1 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3 PR rtl-optimization/60292
4 * sel-sched.c (fill_vec_av_set): Do not reset target availability
5 bit fot the fence instruction.
7 2014-02-24 Alangi Derick <alangiderick@gmail.com>
9 * calls.h: Fix typo in comment.
11 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
13 * config/pa/pa.c (pa_output_move_double): Don't valididate when
14 adjusting offsetable addresses.
16 2014-02-24 Guozhi Wei <carrot@google.com>
18 * sparseset.h (sparseset_pop): Fix the wrong index.
20 2014-02-24 Walter Lee <walt@tilera.com>
22 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
23 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
25 * common/config/tilegx/tilegx-common.c
26 (TARGET_DEFAULT_TARGET_FLAGS): Define.
27 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
29 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
30 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
31 (tilegx_gimplify_va_arg_expr): Handle big endian.
32 (tilegx_expand_unaligned_load): Ditto.
33 (tilegx_expand_unaligned_store): Ditto.
34 (TARGET_RETURN_IN_MSB): New.
35 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
36 (TARGET_ENDIAN_DEFAULT): New.
37 (TARGET_BIG_ENDIAN): Handle big endian.
38 (BYTES_BIG_ENDIAN): Ditto.
39 (WORDS_BIG_ENDIAN): Ditto.
40 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
43 * config/tilegx/tilegx.md (extv): Handle big endian.
46 (insn_st<n>_add<bitsuffix>): Ditto.
47 (insn_stnt<n>): Ditto.
48 (insn_stnt<n>_add<bitsuffix>):Ditto.
49 (vec_interleave_highv8qi): Handle big endian.
50 (vec_interleave_highv8qi_be): New.
51 (vec_interleave_highv8qi_le): New.
52 (insn_v1int_h): Handle big endian.
53 (vec_interleave_lowv8qi): Handle big endian.
54 (vec_interleave_lowv8qi_be): New.
55 (vec_interleave_lowv8qi_le): New.
56 (insn_v1int_l): Handle big endian.
57 (vec_interleave_highv4hi): Handle big endian.
58 (vec_interleave_highv4hi_be): New.
59 (vec_interleave_highv4hi_le): New.
60 (insn_v2int_h): Handle big endian.
61 (vec_interleave_lowv4hi): Handle big endian.
62 (vec_interleave_lowv4hi_be): New.
63 (vec_interleave_lowv4hi_le): New.
64 (insn_v2int_l): Handle big endian.
65 (vec_interleave_highv2si): Handle big endian.
66 (vec_interleave_highv2si_be): New.
67 (vec_interleave_highv2si_le): New.
68 (insn_v4int_h): Handle big endian.
69 (vec_interleave_lowv2si): Handle big endian.
70 (vec_interleave_lowv2si_be): New.
71 (vec_interleave_lowv2si_le): New.
72 (insn_v4int_l): Handle big endian.
73 * config/tilegx/tilegx.opt (mbig-endian): New option.
74 (mlittle-endian): New option.
75 * doc/install.texi: Document tilegxbe-linux.
76 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
78 2014-02-24 Martin Jambor <mjambor@suse.cz>
81 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
82 there are no parameter descriptors.
84 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
86 PR rtl-optimization/60268
87 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
89 (sched_rgn_init): ... here.
90 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
92 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
94 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
97 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
99 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
102 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
104 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk
105 and define TARGET_ASM_OUTPUT_MI_THUNK and
106 TARGET_ASM_CAN_OUTPUT_MI_THUNK.
108 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
110 * config/microblaze/predicates.md: Add cmp_op predicate.
111 * config/microblaze/microblaze.md: Add branch_compare instruction
112 which uses cmp_op predicate and emits cmp insn before branch.
113 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
114 to microblaze_expand_conditional_branch and consolidate logic.
115 (microblaze_expand_conditional_branch): emit branch_compare
116 insn instead of handling cmp op separate from branch insn.
118 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
120 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
123 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
125 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
126 define_insn with define_expand and new define_insn
127 *altivec_lve<VI_char>x_internal.
128 (altivec_stve<VI_char>x): Replace define_insn with define_expand
129 and new define_insn *altivec_stve<VI_char>x_internal.
130 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
132 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
134 (altivec_expand_stvex_be): New function.
136 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
138 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
140 to STACK_POINTER_REGNUM if !frame_pointer_needed.
141 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
142 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
144 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
147 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
148 instead of emit_move_insn.
150 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
152 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
154 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
157 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
159 * config/rs6000/altivec.md (altivec_lvxl): Rename as
160 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
162 (altivec_lvxl_<mode>): New define_expand incorporating
163 -maltivec=be semantics where needed.
164 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
165 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
166 semantics where needed.
167 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
168 (altivec_stvx_<mode>): New define_expand incorporating
169 -maltivec=be semantics where needed.
170 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
171 VM2 iterator instead of V4SI.
172 (altivec_stvxl_<mode>): New define_expand incorporating
173 -maltivec=be semantics where needed.
174 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
175 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
176 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
177 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
178 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
180 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
181 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
182 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
183 ALTIVEC_BUILTIN_STVXL.
184 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
186 (altivec_expand_stvx_be): Likewise.
187 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
188 (altivec_expand_lvx_be): Likewise.
189 (altivec_expand_stvx_be): Likewise.
190 (altivec_expand_builtin): Add cases for
191 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
192 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
193 (altivec_init_builtins): Add definitions for
194 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
195 __builtin_altivec_stvx_<mode>, and
196 __builtin_altivec_stvxl_<mode>.
198 2014-02-21 Catherine Moore <clm@codesourcery.com>
200 * doc/invoke.texi (mvirt, mno-virt): Document.
201 * config/mips/mips.opt (mvirt): New option.
202 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
204 2014-02-21 Richard Biener <rguenther@suse.de>
206 PR tree-optimization/60276
207 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
208 (STMT_VINFO_MIN_NEG_DIST): New macro.
209 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
210 STMT_VINFO_MIN_NEG_DIST.
211 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
212 made for negative dependence distances still hold.
214 2014-02-21 Richard Biener <rguenther@suse.de>
217 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
218 DECL_INITIAL for globals not in the current function context.
220 2014-02-21 Jakub Jelinek <jakub@redhat.com>
222 PR tree-optimization/56490
223 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
224 * tree-ssa-uninit.c: Include params.h.
225 (compute_control_dep_chain): Add num_calls argument, return false
226 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
227 num_calls to recursive call.
228 (find_predicates): Change dep_chain into normal array,
229 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
230 variable and adjust compute_control_dep_chain caller.
231 (find_def_preds): Likewise.
233 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
235 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
236 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
238 2014-02-21 Nick Clifton <nickc@redhat.com>
240 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
242 (popqi1): Add mode to pre_dec.
245 2014-02-21 Jakub Jelinek <jakub@redhat.com>
247 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
248 mode for mask of V8SFmode permutation.
250 2014-02-20 Richard Henderson <rth@redhat.com>
253 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
254 a new pseudo for OLDVAL.
256 2014-02-20 Jakub Jelinek <jakub@redhat.com>
259 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
260 gen_reg_rtx if d->testing_p.
261 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
262 if d->testing_p and we will certainly return true.
263 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
266 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
268 * emit-rtl.c (gen_reg_rtx): Assert that
269 crtl->emit.regno_pointer_align_length is non-zero.
271 2014-02-20 Richard Henderson <rth@redhat.com>
274 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
275 on failure the store back into EXPECT.
277 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
278 Sandra Loosemore <sandra@codesourcery.com>
280 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
281 * config/nios2/nios2.c (nios2_function_profiler):
282 Add -fPIC (flag_pic == 2) support.
283 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
284 (nios2_large_offset_p): New function.
285 (nios2_unspec_reloc_p): Move up position, update to use
286 nios2_large_offset_p.
287 (nios2_unspec_address): Remove function.
288 (nios2_unspec_offset): New function.
289 (nios2_large_got_address): New function.
290 (nios2_got_address): Add large offset support.
291 (nios2_legitimize_tls_address): Update usage of removed and new
293 (nios2_symbol_binds_local_p): New function.
294 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
295 (nios2_legitimize_address): Update to use nios2_large_offset_p.
296 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
297 (nios2_print_operand): Merge H/L processing, add hiadj/lo
298 processing for (const (unspec ...)).
299 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
301 2014-02-20 Richard Biener <rguenther@suse.de>
303 * tree-cfg.c (replace_uses_by): Mark altered BBs before
304 doing the substitution.
305 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
307 2014-02-20 Martin Jambor <mjambor@suse.cz>
310 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
311 info when checking whether lattices are bottom.
313 2014-02-20 Richard Biener <rguenther@suse.de>
316 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
319 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
322 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
323 parameter specifying the scaling.
324 (inline_call): Update.
325 (want_inline_recursively): Guard division by zero.
326 (recursive_inlining): Update.
327 * ipa-inline.h (clone_inlined_nodes): Update.
329 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
332 * config/i386/i386.c (classify_argument): Pass structures of size
333 64 bytes or less in register.
335 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
336 Kirill Yukhin <kirill.yukhin@intel.com>
338 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
339 (_mm_rcp28_round_ss): Ditto.
340 (_mm_rsqrt28_round_sd): Ditto.
341 (_mm_rsqrt28_round_ss): Ditto.
342 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
343 (_mm_rcp14_round_ss): Ditto.
344 (_mm_rsqrt14_round_sd): Ditto.
345 (_mm_rsqrt14_round_ss): Ditto.
346 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
347 the first input operand, get rid of match_dup.
348 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
350 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
352 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
353 operand as the first input operand, set type attribute.
354 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
356 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
357 operand as the first input operand, set type attribute.
359 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
361 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
364 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
367 * config/i386/i386.c (construct_container): Remove TFmode check
368 for X86_64_INTEGER_CLASS.
370 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
373 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
374 only when -Wpsabi is enabled.
376 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
379 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
380 passing arrays in registers are the same as for structs, so remove the
381 special case for them.
383 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
385 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
386 destination type, extract only the valid bits if the source type is not
387 integral and has a different mode.
389 2014-02-19 Richard Biener <rguenther@suse.de>
392 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
395 2014-02-19 Richard Biener <rguenther@suse.de>
398 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
399 (ipa_modify_call_arguments): Emit an argument load explicitely and
400 preserve virtual SSA form there and for the replacement call.
401 Do not update SSA form nor free dominance info.
403 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
405 * ipa.c (function_and_variable_visibility): Also clear WEAK
406 flag when disolving COMDAT_GROUP.
408 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
410 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
411 * ipa-prop.c (ipa_set_jf_known_type): Return early when
413 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
414 do more sanity checks.
415 (detect_type_change): Return true when giving up early.
416 (compute_complex_assign_jump_func): Fix type parameter of
418 (compute_complex_ancestor_jump_func): Likewise.
419 (update_jump_functions_after_inlining): Fix updating of
421 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
423 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
425 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
426 inline clones when edge disappears.
428 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
431 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
432 Split 64-bit moves into 2 patterns. Do not allow the use of
433 direct move for TDmode in little endian, since the decimal value
434 has little endian bytes within a word, but the 64-bit pieces are
435 ordered in a big endian fashion, and normal subreg's of TDmode are
437 (mov<mode>_64bit_dm): Likewise.
438 (movtd_64bit_nodm): Likewise.
440 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
442 PR tree-optimization/60174
443 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
444 statement of an SSA_NAME that occurs in an abnormal PHI node.
446 2014-02-18 Jakub Jelinek <jakub@redhat.com>
449 * final.c (SEEN_BB): Remove.
450 (SEEN_NOTE, SEEN_EMITTED): Renumber.
451 (final_scan_insn): Don't force_source_line on second
452 NOTE_INSN_BASIC_BLOCK.
454 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
457 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
458 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
459 (type_natural_mode): Warn ABI change when %zmm register is not
460 available for AVX512F vector value passing.
462 2014-02-18 Kai Tietz <ktietz@redhat.com>
465 * config/i386/i386.c (ix86_expand_prologue): Use value in
466 rax register as displacement when restoring %r10 or %rax.
467 Fix wrong offset when restoring both registers.
469 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
471 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
472 assertion with conditional return.
474 2014-02-18 Jakub Jelinek <jakub@redhat.com>
475 Uros Bizjak <ubizjak@gmail.com>
478 * config/i386/driver-i386.c (host_detect_local_cpu): If
479 YMM state is not saved by the OS, also clear has_f16c. Move
480 CPUID 0x80000001 handling before YMM state saving checking.
482 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
484 PR rtl-optimization/58960
485 * haifa-sched.c (alloc_global_sched_pressure_data): New,
486 factored out from ...
487 (sched_init): ... here.
488 (free_global_sched_pressure_data): New, factored out from ...
489 (sched_finish): ... here.
490 * sched-int.h (free_global_sched_pressure_data): Declare.
491 * sched-rgn.c (nr_regions_initial): New static global.
492 (haifa_find_rgns): Initialize it.
493 (schedule_region): Disable sched-pressure for the newly
496 2014-02-17 Richard Biener <rguenther@suse.de>
498 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
499 release SSA defs of pattern stmts.
501 2014-02-17 Richard Biener <rguenther@suse.de>
503 * tree-inline.c (expand_call_inline): Release the virtual
504 operand defined by the call we are about to inline.
506 2014-02-17 Richard Biener <rguenther@suse.de>
508 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
510 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
511 Ilya Tocar <ilya.tocar@intel.com>
513 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
514 arguments order in builtin.
515 (_mm512_permutexvar_epi64): Ditto.
516 (_mm512_mask_permutexvar_epi64): Ditto
517 (_mm512_maskz_permutexvar_epi32): Ditto
518 (_mm512_permutexvar_epi32): Ditto
519 (_mm512_mask_permutexvar_epi32): Ditto
521 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
523 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
524 (p8_vmrgow): Likewise.
526 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
528 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
531 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
534 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
535 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
536 into 64-bit and 32-bit moves. On 64-bit moves, add support for
537 using direct move instructions on ISA 2.07. Also adjust
538 instruction length for 64-bit.
539 (mov<mode>_64bit, TFmode/TDmode): Likewise.
540 (mov<mode>_32bit, TFmode/TDmode): Likewise.
542 2014-02-15 Alan Modra <amodra@gmail.com>
546 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
547 find_replacement on parts of insn rtl that might be reloaded.
549 2014-02-15 Richard Biener <rguenther@suse.de>
551 PR tree-optimization/60183
552 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
553 (tree_ssa_phiprop): Calculate and free post-dominators.
555 2014-02-14 Jeff Law <law@redhat.com>
557 PR rtl-optimization/60131
558 * ree.c (get_extended_src_reg): New function.
559 (combine_reaching_defs): Use it rather than assuming location of REG.
560 (find_and_remove_re): Verify first operand of extension is
561 a REG before adding the insns to the copy list.
563 2014-02-14 Roland McGrath <mcgrathr@google.com>
565 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
566 * configure: Regenerated.
567 * config.in: Regenerated.
568 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
569 instead of ASM_SHORT.
571 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
572 Richard Earnshaw <rearnsha@arm.com>
574 PR rtl-optimization/59535
575 * lra-constraints.c (process_alt_operands): Encourage alternative
576 when unassigned pseudo class is superset of the alternative class.
577 (inherit_reload_reg): Don't inherit when optimizing for code size.
578 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
579 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
580 modes not less than 4 for Thumb1.
582 2014-02-14 Kyle McMartin <kyle@redhat.com>
585 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
587 2014-02-14 Richard Biener <rguenther@suse.de>
589 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
590 (get_frame_arg): Drop the assert with langhook types_compatible_p.
591 Do not strip INDIRECT_REFs.
593 2014-02-14 Richard Biener <rguenther@suse.de>
596 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
597 DECL_FUNCTION_SPECIFIC_TARGET.
598 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
599 * tree-streamer-out.c (pack_ts_target_option): Remove.
600 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
601 (write_ts_function_decl_tree_pointers): Do not stream
602 DECL_FUNCTION_SPECIFIC_TARGET.
603 * tree-streamer-in.c (unpack_ts_target_option): Remove.
604 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
605 (lto_input_ts_function_decl_tree_pointers): Do not stream
606 DECL_FUNCTION_SPECIFIC_TARGET.
608 2014-02-14 Jakub Jelinek <jakub@redhat.com>
610 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
611 (get_initial_def_for_induction, vectorizable_induction): Ignore
612 debug stmts when looking for exit_phi.
613 (vectorizable_live_operation): Fix up condition.
615 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
617 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
618 nreverse() because it changes the content of original tree list.
620 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
622 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
623 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
625 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
627 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
628 GNU coding standards.
630 2014-02-13 Jakub Jelinek <jakub@redhat.com>
633 * dwarf2out.c (gen_subprogram_die): Don't call
634 add_calling_convention_attribute if subr_die is old_die.
636 2014-02-13 Sharad Singhai <singhai@google.com>
638 * doc/optinfo.texi: Fix order of nodes.
640 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
642 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
643 operands[2], not operands[3].
645 2014-02-13 Richard Biener <rguenther@suse.de>
648 * doc/install.texi (ISL): Update recommended version to 0.12.2,
649 mention the possibility of an in-tree build.
650 (CLooG): Update recommended version to 0.18.1, mention the
651 possibility of an in-tree build and clarify that the ISL
652 bundled with CLooG does not work.
654 2014-02-13 Jakub Jelinek <jakub@redhat.com>
657 * expr.c (compress_float_constant): If x is a hard register,
658 extend into a pseudo and then move to x.
660 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
662 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
663 caused by bad second argument to warning_at() with -mhotpatch and
664 nested functions (e.g. with gfortran).
666 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
668 * opts.c (option_name): Remove "enabled by default" rider.
670 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
672 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
674 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
675 Uros Bizjak <ubizjak@gmail.com>
678 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
679 * configure: Regenerated.
681 2014-02-12 Richard Biener <rguenther@suse.de>
683 * vec.c (vec_prefix::calculate_allocation): Move as
684 inline variant to vec.h.
685 (vec_prefix::calculate_allocation_1): New out-of-line version.
686 * vec.h (vec_prefix::calculate_allocation_1): Declare.
687 (vec_prefix::m_has_auto_buf): Rename to ...
688 (vec_prefix::m_using_auto_storage): ... this.
689 (vec_prefix::calculate_allocation): Inline the easy cases
690 and dispatch to calculate_allocation_1 which doesn't need the
692 (va_heap::reserve): Use gcc_checking_assert.
693 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
694 m_using_auto_storage.
695 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
697 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
698 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
699 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
701 2014-02-12 Richard Biener <rguenther@suse.de>
703 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
704 when we found a dependence.
706 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
708 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
710 (maybe_fold_stmt): ... into this new function.
711 * omp-low.c (lower_omp): Update comment.
713 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
716 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
719 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
721 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
722 identifiers in comments.
723 (cortexa53_extra_costs): Likewise.
724 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
725 (cortexa7_extra_costs): Likewise.
726 (cortexa12_extra_costs): Likewise.
727 (cortexa15_extra_costs): Likewise.
728 (v7m_extra_costs): Likewise.
730 2014-02-12 Richard Biener <rguenther@suse.de>
733 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
734 of posix_memalign being successful.
735 (lower_stmt): Restrict lowering of posix_memalign to when
736 -ftree-bit-ccp is enabled.
738 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
740 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
742 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
744 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
746 PR rtl-optimization/60116
747 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
748 other_insn once the combination has been validated.
750 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
753 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
755 * ipa-devirt.c: Include demangle.h
756 (odr_violation_reported): New static variable.
757 (add_type_duplicate): Update odr_violations.
758 (maybe_record_node): Add completep parameter; update it.
759 (record_target_from_binfo): Add COMPLETEP parameter;
761 (possible_polymorphic_call_targets_1): Likewise.
762 (struct polymorphic_call_target_d): Add nonconstruction_targets;
763 rename FINAL to COMPLETE.
764 (record_targets_from_bases): Sanity check we found the binfo;
765 fix COMPLETEP updating.
766 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
767 parameter, fix computing of COMPLETEP.
768 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
769 at LTO time do demangling.
770 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
771 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
773 (gimple_get_virt_method_for_binfo): Likewise.
774 * gimple-fold.h (gimple_get_virt_method_for_binfo,
775 gimple_get_virt_method_for_vtable): Update prototypes.
777 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
780 * genautomata.c (add_presence_absence): Fix typo with
781 {final_}presence_list.
783 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
786 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
787 for VSX/Altivec vectors that land in GPR registers.
789 2014-02-11 Richard Henderson <rth@redhat.com>
790 Jakub Jelinek <jakub@redhat.com>
793 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
794 around drhs if type conversion to lacc->type is not useless.
796 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
798 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
800 (cortex-a57.cortex-a53): Likewise.
801 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
803 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
805 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
808 2014-02-11 Renlin Li <Renlin.Li@arm.com>
810 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
811 add_options_for_arm_vfp3.
813 2014-02-11 Jeff Law <law@redhat.com>
816 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
817 object with an undesirable mode.
819 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
822 * config/i386/sol2-9.h: New file.
823 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
824 *-*-solaris2.9*): Use it.
826 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
828 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
829 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
831 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
833 * config/microblaze/microblaze.c: Extend mcpu version format
835 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
837 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
839 2014-02-10 Richard Henderson <rth@redhat.com>
842 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
843 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
844 ms-abi vs -mno-accumulate-outgoing-args.
845 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
846 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
849 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
852 * cfgexpand.c (expand_asm_operands): Attach source location to
853 ASM_INPUT rtx objects.
854 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
856 2014-02-10 Nick Clifton <nickc@redhat.com>
858 * config/mn10300/mn10300.c (popcount): New function.
859 (mn10300_expand_prologue): Include saved registers in stack usage
862 2014-02-10 Jeff Law <law@redhat.com>
865 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
866 when changing the SET_DEST of a prior insn to avoid an input reload.
868 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
870 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
871 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
872 -mcall-openbsd, or -mcall-linux.
873 (CC1_ENDIAN_BIG_SPEC): Remove.
874 (CC1_ENDIAN_LITTLE_SPEC): Remove.
875 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
876 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
877 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
878 and %cc1_endian_default.
879 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
881 2014-02-10 Richard Biener <rguenther@suse.de>
883 PR tree-optimization/60115
884 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
885 MEM_REF handling. Properly verify that the accesses are not
886 out of the objects bound.
888 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
890 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
893 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
895 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
896 proper constants and fix formatting.
897 (possible_polymorphic_call_targets): Fix formatting.
899 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
900 Ilya Tocar <ilya.tocar@intel.com>
902 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
903 (_mm512_loadu_epi32): Renamed into...
904 (_mm512_loadu_si512): This.
905 (_mm512_storeu_epi32): Renamed into...
906 (_mm512_storeu_si512): This.
907 (_mm512_maskz_ceil_ps): Removed.
908 (_mm512_maskz_ceil_pd): Ditto.
909 (_mm512_maskz_floor_ps): Ditto.
910 (_mm512_maskz_floor_pd): Ditto.
911 (_mm512_floor_round_ps): Ditto.
912 (_mm512_floor_round_pd): Ditto.
913 (_mm512_ceil_round_ps): Ditto.
914 (_mm512_ceil_round_pd): Ditto.
915 (_mm512_mask_floor_round_ps): Ditto.
916 (_mm512_mask_floor_round_pd): Ditto.
917 (_mm512_mask_ceil_round_ps): Ditto.
918 (_mm512_mask_ceil_round_pd): Ditto.
919 (_mm512_maskz_floor_round_ps): Ditto.
920 (_mm512_maskz_floor_round_pd): Ditto.
921 (_mm512_maskz_ceil_round_ps): Ditto.
922 (_mm512_maskz_ceil_round_pd): Ditto.
923 (_mm512_expand_pd): Ditto.
924 (_mm512_expand_ps): Ditto.
925 * config/i386/i386.c (ix86_builtins): Remove
926 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
928 * config/i386/predicates.md (const1256_operand): New.
929 (const_1_to_2_operand): Ditto.
930 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
931 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
932 (*avx512pf_gatherpf<mode>sf): Ditto.
933 (avx512pf_gatherpf<mode>df): Ditto.
934 (*avx512pf_gatherpf<mode>df_mask): Ditto.
935 (*avx512pf_gatherpf<mode>df): Ditto.
936 (avx512pf_scatterpf<mode>sf): Ditto.
937 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
938 (*avx512pf_scatterpf<mode>sf): Ditto.
939 (avx512pf_scatterpf<mode>df): Ditto.
940 (*avx512pf_scatterpf<mode>df_mask): Ditto.
941 (*avx512pf_scatterpf<mode>df): Ditto.
942 (avx512f_expand<mode>): Removed.
943 (<shift_insn><mode>3<mask_name>): Change predicate type.
945 2014-02-08 Jakub Jelinek <jakub@redhat.com>
947 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
948 not at the end of datarefs vector use ordered_remove to avoid
949 reordering datarefs vector.
952 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
953 mark local addressable non-static vars as GOVD_PRIVATE
954 instead of GOVD_LOCAL.
955 * omp-low.c (lower_omp_for): Move gimple_bind_vars
956 and BLOCK_VARS of gimple_bind_block to new_stmt rather
960 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
961 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
962 assume_aligned or alloc_align attributes.
963 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
964 arguments. Handle also assume_aligned and alloc_align attributes.
965 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
966 calls to functions with assume_aligned or alloc_align attributes.
967 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
969 2014-02-08 Terry Guo <terry.guo@arm.com>
971 * doc/invoke.texi: Document ARM -march=armv7e-m.
973 2014-02-08 Jakub Jelinek <jakub@redhat.com>
975 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
976 flag on __cilkrts_rethrow builtin.
979 * ipa-cp.c (determine_versionability): Fail at -O0
980 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
981 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
984 2014-02-04 Jakub Jelinek <jakub@redhat.com>
987 * tree-inline.c (copy_forbidden): Fail for
988 __attribute__((optimize (0))) functions.
990 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
992 * varpool.c: Include pointer-set.h.
993 (varpool_remove_unreferenced_decls): Variables in other partitions
994 will not be output; be however careful to not lose information
997 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
999 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1000 lookup in the vtable constructor.
1002 2014-02-07 Jeff Law <law@redhat.com>
1005 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1006 define_insn_and_split.
1008 * ipa-inline.c (inline_small_functions): Fix typos.
1010 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1012 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1013 (s390_can_use_return_insn): Declare.
1014 * config/s390/s390.h (EPILOGUE_USES): Define.
1015 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1017 (s390_chunkify_start): Handle return JUMP_LABELs.
1018 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1019 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1020 (s390_can_use_return_insn): New functions.
1021 (s390_fix_long_loop_prediction): Handle conditional returns.
1022 (TARGET_SET_UP_BY_PROLOGUE): Define.
1023 * config/s390/s390.md (ANY_RETURN): New code iterator.
1024 (*creturn, *csimple_return, return, simple_return): New patterns.
1026 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1028 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1029 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1030 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1031 REG_CFA_RESTORE list when deciding not to restore a register.
1033 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1035 * config/s390/s390.c: Include tree-pass.h and context.h.
1036 (s390_early_mach): New function, split out from...
1037 (s390_emit_prologue): ...here.
1038 (pass_data_s390_early_mach): New pass structure.
1039 (pass_s390_early_mach): New class.
1040 (s390_option_override): Create and register early_mach pass.
1041 Move to end of file.
1043 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1045 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1046 to match for the exit block.
1048 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1050 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1051 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1052 Reject misaligned operands.
1054 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1056 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1058 2014-02-07 Richard Biener <rguenther@suse.de>
1061 * gimple-low.c (lower_builtin_posix_memalign): New function.
1062 (lower_stmt): Call it to lower posix_memalign in a way
1063 to make alignment info accessible.
1065 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1068 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1069 __builtin_setjmp_receiver.
1071 2014-02-07 Richard Biener <rguenther@suse.de>
1074 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1075 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1076 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1077 Handle BUILT_IN_POSIX_MEMALIGN.
1078 (find_func_clobbers): Likewise.
1079 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1080 (call_may_clobber_ref_p_1): Likewise.
1082 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1085 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1088 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1091 * lto-cgraph.c (lto_output_node): Use
1092 symtab_get_symbol_partitioning_class.
1093 (lto_output_varpool_node): likewise.
1094 (symtab_get_symbol_partitioning_class): Move here from
1096 * cgraph.h (symbol_partitioning_class): Likewise.
1097 (symtab_get_symbol_partitioning_class): Declare.
1099 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1101 * ggc.h (ggc_internal_cleared_alloc): New macro.
1102 * vec.h (vec_safe_copy): Handle memory stats.
1103 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1104 * target-globals.c (save_target_globals): Likewise.
1106 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1109 * expr.c (emit_move_resolve_push): Export; be bit more selective
1110 on when to clear alias set.
1111 * expr.h (emit_move_resolve_push): Declare.
1112 * function.h (struct function): Add tail_call_marked.
1113 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1114 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1115 * config/i386/i386.md (TImode move expander): De not call
1117 (FP push expanders): Preserve memory attributes.
1118 * config/i386/sse.md (push<mode>1): Remove.
1119 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1120 (ix86_expand_push): Remove.
1121 * config/i386/mmx.md (push<mode>1): Remove.
1123 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1125 PR rtl-optimization/60030
1126 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1127 lopart with paradoxical subreg before shifting it up by hprec.
1129 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1131 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1132 Remove extra newline at end of file.
1133 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1134 (arm_issue_rate): Handle cortexa57.
1135 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1136 (cortex-a57.cortex-a53): Likewise.
1138 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1141 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1142 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1144 (arm_expand_prologue): Adjust all callers.
1145 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1146 info, registers also at the lowest numbered registers side. Use
1147 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1151 * var-tracking.c (adjust_mems): Before adding a SET to
1152 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1154 2014-02-06 Alan Modra <amodra@gmail.com>
1157 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1158 change SDmode to DDmode when lra_in_progress.
1160 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1163 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1164 free_data_ref on the dr first, and before goto again also set dr
1165 to the next dr. For simd_lane_access, free old datarefs[i] before
1166 overwriting it. For get_vectype_for_scalar_type failure, don't
1167 free_data_ref if simd_lane_access.
1169 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1172 * tree.h (opts_for_fn): New inline function.
1173 (opt_for_fn): Define.
1174 * config/i386/i386.c (ix86_function_regparm): Use
1175 opt_for_fn (decl, optimize) instead of optimize.
1177 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1179 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1180 for SYMBOL_REF in large memory model.
1182 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1184 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1186 (cortex-a57): Likewise.
1187 (cortex-a57.cortex-a53): Likewise.
1189 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1190 Kugan Vivekanandarajah <kuganv@linaro.org>
1192 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1194 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1196 2014-02-06 Richard Biener <rguenther@suse.de>
1198 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1199 set_loop_copy and initialize_original_copy_tables.
1201 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1203 * config/aarch64/aarch64-simd.md
1204 (aarch64_ashr_simddi): Change QI to SI.
1206 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1207 Jakub Jelinek <jakub@redhat.com>
1210 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1213 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1215 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1216 CODE_FOR_altivec_vpku[hw]um to
1217 CODE_FOR_altivec_vpku[hw]um_direct.
1218 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1219 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1220 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1221 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1223 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1225 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1226 generation for -maltivec=be.
1227 (altivec_vsumsws): Simplify redundant test.
1229 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1231 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1232 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1233 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1234 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1235 gen_altivec_vpkuwum.
1236 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1238 (altivec_vpks<VI_char>ss): Likewise.
1239 (altivec_vpks<VI_char>us): Likewise.
1240 (altivec_vpku<VI_char>us): Likewise.
1241 (altivec_vpku<VI_char>um): Likewise.
1242 (altivec_vpku<VI_char>um_direct): New (copy of
1243 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1245 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1246 target is little endian and -maltivec=be is not specified.
1247 (*altivec_vupkhs<VU_char>_direct): New (copy of
1248 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1249 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1250 target is little endian and -maltivec=be is not specified.
1251 (*altivec_vupkls<VU_char>_direct): New (copy of
1252 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1253 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1254 little endian and -maltivec=be is not specified.
1255 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1256 little endian and -maltivec=be is not specified.
1258 2014-02-05 Richard Henderson <rth@redhat.com>
1261 * combine-stack-adj.c: Revert r206943.
1262 * sched-int.h (struct deps_desc): Add last_args_size.
1263 * sched-deps.c (init_deps): Initialize it.
1264 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1265 contain REG_ARGS_SIZE notes.
1267 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1269 * lto-cgraph.c (asm_nodes_output): Make global.
1270 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1271 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1272 (driver_handle_option): Handle OPT_fwpa.
1274 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1277 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1278 a comment typo and formatting issue. If odr_hash hasn't been
1279 created, return vNULL and set *completep to false.
1282 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1283 bb with no successors.
1285 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1288 * doc/invoke.texi (-march): Clarify documentation for ARM.
1292 2014-02-05 Richard Biener <rguenther@suse.de>
1294 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1295 when not vectorizing because of too many alias checks.
1296 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1297 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1299 2014-02-05 Nick Clifton <nickc@redhat.com>
1301 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1302 accept extended registers in any mode when compiling for the MN10300.
1304 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1306 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1307 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1308 sanitization attributes.
1309 (can_inline_edge_p): Likewise.
1310 (sanitize_attrs_match_for_inline_p): New function.
1312 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1314 * ipa-prop.c (detect_type_change): Shor circuit testing of
1315 type changes on THIS pointer.
1317 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1320 * config/pa/pa.c (legitimize_tls_address): Return original address
1321 if not passed a SYMBOL_REF rtx.
1322 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1324 (pa_emit_move_sequence): Simplify TLS source operands.
1325 (pa_legitimate_constant_p): Reject all TLS constants.
1326 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1327 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1329 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1331 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1332 groups when we know they are controlled by LTO.
1333 * varasm.c (default_binds_local_p_1): If object is in other partition,
1334 it will be resolved locally.
1336 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1338 * config/host-linux.c (linux_gt_pch_use_address): Don't
1339 use SSIZE_MAX because it is not always defined.
1341 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1344 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1345 threshold for pseudo splitting.
1346 (update_ebb_live_info): Process call argument hard registers and
1347 hard registers from insn definition too.
1348 (max_small_class_regs_num): New constant.
1349 (inherit_in_ebb): Update live hard regs through EBBs. Update
1350 reloads_num only for small register classes. Don't split for
1353 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1356 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1359 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1361 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1364 2014-02-04 Marek Polacek <polacek@redhat.com>
1366 * gdbinit.in (pel): Define.
1368 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1370 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1373 2014-02-04 Richard Biener <rguenther@suse.de>
1376 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1377 in function context local.
1378 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1379 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1380 similar to LTO_imported_decl_ref.
1382 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1384 PR tree-optimization/60002
1385 * cgraphclones.c (build_function_decl_skip_args): Clear
1388 PR tree-optimization/60023
1389 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1390 false to gsi_replace.
1391 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1392 has been in some EH region and vec_stmt could throw, add
1393 vec_stmt into the same EH region.
1394 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1395 has no lhs, ignore it.
1396 * internal-fn.c (expand_MASK_LOAD): Likewise.
1399 * tree-inline.c (copy_forbidden): Fail for
1400 __attribute__((optimize (0))) functions.
1403 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1404 is set, copy one less argument.
1405 (expand_simd_clones): Don't subtract clone_info->inbranch
1406 from simd_clone_struct_alloc argument.
1408 PR rtl-optimization/57915
1409 * recog.c (simplify_while_replacing): If all unary/binary/relational
1410 operation arguments are constant, attempt to simplify those.
1413 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1414 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1416 2014-02-04 Richard Biener <rguenther@suse.de>
1418 PR tree-optimization/60012
1419 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1420 TBAA disambiguation to all DDRs.
1422 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1425 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1426 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1428 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1431 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1433 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1435 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1436 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1438 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1441 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1442 to figure out targets of polymorphic calls with known decl.
1443 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1444 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1445 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1446 (get_polymorphic_call_info): ... here.
1447 (get_polymorphic_call_info_from_invariant): New function.
1449 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1451 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1452 lookup via vtable pointer; check for type consistency
1453 and turn inconsitent facts into UNREACHABLE.
1454 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1455 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1456 type inconsistent querries; return UNREACHABLE instead.
1458 2014-02-03 Richard Henderson <rth@twiddle.net>
1461 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1462 already processed this node.
1463 (normalize_one_pred_1): Pass along mark_set.
1464 (normalize_one_pred): Create and destroy a pointer_set_t.
1465 (normalize_one_pred_chain): Likewise.
1467 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1469 PR gcov-profile/58602
1470 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1472 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1475 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1476 -fno-devirtualize; try to devirtualize by the knowledge of
1477 virtual table pointer given by aggregate propagation.
1478 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1479 (ipa_print_node_jump_functions): Dump also offset that
1480 is relevant for polymorphic calls.
1481 (determine_known_aggregate_parts): Add arg_type parameter; use it
1482 instead of determining the type from pointer type.
1483 (ipa_compute_jump_functions_for_edge): Update call of
1484 determine_known_aggregate_parts.
1485 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1486 (gimple_get_virt_method_for_binfo): ... here; simplify using
1487 vtable_pointer_value_to_vtable.
1488 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1489 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1490 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1491 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1493 (vtable_pointer_value_to_binfo): ... here.
1494 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1496 2014-02-03 Teresa Johnson <tejohnson@google.com>
1498 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1499 redef of outer loop index variable.
1501 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1505 * doc/extend.texi (Function Attributes): Typo.
1507 2014-02-03 Cong Hou <congh@google.com>
1509 PR tree-optimization/60000
1510 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1511 if the vectorized statement is a store. A store statement can only
1512 appear at the end of pattern statements.
1514 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1516 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1517 (ix86_option_override_internal): Default long double to 64-bit for
1518 32-bit Bionic and to 128-bit for 64-bit Bionic.
1520 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1521 TARGET_LONG_DOUBLE_128 is true.
1522 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1524 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1525 (mlong-double-64): Negate -mlong-double-128.
1526 (mlong-double-128): New option.
1528 * config/i386/i386-c.c (ix86_target_macros): Define
1529 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1531 * doc/invoke.texi: Document -mlong-double-128.
1533 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1535 PR rtl-optimization/60024
1536 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1538 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1540 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1542 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1544 PR rtl-optimization/57662
1545 * sel-sched.c (code_motion_path_driver): Do not mark already not
1546 existing blocks in the visiting bitmap.
1548 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1550 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1551 on the insn being emitted.
1553 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1554 Will Deacon <will.deacon@arm.com>
1556 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1558 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1560 * config/arm/arm-tables.opt: Regenerate.
1562 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1564 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1565 for vector types other than V16QImode.
1566 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1567 define_expand, and call altivec_expand_vec_perm_le when producing
1568 code with little endian element order.
1569 (*altivec_vperm_<mode>_internal): New insn having previous
1570 behavior of altivec_vperm_<mode>.
1571 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1572 altivec_expand_vec_perm_le when producing code with little endian
1574 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1575 behavior of altivec_vperm_<mode>_uns.
1577 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1579 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1580 (altivec_vsumsws): Add handling for -maltivec=be with a little
1582 (altivec_vsumsws_direct): New.
1583 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1584 gen_altivec_vsumsws.
1586 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1588 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1589 vtable_pointer_value_to_binfo): New functions.
1590 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1591 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1593 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1595 * config/nios2/nios2.md (load_got_register): Initialize GOT
1596 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1597 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1599 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1601 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1602 preserverd by passthrough, do not propagate the type.
1604 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1606 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1607 (mips_atomic_assign_expand_fenv): New function.
1608 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1610 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1612 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1613 (__builtin_mips_set_fcsr): Likewise.
1614 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1615 MIPS_USI_FTYPE_VOID.
1616 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1617 (mips16_expand_set_fcsr): Likewise.
1618 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1619 (mips16_set_fcsr_stub): Likewise.
1620 (mips16_get_fcsr_one_only_stub): New class.
1621 (mips16_set_fcsr_one_only_stub): Likewise.
1622 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1623 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1624 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1625 (hard_float): New availability predicate.
1626 (mips_builtins): Add get_fcsr and set_fcsr.
1627 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1628 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1629 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1630 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1631 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1634 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1636 * config/mips/mips.c (mips_one_only_stub): New class.
1637 (mips_need_mips16_rdhwr_p): Replace with...
1638 (mips16_rdhwr_stub): ...this new variable.
1639 (mips16_stub_call_address): New function.
1640 (mips16_rdhwr_one_only_stub): New class.
1641 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1642 (mips_output_mips16_rdhwr): Delete.
1643 (mips_finish_stub): New function.
1644 (mips_code_end): Use it to handle rdhwr stubs.
1646 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1649 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1650 when calculating size of integer atomic types.
1652 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1654 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1656 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1658 PR tree-optimization/60003
1659 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1660 * profile.c (branch_prob): Use gimple_call_builtin_p
1661 to check for BUILT_IN_SETJMP_RECEIVER.
1662 * tree-inline.c (copy_bb): Call notice_special_calls.
1664 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1667 * lra-constraints.c (process_alt_operands): Update reload_sum only
1670 2014-01-31 Richard Henderson <rth@redhat.com>
1673 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1674 until after else_eh is processed.
1676 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1678 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1679 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1680 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1681 in smmintrin.h, remove them.
1682 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1683 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1684 * config/i386/i386.md (ROUND_SAE): Fix value.
1685 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1686 (const48_operand): New.
1687 * config/i386/subst.md (round), (round_expand): Use
1688 const_4_or_8_to_11_operand.
1689 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1691 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1693 * config/i386/constraints.md (Yk): Swap meaning with k.
1694 * config/i386/i386.md (movhi_internal): Change Yk to k.
1695 (movqi_internal): Ditto.
1696 (*k<logic><mode>): Ditto.
1699 (kandn<mode>): Ditto.
1700 (*<code>hi_1): Ditto.
1701 (*<code>qi_1): Ditto.
1702 (kxnor<mode>): Ditto.
1703 (kortestzhi): Ditto.
1704 (kortestchi): Ditto.
1706 (*one_cmplhi2_1): Ditto.
1707 (*one_cmplqi2_1): Ditto.
1708 * config/i386/sse.md (): Change k to Yk.
1709 (avx512f_load<mode>_mask): Ditto.
1710 (avx512f_blendm<mode>): Ditto.
1711 (avx512f_store<mode>_mask): Ditto.
1712 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1713 (avx512f_storedqu<mode>_mask): Ditto.
1714 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1716 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1717 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1718 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1719 (avx512f_maskcmp<mode>3): Ditto.
1720 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1721 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1722 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1723 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1724 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1725 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1726 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1727 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1728 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1729 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1730 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1731 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1732 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1733 (vec_extract_lo_<mode>_maskm): Ditto.
1734 (vec_extract_hi_<mode>_maskm): Ditto.
1735 (avx512f_vternlog<mode>_mask): Ditto.
1736 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1737 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1738 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1739 (avx512f_<code>v8div16qi2_mask): Ditto.
1740 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1741 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1742 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1743 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1744 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1745 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1746 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1747 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1748 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1749 (avx512cd_maskb_vec_dupv8di): Ditto.
1750 (avx512cd_maskw_vec_dupv16si): Ditto.
1751 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1752 (avx512f_vpermi2var<mode>3_mask): Ditto.
1753 (avx512f_vpermi2var<mode>3_mask): Ditto.
1754 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1755 (*avx512f_gathersi<mode>): Ditto.
1756 (*avx512f_gathersi<mode>_2): Ditto.
1757 (*avx512f_gatherdi<mode>): Ditto.
1758 (*avx512f_gatherdi<mode>_2): Ditto.
1759 (*avx512f_scattersi<mode>): Ditto.
1760 (*avx512f_scatterdi<mode>): Ditto.
1761 (avx512f_compress<mode>_mask): Ditto.
1762 (avx512f_compressstore<mode>_mask): Ditto.
1763 (avx512f_expand<mode>_mask): Ditto.
1764 * config/i386/subst.md (mask): Change k to Yk.
1765 (mask_scalar_merge): Ditto.
1768 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1770 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1772 2014-01-31 Richard Biener <rguenther@suse.de>
1775 * builtins.c (fold_builtin_memory_op): Make sure to not
1776 use a floating-point mode or a boolean or enumeral type for
1779 2014-01-30 DJ Delorie <dj@redhat.com>
1781 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1782 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1783 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1784 whenever main() has an epilogue.
1786 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1788 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1789 unused variable "field".
1790 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1791 (vsx_mergeh_<mode>): Likewise.
1792 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1793 (altivec_vmrghh): Likewise.
1794 (altivec_vmrghw): Likewise.
1795 (altivec_vmrglb): Likewise.
1796 (altivec_vmrglh): Likewise.
1797 (altivec_vmrglw): Likewise.
1798 (altivec_vspltb): Add missing uses.
1799 (altivec_vsplth): Likewise.
1800 (altivec_vspltw): Likewise.
1801 (altivec_vspltsf): Likewise.
1803 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1806 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1807 frame related instructions.
1809 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1811 PR rtl-optimization/59959
1812 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1813 any reload of register whose subreg is invalid.
1815 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1817 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1818 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1819 Add missing return type - void.
1821 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1823 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1824 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1825 remove element index adjustment for endian (now handled in vsx.md
1827 (altivec_expand_vec_perm_const): Use
1828 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1829 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1830 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1831 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1832 define_expand and a new define_insn *altivec_vspltb_internal;
1833 adjust for -maltivec=be on a little endian target.
1834 (altivec_vspltb_direct): New.
1835 (altivec_vsplth): Divide into a define_expand and a new
1836 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1837 little endian target.
1838 (altivec_vsplth_direct): New.
1839 (altivec_vspltw): Divide into a define_expand and a new
1840 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1841 little endian target.
1842 (altivec_vspltw_direct): New.
1843 (altivec_vspltsf): Divide into a define_expand and a new
1844 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1845 a little endian target.
1847 2014-01-30 Richard Biener <rguenther@suse.de>
1849 PR tree-optimization/59993
1850 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1851 can propagate form the earlier stmt and avoid the transform
1852 when the intermediate result is needed.
1854 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1856 * README.Portability: Fix typo.
1858 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1860 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1861 comparison_operator with ordered_comparison_operator.
1863 2014-01-30 Nick Clifton <nickc@redhat.com>
1865 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1866 Rename to mn10300_store_multiple_regs.
1867 * config/mn10300/mn10300.c: Likewise.
1868 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1869 store_multiple_regs.
1870 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1871 Call mn10300_store_multiple_regs.
1873 2014-01-30 Nick Clifton <nickc@redhat.com>
1874 DJ Delorie <dj@redhat.com>
1876 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1877 %fp 2 to keep registers after it properly word-aligned.
1878 (rl78_alloc_physical_registers_umul): Handle the case where both
1879 input operands are the same.
1881 2014-01-30 Richard Biener <rguenther@suse.de>
1883 PR tree-optimization/59903
1884 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1887 2014-01-30 Jason Merrill <jason@redhat.com>
1890 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1893 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1895 2014-01-30 Richard Biener <rguenther@suse.de>
1897 PR tree-optimization/59951
1898 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1900 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1903 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1904 SFmode to DFmode case.
1906 2014-01-29 DJ Delorie <dj@redhat.com>
1908 * config/msp430/msp430.opt (-minrt): New.
1909 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1911 (ENDFILE_SPEC): Likewise.
1913 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1915 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1916 (estimate_function_body_sizes): Use it.
1918 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1921 * dwarf2out.c (is_cxx_auto): New.
1922 (is_base_type): Use it.
1923 (gen_type_die_with_usage): Likewise.
1925 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1927 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1928 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1929 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1930 -maltivec=be with LE targets.
1931 (vsx_mergeh_<mode>): Likewise.
1932 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1933 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1934 (altivec_vmrghb): Replace with define_expand and new
1935 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1936 (altivec_vmrghb_direct): New define_insn.
1937 (altivec_vmrghh): Replace with define_expand and new
1938 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1939 (altivec_vmrghh_direct): New define_insn.
1940 (altivec_vmrghw): Replace with define_expand and new
1941 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1942 (altivec_vmrghw_direct): New define_insn.
1943 (*altivec_vmrghsf): Adjust for endianness.
1944 (altivec_vmrglb): Replace with define_expand and new
1945 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1946 (altivec_vmrglb_direct): New define_insn.
1947 (altivec_vmrglh): Replace with define_expand and new
1948 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1949 (altivec_vmrglh_direct): New define_insn.
1950 (altivec_vmrglw): Replace with define_expand and new
1951 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1952 (altivec_vmrglw_direct): New define_insn.
1953 (*altivec_vmrglsf): Adjust for endianness.
1954 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1955 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1956 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1957 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1958 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1959 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1960 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1961 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1963 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1965 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1966 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1969 2014-01-29 Richard Biener <rguenther@suse.de>
1971 PR tree-optimization/58742
1972 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1973 associate_pointerplus_align.
1974 (associate_pointerplus_diff): New function.
1975 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1976 and associate_pointerplus_diff.
1978 2014-01-29 Richard Biener <rguenther@suse.de>
1980 * lto-streamer.h (LTO_major_version): Bump to 3.
1981 (LTO_minor_version): Reset to 0.
1983 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1985 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1986 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1987 (arm_file_start): Generate correct asm header for armv7ve.
1988 * config/arm/bpabi.h: Add multilib support for armv7ve.
1989 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1990 and cortex-a15 to armv7ve.
1991 * config/arm/t-aprofile: Add multilib support for armv7ve.
1992 * doc/invoke.texi: Document -march=armv7ve.
1994 2014-01-29 Richard Biener <rguenther@suse.de>
1996 PR tree-optimization/58742
1997 * tree-ssa-forwprop.c (associate_plusminus): Return true
1998 if we changed sth, defer EH cleanup to ...
1999 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2000 (simplify_mult): New function.
2002 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2005 PR tree-optimization/59920
2006 * tree.c (build_common_builtin_nodes): Remove
2007 __builtin_setjmp_dispatcher initialization.
2008 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2009 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2010 instead of gsi_after_labels + manually skipping debug stmts.
2011 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2012 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2013 * tree-inline.c (copy_edges_for_bb): Remove
2014 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2015 argument. Ignore computed_goto_p stmts. Don't call
2016 make_abnormal_goto_edges. If a call might need abnormal edges
2017 for non-local gotos, see if it already has an edge to
2018 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2019 with true argument, don't do anything then, otherwise add
2020 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2021 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2023 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2024 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2025 (lower_stmt): Don't set data->calls_builtin_setjmp.
2026 (lower_builtin_setjmp): Adjust comment.
2027 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2028 * tree-cfg.c (found_computed_goto): Remove.
2029 (factor_computed_gotos): Remove.
2030 (make_goto_expr_edges): Return bool, true for computed gotos.
2031 Don't call make_abnormal_goto_edges.
2032 (build_gimple_cfg): Don't set found_computed_goto, don't call
2033 factor_computed_gotos.
2034 (computed_goto_p): No longer static.
2035 (make_blocks): Don't set found_computed_goto.
2036 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2037 (make_edges): If make_goto_expr_edges returns true, push bb
2038 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2039 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2040 vector. Record mapping between bbs and OpenMP regions if there
2041 are any, adjust make_gimple_omp_edges caller. Call
2042 handle_abnormal_edges.
2043 (make_abnormal_goto_edges): Remove.
2044 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2045 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2046 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2047 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2048 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2049 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2050 filling *region also set *region_idx to (*region)->entry->index.
2053 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2054 For REGs set ORIGINAL_REGNO.
2056 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2058 * doc/md.texi: Mention that a target shouldn't implement
2059 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2062 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2064 PR tree-optimization/59594
2065 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2066 a copy of the datarefs vector rather than the vector itself.
2068 2014-01-28 Jason Merrill <jason@redhat.com>
2071 * dwarf2out.c (auto_die): New static.
2072 (gen_type_die_with_usage): Handle C++1y 'auto'.
2073 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2076 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2079 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2080 (SPEC_X32): Likewise.
2081 (SPEC_64): Likewise.
2082 * config/i386/i386.c (ix86_option_override_internal): Turn off
2083 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2085 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2086 * config/i386/i386.h (TARGET_16BIT): New macro.
2087 (TARGET_16BIT_P): Likewise.
2088 * config/i386/i386.opt: Add m16.
2089 * doc/invoke.texi: Document -m16.
2091 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2093 PR preprocessor/59935
2094 * input.c (location_get_source_line): Bail out on when line number
2095 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2097 2014-01-28 Richard Biener <rguenther@suse.de>
2099 PR tree-optimization/58742
2100 * tree-ssa-forwprop.c (associate_plusminus): Handle
2101 pointer subtraction of the form (T)(P + A) - (T)P.
2103 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2105 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2108 2014-01-28 Richard Biener <rguenther@suse.de>
2111 2014-01-28 Richard Biener <rguenther@suse.de>
2113 PR rtl-optimization/45364
2114 PR rtl-optimization/59890
2115 * var-tracking.c (local_get_addr_clear_given_value): Handle
2116 already cleared slot.
2117 (val_reset): Handle not allocated local_get_addr_cache.
2118 (vt_find_locations): Use post-order on the inverted CFG.
2120 2014-01-28 Richard Biener <rguenther@suse.de>
2122 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2124 2014-01-28 Richard Biener <rguenther@suse.de>
2126 PR rtl-optimization/45364
2127 PR rtl-optimization/59890
2128 * var-tracking.c (local_get_addr_clear_given_value): Handle
2129 already cleared slot.
2130 (val_reset): Handle not allocated local_get_addr_cache.
2131 (vt_find_locations): Use post-order on the inverted CFG.
2133 2014-01-28 Alan Modra <amodra@gmail.com>
2135 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2136 * configure.ac <recursive call for build != host>: Define
2137 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2138 and LD_FOR_BUILD too.
2139 * configure: Regenerate.
2141 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2143 * config/i386/i386.c (get_builtin_code_for_version): Separate
2144 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2145 Broadwell from Haswell.
2147 2014-01-27 Steve Ellcey <sellcey@mips.com>
2149 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2150 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2151 * config/mips/mips.c (mips_option_override): Change setting
2153 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2154 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2155 Change from Mask to Var.
2157 2014-01-27 Jeff Law <law@redhat.com>
2159 * ipa-inline.c (inline_small_functions): Fix typo.
2161 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2163 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2164 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2165 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2166 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2167 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2168 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2169 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2170 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2171 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2172 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2173 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2174 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2175 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2176 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2177 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2178 (_mm512_storeu_epi64): Ditto.
2179 (_mm512_cmpge_epi32_mask): Ditto.
2180 (_mm512_cmpge_epu32_mask): Ditto.
2181 (_mm512_cmpge_epi64_mask): Ditto.
2182 (_mm512_cmpge_epu64_mask): Ditto.
2183 (_mm512_cmple_epi32_mask): Ditto.
2184 (_mm512_cmple_epu32_mask): Ditto.
2185 (_mm512_cmple_epi64_mask): Ditto.
2186 (_mm512_cmple_epu64_mask): Ditto.
2187 (_mm512_cmplt_epi32_mask): Ditto.
2188 (_mm512_cmplt_epu32_mask): Ditto.
2189 (_mm512_cmplt_epi64_mask): Ditto.
2190 (_mm512_cmplt_epu64_mask): Ditto.
2191 (_mm512_cmpneq_epi32_mask): Ditto.
2192 (_mm512_cmpneq_epu32_mask): Ditto.
2193 (_mm512_cmpneq_epi64_mask): Ditto.
2194 (_mm512_cmpneq_epu64_mask): Ditto.
2195 (_mm512_expand_pd): Ditto.
2196 (_mm512_expand_ps): Ditto.
2197 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2198 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2199 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2200 * config/i386/i386.c (ix86_builtins): Add
2201 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2202 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2203 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2204 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2205 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2206 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2207 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2208 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2209 IX86_BUILTIN_PMOVUSQW512_MEM.
2210 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2211 __builtin_ia32_pmovsqd512mem_mask,
2212 __builtin_ia32_pmovqd512mem_mask,
2213 __builtin_ia32_pmovusqw512mem_mask,
2214 __builtin_ia32_pmovsqw512mem_mask,
2215 __builtin_ia32_pmovqw512mem_mask,
2216 __builtin_ia32_pmovusdw512mem_mask,
2217 __builtin_ia32_pmovsdw512mem_mask,
2218 __builtin_ia32_pmovdw512mem_mask,
2219 __builtin_ia32_pmovqb512mem_mask,
2220 __builtin_ia32_pmovusqb512mem_mask,
2221 __builtin_ia32_pmovsqb512mem_mask,
2222 __builtin_ia32_pmovusdb512mem_mask,
2223 __builtin_ia32_pmovsdb512mem_mask,
2224 __builtin_ia32_pmovdb512mem_mask.
2225 (bdesc_args): Add __builtin_ia32_expanddf512,
2226 __builtin_ia32_expandsf512.
2227 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2228 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2229 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2230 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2231 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2232 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2233 (avx512f_<code>v8div16qi2_mask_store): This.
2234 (avx512f_expand<mode>): New.
2236 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2238 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2240 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2241 (_mm512_prefetch_i32scatter_pd): Ditto.
2242 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2243 (_mm512_prefetch_i64scatter_pd): Ditto.
2244 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2245 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2246 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2247 (_mm512_prefetch_i32scatter_ps): Ditto.
2248 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2249 (_mm512_prefetch_i64scatter_ps): Ditto.
2250 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2251 * config/i386/i386-builtin-types.def: Define
2252 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2253 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2254 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2255 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2256 IX86_BUILTIN_SCATTERPFQPD.
2257 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2258 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2259 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2260 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2261 __builtin_ia32_scatterpfqps.
2262 (ix86_expand_builtin): Expand new built-ins.
2263 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2264 fix memory access data type.
2265 (*avx512pf_gatherpf<mode>_mask): Ditto.
2266 (*avx512pf_gatherpf<mode>): Ditto.
2267 (avx512pf_scatterpf<mode>): Ditto.
2268 (*avx512pf_scatterpf<mode>_mask): Ditto.
2269 (*avx512pf_scatterpf<mode>): Ditto.
2270 (GATHER_SCATTER_SF_MEM_MODE): New.
2271 (avx512pf_gatherpf<mode>df): Ditto.
2272 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2273 (*avx512pf_scatterpf<mode>df): Ditto.
2275 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2278 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2279 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2282 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2284 * common/config/arm/arm-common.c
2285 (arm_rewrite_mcpu): Handle multiple names.
2287 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2289 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2291 * gimple-builder.h (create_gimple_tmp): Delete.
2293 2014-01-27 Christian Bruel <christian.bruel@st.com>
2295 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2298 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2300 * config/pa/pa.md (call): Generate indirect long calls to non-local
2301 functions when outputing 32-bit code.
2302 (call_value): Likewise except for special call to buggy powf function.
2304 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2305 portable runtime and PIC indirect calls.
2306 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2307 and PIC call sequences. Use ldo instead of blr to set return register
2308 in PIC call sequence.
2310 2014-01-25 Walter Lee <walt@tilera.com>
2312 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2313 avoid clobbering a live register.
2315 2014-01-25 Walter Lee <walt@tilera.com>
2317 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2318 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2319 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2320 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2322 2014-01-25 Walter Lee <walt@tilera.com>
2324 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2325 arguments on even registers.
2326 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2328 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2329 (BIGGEST_ALIGNMENT): Ditto.
2330 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2332 2014-01-25 Walter Lee <walt@tilera.com>
2334 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2335 insns before bundling.
2336 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2338 2014-01-25 Walter Lee <walt@tilera.com>
2340 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2341 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2342 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2344 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2346 * config/mips/constraints.md (kl): Delete.
2347 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2348 define expands, using...
2349 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2350 instructions for MIPS16.
2351 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2352 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2354 2014-01-25 Walter Lee <walt@tilera.com>
2356 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2360 2014-01-25 Walter Lee <walt@tilera.com>
2362 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2363 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2365 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2367 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2370 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2372 * print-rtl.c (in_call_function_usage): New var.
2373 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2374 EXPR_LIST mode as mode and not as reg note name.
2377 * cfgloopmanip.c (copy_loop_info): If
2378 loop->warned_aggressive_loop_optimizations, make sure
2379 the flag is set in target loop too.
2381 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2383 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2385 * builtins.def: Likewise.
2386 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2387 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2388 * ira.c (ira_setup_eliminable_regset): Likewise.
2389 * omp-low.c (gate_expand_omp): Likewise.
2390 (execute_lower_omp): Likewise.
2391 (diagnose_sb_0): Likewise.
2392 (gate_diagnose_omp_blocks): Likewise.
2393 (simd_clone_clauses_extract): Likewise.
2396 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2398 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2399 correction for little endian...
2400 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2403 2014-01-24 Jeff Law <law@redhat.com>
2405 PR tree-optimization/59919
2406 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2407 for non-returning calls.
2409 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2411 * common/config/aarch64/aarch64-common.c
2412 (aarch64_rewrite_mcpu): Handle multiple names.
2413 * config/aarch64/aarch64.h
2414 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2416 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2418 * input.c (add_file_to_cache_tab): Handle the case where fopen
2421 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2424 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2425 from push operand if code of push isn't PRE_DEC.
2427 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2430 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2431 -mquad-memory-atomic. Update -mquad-memory documentation to say
2432 it is only used for non-atomic loads/stores.
2434 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2435 -mquad-memory or -mquad-memory-atomic switches.
2437 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2438 -mquad-memory-atomic to ISA 2.07 support.
2440 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2441 to separate support of normal quad word memory operations (ldq, stq)
2442 from the atomic quad word memory operations.
2444 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2445 support to separate non-atomic quad word operations from atomic
2446 quad word operations. Disable non-atomic quad word operations in
2447 little endian mode so that we don't have to swap words after the
2448 load and before the store.
2449 (quad_load_store_p): Add comment about atomic quad word support.
2450 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2451 options printed with -mdebug=reg.
2453 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2454 -mquad-memory-atomic as the test for whether we have quad word
2455 atomic instructions.
2456 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2457 or -mp8-vector are used, allow byte/half-word atomic operations.
2459 * config/rs6000/sync.md (load_lockedti): Insure that the address
2460 is a proper indexed or indirect address for the lqarx instruction.
2461 On little endian systems, swap the hi/lo registers after the lqarx
2463 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2464 insure the address is valid for the lqarx instruction.
2465 (store_conditionalti): Insure that the address is a proper indexed
2466 or indirect address for the stqcrx. instruction. On little endian
2467 systems, swap the hi/lo registers before doing the stqcrx.
2469 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2470 insure the address is valid for the stqcrx. instruction.
2472 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2473 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2474 type of quad memory support is available.
2476 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2479 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2480 there is a danger of looping.
2482 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2485 force flag_ira_loop_pressure if set via command line.
2487 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2489 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2490 (ashr_simd): New builtin handling DI mode.
2491 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2492 (aarch64_sshr_simddi): New match pattern.
2493 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2494 (vshrd_n_s64): Likewise.
2495 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2497 2014-01-23 Nick Clifton <nickc@redhat.com>
2499 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2500 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2501 favour of mcu specific scripts.
2502 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2505 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2506 Alex Velenko <Alex.Velenko@arm.com>
2508 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2509 (vaddv_s16): Likewise.
2510 (vaddv_s32): Likewise.
2511 (vaddv_u8): Likewise.
2512 (vaddv_u16): Likewise.
2513 (vaddv_u32): Likewise.
2514 (vaddvq_s8): Likewise.
2515 (vaddvq_s16): Likewise.
2516 (vaddvq_s32): Likewise.
2517 (vaddvq_s64): Likewise.
2518 (vaddvq_u8): Likewise.
2519 (vaddvq_u16): Likewise.
2520 (vaddvq_u32): Likewise.
2521 (vaddvq_u64): Likewise.
2522 (vaddv_f32): Likewise.
2523 (vaddvq_f32): Likewise.
2524 (vaddvq_f64): Likewise.
2525 (vmaxv_f32): Likewise.
2526 (vmaxv_s8): Likewise.
2527 (vmaxv_s16): Likewise.
2528 (vmaxv_s32): Likewise.
2529 (vmaxv_u8): Likewise.
2530 (vmaxv_u16): Likewise.
2531 (vmaxv_u32): Likewise.
2532 (vmaxvq_f32): Likewise.
2533 (vmaxvq_f64): Likewise.
2534 (vmaxvq_s8): Likewise.
2535 (vmaxvq_s16): Likewise.
2536 (vmaxvq_s32): Likewise.
2537 (vmaxvq_u8): Likewise.
2538 (vmaxvq_u16): Likewise.
2539 (vmaxvq_u32): Likewise.
2540 (vmaxnmv_f32): Likewise.
2541 (vmaxnmvq_f32): Likewise.
2542 (vmaxnmvq_f64): Likewise.
2543 (vminv_f32): Likewise.
2544 (vminv_s8): Likewise.
2545 (vminv_s16): Likewise.
2546 (vminv_s32): Likewise.
2547 (vminv_u8): Likewise.
2548 (vminv_u16): Likewise.
2549 (vminv_u32): Likewise.
2550 (vminvq_f32): Likewise.
2551 (vminvq_f64): Likewise.
2552 (vminvq_s8): Likewise.
2553 (vminvq_s16): Likewise.
2554 (vminvq_s32): Likewise.
2555 (vminvq_u8): Likewise.
2556 (vminvq_u16): Likewise.
2557 (vminvq_u32): Likewise.
2558 (vminnmv_f32): Likewise.
2559 (vminnmvq_f32): Likewise.
2560 (vminnmvq_f64): Likewise.
2562 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2564 * config/aarch64/aarch64-simd.md
2565 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2566 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2567 (*aarch64_mul3_elt<mode>): Likewise.
2568 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2569 (*aarch64_mul3_elt_to_64v2df): Likewise.
2570 (*aarch64_mla_elt<mode>): Likewise.
2571 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2572 (*aarch64_mls_elt<mode>): Likewise.
2573 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2574 (*aarch64_fma4_elt<mode>): Likewise.
2575 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2576 (*aarch64_fma4_elt_to_64v2df): Likewise.
2577 (*aarch64_fnma4_elt<mode>): Likewise.
2578 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2579 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2580 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2581 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2582 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2583 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2584 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2585 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2586 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2588 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2590 * config/aarch64/aarch64-simd.md
2591 (aarch64_be_checked_get_lane<mode>): New define_expand.
2592 * config/aarch64/aarch64-simd-builtins.def
2593 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2594 New builtin definition.
2595 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2596 Use new safe be builtin.
2598 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2600 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2602 (aarch64_be_st1<mode>): Likewise.
2603 (aarch_ld1<VALL:mode>): Define_expand modified.
2604 (aarch_st1<VALL:mode>): Likewise.
2605 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2606 (UNSPEC_ST1): Likewise.
2608 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2610 * config/microblaze/microblaze.md: Add trap insn and attribute
2612 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2614 PR preprocessor/58580
2615 * input.h (location_get_source_line): Take an additional line_size
2617 (void diagnostics_file_cache_fini): Declare new function.
2618 * input.c (struct fcache): New type.
2619 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2620 New static constants.
2621 (diagnostic_file_cache_init, total_lines_num)
2622 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2623 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2624 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2625 (get_next_line, read_next_line, goto_next_line, read_line_num):
2626 New static function definitions.
2627 (diagnostic_file_cache_fini): New function.
2628 (location_get_source_line): Take an additional output line_len
2629 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2631 * diagnostic.c (diagnostic_finish): Call
2632 diagnostic_file_cache_fini.
2633 (adjust_line): Take an additional input parameter for the length
2634 of the line, rather than calculating it with strlen.
2635 (diagnostic_show_locus): Adjust the use of
2636 location_get_source_line and adjust_line with respect to their new
2637 signature. While displaying a line now, do not stop at the first
2638 null byte. Rather, display the zero byte as a space and keep
2639 going until we reach the size of the line.
2640 * Makefile.in: Add vec.o to OBJS-libcommon
2642 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2643 Ilya Tocar <ilya.tocar@intel.com>
2645 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2646 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2647 (__builtin_ia32_kmov16): Ditto.
2648 * config/i386/i386.md (UNSPEC_KMOV): New.
2651 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2653 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2654 (_mm512_storeu_si512): Ditto.
2656 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2659 * rtl.h (get_referenced_operands): Declare.
2660 * recog.c (get_referenced_operands): New function.
2661 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2662 operands have been referenced when recording LO_SUM references.
2664 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2666 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2668 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2670 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2671 Enable for generic and recent AMD targets.
2673 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2675 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2676 ARG_SIZE note when adjustment was eliminated.
2678 2014-01-22 Jeff Law <law@redhat.com>
2680 PR tree-optimization/59597
2681 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2682 in file. Accept new argument REGISTERING and use it to modify
2683 dump output appropriately.
2684 (register_jump_thread): Corresponding changes.
2685 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2686 thread paths involving joiner blocks. Add code to dump cancelled
2687 jump threading paths.
2689 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2691 PR rtl-optimization/59477
2692 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2693 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2695 2014-01-22 Tom Tromey <tromey@redhat.com>
2697 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2699 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2701 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2703 PR rtl-optimization/59896
2704 * lra-constraints.c (process_alt_operands): Check unused note for
2705 matched operands of insn with no output reloads.
2707 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2709 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2710 (mips_move_from_gpr_cost): Likewise.
2712 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2714 PR rtl-optimization/59858
2715 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2716 ira_class_hard_regs_num.
2717 (process_alt_operands): Increase reject for dying matched operand.
2719 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2722 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2723 smaller than size, perform several stores or loads and stores
2724 at dst + count - size to store or copy all of size bytes, rather
2725 than just last modesize bytes.
2727 2014-01-20 DJ Delorie <dj@redhat.com>
2729 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2730 that CLOBBERs are REGs before propogating their values.
2732 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2735 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2736 (cgraph_inline_failed_type): New function.
2737 * cgraph.h (DEFCIFCODE): Add type.
2738 (cgraph_inline_failed_type_t): New enum.
2739 (cgraph_inline_failed_type): New prototype.
2740 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2741 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2742 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2743 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2744 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2745 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2746 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2747 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2748 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2749 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2750 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2751 OPTIMIZATION_MISMATCH.
2752 * tree-inline.c (expand_call_inline): Emit errors during
2753 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2755 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2758 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2759 mode attribute in insn output.
2761 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2763 * output.h (output_constant): Delete.
2764 * varasm.c (output_constant): Make private.
2766 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2768 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2770 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2773 * tree.h (fold_builtin_strcat): New prototype.
2774 * builtins.c (fold_builtin_strcat): No longer static. Add len
2775 argument, if non-NULL, don't call c_strlen. Optimize
2776 directly into __builtin_memcpy instead of __builtin_strcpy.
2777 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2778 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2780 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2782 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2783 for SImode_address_operand operands, having only a REG argument.
2785 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2787 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2788 loader name using mbig-endian.
2789 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2791 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2793 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2797 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2799 * config/aarch64/aarch64-protos.h
2800 (aarch64_cannot_change_mode_class_ptr): Declare.
2801 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2802 aarch64_cannot_change_mode_class_ptr): New.
2803 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2804 backend hook aarch64_cannot_change_mode_class.
2806 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2808 * common/config/aarch64/aarch64-common.c
2809 (aarch64_handle_option): Don't handle any option order logic here.
2810 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2811 selected_cpu, warn on architecture version mismatch.
2812 (aarch64_override_options): Fix parsing order for option strings.
2814 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2815 Iain Sandoe <iain@codesourcery.com>
2818 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2819 warning. Amend comment to reflect current functionality.
2821 2014-01-20 Richard Biener <rguenther@suse.de>
2824 * builtins.c (fold_builtin_strcat): Remove case better handled
2825 by tree-ssa-strlen.c.
2827 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2829 * config/aarch64/aarch64.opt
2830 (mcpu, march, mtune): Make case-insensitive.
2832 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2835 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2836 if operands[1] is a REG or ZERO_EXTEND of a REG.
2838 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2840 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2842 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2844 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2845 long non-pic millicode calls.
2847 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2849 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2851 2014-01-19 Kito Cheng <kito@0xlab.org>
2853 * builtins.c (expand_movstr): Check movstr expand done or fail.
2855 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2856 H.J. Lu <hongjiu.lu@intel.com>
2859 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2860 to DImode for zero-extended addresses.
2862 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2864 PR rtl-optimization/57763
2865 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2866 on the new indirect jump_insn and increment LABEL_NUSES (label).
2868 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2872 * config.gcc (x86_archs): New variable.
2873 (x86_64_archs): Likewise.
2874 (x86_cpus): Likewise.
2875 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2876 --with-arch/--with-cpu= options.
2877 Support --with-arch=/--with-cpu={nehalem,westmere,
2878 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2880 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2882 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2883 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2885 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2887 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2889 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2892 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2893 clear cpp_get_options (parse_in)->warn_unused_macros for
2894 ix86_target_macros_internal with cpp_define.
2896 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2898 * jump.c (delete_related_insns): Keep (use (insn))s.
2899 * reorg.c (redundant_insn): Check for barriers too.
2901 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2903 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2905 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2907 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2908 call to $$dyncall when TARGET_LONG_CALLS is true.
2910 2014-01-17 Jeff Law <law@redhat.com>
2912 * ree.c (combine_set_extension): Temporarily disable test for
2913 changing number of hard registers.
2915 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2918 * ipa-inline-analysis.c (inline_free_summary):
2919 Do not free summary of aliases.
2921 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2924 * gimplify.c (gimplify_expr): Use create_tmp_var
2925 instead of create_tmp_var_raw. If cond doesn't have
2926 integral type, don't add the IFN_ANNOTATE builtin at all.
2928 2014-01-17 Martin Jambor <mjambor@suse.cz>
2931 * ipa-cp.c (prev_edge_clone): New variable.
2932 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2933 Also resize prev_edge_clone vector.
2934 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2935 (ipcp_edge_removal_hook): New function.
2936 (ipcp_driver): Register ipcp_edge_removal_hook.
2938 2014-01-17 Andrew Pinski <apinski@cavium.com>
2939 Steve Ellcey <sellcey@mips.com>
2942 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2945 2014-01-17 Jeff Law <law@redhat.com>
2948 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2949 so that pass_ccp runs first.
2951 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2953 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2954 (ix86_adjust_cost): Use !TARGET_XXX.
2955 (do_reorder_for_imul): Likewise.
2956 (swap_top_of_ready_list): Likewise.
2957 (ix86_sched_reorder): Likewise.
2959 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2961 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2962 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2963 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2964 (intel_memset): New. Duplicate slm_memset.
2965 (intel_cost): New. Duplicate slm_cost.
2966 (m_INTEL): New macro.
2967 (processor_target_table): Add "intel".
2968 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2969 with PROCESSOR_INTEL for "intel".
2970 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2971 PROCESSOR_SILVERMONT.
2972 (ix86_issue_rate): Likewise.
2973 (ix86_adjust_cost): Likewise.
2974 (ia32_multipass_dfa_lookahead): Likewise.
2975 (swap_top_of_ready_list): Likewise.
2976 (ix86_sched_reorder): Likewise.
2977 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2978 instead of TARGET_OPT_AGU.
2979 * config/i386/i386.h (TARGET_INTEL): New.
2980 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2981 (processor_type): Add PROCESSOR_INTEL.
2982 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2983 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2985 2014-01-17 Marek Polacek <polacek@redhat.com>
2988 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2991 2014-01-17 Richard Biener <rguenther@suse.de>
2993 PR tree-optimization/46590
2994 * opts.c (default_options_table): Add entries for
2995 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2996 all enabled at -O1 but not for -Og.
2997 * common.opt (fbranch-count-reg): Remove Init(1).
2998 (fmove-loop-invariants): Likewise.
2999 (ftree-pta): Likewise.
3001 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3003 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3004 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3005 decls to at least the GCC 4.8 used alignments.
3008 * tree-nested.c (convert_nonlocal_reference_stmt,
3009 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3010 of GIMPLE_BIND stmts, adjust associated decls.
3012 2014-01-17 Richard Biener <rguenther@suse.de>
3014 PR tree-optimization/46590
3015 * vec.h (vec<>::bseach): New member function implementing
3016 binary search according to C89 bsearch.
3017 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3018 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3019 bitmap pointer again. Make accesses_in_loop a flat array.
3020 (mem_ref_obstack): New global.
3021 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3022 (mark_ref_stored): Likewise.
3023 (ref_indep_loop_p_2): Likewise.
3024 (set_ref_stored_in_loop): New helper function.
3025 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3026 (memref_free): Adjust.
3027 (record_mem_ref_loc): Simplify.
3028 (gather_mem_refs_stmt): Adjust.
3029 (sort_locs_in_loop_postorder_cmp): New function.
3030 (analyze_memory_references): Sort accesses_in_loop after
3031 loop postorder number.
3032 (find_ref_loc_in_loop_cmp): New function.
3033 (for_all_locs_in_loop): Find relevant cluster of locs in
3034 accesses_in_loop and iterate without recursion.
3035 (execute_sm): Avoid uninit warning.
3036 (struct ref_always_accessed): Simplify.
3037 (ref_always_accessed::operator ()): Likewise.
3038 (ref_always_accessed_p): Likewise.
3039 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3040 loop postorder numbers here.
3041 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3044 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3047 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3048 on decls for which assemble_alias has been called.
3050 2014-01-17 Nick Clifton <nickc@redhat.com>
3052 * config/msp430/msp430.opt: (mcpu): New option.
3053 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3054 (msp430_option_override): Parse target_cpu. If the MCU name
3055 matches a generic string, clear target_mcu.
3056 (msp430_attr): Allow numeric interrupt values up to 63.
3057 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3058 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3060 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3062 * config/msp430/msp430.md (popm): Use %J rather than %I.
3063 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3064 (addhi_cy_i): Use immediate_operand for operand 2.
3065 * doc/invoke.texi: Document -mcpu option.
3067 2014-01-17 Richard Biener <rguenther@suse.de>
3069 PR rtl-optimization/38518
3070 * df.h (df_analyze_loop): Declare.
3071 * df-core.c: Include cfgloop.h.
3072 (df_analyze_1): Split out main part of df_analyze.
3073 (df_analyze): Adjust.
3074 (loop_inverted_post_order_compute): New function.
3075 (loop_post_order_compute): Likewise.
3076 (df_analyze_loop): New function avoiding whole-function
3078 * loop-invariant.c (find_defs): Use df_analyze_loop.
3079 (find_invariants): Adjust.
3080 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3082 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3084 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3085 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3087 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3089 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3090 traversal when removing references.
3092 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3095 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3097 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3100 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3101 pushing a reload for an autoinc when we had previously reloaded an
3102 inner part of the address.
3104 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3106 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3108 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3109 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3110 when not giving up or versioning for alias only because of
3112 (vect_analyze_data_ref_dependences): Set to true.
3113 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3115 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3116 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3120 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3123 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3124 operand 0 predicate for gathers, use a new pseudo as subtarget.
3126 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3129 * lra-constraints.c (process_alt_operands): Add printing debug info.
3130 Check absence of input/output reloads for matched operands too.
3132 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3134 PR rtl-optimization/59835
3135 * ira.c (ira_init_register_move_cost): Increase cost for
3138 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3140 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3142 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3145 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3146 non-register objects. Use gen_(high/low)part more consistently.
3149 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3152 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3153 endian support, remove tests for WORDS_BIG_ENDIAN.
3154 (p8_mfvsrd_3_<mode>): Likewise.
3155 (reload_gpr_from_vsx<mode>): Likewise.
3156 (reload_gpr_from_vsxsf): Likewise.
3157 (p8_mfvsrd_4_disf): Likewise.
3159 2014-01-16 Richard Biener <rguenther@suse.de>
3161 PR rtl-optimization/46590
3162 * lcm.c (compute_antinout_edge): Use postorder iteration.
3163 (compute_laterin): Use inverted postorder iteration.
3165 2014-01-16 Nick Clifton <nickc@redhat.com>
3168 * varasm.c (output_constant): Return the number of bytes actually
3170 (output_constructor_array_range): Update the field size with the
3171 number of bytes emitted by output_constant.
3172 (output_constructor_regular_field): Likewise. Also do not
3173 complain if the total number of bytes emitted is now greater
3174 than the expected fieldpos.
3175 * output.h (output_constant): Update prototype and descriptive comment.
3177 2014-01-16 Marek Polacek <polacek@redhat.com>
3180 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3181 it is error_mark_node.
3183 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3185 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3186 VALID_AVX256_REG_OR_OI_MODE.
3188 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3190 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3191 current procedure should be profiled.
3193 2014-01-15 Andrew Pinski <apinski@cavium.com>
3195 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3196 of moving from/to the STACK_REG register class.
3198 2014-01-15 Richard Henderson <rth@redhat.com>
3201 * reginfo.c (global_regs_decl): Globalize.
3202 * rtl.h (global_regs_decl): Declare.
3203 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3204 reserved via global_regs.
3206 2014-01-15 Teresa Johnson <tejohnson@google.com>
3208 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3210 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3212 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3213 and vmulosh rather than call gen_vec_widen_smult_*.
3214 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3215 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3216 (vec_widen_smult_even_v16qi): Likewise.
3217 (vec_widen_umult_even_v8hi): Likewise.
3218 (vec_widen_smult_even_v8hi): Likewise.
3219 (vec_widen_umult_odd_v16qi): Likewise.
3220 (vec_widen_smult_odd_v16qi): Likewise.
3221 (vec_widen_umult_odd_v8hi): Likewise.
3222 (vec_widen_smult_odd_v8hi): Likewise.
3223 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3224 vmuloub rather than call gen_vec_widen_umult_*.
3225 (vec_widen_umult_lo_v16qi): Likewise.
3226 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3227 vmulosb rather than call gen_vec_widen_smult_*.
3228 (vec_widen_smult_lo_v16qi): Likewise.
3229 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3230 rather than call gen_vec_widen_umult_*.
3231 (vec_widen_umult_lo_v8hi): Likewise.
3232 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3233 rather than call gen_vec_widen_smult_*.
3234 (vec_widen_smult_lo_v8hi): Likewise.
3236 2014-01-15 Jeff Law <law@redhat.com>
3238 PR tree-optimization/59747
3239 * ree.c (find_and_remove_re): Properly handle case where a second
3240 eliminated extension requires widening a copy created for elimination
3241 of a prior extension.
3242 (combine_set_extension): Ensure that the number of hard regs needed
3243 for a destination register does not change when we widen it.
3245 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3247 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3248 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3249 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3250 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3251 (avr-*-rtems*): Likewise.
3252 (bfin*-rtems*): Likewise.
3253 (moxie-*-rtems*): Likewise.
3254 (h8300-*-rtems*): Likewise.
3255 (i[34567]86-*-rtems*): Likewise.
3256 (lm32-*-rtems*): Likewise.
3257 (m32r-*-rtems*): Likewise.
3258 (m68k-*-rtems*): Likewise.
3259 (microblaze*-*-rtems*): Likewise.
3260 (mips*-*-rtems*): Likewise.
3261 (powerpc-*-rtems*): Likewise.
3262 (sh-*-rtems*): Likewise.
3263 (sparc-*-rtems*): Likewise.
3264 (sparc64-*-rtems*): Likewise.
3265 (v850-*-rtems*): Likewise.
3266 (m32c-*-rtems*): Likewise.
3268 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3270 PR rtl-optimization/59511
3271 * ira.c (ira_init_register_move_cost): Use memory costs for some
3272 cases of register move cost calculations.
3273 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3274 instead of BB frequency.
3275 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3276 * lra-assigns.c (find_hard_regno_for): Ditto.
3278 2014-01-15 Richard Biener <rguenther@suse.de>
3280 PR tree-optimization/59822
3281 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3282 (vectorizable_load): Use it to hoist defs of uses of invariant
3283 loads out of the loop.
3285 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3286 Kugan Vivekanandarajah <kuganv@linaro.org>
3289 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3292 2014-01-15 Richard Biener <rguenther@suse.de>
3294 PR rtl-optimization/59802
3295 * lcm.c (compute_available): Use inverted postorder to seed
3296 the initial worklist.
3298 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3301 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3302 ADDR_REGS for invalid symrefs in non-PIC code.
3304 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3307 * builtins.c (determine_block_size): Initialize *probable_max_size
3308 even if len_rtx is CONST_INT.
3310 2014-01-14 Andrew Pinski <apinski@cavium.com>
3312 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3313 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3314 (cortexa53_tunings): Likewise.
3315 (aarch64_sched_issue_rate): New function.
3316 (TARGET_SCHED_ISSUE_RATE): Define.
3318 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3320 * ira-costs.c (find_costs_and_classes): Add missed
3321 ira_init_register_move_cost_if_necessary.
3323 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3326 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3328 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3331 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3332 to indicate if type is used for function return value. Warn ABI
3333 change if the vector mode isn't available for function return value.
3334 (ix86_function_arg_advance): Pass false to type_natural_mode.
3335 (ix86_function_arg): Likewise.
3336 (ix86_gimplify_va_arg): Likewise.
3337 (function_arg_32): Don't warn ABI change.
3338 (ix86_function_value): Pass true to type_natural_mode.
3339 (ix86_return_in_memory): Likewise.
3340 (ix86_struct_value_rtx): Removed.
3341 (TARGET_STRUCT_VALUE_RTX): Likewise.
3343 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3345 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3346 converting a conditional jump into a conditional return.
3348 2014-01-14 Richard Biener <rguenther@suse.de>
3350 PR tree-optimization/58921
3351 PR tree-optimization/59006
3352 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3353 hoisting invariant stmts.
3354 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3355 invariant loads on the preheader edge if possible.
3357 2014-01-14 Joey Ye <joey.ye@arm.com>
3359 * doc/plugin.texi (Building GCC plugins): Update to C++.
3361 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3363 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3364 (_mm_rcp28_round_ss): Ditto.
3365 (_mm_rsqrt28_round_sd): Ditto.
3366 (_mm_rsqrt28_round_ss): Ditto.
3367 (_mm_rcp28_sd): Ditto.
3368 (_mm_rcp28_ss): Ditto.
3369 (_mm_rsqrt28_sd): Ditto.
3370 (_mm_rsqrt28_ss): Ditto.
3371 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3372 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3373 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3374 (IX86_BUILTIN_RCP28SD): Ditto.
3375 (IX86_BUILTIN_RCP28SS): Ditto.
3376 (IX86_BUILTIN_RSQRT28SD): Ditto.
3377 (IX86_BUILTIN_RSQRT28SS): Ditto.
3378 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3379 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3380 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3381 (ix86_expand_special_args_builtin): Expand new FTYPE.
3382 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3383 (srcp14<mode>): Make insn unary.
3384 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3385 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3386 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3387 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3388 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3389 Fix rounding: make it SAE only.
3390 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3392 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3394 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3395 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3396 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3397 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3398 (round_saeonly_mask_scalar_operand4): Ditto.
3399 (round_saeonly_mask_scalar_op3): Ditto.
3400 (round_saeonly_mask_scalar_op4): Ditto.
3402 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3404 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3405 Implement -maltivec=be for vec_insert and vec_extract.
3407 2014-01-10 DJ Delorie <dj@redhat.com>
3409 * config/msp430/msp430.md (call_internal): Don't allow memory
3410 references with SP as the base register.
3411 (call_value_internal): Likewise.
3412 * config/msp430/constraints.md (Yc): New. For memory references
3413 that don't use SP as a base register.
3415 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3416 "an integer without a # prefix"
3417 * config/msp430/msp430.md (epilogue_helper): Use it.
3419 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3422 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3423 AVX512F gather builtins.
3424 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3425 on gather decls with INTEGER_TYPE masktype.
3426 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3427 directly into the builtin rather than hoisting it before loop.
3429 PR tree-optimization/59387
3430 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3431 (scev_const_prop): If folded_casts and type has undefined overflow,
3432 use force_gimple_operand instead of force_gimple_operand_gsi and
3433 for each added stmt if it is assign with
3434 arith_code_with_undefined_signed_overflow, call
3435 rewrite_to_defined_overflow.
3436 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3437 gimple-fold.h instead.
3438 (arith_code_with_undefined_signed_overflow,
3439 rewrite_to_defined_overflow): Moved to ...
3440 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3441 rewrite_to_defined_overflow): ... here. No longer static.
3442 Include gimplify-me.h.
3443 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3444 rewrite_to_defined_overflow): New prototypes.
3446 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3448 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3450 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3452 * builtins.c (get_object_alignment_2): Minor tweak.
3453 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3455 2014-01-13 Christian Bruel <christian.bruel@st.com>
3457 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3458 optimized non constant lengths.
3460 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3463 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3464 load as __atomic_load_N if possible.
3466 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3468 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3470 (rs6000_expand_builtin): Adjust call.
3472 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3475 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3476 * config/rs6000/rs6000.c: Include target-globals.h.
3477 (rs6000_set_current_function): Instead of doing target_reinit
3478 unconditionally, use save_target_globals_default_opts and
3479 restore_target_globals.
3481 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3483 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3484 (rs6000_expand_builtin): Handle mffs and mtfsf.
3485 (rs6000_init_builtins): Define mffs and mtfsf.
3486 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3487 (rs6000_mffs): New pattern.
3488 (rs6000_mtfsf): New pattern.
3490 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3492 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3493 Start narrowing with START. Apply candidate-use pair
3494 and check overall cost in narrowing.
3495 (iv_ca_prune): Pass new argument.
3497 2014-01-10 Jeff Law <law@redhat.com>
3500 * ree.c (combine_reaching_defs): Ensure the defining statement
3501 occurs before the extension when optimizing extensions with
3502 different source and destination hard registers.
3504 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3507 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3508 vtables into the type inheritance graph.
3510 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3512 PR rtl-optimization/59754
3513 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3514 modes in the REGNO != REGNO case.
3516 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3518 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3520 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3522 PR tree-optimization/59745
3523 * tree-predcom.c (tree_predictive_commoning_loop): Call
3524 free_affine_expand_cache if giving up because components is NULL.
3526 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3527 GC in payload of target_globals struct instead of allocating them on
3528 the heap and the larger structs separately using GC.
3529 * target-globals.h (struct target_globals): Make regs, hard_regs,
3530 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3531 of GTY((skip)) and change type to void *.
3532 (reset_target_globals): Cast loads from those fields to corresponding
3535 2014-01-10 Steve Ellcey <sellcey@mips.com>
3538 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3539 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3540 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3542 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3545 * aarch64-modes.def (CC_Zmode): New flags mode.
3546 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3547 represents an equality.
3548 (aarch64_get_condition_code): Handle CC_Zmode.
3549 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3551 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3553 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3554 extraction in good case.
3556 2014-01-10 Richard Biener <rguenther@suse.de>
3558 PR tree-optimization/59374
3559 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3560 checking after SLP discovery. Mark stmts not participating
3561 in any SLP instance properly.
3563 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3565 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3566 when handling a SET rtx.
3568 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3570 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3571 (cortex-a57): Likewise.
3572 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3574 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3576 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3577 non-iwmmxt builtins.
3579 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3583 * ipa-devirt.c record_target_from_binfo): Take as argument
3584 stack of binfos and lookup matching one for virtual inheritance.
3585 (possible_polymorphic_call_targets_1): Update.
3587 2014-01-10 Huacai Chen <chenhc@lemote.com>
3589 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3590 kernel strings for Loongson-2E/2F/3A.
3592 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3595 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3596 is_gimple_call before calling gimple_call_internal_p.
3598 2014-01-09 Steve Ellcey <sellcey@mips.com>
3600 * Makefile.in (TREE_FLOW_H): Remove.
3601 (TREE_SSA_H): Add file names from tree-flow.h.
3602 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3603 * tree.h: Remove tree-flow.h reference.
3604 * hash-table.h: Remove tree-flow.h reference.
3605 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3606 reference with tree-ssa-loop.h.
3608 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3610 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3611 default element-order behavior for -maltivec.
3612 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3614 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3615 when targeting big endian, at least for now.
3616 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3618 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3621 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3622 var satisfies use_register_for_decl, just take into account type
3623 alignment, rather than decl alignment.
3625 PR tree-optimization/59622
3626 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3627 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3628 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3629 Don't devirtualize for inplace at all. For targets.length () == 1,
3630 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3632 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3634 * config/i386/i386.md (cpu): Remove the unused btver1.
3636 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3638 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3640 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3643 * tree-core.h (struct target_globals): New forward declaration.
3644 (struct tree_target_option): Add globals field.
3645 * tree.h (TREE_TARGET_GLOBALS): Define.
3646 (prepare_target_option_nodes_for_pch): New prototype.
3647 * target-globals.h (struct target_globals): Define even if
3649 * tree.c (prepare_target_option_node_for_pch,
3650 prepare_target_option_nodes_for_pch): New functions.
3651 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3652 * config/i386/i386.c: Include target-globals.h.
3653 (ix86_set_current_function): Instead of doing target_reinit
3654 unconditionally, use save_target_globals_default_opts and
3655 restore_target_globals.
3657 2014-01-09 Richard Biener <rguenther@suse.de>
3659 PR tree-optimization/59715
3660 * tree-cfg.h (split_critical_edges): Declare.
3661 * tree-cfg.c (split_critical_edges): Export.
3662 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3664 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3666 * cfgexpand.c (expand_stack_vars): Optionally disable
3667 asan stack protection.
3668 (expand_used_vars): Likewise.
3669 (partition_stack_vars): Likewise.
3670 * asan.c (asan_emit_stack_protection): Optionally disable
3671 after return stack usage.
3672 (instrument_derefs): Optionally disable memory access instrumentation.
3673 (instrument_builtin_call): Likewise.
3674 (instrument_strlen_call): Likewise.
3675 (asan_protect_global): Optionally disable global variables protection.
3676 * doc/invoke.texi: Added doc for new options.
3677 * params.def: Added new options.
3678 * params.h: Likewise.
3680 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3682 PR rtl-optimization/59724
3683 * ifcvt.c (cond_exec_process_if_block): Don't call
3684 flow_find_head_matching_sequence with 0 longest_match.
3685 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3686 non-active insns if !stop_after.
3687 (try_head_merge_bb): Revert 2014-01-07 changes.
3689 2014-01-08 Jeff Law <law@redhat.com>
3691 * ree.c (get_sub_rtx): New function, extracted from...
3692 (merge_def_and_ext): Here.
3693 (combine_reaching_defs): Use get_sub_rtx.
3695 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3697 * cgraph.h (varpool_variable_node): Do not choke on null node.
3699 2014-01-08 Catherine Moore <clm@codesourcery.com>
3701 * config/mips/mips.md (simple_return): Attempt to use JRC
3703 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3705 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3707 PR rtl-optimization/59137
3708 * reorg.c (steal_delay_list_from_target): Call update_block for
3710 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3712 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3714 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3715 two duplicate entries.
3717 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3720 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3722 * config/mips/mips.c (mips_truncated_op_cost): New function.
3723 (mips_rtx_costs): Adjust test for BADDU.
3724 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3726 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3728 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3729 (*baddu_si): ...this new pattern.
3731 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3734 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3736 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3739 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3741 (expand_expr, expand_normal): Adjust.
3742 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3743 inner_reference_p. Use inner_reference_p to expand inner references.
3744 (store_expr): Adjust.
3745 * cfgexpand.c (expand_call_stmt): Adjust.
3747 2014-01-08 Rong Xu <xur@google.com>
3749 * gcov-io.c (gcov_var): Move from gcov-io.h.
3750 (gcov_position): Ditto.
3751 (gcov_is_error): Ditto.
3752 (gcov_rewrite): Ditto.
3753 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3754 only part to libgcc/libgcov.h.
3756 2014-01-08 Marek Polacek <polacek@redhat.com>
3759 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3761 2014-01-08 Marek Polacek <polacek@redhat.com>
3764 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3766 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3768 PR rtl-optimization/59649
3769 * stor-layout.c (get_mode_bounds): For BImode return
3770 0 and STORE_FLAG_VALUE.
3772 2014-01-08 Richard Biener <rguenther@suse.de>
3775 * gimple.h (is_gimple_builtin_call): Remove.
3776 (gimple_builtin_call_types_compatible_p): New.
3777 (gimple_call_builtin_p): New overload.
3778 * gimple.c (is_gimple_builtin_call): Remove.
3779 (validate_call): Rename to ...
3780 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3782 (validate_type): New static function.
3783 (gimple_call_builtin_p): New overload and adjust.
3784 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3785 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3786 (gimple_fold_stmt_to_constant_1): Likewise.
3787 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3789 2014-01-08 Richard Biener <rguenther@suse.de>
3792 * gimplify.c (gimplify_expr): Gimplify register-register type
3793 VIEW_CONVERT_EXPRs to separate stmts.
3795 2014-01-07 Jeff Law <law@redhat.com>
3798 * ree.c (combine_set_extension): Handle case where source
3799 and destination registers in an extension insn are different.
3800 (combine_reaching_defs): Allow source and destination registers
3801 in extension to be different under limited circumstances.
3802 (add_removable_extension): Remove restriction that the
3803 source and destination registers in the extension are the same.
3804 (find_and_remove_re): Emit a copy from the extension's
3805 destination to its source after the defining insn if
3806 the source and destination registers are different.
3809 * ifcvt.c (merge_if_block): If we are merging a block with more than
3810 one successor with a block with no successors, remove any BARRIER
3811 after the second block.
3813 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3815 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3817 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3820 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3821 for 14-bit register offsets when INT14_OK_STRICT is false.
3823 2014-01-07 Roland Stigge <stigge@antcom.de>
3824 Michael Meissner <meissner@linux.vnet.ibm.com>
3827 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3828 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3830 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3832 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3835 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3837 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3838 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3839 rtx is const0_rtx or not.
3841 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3844 * target-globals.c (save_target_globals): Remove this_fn_optab
3846 * toplev.c: Include optabs.h.
3847 (target_reinit): Temporarily restore the global options if another
3848 set of options are in force.
3850 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3852 PR rtl-optimization/58668
3853 * cfgcleanup.c (flow_find_cross_jump): Don't count
3854 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3855 to determine what is counted.
3856 (flow_find_head_matching_sequence): Use active_insn_p to determine
3858 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3860 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3861 determine what is counted.
3863 PR tree-optimization/59643
3864 * tree-predcom.c (split_data_refs_to_components): If one dr is
3865 read and one write, determine_offset fails and the write isn't
3866 in the bad component, just put the read into the bad component.
3868 2014-01-07 Mike Stump <mikestump@comcast.net>
3869 Jakub Jelinek <jakub@redhat.com>
3872 * tree-core.h (struct tree_optimization_option): Change optabs
3873 type from unsigned char * to void *.
3874 * optabs.c (init_tree_optimization_optabs): Adjust
3875 TREE_OPTIMIZATION_OPTABS initialization.
3877 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3880 * config/i386/i386.h (struct machine_function): Add
3881 no_drap_save_restore field.
3882 * config/i386/i386.c (ix86_save_reg): Use
3883 !cfun->machine->no_drap_save_restore instead of
3884 crtl->stack_realign_needed.
3885 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3886 this function clears frame_pointer_needed. Set
3887 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3888 and DRAP reg is needed.
3890 2014-01-06 Marek Polacek <polacek@redhat.com>
3893 * doc/implement-c.texi: Mention that other integer types are
3894 permitted as bit-field types in strictly conforming mode.
3896 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3898 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3901 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3903 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3905 2014-01-06 Martin Jambor <mjambor@suse.cz>
3908 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3910 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3912 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3916 * var-tracking.c (add_stores): Preserve the value of the source even if
3917 we don't record the store.
3919 2014-01-06 Terry Guo <terry.guo@arm.com>
3921 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3923 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3926 * config/darwin.c (darwin_function_section): Adjust return values to
3927 correspond to optimisation changes made in r206070.
3929 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3931 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3932 from prefetch_block tune setting.
3933 (nocona_cost): Correct size of prefetch block to 64.
3935 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3937 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3938 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3939 used to save the static chain register in the computation of the offset
3940 from which the FP registers need to be restored.
3942 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3944 PR tree-optimization/59519
3945 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3946 ICE if get_current_def (current_new_name) is already non-NULL, as long
3947 as it is a phi result of some other phi in *new_exit_bb that has
3950 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3951 or vmovdqu* for misaligned_operand.
3952 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3953 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3954 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3955 aligned_mem for AVX512F masked aligned load and store builtins and for
3958 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3960 PR tree-optimization/59651
3961 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3962 Address range for negative step should be added by TYPE_SIZE_UNIT.
3964 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3966 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3967 overlapping registers also for registers other than the stack pointer.
3969 2014-01-03 Marek Polacek <polacek@redhat.com>
3972 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3975 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3978 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3981 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3982 (push splitter): Use <P:MODE_SIZE> instead of
3983 GET_MODE_SIZE (<P:MODE>mode).
3984 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3985 (mov -1, reg peephole2): Likewise.
3986 * config/i386/sse.md (*mov<mode>_internal,
3987 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3988 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3989 *<code><mode>3, *andnot<mode>3<mask_name>,
3990 <mask_codefor><code><mode>3<mask_name>): Likewise.
3991 * config/i386/subst.md (mask_mode512bit_condition,
3992 sd_mask_mode512bit_condition): Likewise.
3994 2014-01-02 Xinliang David Li <davidxl@google.com>
3996 PR tree-optimization/59303
3997 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3998 (dump_predicates): Better output format.
3999 (pred_equal_p): New function.
4000 (is_neq_relop_p): Ditto.
4001 (is_neq_zero_form_p): Ditto.
4002 (pred_expr_equal_p): Ditto.
4003 (pred_neg_p): Ditto.
4004 (simplify_pred): Ditto.
4005 (simplify_preds_2): Ditto.
4006 (simplify_preds_3): Ditto.
4007 (simplify_preds_4): Ditto.
4008 (simplify_preds): Ditto.
4010 (push_to_worklist): Ditto.
4011 (get_pred_info_from_cmp): Ditto.
4012 (is_degenerated_phi): Ditto.
4013 (normalize_one_pred_1): Ditto.
4014 (normalize_one_pred): Ditto.
4015 (normalize_one_pred_chain): Ditto.
4016 (normalize_preds): Ditto.
4017 (normalize_cond_1): Remove function.
4018 (normalize_cond): Ditto.
4019 (is_gcond_subset_of): Ditto.
4020 (is_subset_of_any): Ditto.
4021 (is_or_set_subset_of): Ditto.
4022 (is_and_set_subset_of): Ditto.
4023 (is_norm_cond_subset_of): Ditto.
4024 (pred_chain_length_cmp): Ditto.
4025 (convert_control_dep_chain_into_preds): Type change.
4026 (find_predicates): Ditto.
4027 (find_def_preds): Ditto.
4028 (destroy_predicates_vecs): Ditto.
4029 (find_matching_predicates_in_rest_chains): Ditto.
4030 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4031 (is_pred_expr_subset): Ditto.
4032 (is_pred_chain_subset_of): Ditto.
4033 (is_included_in): Ditto.
4034 (is_superset_of): Ditto.
4036 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4038 Update copyright years.
4040 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4042 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4043 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4044 config/arc/arc.md, config/arc/arc.opt,
4045 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4046 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4047 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4048 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4049 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4050 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4051 vtable-verify.c, vtable-verify.h: Use the standard form for the
4054 2014-01-02 Tobias Burnus <burnus@net-b.de>
4056 * gcc.c (process_command): Update copyright notice dates.
4057 * gcov-dump.c: Ditto.
4059 * doc/cpp.texi: Bump @copying's copyright year.
4060 * doc/cppinternals.texi: Ditto.
4061 * doc/gcc.texi: Ditto.
4062 * doc/gccint.texi: Ditto.
4063 * doc/gcov.texi: Ditto.
4064 * doc/install.texi: Ditto.
4065 * doc/invoke.texi: Ditto.
4067 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4069 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4071 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4073 * config/i386/sse.md (*mov<mode>_internal): Guard
4074 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4076 PR rtl-optimization/59647
4077 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4078 new_rtx into UNSIGNED_FLOAT rtxes.
4080 Copyright (C) 2014 Free Software Foundation, Inc.
4082 Copying and distribution of this file, with or without modification,
4083 are permitted in any medium without royalty provided the copyright
4084 notice and this notice are preserved.