loop postorder number.
3057 (find_ref_loc_in_loop_cmp): New function.
3058 (for_all_locs_in_loop): Find relevant cluster of locs in
3059 accesses_in_loop and iterate without recursion.
3060 (execute_sm): Avoid uninit warning.
3061 (struct ref_always_accessed): Simplify.
3062 (ref_always_accessed::operator ()): Likewise.
3063 (ref_always_accessed_p): Likewise.
3064 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3065 loop postorder numbers here.
3066 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3069 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3072 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3073 on decls for which assemble_alias has been called.
3075 2014-01-17 Nick Clifton <nickc@redhat.com>
3077 * config/msp430/msp430.opt: (mcpu): New option.
3078 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3079 (msp430_option_override): Parse target_cpu. If the MCU name
3080 matches a generic string, clear target_mcu.
3081 (msp430_attr): Allow numeric interrupt values up to 63.
3082 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3083 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3085 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3087 * config/msp430/msp430.md (popm): Use %J rather than %I.
3088 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3089 (addhi_cy_i): Use immediate_operand for operand 2.
3090 * doc/invoke.texi: Document -mcpu option.
3092 2014-01-17 Richard Biener <rguenther@suse.de>
3094 PR rtl-optimization/38518
3095 * df.h (df_analyze_loop): Declare.
3096 * df-core.c: Include cfgloop.h.
3097 (df_analyze_1): Split out main part of df_analyze.
3098 (df_analyze): Adjust.
3099 (loop_inverted_post_order_compute): New function.
3100 (loop_post_order_compute): Likewise.
3101 (df_analyze_loop): New function avoiding whole-function
3103 * loop-invariant.c (find_defs): Use df_analyze_loop.
3104 (find_invariants): Adjust.
3105 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3107 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3109 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3110 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3112 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3114 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3115 traversal when removing references.
3117 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3120 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3122 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3125 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3126 pushing a reload for an autoinc when we had previously reloaded an
3127 inner part of the address.
3129 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3131 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3133 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3134 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3135 when not giving up or versioning for alias only because of
3137 (vect_analyze_data_ref_dependences): Set to true.
3138 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3140 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3141 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3145 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3148 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3149 operand 0 predicate for gathers, use a new pseudo as subtarget.
3151 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3154 * lra-constraints.c (process_alt_operands): Add printing debug info.
3155 Check absence of input/output reloads for matched operands too.
3157 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3159 PR rtl-optimization/59835
3160 * ira.c (ira_init_register_move_cost): Increase cost for
3163 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3165 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3167 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3170 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3171 non-register objects. Use gen_(high/low)part more consistently.
3174 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3177 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3178 endian support, remove tests for WORDS_BIG_ENDIAN.
3179 (p8_mfvsrd_3_<mode>): Likewise.
3180 (reload_gpr_from_vsx<mode>): Likewise.
3181 (reload_gpr_from_vsxsf): Likewise.
3182 (p8_mfvsrd_4_disf): Likewise.
3184 2014-01-16 Richard Biener <rguenther@suse.de>
3186 PR rtl-optimization/46590
3187 * lcm.c (compute_antinout_edge): Use postorder iteration.
3188 (compute_laterin): Use inverted postorder iteration.
3190 2014-01-16 Nick Clifton <nickc@redhat.com>
3193 * varasm.c (output_constant): Return the number of bytes actually
3195 (output_constructor_array_range): Update the field size with the
3196 number of bytes emitted by output_constant.
3197 (output_constructor_regular_field): Likewise. Also do not
3198 complain if the total number of bytes emitted is now greater
3199 than the expected fieldpos.
3200 * output.h (output_constant): Update prototype and descriptive comment.
3202 2014-01-16 Marek Polacek <polacek@redhat.com>
3205 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3206 it is error_mark_node.
3208 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3210 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3211 VALID_AVX256_REG_OR_OI_MODE.
3213 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3215 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3216 current procedure should be profiled.
3218 2014-01-15 Andrew Pinski <apinski@cavium.com>
3220 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3221 of moving from/to the STACK_REG register class.
3223 2014-01-15 Richard Henderson <rth@redhat.com>
3226 * reginfo.c (global_regs_decl): Globalize.
3227 * rtl.h (global_regs_decl): Declare.
3228 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3229 reserved via global_regs.
3231 2014-01-15 Teresa Johnson <tejohnson@google.com>
3233 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3235 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3237 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3238 and vmulosh rather than call gen_vec_widen_smult_*.
3239 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3240 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3241 (vec_widen_smult_even_v16qi): Likewise.
3242 (vec_widen_umult_even_v8hi): Likewise.
3243 (vec_widen_smult_even_v8hi): Likewise.
3244 (vec_widen_umult_odd_v16qi): Likewise.
3245 (vec_widen_smult_odd_v16qi): Likewise.
3246 (vec_widen_umult_odd_v8hi): Likewise.
3247 (vec_widen_smult_odd_v8hi): Likewise.
3248 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3249 vmuloub rather than call gen_vec_widen_umult_*.
3250 (vec_widen_umult_lo_v16qi): Likewise.
3251 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3252 vmulosb rather than call gen_vec_widen_smult_*.
3253 (vec_widen_smult_lo_v16qi): Likewise.
3254 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3255 rather than call gen_vec_widen_umult_*.
3256 (vec_widen_umult_lo_v8hi): Likewise.
3257 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3258 rather than call gen_vec_widen_smult_*.
3259 (vec_widen_smult_lo_v8hi): Likewise.
3261 2014-01-15 Jeff Law <law@redhat.com>
3263 PR tree-optimization/59747
3264 * ree.c (find_and_remove_re): Properly handle case where a second
3265 eliminated extension requires widening a copy created for elimination
3266 of a prior extension.
3267 (combine_set_extension): Ensure that the number of hard regs needed
3268 for a destination register does not change when we widen it.
3270 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3272 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3273 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3274 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3275 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3276 (avr-*-rtems*): Likewise.
3277 (bfin*-rtems*): Likewise.
3278 (moxie-*-rtems*): Likewise.
3279 (h8300-*-rtems*): Likewise.
3280 (i[34567]86-*-rtems*): Likewise.
3281 (lm32-*-rtems*): Likewise.
3282 (m32r-*-rtems*): Likewise.
3283 (m68k-*-rtems*): Likewise.
3284 (microblaze*-*-rtems*): Likewise.
3285 (mips*-*-rtems*): Likewise.
3286 (powerpc-*-rtems*): Likewise.
3287 (sh-*-rtems*): Likewise.
3288 (sparc-*-rtems*): Likewise.
3289 (sparc64-*-rtems*): Likewise.
3290 (v850-*-rtems*): Likewise.
3291 (m32c-*-rtems*): Likewise.
3293 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3295 PR rtl-optimization/59511
3296 * ira.c (ira_init_register_move_cost): Use memory costs for some
3297 cases of register move cost calculations.
3298 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3299 instead of BB frequency.
3300 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3301 * lra-assigns.c (find_hard_regno_for): Ditto.
3303 2014-01-15 Richard Biener <rguenther@suse.de>
3305 PR tree-optimization/59822
3306 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3307 (vectorizable_load): Use it to hoist defs of uses of invariant
3308 loads out of the loop.
3310 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3311 Kugan Vivekanandarajah <kuganv@linaro.org>
3314 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3317 2014-01-15 Richard Biener <rguenther@suse.de>
3319 PR rtl-optimization/59802
3320 * lcm.c (compute_available): Use inverted postorder to seed
3321 the initial worklist.
3323 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3326 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3327 ADDR_REGS for invalid symrefs in non-PIC code.
3329 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3332 * builtins.c (determine_block_size): Initialize *probable_max_size
3333 even if len_rtx is CONST_INT.
3335 2014-01-14 Andrew Pinski <apinski@cavium.com>
3337 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3338 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3339 (cortexa53_tunings): Likewise.
3340 (aarch64_sched_issue_rate): New function.
3341 (TARGET_SCHED_ISSUE_RATE): Define.
3343 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3345 * ira-costs.c (find_costs_and_classes): Add missed
3346 ira_init_register_move_cost_if_necessary.
3348 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3351 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3353 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3356 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3357 to indicate if type is used for function return value. Warn ABI
3358 change if the vector mode isn't available for function return value.
3359 (ix86_function_arg_advance): Pass false to type_natural_mode.
3360 (ix86_function_arg): Likewise.
3361 (ix86_gimplify_va_arg): Likewise.
3362 (function_arg_32): Don't warn ABI change.
3363 (ix86_function_value): Pass true to type_natural_mode.
3364 (ix86_return_in_memory): Likewise.
3365 (ix86_struct_value_rtx): Removed.
3366 (TARGET_STRUCT_VALUE_RTX): Likewise.
3368 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3370 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3371 converting a conditional jump into a conditional return.
3373 2014-01-14 Richard Biener <rguenther@suse.de>
3375 PR tree-optimization/58921
3376 PR tree-optimization/59006
3377 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3378 hoisting invariant stmts.
3379 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3380 invariant loads on the preheader edge if possible.
3382 2014-01-14 Joey Ye <joey.ye@arm.com>
3384 * doc/plugin.texi (Building GCC plugins): Update to C++.
3386 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3388 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3389 (_mm_rcp28_round_ss): Ditto.
3390 (_mm_rsqrt28_round_sd): Ditto.
3391 (_mm_rsqrt28_round_ss): Ditto.
3392 (_mm_rcp28_sd): Ditto.
3393 (_mm_rcp28_ss): Ditto.
3394 (_mm_rsqrt28_sd): Ditto.
3395 (_mm_rsqrt28_ss): Ditto.
3396 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3397 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3398 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3399 (IX86_BUILTIN_RCP28SD): Ditto.
3400 (IX86_BUILTIN_RCP28SS): Ditto.
3401 (IX86_BUILTIN_RSQRT28SD): Ditto.
3402 (IX86_BUILTIN_RSQRT28SS): Ditto.
3403 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3404 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3405 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3406 (ix86_expand_special_args_builtin): Expand new FTYPE.
3407 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3408 (srcp14<mode>): Make insn unary.
3409 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3410 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3411 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3412 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3413 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3414 Fix rounding: make it SAE only.
3415 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3417 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3419 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3420 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3421 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3422 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3423 (round_saeonly_mask_scalar_operand4): Ditto.
3424 (round_saeonly_mask_scalar_op3): Ditto.
3425 (round_saeonly_mask_scalar_op4): Ditto.
3427 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3429 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3430 Implement -maltivec=be for vec_insert and vec_extract.
3432 2014-01-10 DJ Delorie <dj@redhat.com>
3434 * config/msp430/msp430.md (call_internal): Don't allow memory
3435 references with SP as the base register.
3436 (call_value_internal): Likewise.
3437 * config/msp430/constraints.md (Yc): New. For memory references
3438 that don't use SP as a base register.
3440 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3441 "an integer without a # prefix"
3442 * config/msp430/msp430.md (epilogue_helper): Use it.
3444 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3447 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3448 AVX512F gather builtins.
3449 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3450 on gather decls with INTEGER_TYPE masktype.
3451 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3452 directly into the builtin rather than hoisting it before loop.
3454 PR tree-optimization/59387
3455 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3456 (scev_const_prop): If folded_casts and type has undefined overflow,
3457 use force_gimple_operand instead of force_gimple_operand_gsi and
3458 for each added stmt if it is assign with
3459 arith_code_with_undefined_signed_overflow, call
3460 rewrite_to_defined_overflow.
3461 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3462 gimple-fold.h instead.
3463 (arith_code_with_undefined_signed_overflow,
3464 rewrite_to_defined_overflow): Moved to ...
3465 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3466 rewrite_to_defined_overflow): ... here. No longer static.
3467 Include gimplify-me.h.
3468 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3469 rewrite_to_defined_overflow): New prototypes.
3471 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3473 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3475 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3477 * builtins.c (get_object_alignment_2): Minor tweak.
3478 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3480 2014-01-13 Christian Bruel <christian.bruel@st.com>
3482 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3483 optimized non constant lengths.
3485 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3488 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3489 load as __atomic_load_N if possible.
3491 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3493 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3495 (rs6000_expand_builtin): Adjust call.
3497 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3500 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3501 * config/rs6000/rs6000.c: Include target-globals.h.
3502 (rs6000_set_current_function): Instead of doing target_reinit
3503 unconditionally, use save_target_globals_default_opts and
3504 restore_target_globals.
3506 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3508 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3509 (rs6000_expand_builtin): Handle mffs and mtfsf.
3510 (rs6000_init_builtins): Define mffs and mtfsf.
3511 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3512 (rs6000_mffs): New pattern.
3513 (rs6000_mtfsf): New pattern.
3515 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3517 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3518 Start narrowing with START. Apply candidate-use pair
3519 and check overall cost in narrowing.
3520 (iv_ca_prune): Pass new argument.
3522 2014-01-10 Jeff Law <law@redhat.com>
3525 * ree.c (combine_reaching_defs): Ensure the defining statement
3526 occurs before the extension when optimizing extensions with
3527 different source and destination hard registers.
3529 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3532 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3533 vtables into the type inheritance graph.
3535 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3537 PR rtl-optimization/59754
3538 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3539 modes in the REGNO != REGNO case.
3541 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3543 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3545 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3547 PR tree-optimization/59745
3548 * tree-predcom.c (tree_predictive_commoning_loop): Call
3549 free_affine_expand_cache if giving up because components is NULL.
3551 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3552 GC in payload of target_globals struct instead of allocating them on
3553 the heap and the larger structs separately using GC.
3554 * target-globals.h (struct target_globals): Make regs, hard_regs,
3555 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3556 of GTY((skip)) and change type to void *.
3557 (reset_target_globals): Cast loads from those fields to corresponding
3560 2014-01-10 Steve Ellcey <sellcey@mips.com>
3563 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3564 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3565 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3567 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3570 * aarch64-modes.def (CC_Zmode): New flags mode.
3571 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3572 represents an equality.
3573 (aarch64_get_condition_code): Handle CC_Zmode.
3574 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3576 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3578 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3579 extraction in good case.
3581 2014-01-10 Richard Biener <rguenther@suse.de>
3583 PR tree-optimization/59374
3584 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3585 checking after SLP discovery. Mark stmts not participating
3586 in any SLP instance properly.
3588 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3590 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3591 when handling a SET rtx.
3593 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3595 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3596 (cortex-a57): Likewise.
3597 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3599 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3601 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3602 non-iwmmxt builtins.
3604 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3608 * ipa-devirt.c record_target_from_binfo): Take as argument
3609 stack of binfos and lookup matching one for virtual inheritance.
3610 (possible_polymorphic_call_targets_1): Update.
3612 2014-01-10 Huacai Chen <chenhc@lemote.com>
3614 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3615 kernel strings for Loongson-2E/2F/3A.
3617 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3620 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3621 is_gimple_call before calling gimple_call_internal_p.
3623 2014-01-09 Steve Ellcey <sellcey@mips.com>
3625 * Makefile.in (TREE_FLOW_H): Remove.
3626 (TREE_SSA_H): Add file names from tree-flow.h.
3627 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3628 * tree.h: Remove tree-flow.h reference.
3629 * hash-table.h: Remove tree-flow.h reference.
3630 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3631 reference with tree-ssa-loop.h.
3633 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3635 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3636 default element-order behavior for -maltivec.
3637 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3638 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3639 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3640 when targeting big endian, at least for now.
3641 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3643 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3646 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3647 var satisfies use_register_for_decl, just take into account type
3648 alignment, rather than decl alignment.
3650 PR tree-optimization/59622
3651 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3652 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3653 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3654 Don't devirtualize for inplace at all. For targets.length () == 1,
3655 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3657 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3659 * config/i386/i386.md (cpu): Remove the unused btver1.
3661 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3663 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3665 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3668 * tree-core.h (struct target_globals): New forward declaration.
3669 (struct tree_target_option): Add globals field.
3670 * tree.h (TREE_TARGET_GLOBALS): Define.
3671 (prepare_target_option_nodes_for_pch): New prototype.
3672 * target-globals.h (struct target_globals): Define even if
3674 * tree.c (prepare_target_option_node_for_pch,
3675 prepare_target_option_nodes_for_pch): New functions.
3676 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3677 * config/i386/i386.c: Include target-globals.h.
3678 (ix86_set_current_function): Instead of doing target_reinit
3679 unconditionally, use save_target_globals_default_opts and
3680 restore_target_globals.
3682 2014-01-09 Richard Biener <rguenther@suse.de>
3684 PR tree-optimization/59715
3685 * tree-cfg.h (split_critical_edges): Declare.
3686 * tree-cfg.c (split_critical_edges): Export.
3687 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3689 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3691 * cfgexpand.c (expand_stack_vars): Optionally disable
3692 asan stack protection.
3693 (expand_used_vars): Likewise.
3694 (partition_stack_vars): Likewise.
3695 * asan.c (asan_emit_stack_protection): Optionally disable
3696 after return stack usage.
3697 (instrument_derefs): Optionally disable memory access instrumentation.
3698 (instrument_builtin_call): Likewise.
3699 (instrument_strlen_call): Likewise.
3700 (asan_protect_global): Optionally disable global variables protection.
3701 * doc/invoke.texi: Added doc for new options.
3702 * params.def: Added new options.
3703 * params.h: Likewise.
3705 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3707 PR rtl-optimization/59724
3708 * ifcvt.c (cond_exec_process_if_block): Don't call
3709 flow_find_head_matching_sequence with 0 longest_match.
3710 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3711 non-active insns if !stop_after.
3712 (try_head_merge_bb): Revert 2014-01-07 changes.
3714 2014-01-08 Jeff Law <law@redhat.com>
3716 * ree.c (get_sub_rtx): New function, extracted from...
3717 (merge_def_and_ext): Here.
3718 (combine_reaching_defs): Use get_sub_rtx.
3720 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3722 * cgraph.h (varpool_variable_node): Do not choke on null node.
3724 2014-01-08 Catherine Moore <clm@codesourcery.com>
3726 * config/mips/mips.md (simple_return): Attempt to use JRC
3728 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3730 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3732 PR rtl-optimization/59137
3733 * reorg.c (steal_delay_list_from_target): Call update_block for
3735 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3737 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3739 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3740 two duplicate entries.
3742 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3745 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3747 * config/mips/mips.c (mips_truncated_op_cost): New function.
3748 (mips_rtx_costs): Adjust test for BADDU.
3749 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3751 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3753 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3754 (*baddu_si): ...this new pattern.
3756 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3759 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3761 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3764 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3766 (expand_expr, expand_normal): Adjust.
3767 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3768 inner_reference_p. Use inner_reference_p to expand inner references.
3769 (store_expr): Adjust.
3770 * cfgexpand.c (expand_call_stmt): Adjust.
3772 2014-01-08 Rong Xu <xur@google.com>
3774 * gcov-io.c (gcov_var): Move from gcov-io.h.
3775 (gcov_position): Ditto.
3776 (gcov_is_error): Ditto.
3777 (gcov_rewrite): Ditto.
3778 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3779 only part to libgcc/libgcov.h.
3781 2014-01-08 Marek Polacek <polacek@redhat.com>
3784 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3786 2014-01-08 Marek Polacek <polacek@redhat.com>
3789 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3791 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3793 PR rtl-optimization/59649
3794 * stor-layout.c (get_mode_bounds): For BImode return
3795 0 and STORE_FLAG_VALUE.
3797 2014-01-08 Richard Biener <rguenther@suse.de>
3800 * gimple.h (is_gimple_builtin_call): Remove.
3801 (gimple_builtin_call_types_compatible_p): New.
3802 (gimple_call_builtin_p): New overload.
3803 * gimple.c (is_gimple_builtin_call): Remove.
3804 (validate_call): Rename to ...
3805 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3807 (validate_type): New static function.
3808 (gimple_call_builtin_p): New overload and adjust.
3809 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3810 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3811 (gimple_fold_stmt_to_constant_1): Likewise.
3812 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3814 2014-01-08 Richard Biener <rguenther@suse.de>
3817 * gimplify.c (gimplify_expr): Gimplify register-register type
3818 VIEW_CONVERT_EXPRs to separate stmts.
3820 2014-01-07 Jeff Law <law@redhat.com>
3823 * ree.c (combine_set_extension): Handle case where source
3824 and destination registers in an extension insn are different.
3825 (combine_reaching_defs): Allow source and destination registers
3826 in extension to be different under limited circumstances.
3827 (add_removable_extension): Remove restriction that the
3828 source and destination registers in the extension are the same.
3829 (find_and_remove_re): Emit a copy from the extension's
3830 destination to its source after the defining insn if
3831 the source and destination registers are different.
3834 * ifcvt.c (merge_if_block): If we are merging a block with more than
3835 one successor with a block with no successors, remove any BARRIER
3836 after the second block.
3838 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3840 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3842 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3845 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3846 for 14-bit register offsets when INT14_OK_STRICT is false.
3848 2014-01-07 Roland Stigge <stigge@antcom.de>
3849 Michael Meissner <meissner@linux.vnet.ibm.com>
3852 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3853 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3855 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3857 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3860 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3862 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3863 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3864 rtx is const0_rtx or not.
3866 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3869 * target-globals.c (save_target_globals): Remove this_fn_optab
3871 * toplev.c: Include optabs.h.
3872 (target_reinit): Temporarily restore the global options if another
3873 set of options are in force.
3875 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3877 PR rtl-optimization/58668
3878 * cfgcleanup.c (flow_find_cross_jump): Don't count
3879 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3880 to determine what is counted.
3881 (flow_find_head_matching_sequence): Use active_insn_p to determine
3883 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3885 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3886 determine what is counted.
3888 PR tree-optimization/59643
3889 * tree-predcom.c (split_data_refs_to_components): If one dr is
3890 read and one write, determine_offset fails and the write isn't
3891 in the bad component, just put the read into the bad component.
3893 2014-01-07 Mike Stump <mikestump@comcast.net>
3894 Jakub Jelinek <jakub@redhat.com>
3897 * tree-core.h (struct tree_optimization_option): Change optabs
3898 type from unsigned char * to void *.
3899 * optabs.c (init_tree_optimization_optabs): Adjust
3900 TREE_OPTIMIZATION_OPTABS initialization.
3902 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3905 * config/i386/i386.h (struct machine_function): Add
3906 no_drap_save_restore field.
3907 * config/i386/i386.c (ix86_save_reg): Use
3908 !cfun->machine->no_drap_save_restore instead of
3909 crtl->stack_realign_needed.
3910 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3911 this function clears frame_pointer_needed. Set
3912 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3913 and DRAP reg is needed.
3915 2014-01-06 Marek Polacek <polacek@redhat.com>
3918 * doc/implement-c.texi: Mention that other integer types are
3919 permitted as bit-field types in strictly conforming mode.
3921 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3923 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3926 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3928 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3930 2014-01-06 Martin Jambor <mjambor@suse.cz>
3933 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3935 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3937 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3941 * var-tracking.c (add_stores): Preserve the value of the source even if
3942 we don't record the store.
3944 2014-01-06 Terry Guo <terry.guo@arm.com>
3946 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3948 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3951 * config/darwin.c (darwin_function_section): Adjust return values to
3952 correspond to optimisation changes made in r206070.
3954 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3956 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3957 from prefetch_block tune setting.
3958 (nocona_cost): Correct size of prefetch block to 64.
3960 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3962 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3963 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3964 used to save the static chain register in the computation of the offset
3965 from which the FP registers need to be restored.
3967 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3969 PR tree-optimization/59519
3970 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3971 ICE if get_current_def (current_new_name) is already non-NULL, as long
3972 as it is a phi result of some other phi in *new_exit_bb that has
3975 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3976 or vmovdqu* for misaligned_operand.
3977 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3978 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3979 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3980 aligned_mem for AVX512F masked aligned load and store builtins and for
3983 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3985 PR tree-optimization/59651
3986 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3987 Address range for negative step should be added by TYPE_SIZE_UNIT.
3989 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3991 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3992 overlapping registers also for registers other than the stack pointer.
3994 2014-01-03 Marek Polacek <polacek@redhat.com>
3997 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4000 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4003 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4006 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4007 (push splitter): Use <P:MODE_SIZE> instead of
4008 GET_MODE_SIZE (<P:MODE>mode).
4009 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4010 (mov -1, reg peephole2): Likewise.
4011 * config/i386/sse.md (*mov<mode>_internal,
4012 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4013 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4014 *<code><mode>3, *andnot<mode>3<mask_name>,
4015 <mask_codefor><code><mode>3<mask_name>): Likewise.
4016 * config/i386/subst.md (mask_mode512bit_condition,
4017 sd_mask_mode512bit_condition): Likewise.
4019 2014-01-02 Xinliang David Li <davidxl@google.com>
4021 PR tree-optimization/59303
4022 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4023 (dump_predicates): Better output format.
4024 (pred_equal_p): New function.
4025 (is_neq_relop_p): Ditto.
4026 (is_neq_zero_form_p): Ditto.
4027 (pred_expr_equal_p): Ditto.
4028 (pred_neg_p): Ditto.
4029 (simplify_pred): Ditto.
4030 (simplify_preds_2): Ditto.
4031 (simplify_preds_3): Ditto.
4032 (simplify_preds_4): Ditto.
4033 (simplify_preds): Ditto.
4035 (push_to_worklist): Ditto.
4036 (get_pred_info_from_cmp): Ditto.
4037 (is_degenerated_phi): Ditto.
4038 (normalize_one_pred_1): Ditto.
4039 (normalize_one_pred): Ditto.
4040 (normalize_one_pred_chain): Ditto.
4041 (normalize_preds): Ditto.
4042 (normalize_cond_1): Remove function.
4043 (normalize_cond): Ditto.
4044 (is_gcond_subset_of): Ditto.
4045 (is_subset_of_any): Ditto.
4046 (is_or_set_subset_of): Ditto.
4047 (is_and_set_subset_of): Ditto.
4048 (is_norm_cond_subset_of): Ditto.
4049 (pred_chain_length_cmp): Ditto.
4050 (convert_control_dep_chain_into_preds): Type change.
4051 (find_predicates): Ditto.
4052 (find_def_preds): Ditto.
4053 (destroy_predicates_vecs): Ditto.
4054 (find_matching_predicates_in_rest_chains): Ditto.
4055 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4056 (is_pred_expr_subset): Ditto.
4057 (is_pred_chain_subset_of): Ditto.
4058 (is_included_in): Ditto.
4059 (is_superset_of): Ditto.
4061 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4063 Update copyright years.
4065 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4067 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4068 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4069 config/arc/arc.md, config/arc/arc.opt,
4070 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4071 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4072 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4073 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4074 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4075 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4076 vtable-verify.c, vtable-verify.h: Use the standard form for the
4079 2014-01-02 Tobias Burnus <burnus@net-b.de>
4081 * gcc.c (process_command): Update copyright notice dates.
4082 * gcov-dump.c: Ditto.
4084 * doc/cpp.texi: Bump @copying's copyright year.
4085 * doc/cppinternals.texi: Ditto.
4086 * doc/gcc.texi: Ditto.
4087 * doc/gccint.texi: Ditto.
4088 * doc/gcov.texi: Ditto.
4089 * doc/install.texi: Ditto.
4090 * doc/invoke.texi: Ditto.
4092 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4094 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4096 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4098 * config/i386/sse.md (*mov<mode>_internal): Guard
4099 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4101 PR rtl-optimization/59647
4102 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4103 new_rtx into UNSIGNED_FLOAT rtxes.
4105 Copyright (C) 2014 Free Software Foundation, Inc.
4107 Copying and distribution of this file, with or without modification,
4108 are permitted in any medium without royalty provided the copyright
4109 notice and this notice are preserved.