1 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk
4 and define TARGET_ASM_OUTPUT_MI_THUNK and
5 TARGET_ASM_CAN_OUTPUT_MI_THUNK.
7 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
9 * config/microblaze/predicates.md: Add cmp_op predicate.
10 * config/microblaze/microblaze.md: Add branch_compare instruction
11 which uses cmp_op predicate and emits cmp insn before branch.
12 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
13 to microblaze_expand_conditional_branch and consolidate logic.
14 (microblaze_expand_conditional_branch): emit branch_compare
15 insn instead of handling cmp op separate from branch insn.
17 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
22 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
25 define_insn with define_expand and new define_insn
26 *altivec_lve<VI_char>x_internal.
27 (altivec_stve<VI_char>x): Replace define_insn with define_expand
28 and new define_insn *altivec_stve<VI_char>x_internal.
29 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
31 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
33 (altivec_expand_stvex_be): New function.
35 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
37 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
39 to STACK_POINTER_REGNUM if !frame_pointer_needed.
40 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
41 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
43 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
46 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
47 instead of emit_move_insn.
49 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
51 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
53 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
56 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
58 * config/rs6000/altivec.md (altivec_lvxl): Rename as
59 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
61 (altivec_lvxl_<mode>): New define_expand incorporating
62 -maltivec=be semantics where needed.
63 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
64 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
65 semantics where needed.
66 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
67 (altivec_stvx_<mode>): New define_expand incorporating
68 -maltivec=be semantics where needed.
69 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
70 VM2 iterator instead of V4SI.
71 (altivec_stvxl_<mode>): New define_expand incorporating
72 -maltivec=be semantics where needed.
73 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
74 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
75 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
76 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
77 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
79 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
80 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
81 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
82 ALTIVEC_BUILTIN_STVXL.
83 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
85 (altivec_expand_stvx_be): Likewise.
86 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
87 (altivec_expand_lvx_be): Likewise.
88 (altivec_expand_stvx_be): Likewise.
89 (altivec_expand_builtin): Add cases for
90 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
91 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
92 (altivec_init_builtins): Add definitions for
93 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
94 __builtin_altivec_stvx_<mode>, and
95 __builtin_altivec_stvxl_<mode>.
97 2014-02-21 Catherine Moore <clm@codesourcery.com>
99 * doc/invoke.texi (mvirt, mno-virt): Document.
100 * config/mips/mips.opt (mvirt): New option.
101 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
103 2014-02-21 Richard Biener <rguenther@suse.de>
105 PR tree-optimization/60276
106 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
107 (STMT_VINFO_MIN_NEG_DIST): New macro.
108 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
109 STMT_VINFO_MIN_NEG_DIST.
110 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
111 made for negative dependence distances still hold.
113 2014-02-21 Richard Biener <rguenther@suse.de>
116 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
117 DECL_INITIAL for globals not in the current function context.
119 2014-02-21 Jakub Jelinek <jakub@redhat.com>
121 PR tree-optimization/56490
122 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
123 * tree-ssa-uninit.c: Include params.h.
124 (compute_control_dep_chain): Add num_calls argument, return false
125 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
126 num_calls to recursive call.
127 (find_predicates): Change dep_chain into normal array,
128 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
129 variable and adjust compute_control_dep_chain caller.
130 (find_def_preds): Likewise.
132 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
134 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
135 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
137 2014-02-21 Nick Clifton <nickc@redhat.com>
139 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
141 (popqi1): Add mode to pre_dec.
144 2014-02-21 Jakub Jelinek <jakub@redhat.com>
146 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
147 mode for mask of V8SFmode permutation.
149 2014-02-20 Richard Henderson <rth@redhat.com>
152 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
153 a new pseudo for OLDVAL.
155 2014-02-20 Jakub Jelinek <jakub@redhat.com>
158 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
159 gen_reg_rtx if d->testing_p.
160 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
161 if d->testing_p and we will certainly return true.
162 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
165 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
167 * emit-rtl.c (gen_reg_rtx): Assert that
168 crtl->emit.regno_pointer_align_length is non-zero.
170 2014-02-20 Richard Henderson <rth@redhat.com>
173 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
174 on failure the store back into EXPECT.
176 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
177 Sandra Loosemore <sandra@codesourcery.com>
179 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
180 * config/nios2/nios2.c (nios2_function_profiler):
181 Add -fPIC (flag_pic == 2) support.
182 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
183 (nios2_large_offset_p): New function.
184 (nios2_unspec_reloc_p): Move up position, update to use
185 nios2_large_offset_p.
186 (nios2_unspec_address): Remove function.
187 (nios2_unspec_offset): New function.
188 (nios2_large_got_address): New function.
189 (nios2_got_address): Add large offset support.
190 (nios2_legitimize_tls_address): Update usage of removed and new
192 (nios2_symbol_binds_local_p): New function.
193 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
194 (nios2_legitimize_address): Update to use nios2_large_offset_p.
195 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
196 (nios2_print_operand): Merge H/L processing, add hiadj/lo
197 processing for (const (unspec ...)).
198 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
200 2014-02-20 Richard Biener <rguenther@suse.de>
202 * tree-cfg.c (replace_uses_by): Mark altered BBs before
203 doing the substitution.
204 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
206 2014-02-20 Martin Jambor <mjambor@suse.cz>
209 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
210 info when checking whether lattices are bottom.
212 2014-02-20 Richard Biener <rguenther@suse.de>
215 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
218 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
221 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
222 parameter specifying the scaling.
223 (inline_call): Update.
224 (want_inline_recursively): Guard division by zero.
225 (recursive_inlining): Update.
226 * ipa-inline.h (clone_inlined_nodes): Update.
228 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
231 * config/i386/i386.c (classify_argument): Pass structures of size
232 64 bytes or less in register.
234 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
235 Kirill Yukhin <kirill.yukhin@intel.com>
237 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
238 (_mm_rcp28_round_ss): Ditto.
239 (_mm_rsqrt28_round_sd): Ditto.
240 (_mm_rsqrt28_round_ss): Ditto.
241 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
242 (_mm_rcp14_round_ss): Ditto.
243 (_mm_rsqrt14_round_sd): Ditto.
244 (_mm_rsqrt14_round_ss): Ditto.
245 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
246 the first input operand, get rid of match_dup.
247 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
249 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
251 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
252 operand as the first input operand, set type attribute.
253 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
255 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
256 operand as the first input operand, set type attribute.
258 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
260 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
263 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
266 * config/i386/i386.c (construct_container): Remove TFmode check
267 for X86_64_INTEGER_CLASS.
269 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
272 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
273 only when -Wpsabi is enabled.
275 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
278 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
279 passing arrays in registers are the same as for structs, so remove the
280 special case for them.
282 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
284 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
285 destination type, extract only the valid bits if the source type is not
286 integral and has a different mode.
288 2014-02-19 Richard Biener <rguenther@suse.de>
291 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
294 2014-02-19 Richard Biener <rguenther@suse.de>
297 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
298 (ipa_modify_call_arguments): Emit an argument load explicitely and
299 preserve virtual SSA form there and for the replacement call.
300 Do not update SSA form nor free dominance info.
302 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
304 * ipa.c (function_and_variable_visibility): Also clear WEAK
305 flag when disolving COMDAT_GROUP.
307 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
309 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
310 * ipa-prop.c (ipa_set_jf_known_type): Return early when
312 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
313 do more sanity checks.
314 (detect_type_change): Return true when giving up early.
315 (compute_complex_assign_jump_func): Fix type parameter of
317 (compute_complex_ancestor_jump_func): Likewise.
318 (update_jump_functions_after_inlining): Fix updating of
320 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
322 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
324 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
325 inline clones when edge disappears.
327 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
330 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
331 Split 64-bit moves into 2 patterns. Do not allow the use of
332 direct move for TDmode in little endian, since the decimal value
333 has little endian bytes within a word, but the 64-bit pieces are
334 ordered in a big endian fashion, and normal subreg's of TDmode are
336 (mov<mode>_64bit_dm): Likewise.
337 (movtd_64bit_nodm): Likewise.
339 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
341 PR tree-optimization/60174
342 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
343 statement of an SSA_NAME that occurs in an abnormal PHI node.
345 2014-02-18 Jakub Jelinek <jakub@redhat.com>
348 * final.c (SEEN_BB): Remove.
349 (SEEN_NOTE, SEEN_EMITTED): Renumber.
350 (final_scan_insn): Don't force_source_line on second
351 NOTE_INSN_BASIC_BLOCK.
353 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
356 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
357 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
358 (type_natural_mode): Warn ABI change when %zmm register is not
359 available for AVX512F vector value passing.
361 2014-02-18 Kai Tietz <ktietz@redhat.com>
364 * config/i386/i386.c (ix86_expand_prologue): Use value in
365 rax register as displacement when restoring %r10 or %rax.
366 Fix wrong offset when restoring both registers.
368 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
370 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
371 assertion with conditional return.
373 2014-02-18 Jakub Jelinek <jakub@redhat.com>
374 Uros Bizjak <ubizjak@gmail.com>
377 * config/i386/driver-i386.c (host_detect_local_cpu): If
378 YMM state is not saved by the OS, also clear has_f16c. Move
379 CPUID 0x80000001 handling before YMM state saving checking.
381 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
383 PR rtl-optimization/58960
384 * haifa-sched.c (alloc_global_sched_pressure_data): New,
385 factored out from ...
386 (sched_init): ... here.
387 (free_global_sched_pressure_data): New, factored out from ...
388 (sched_finish): ... here.
389 * sched-int.h (free_global_sched_pressure_data): Declare.
390 * sched-rgn.c (nr_regions_initial): New static global.
391 (haifa_find_rgns): Initialize it.
392 (schedule_region): Disable sched-pressure for the newly
395 2014-02-17 Richard Biener <rguenther@suse.de>
397 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
398 release SSA defs of pattern stmts.
400 2014-02-17 Richard Biener <rguenther@suse.de>
402 * tree-inline.c (expand_call_inline): Release the virtual
403 operand defined by the call we are about to inline.
405 2014-02-17 Richard Biener <rguenther@suse.de>
407 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
409 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
410 Ilya Tocar <ilya.tocar@intel.com>
412 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
413 arguments order in builtin.
414 (_mm512_permutexvar_epi64): Ditto.
415 (_mm512_mask_permutexvar_epi64): Ditto
416 (_mm512_maskz_permutexvar_epi32): Ditto
417 (_mm512_permutexvar_epi32): Ditto
418 (_mm512_mask_permutexvar_epi32): Ditto
420 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
422 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
423 (p8_vmrgow): Likewise.
425 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
427 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
430 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
433 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
434 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
435 into 64-bit and 32-bit moves. On 64-bit moves, add support for
436 using direct move instructions on ISA 2.07. Also adjust
437 instruction length for 64-bit.
438 (mov<mode>_64bit, TFmode/TDmode): Likewise.
439 (mov<mode>_32bit, TFmode/TDmode): Likewise.
441 2014-02-15 Alan Modra <amodra@gmail.com>
445 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
446 find_replacement on parts of insn rtl that might be reloaded.
448 2014-02-15 Richard Biener <rguenther@suse.de>
450 PR tree-optimization/60183
451 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
452 (tree_ssa_phiprop): Calculate and free post-dominators.
454 2014-02-14 Jeff Law <law@redhat.com>
456 PR rtl-optimization/60131
457 * ree.c (get_extended_src_reg): New function.
458 (combine_reaching_defs): Use it rather than assuming location of REG.
459 (find_and_remove_re): Verify first operand of extension is
460 a REG before adding the insns to the copy list.
462 2014-02-14 Roland McGrath <mcgrathr@google.com>
464 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
465 * configure: Regenerated.
466 * config.in: Regenerated.
467 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
468 instead of ASM_SHORT.
470 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
471 Richard Earnshaw <rearnsha@arm.com>
473 PR rtl-optimization/59535
474 * lra-constraints.c (process_alt_operands): Encourage alternative
475 when unassigned pseudo class is superset of the alternative class.
476 (inherit_reload_reg): Don't inherit when optimizing for code size.
477 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
478 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
479 modes not less than 4 for Thumb1.
481 2014-02-14 Kyle McMartin <kyle@redhat.com>
484 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
486 2014-02-14 Richard Biener <rguenther@suse.de>
488 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
489 (get_frame_arg): Drop the assert with langhook types_compatible_p.
490 Do not strip INDIRECT_REFs.
492 2014-02-14 Richard Biener <rguenther@suse.de>
495 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
496 DECL_FUNCTION_SPECIFIC_TARGET.
497 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
498 * tree-streamer-out.c (pack_ts_target_option): Remove.
499 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
500 (write_ts_function_decl_tree_pointers): Do not stream
501 DECL_FUNCTION_SPECIFIC_TARGET.
502 * tree-streamer-in.c (unpack_ts_target_option): Remove.
503 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
504 (lto_input_ts_function_decl_tree_pointers): Do not stream
505 DECL_FUNCTION_SPECIFIC_TARGET.
507 2014-02-14 Jakub Jelinek <jakub@redhat.com>
509 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
510 (get_initial_def_for_induction, vectorizable_induction): Ignore
511 debug stmts when looking for exit_phi.
512 (vectorizable_live_operation): Fix up condition.
514 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
516 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
517 nreverse() because it changes the content of original tree list.
519 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
521 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
522 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
524 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
526 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
527 GNU coding standards.
529 2014-02-13 Jakub Jelinek <jakub@redhat.com>
532 * dwarf2out.c (gen_subprogram_die): Don't call
533 add_calling_convention_attribute if subr_die is old_die.
535 2014-02-13 Sharad Singhai <singhai@google.com>
537 * doc/optinfo.texi: Fix order of nodes.
539 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
541 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
542 operands[2], not operands[3].
544 2014-02-13 Richard Biener <rguenther@suse.de>
547 * doc/install.texi (ISL): Update recommended version to 0.12.2,
548 mention the possibility of an in-tree build.
549 (CLooG): Update recommended version to 0.18.1, mention the
550 possibility of an in-tree build and clarify that the ISL
551 bundled with CLooG does not work.
553 2014-02-13 Jakub Jelinek <jakub@redhat.com>
556 * expr.c (compress_float_constant): If x is a hard register,
557 extend into a pseudo and then move to x.
559 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
561 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
562 caused by bad second argument to warning_at() with -mhotpatch and
563 nested functions (e.g. with gfortran).
565 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
567 * opts.c (option_name): Remove "enabled by default" rider.
569 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
571 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
573 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
574 Uros Bizjak <ubizjak@gmail.com>
577 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
578 * configure: Regenerated.
580 2014-02-12 Richard Biener <rguenther@suse.de>
582 * vec.c (vec_prefix::calculate_allocation): Move as
583 inline variant to vec.h.
584 (vec_prefix::calculate_allocation_1): New out-of-line version.
585 * vec.h (vec_prefix::calculate_allocation_1): Declare.
586 (vec_prefix::m_has_auto_buf): Rename to ...
587 (vec_prefix::m_using_auto_storage): ... this.
588 (vec_prefix::calculate_allocation): Inline the easy cases
589 and dispatch to calculate_allocation_1 which doesn't need the
591 (va_heap::reserve): Use gcc_checking_assert.
592 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
593 m_using_auto_storage.
594 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
596 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
597 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
598 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
600 2014-02-12 Richard Biener <rguenther@suse.de>
602 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
603 when we found a dependence.
605 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
607 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
609 (maybe_fold_stmt): ... into this new function.
610 * omp-low.c (lower_omp): Update comment.
612 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
615 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
618 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
620 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
621 identifiers in comments.
622 (cortexa53_extra_costs): Likewise.
623 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
624 (cortexa7_extra_costs): Likewise.
625 (cortexa12_extra_costs): Likewise.
626 (cortexa15_extra_costs): Likewise.
627 (v7m_extra_costs): Likewise.
629 2014-02-12 Richard Biener <rguenther@suse.de>
632 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
633 of posix_memalign being successful.
634 (lower_stmt): Restrict lowering of posix_memalign to when
635 -ftree-bit-ccp is enabled.
637 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
639 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
641 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
643 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
645 PR rtl-optimization/60116
646 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
647 other_insn once the combination has been validated.
649 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
652 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
654 * ipa-devirt.c: Include demangle.h
655 (odr_violation_reported): New static variable.
656 (add_type_duplicate): Update odr_violations.
657 (maybe_record_node): Add completep parameter; update it.
658 (record_target_from_binfo): Add COMPLETEP parameter;
660 (possible_polymorphic_call_targets_1): Likewise.
661 (struct polymorphic_call_target_d): Add nonconstruction_targets;
662 rename FINAL to COMPLETE.
663 (record_targets_from_bases): Sanity check we found the binfo;
664 fix COMPLETEP updating.
665 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
666 parameter, fix computing of COMPLETEP.
667 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
668 at LTO time do demangling.
669 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
670 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
672 (gimple_get_virt_method_for_binfo): Likewise.
673 * gimple-fold.h (gimple_get_virt_method_for_binfo,
674 gimple_get_virt_method_for_vtable): Update prototypes.
676 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
679 * genautomata.c (add_presence_absence): Fix typo with
680 {final_}presence_list.
682 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
685 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
686 for VSX/Altivec vectors that land in GPR registers.
688 2014-02-11 Richard Henderson <rth@redhat.com>
689 Jakub Jelinek <jakub@redhat.com>
692 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
693 around drhs if type conversion to lacc->type is not useless.
695 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
697 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
699 (cortex-a57.cortex-a53): Likewise.
700 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
702 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
704 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
707 2014-02-11 Renlin Li <Renlin.Li@arm.com>
709 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
710 add_options_for_arm_vfp3.
712 2014-02-11 Jeff Law <law@redhat.com>
715 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
716 object with an undesirable mode.
718 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
721 * config/i386/sol2-9.h: New file.
722 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
723 *-*-solaris2.9*): Use it.
725 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
727 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
728 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
730 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
732 * config/microblaze/microblaze.c: Extend mcpu version format
734 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
736 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
738 2014-02-10 Richard Henderson <rth@redhat.com>
741 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
742 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
743 ms-abi vs -mno-accumulate-outgoing-args.
744 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
745 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
748 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
751 * cfgexpand.c (expand_asm_operands): Attach source location to
752 ASM_INPUT rtx objects.
753 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
755 2014-02-10 Nick Clifton <nickc@redhat.com>
757 * config/mn10300/mn10300.c (popcount): New function.
758 (mn10300_expand_prologue): Include saved registers in stack usage
761 2014-02-10 Jeff Law <law@redhat.com>
764 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
765 when changing the SET_DEST of a prior insn to avoid an input reload.
767 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
769 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
770 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
771 -mcall-openbsd, or -mcall-linux.
772 (CC1_ENDIAN_BIG_SPEC): Remove.
773 (CC1_ENDIAN_LITTLE_SPEC): Remove.
774 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
775 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
776 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
777 and %cc1_endian_default.
778 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
780 2014-02-10 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/60115
783 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
784 MEM_REF handling. Properly verify that the accesses are not
785 out of the objects bound.
787 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
789 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
792 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
794 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
795 proper constants and fix formatting.
796 (possible_polymorphic_call_targets): Fix formatting.
798 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
799 Ilya Tocar <ilya.tocar@intel.com>
801 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
802 (_mm512_loadu_epi32): Renamed into...
803 (_mm512_loadu_si512): This.
804 (_mm512_storeu_epi32): Renamed into...
805 (_mm512_storeu_si512): This.
806 (_mm512_maskz_ceil_ps): Removed.
807 (_mm512_maskz_ceil_pd): Ditto.
808 (_mm512_maskz_floor_ps): Ditto.
809 (_mm512_maskz_floor_pd): Ditto.
810 (_mm512_floor_round_ps): Ditto.
811 (_mm512_floor_round_pd): Ditto.
812 (_mm512_ceil_round_ps): Ditto.
813 (_mm512_ceil_round_pd): Ditto.
814 (_mm512_mask_floor_round_ps): Ditto.
815 (_mm512_mask_floor_round_pd): Ditto.
816 (_mm512_mask_ceil_round_ps): Ditto.
817 (_mm512_mask_ceil_round_pd): Ditto.
818 (_mm512_maskz_floor_round_ps): Ditto.
819 (_mm512_maskz_floor_round_pd): Ditto.
820 (_mm512_maskz_ceil_round_ps): Ditto.
821 (_mm512_maskz_ceil_round_pd): Ditto.
822 (_mm512_expand_pd): Ditto.
823 (_mm512_expand_ps): Ditto.
824 * config/i386/i386.c (ix86_builtins): Remove
825 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
827 * config/i386/predicates.md (const1256_operand): New.
828 (const_1_to_2_operand): Ditto.
829 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
830 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
831 (*avx512pf_gatherpf<mode>sf): Ditto.
832 (avx512pf_gatherpf<mode>df): Ditto.
833 (*avx512pf_gatherpf<mode>df_mask): Ditto.
834 (*avx512pf_gatherpf<mode>df): Ditto.
835 (avx512pf_scatterpf<mode>sf): Ditto.
836 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
837 (*avx512pf_scatterpf<mode>sf): Ditto.
838 (avx512pf_scatterpf<mode>df): Ditto.
839 (*avx512pf_scatterpf<mode>df_mask): Ditto.
840 (*avx512pf_scatterpf<mode>df): Ditto.
841 (avx512f_expand<mode>): Removed.
842 (<shift_insn><mode>3<mask_name>): Change predicate type.
844 2014-02-08 Jakub Jelinek <jakub@redhat.com>
846 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
847 not at the end of datarefs vector use ordered_remove to avoid
848 reordering datarefs vector.
851 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
852 mark local addressable non-static vars as GOVD_PRIVATE
853 instead of GOVD_LOCAL.
854 * omp-low.c (lower_omp_for): Move gimple_bind_vars
855 and BLOCK_VARS of gimple_bind_block to new_stmt rather
859 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
860 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
861 assume_aligned or alloc_align attributes.
862 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
863 arguments. Handle also assume_aligned and alloc_align attributes.
864 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
865 calls to functions with assume_aligned or alloc_align attributes.
866 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
868 2014-02-08 Terry Guo <terry.guo@arm.com>
870 * doc/invoke.texi: Document ARM -march=armv7e-m.
872 2014-02-08 Jakub Jelinek <jakub@redhat.com>
874 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
875 flag on __cilkrts_rethrow builtin.
878 * ipa-cp.c (determine_versionability): Fail at -O0
879 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
880 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
883 2014-02-04 Jakub Jelinek <jakub@redhat.com>
886 * tree-inline.c (copy_forbidden): Fail for
887 __attribute__((optimize (0))) functions.
889 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
891 * varpool.c: Include pointer-set.h.
892 (varpool_remove_unreferenced_decls): Variables in other partitions
893 will not be output; be however careful to not lose information
896 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
898 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
899 lookup in the vtable constructor.
901 2014-02-07 Jeff Law <law@redhat.com>
904 * config/m68k/m68k.md (ashldi_extsi): Turn into a
905 define_insn_and_split.
907 * ipa-inline.c (inline_small_functions): Fix typos.
909 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
911 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
912 (s390_can_use_return_insn): Declare.
913 * config/s390/s390.h (EPILOGUE_USES): Define.
914 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
916 (s390_chunkify_start): Handle return JUMP_LABELs.
917 (s390_early_mach): Emit a main_pool instruction on the entry edge.
918 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
919 (s390_can_use_return_insn): New functions.
920 (s390_fix_long_loop_prediction): Handle conditional returns.
921 (TARGET_SET_UP_BY_PROLOGUE): Define.
922 * config/s390/s390.md (ANY_RETURN): New code iterator.
923 (*creturn, *csimple_return, return, simple_return): New patterns.
925 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
927 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
928 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
929 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
930 REG_CFA_RESTORE list when deciding not to restore a register.
932 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
934 * config/s390/s390.c: Include tree-pass.h and context.h.
935 (s390_early_mach): New function, split out from...
936 (s390_emit_prologue): ...here.
937 (pass_data_s390_early_mach): New pass structure.
938 (pass_s390_early_mach): New class.
939 (s390_option_override): Create and register early_mach pass.
942 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
944 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
945 to match for the exit block.
947 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
949 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
950 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
951 Reject misaligned operands.
953 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
955 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
957 2014-02-07 Richard Biener <rguenther@suse.de>
960 * gimple-low.c (lower_builtin_posix_memalign): New function.
961 (lower_stmt): Call it to lower posix_memalign in a way
962 to make alignment info accessible.
964 2014-02-07 Jakub Jelinek <jakub@redhat.com>
967 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
968 __builtin_setjmp_receiver.
970 2014-02-07 Richard Biener <rguenther@suse.de>
973 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
974 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
975 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
976 Handle BUILT_IN_POSIX_MEMALIGN.
977 (find_func_clobbers): Likewise.
978 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
979 (call_may_clobber_ref_p_1): Likewise.
981 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
984 * ipa-devirt.c (record_target_from_binfo): Remove overactive
987 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
990 * lto-cgraph.c (lto_output_node): Use
991 symtab_get_symbol_partitioning_class.
992 (lto_output_varpool_node): likewise.
993 (symtab_get_symbol_partitioning_class): Move here from
995 * cgraph.h (symbol_partitioning_class): Likewise.
996 (symtab_get_symbol_partitioning_class): Declare.
998 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1000 * ggc.h (ggc_internal_cleared_alloc): New macro.
1001 * vec.h (vec_safe_copy): Handle memory stats.
1002 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1003 * target-globals.c (save_target_globals): Likewise.
1005 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1008 * expr.c (emit_move_resolve_push): Export; be bit more selective
1009 on when to clear alias set.
1010 * expr.h (emit_move_resolve_push): Declare.
1011 * function.h (struct function): Add tail_call_marked.
1012 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1013 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1014 * config/i386/i386.md (TImode move expander): De not call
1016 (FP push expanders): Preserve memory attributes.
1017 * config/i386/sse.md (push<mode>1): Remove.
1018 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1019 (ix86_expand_push): Remove.
1020 * config/i386/mmx.md (push<mode>1): Remove.
1022 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1024 PR rtl-optimization/60030
1025 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1026 lopart with paradoxical subreg before shifting it up by hprec.
1028 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1030 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1031 Remove extra newline at end of file.
1032 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1033 (arm_issue_rate): Handle cortexa57.
1034 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1035 (cortex-a57.cortex-a53): Likewise.
1037 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1040 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1041 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1043 (arm_expand_prologue): Adjust all callers.
1044 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1045 info, registers also at the lowest numbered registers side. Use
1046 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1050 * var-tracking.c (adjust_mems): Before adding a SET to
1051 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1053 2014-02-06 Alan Modra <amodra@gmail.com>
1056 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1057 change SDmode to DDmode when lra_in_progress.
1059 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1062 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1063 free_data_ref on the dr first, and before goto again also set dr
1064 to the next dr. For simd_lane_access, free old datarefs[i] before
1065 overwriting it. For get_vectype_for_scalar_type failure, don't
1066 free_data_ref if simd_lane_access.
1068 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1071 * tree.h (opts_for_fn): New inline function.
1072 (opt_for_fn): Define.
1073 * config/i386/i386.c (ix86_function_regparm): Use
1074 opt_for_fn (decl, optimize) instead of optimize.
1076 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1078 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1079 for SYMBOL_REF in large memory model.
1081 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1083 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1085 (cortex-a57): Likewise.
1086 (cortex-a57.cortex-a53): Likewise.
1088 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1089 Kugan Vivekanandarajah <kuganv@linaro.org>
1091 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1093 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1095 2014-02-06 Richard Biener <rguenther@suse.de>
1097 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1098 set_loop_copy and initialize_original_copy_tables.
1100 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1102 * config/aarch64/aarch64-simd.md
1103 (aarch64_ashr_simddi): Change QI to SI.
1105 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1106 Jakub Jelinek <jakub@redhat.com>
1109 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1112 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1114 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1115 CODE_FOR_altivec_vpku[hw]um to
1116 CODE_FOR_altivec_vpku[hw]um_direct.
1117 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1118 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1119 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1120 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1122 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1124 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1125 generation for -maltivec=be.
1126 (altivec_vsumsws): Simplify redundant test.
1128 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1130 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1131 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1132 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1133 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1134 gen_altivec_vpkuwum.
1135 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1137 (altivec_vpks<VI_char>ss): Likewise.
1138 (altivec_vpks<VI_char>us): Likewise.
1139 (altivec_vpku<VI_char>us): Likewise.
1140 (altivec_vpku<VI_char>um): Likewise.
1141 (altivec_vpku<VI_char>um_direct): New (copy of
1142 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1144 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1145 target is little endian and -maltivec=be is not specified.
1146 (*altivec_vupkhs<VU_char>_direct): New (copy of
1147 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1148 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1149 target is little endian and -maltivec=be is not specified.
1150 (*altivec_vupkls<VU_char>_direct): New (copy of
1151 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1152 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1153 little endian and -maltivec=be is not specified.
1154 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1155 little endian and -maltivec=be is not specified.
1157 2014-02-05 Richard Henderson <rth@redhat.com>
1160 * combine-stack-adj.c: Revert r206943.
1161 * sched-int.h (struct deps_desc): Add last_args_size.
1162 * sched-deps.c (init_deps): Initialize it.
1163 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1164 contain REG_ARGS_SIZE notes.
1166 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1168 * lto-cgraph.c (asm_nodes_output): Make global.
1169 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1170 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1171 (driver_handle_option): Handle OPT_fwpa.
1173 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1176 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1177 a comment typo and formatting issue. If odr_hash hasn't been
1178 created, return vNULL and set *completep to false.
1181 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1182 bb with no successors.
1184 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1187 * doc/invoke.texi (-march): Clarify documentation for ARM.
1191 2014-02-05 Richard Biener <rguenther@suse.de>
1193 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1194 when not vectorizing because of too many alias checks.
1195 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1196 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1198 2014-02-05 Nick Clifton <nickc@redhat.com>
1200 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1201 accept extended registers in any mode when compiling for the MN10300.
1203 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1205 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1206 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1207 sanitization attributes.
1208 (can_inline_edge_p): Likewise.
1209 (sanitize_attrs_match_for_inline_p): New function.
1211 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1213 * ipa-prop.c (detect_type_change): Shor circuit testing of
1214 type changes on THIS pointer.
1216 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1219 * config/pa/pa.c (legitimize_tls_address): Return original address
1220 if not passed a SYMBOL_REF rtx.
1221 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1223 (pa_emit_move_sequence): Simplify TLS source operands.
1224 (pa_legitimate_constant_p): Reject all TLS constants.
1225 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1226 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1228 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1230 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1231 groups when we know they are controlled by LTO.
1232 * varasm.c (default_binds_local_p_1): If object is in other partition,
1233 it will be resolved locally.
1235 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1237 * config/host-linux.c (linux_gt_pch_use_address): Don't
1238 use SSIZE_MAX because it is not always defined.
1240 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1243 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1244 threshold for pseudo splitting.
1245 (update_ebb_live_info): Process call argument hard registers and
1246 hard registers from insn definition too.
1247 (max_small_class_regs_num): New constant.
1248 (inherit_in_ebb): Update live hard regs through EBBs. Update
1249 reloads_num only for small register classes. Don't split for
1252 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1255 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1258 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1260 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1263 2014-02-04 Marek Polacek <polacek@redhat.com>
1265 * gdbinit.in (pel): Define.
1267 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1269 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1272 2014-02-04 Richard Biener <rguenther@suse.de>
1275 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1276 in function context local.
1277 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1278 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1279 similar to LTO_imported_decl_ref.
1281 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1283 PR tree-optimization/60002
1284 * cgraphclones.c (build_function_decl_skip_args): Clear
1287 PR tree-optimization/60023
1288 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1289 false to gsi_replace.
1290 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1291 has been in some EH region and vec_stmt could throw, add
1292 vec_stmt into the same EH region.
1293 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1294 has no lhs, ignore it.
1295 * internal-fn.c (expand_MASK_LOAD): Likewise.
1298 * tree-inline.c (copy_forbidden): Fail for
1299 __attribute__((optimize (0))) functions.
1302 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1303 is set, copy one less argument.
1304 (expand_simd_clones): Don't subtract clone_info->inbranch
1305 from simd_clone_struct_alloc argument.
1307 PR rtl-optimization/57915
1308 * recog.c (simplify_while_replacing): If all unary/binary/relational
1309 operation arguments are constant, attempt to simplify those.
1312 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1313 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1315 2014-02-04 Richard Biener <rguenther@suse.de>
1317 PR tree-optimization/60012
1318 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1319 TBAA disambiguation to all DDRs.
1321 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1324 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1325 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1327 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1330 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1332 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1334 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1335 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1337 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1340 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1341 to figure out targets of polymorphic calls with known decl.
1342 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1343 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1344 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1345 (get_polymorphic_call_info): ... here.
1346 (get_polymorphic_call_info_from_invariant): New function.
1348 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1350 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1351 lookup via vtable pointer; check for type consistency
1352 and turn inconsitent facts into UNREACHABLE.
1353 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1354 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1355 type inconsistent querries; return UNREACHABLE instead.
1357 2014-02-03 Richard Henderson <rth@twiddle.net>
1360 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1361 already processed this node.
1362 (normalize_one_pred_1): Pass along mark_set.
1363 (normalize_one_pred): Create and destroy a pointer_set_t.
1364 (normalize_one_pred_chain): Likewise.
1366 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1368 PR gcov-profile/58602
1369 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1371 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1374 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1375 -fno-devirtualize; try to devirtualize by the knowledge of
1376 virtual table pointer given by aggregate propagation.
1377 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1378 (ipa_print_node_jump_functions): Dump also offset that
1379 is relevant for polymorphic calls.
1380 (determine_known_aggregate_parts): Add arg_type parameter; use it
1381 instead of determining the type from pointer type.
1382 (ipa_compute_jump_functions_for_edge): Update call of
1383 determine_known_aggregate_parts.
1384 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1385 (gimple_get_virt_method_for_binfo): ... here; simplify using
1386 vtable_pointer_value_to_vtable.
1387 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1388 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1389 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1390 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1392 (vtable_pointer_value_to_binfo): ... here.
1393 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1395 2014-02-03 Teresa Johnson <tejohnson@google.com>
1397 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1398 redef of outer loop index variable.
1400 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1404 * doc/extend.texi (Function Attributes): Typo.
1406 2014-02-03 Cong Hou <congh@google.com>
1408 PR tree-optimization/60000
1409 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1410 if the vectorized statement is a store. A store statement can only
1411 appear at the end of pattern statements.
1413 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1415 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1416 (ix86_option_override_internal): Default long double to 64-bit for
1417 32-bit Bionic and to 128-bit for 64-bit Bionic.
1419 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1420 TARGET_LONG_DOUBLE_128 is true.
1421 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1423 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1424 (mlong-double-64): Negate -mlong-double-128.
1425 (mlong-double-128): New option.
1427 * config/i386/i386-c.c (ix86_target_macros): Define
1428 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1430 * doc/invoke.texi: Document -mlong-double-128.
1432 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1434 PR rtl-optimization/60024
1435 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1437 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1439 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1441 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1443 PR rtl-optimization/57662
1444 * sel-sched.c (code_motion_path_driver): Do not mark already not
1445 existing blocks in the visiting bitmap.
1447 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1449 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1450 on the insn being emitted.
1452 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1453 Will Deacon <will.deacon@arm.com>
1455 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1457 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1459 * config/arm/arm-tables.opt: Regenerate.
1461 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1463 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1464 for vector types other than V16QImode.
1465 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1466 define_expand, and call altivec_expand_vec_perm_le when producing
1467 code with little endian element order.
1468 (*altivec_vperm_<mode>_internal): New insn having previous
1469 behavior of altivec_vperm_<mode>.
1470 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1471 altivec_expand_vec_perm_le when producing code with little endian
1473 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1474 behavior of altivec_vperm_<mode>_uns.
1476 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1478 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1479 (altivec_vsumsws): Add handling for -maltivec=be with a little
1481 (altivec_vsumsws_direct): New.
1482 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1483 gen_altivec_vsumsws.
1485 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1487 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1488 vtable_pointer_value_to_binfo): New functions.
1489 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1490 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1492 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1494 * config/nios2/nios2.md (load_got_register): Initialize GOT
1495 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1496 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1498 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1500 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1501 preserverd by passthrough, do not propagate the type.
1503 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1505 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1506 (mips_atomic_assign_expand_fenv): New function.
1507 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1509 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1511 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1512 (__builtin_mips_set_fcsr): Likewise.
1513 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1514 MIPS_USI_FTYPE_VOID.
1515 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1516 (mips16_expand_set_fcsr): Likewise.
1517 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1518 (mips16_set_fcsr_stub): Likewise.
1519 (mips16_get_fcsr_one_only_stub): New class.
1520 (mips16_set_fcsr_one_only_stub): Likewise.
1521 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1522 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1523 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1524 (hard_float): New availability predicate.
1525 (mips_builtins): Add get_fcsr and set_fcsr.
1526 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1527 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1528 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1529 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1530 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1533 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1535 * config/mips/mips.c (mips_one_only_stub): New class.
1536 (mips_need_mips16_rdhwr_p): Replace with...
1537 (mips16_rdhwr_stub): ...this new variable.
1538 (mips16_stub_call_address): New function.
1539 (mips16_rdhwr_one_only_stub): New class.
1540 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1541 (mips_output_mips16_rdhwr): Delete.
1542 (mips_finish_stub): New function.
1543 (mips_code_end): Use it to handle rdhwr stubs.
1545 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1548 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1549 when calculating size of integer atomic types.
1551 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1553 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1555 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1557 PR tree-optimization/60003
1558 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1559 * profile.c (branch_prob): Use gimple_call_builtin_p
1560 to check for BUILT_IN_SETJMP_RECEIVER.
1561 * tree-inline.c (copy_bb): Call notice_special_calls.
1563 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1566 * lra-constraints.c (process_alt_operands): Update reload_sum only
1569 2014-01-31 Richard Henderson <rth@redhat.com>
1572 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1573 until after else_eh is processed.
1575 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1577 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1578 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1579 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1580 in smmintrin.h, remove them.
1581 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1582 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1583 * config/i386/i386.md (ROUND_SAE): Fix value.
1584 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1585 (const48_operand): New.
1586 * config/i386/subst.md (round), (round_expand): Use
1587 const_4_or_8_to_11_operand.
1588 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1590 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1592 * config/i386/constraints.md (Yk): Swap meaning with k.
1593 * config/i386/i386.md (movhi_internal): Change Yk to k.
1594 (movqi_internal): Ditto.
1595 (*k<logic><mode>): Ditto.
1598 (kandn<mode>): Ditto.
1599 (*<code>hi_1): Ditto.
1600 (*<code>qi_1): Ditto.
1601 (kxnor<mode>): Ditto.
1602 (kortestzhi): Ditto.
1603 (kortestchi): Ditto.
1605 (*one_cmplhi2_1): Ditto.
1606 (*one_cmplqi2_1): Ditto.
1607 * config/i386/sse.md (): Change k to Yk.
1608 (avx512f_load<mode>_mask): Ditto.
1609 (avx512f_blendm<mode>): Ditto.
1610 (avx512f_store<mode>_mask): Ditto.
1611 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1612 (avx512f_storedqu<mode>_mask): Ditto.
1613 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1615 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1616 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1617 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1618 (avx512f_maskcmp<mode>3): Ditto.
1619 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1620 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1621 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1622 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1623 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1624 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1625 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1626 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1627 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1628 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1629 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1630 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1631 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1632 (vec_extract_lo_<mode>_maskm): Ditto.
1633 (vec_extract_hi_<mode>_maskm): Ditto.
1634 (avx512f_vternlog<mode>_mask): Ditto.
1635 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1636 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1637 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1638 (avx512f_<code>v8div16qi2_mask): Ditto.
1639 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1640 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1641 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1642 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1643 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1644 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1645 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1646 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1647 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1648 (avx512cd_maskb_vec_dupv8di): Ditto.
1649 (avx512cd_maskw_vec_dupv16si): Ditto.
1650 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1651 (avx512f_vpermi2var<mode>3_mask): Ditto.
1652 (avx512f_vpermi2var<mode>3_mask): Ditto.
1653 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1654 (*avx512f_gathersi<mode>): Ditto.
1655 (*avx512f_gathersi<mode>_2): Ditto.
1656 (*avx512f_gatherdi<mode>): Ditto.
1657 (*avx512f_gatherdi<mode>_2): Ditto.
1658 (*avx512f_scattersi<mode>): Ditto.
1659 (*avx512f_scatterdi<mode>): Ditto.
1660 (avx512f_compress<mode>_mask): Ditto.
1661 (avx512f_compressstore<mode>_mask): Ditto.
1662 (avx512f_expand<mode>_mask): Ditto.
1663 * config/i386/subst.md (mask): Change k to Yk.
1664 (mask_scalar_merge): Ditto.
1667 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1669 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1671 2014-01-31 Richard Biener <rguenther@suse.de>
1674 * builtins.c (fold_builtin_memory_op): Make sure to not
1675 use a floating-point mode or a boolean or enumeral type for
1678 2014-01-30 DJ Delorie <dj@redhat.com>
1680 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1681 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1682 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1683 whenever main() has an epilogue.
1685 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1687 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1688 unused variable "field".
1689 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1690 (vsx_mergeh_<mode>): Likewise.
1691 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1692 (altivec_vmrghh): Likewise.
1693 (altivec_vmrghw): Likewise.
1694 (altivec_vmrglb): Likewise.
1695 (altivec_vmrglh): Likewise.
1696 (altivec_vmrglw): Likewise.
1697 (altivec_vspltb): Add missing uses.
1698 (altivec_vsplth): Likewise.
1699 (altivec_vspltw): Likewise.
1700 (altivec_vspltsf): Likewise.
1702 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1705 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1706 frame related instructions.
1708 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1710 PR rtl-optimization/59959
1711 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1712 any reload of register whose subreg is invalid.
1714 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1716 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1717 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1718 Add missing return type - void.
1720 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1722 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1723 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1724 remove element index adjustment for endian (now handled in vsx.md
1726 (altivec_expand_vec_perm_const): Use
1727 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1728 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1729 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1730 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1731 define_expand and a new define_insn *altivec_vspltb_internal;
1732 adjust for -maltivec=be on a little endian target.
1733 (altivec_vspltb_direct): New.
1734 (altivec_vsplth): Divide into a define_expand and a new
1735 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1736 little endian target.
1737 (altivec_vsplth_direct): New.
1738 (altivec_vspltw): Divide into a define_expand and a new
1739 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1740 little endian target.
1741 (altivec_vspltw_direct): New.
1742 (altivec_vspltsf): Divide into a define_expand and a new
1743 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1744 a little endian target.
1746 2014-01-30 Richard Biener <rguenther@suse.de>
1748 PR tree-optimization/59993
1749 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1750 can propagate form the earlier stmt and avoid the transform
1751 when the intermediate result is needed.
1753 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1755 * README.Portability: Fix typo.
1757 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1759 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1760 comparison_operator with ordered_comparison_operator.
1762 2014-01-30 Nick Clifton <nickc@redhat.com>
1764 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1765 Rename to mn10300_store_multiple_regs.
1766 * config/mn10300/mn10300.c: Likewise.
1767 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1768 store_multiple_regs.
1769 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1770 Call mn10300_store_multiple_regs.
1772 2014-01-30 Nick Clifton <nickc@redhat.com>
1773 DJ Delorie <dj@redhat.com>
1775 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1776 %fp 2 to keep registers after it properly word-aligned.
1777 (rl78_alloc_physical_registers_umul): Handle the case where both
1778 input operands are the same.
1780 2014-01-30 Richard Biener <rguenther@suse.de>
1782 PR tree-optimization/59903
1783 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1786 2014-01-30 Jason Merrill <jason@redhat.com>
1789 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1792 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1794 2014-01-30 Richard Biener <rguenther@suse.de>
1796 PR tree-optimization/59951
1797 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1799 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1802 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1803 SFmode to DFmode case.
1805 2014-01-29 DJ Delorie <dj@redhat.com>
1807 * config/msp430/msp430.opt (-minrt): New.
1808 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1810 (ENDFILE_SPEC): Likewise.
1812 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1814 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1815 (estimate_function_body_sizes): Use it.
1817 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1820 * dwarf2out.c (is_cxx_auto): New.
1821 (is_base_type): Use it.
1822 (gen_type_die_with_usage): Likewise.
1824 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1826 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1827 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1828 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1829 -maltivec=be with LE targets.
1830 (vsx_mergeh_<mode>): Likewise.
1831 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1832 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1833 (altivec_vmrghb): Replace with define_expand and new
1834 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1835 (altivec_vmrghb_direct): New define_insn.
1836 (altivec_vmrghh): Replace with define_expand and new
1837 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1838 (altivec_vmrghh_direct): New define_insn.
1839 (altivec_vmrghw): Replace with define_expand and new
1840 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1841 (altivec_vmrghw_direct): New define_insn.
1842 (*altivec_vmrghsf): Adjust for endianness.
1843 (altivec_vmrglb): Replace with define_expand and new
1844 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1845 (altivec_vmrglb_direct): New define_insn.
1846 (altivec_vmrglh): Replace with define_expand and new
1847 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1848 (altivec_vmrglh_direct): New define_insn.
1849 (altivec_vmrglw): Replace with define_expand and new
1850 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1851 (altivec_vmrglw_direct): New define_insn.
1852 (*altivec_vmrglsf): Adjust for endianness.
1853 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1854 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1855 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1856 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1857 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1858 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1859 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1860 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1862 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1864 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1865 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1868 2014-01-29 Richard Biener <rguenther@suse.de>
1870 PR tree-optimization/58742
1871 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1872 associate_pointerplus_align.
1873 (associate_pointerplus_diff): New function.
1874 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1875 and associate_pointerplus_diff.
1877 2014-01-29 Richard Biener <rguenther@suse.de>
1879 * lto-streamer.h (LTO_major_version): Bump to 3.
1880 (LTO_minor_version): Reset to 0.
1882 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1884 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1885 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1886 (arm_file_start): Generate correct asm header for armv7ve.
1887 * config/arm/bpabi.h: Add multilib support for armv7ve.
1888 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1889 and cortex-a15 to armv7ve.
1890 * config/arm/t-aprofile: Add multilib support for armv7ve.
1891 * doc/invoke.texi: Document -march=armv7ve.
1893 2014-01-29 Richard Biener <rguenther@suse.de>
1895 PR tree-optimization/58742
1896 * tree-ssa-forwprop.c (associate_plusminus): Return true
1897 if we changed sth, defer EH cleanup to ...
1898 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1899 (simplify_mult): New function.
1901 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1904 PR tree-optimization/59920
1905 * tree.c (build_common_builtin_nodes): Remove
1906 __builtin_setjmp_dispatcher initialization.
1907 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1908 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1909 instead of gsi_after_labels + manually skipping debug stmts.
1910 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1911 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1912 * tree-inline.c (copy_edges_for_bb): Remove
1913 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1914 argument. Ignore computed_goto_p stmts. Don't call
1915 make_abnormal_goto_edges. If a call might need abnormal edges
1916 for non-local gotos, see if it already has an edge to
1917 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1918 with true argument, don't do anything then, otherwise add
1919 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1920 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1922 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1923 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1924 (lower_stmt): Don't set data->calls_builtin_setjmp.
1925 (lower_builtin_setjmp): Adjust comment.
1926 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1927 * tree-cfg.c (found_computed_goto): Remove.
1928 (factor_computed_gotos): Remove.
1929 (make_goto_expr_edges): Return bool, true for computed gotos.
1930 Don't call make_abnormal_goto_edges.
1931 (build_gimple_cfg): Don't set found_computed_goto, don't call
1932 factor_computed_gotos.
1933 (computed_goto_p): No longer static.
1934 (make_blocks): Don't set found_computed_goto.
1935 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1936 (make_edges): If make_goto_expr_edges returns true, push bb
1937 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1938 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1939 vector. Record mapping between bbs and OpenMP regions if there
1940 are any, adjust make_gimple_omp_edges caller. Call
1941 handle_abnormal_edges.
1942 (make_abnormal_goto_edges): Remove.
1943 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1944 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1945 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1946 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1947 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1948 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1949 filling *region also set *region_idx to (*region)->entry->index.
1952 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1953 For REGs set ORIGINAL_REGNO.
1955 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1957 * doc/md.texi: Mention that a target shouldn't implement
1958 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1961 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1963 PR tree-optimization/59594
1964 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1965 a copy of the datarefs vector rather than the vector itself.
1967 2014-01-28 Jason Merrill <jason@redhat.com>
1970 * dwarf2out.c (auto_die): New static.
1971 (gen_type_die_with_usage): Handle C++1y 'auto'.
1972 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1975 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1978 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1979 (SPEC_X32): Likewise.
1980 (SPEC_64): Likewise.
1981 * config/i386/i386.c (ix86_option_override_internal): Turn off
1982 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1984 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1985 * config/i386/i386.h (TARGET_16BIT): New macro.
1986 (TARGET_16BIT_P): Likewise.
1987 * config/i386/i386.opt: Add m16.
1988 * doc/invoke.texi: Document -m16.
1990 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1992 PR preprocessor/59935
1993 * input.c (location_get_source_line): Bail out on when line number
1994 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1996 2014-01-28 Richard Biener <rguenther@suse.de>
1998 PR tree-optimization/58742
1999 * tree-ssa-forwprop.c (associate_plusminus): Handle
2000 pointer subtraction of the form (T)(P + A) - (T)P.
2002 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2004 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2007 2014-01-28 Richard Biener <rguenther@suse.de>
2010 2014-01-28 Richard Biener <rguenther@suse.de>
2012 PR rtl-optimization/45364
2013 PR rtl-optimization/59890
2014 * var-tracking.c (local_get_addr_clear_given_value): Handle
2015 already cleared slot.
2016 (val_reset): Handle not allocated local_get_addr_cache.
2017 (vt_find_locations): Use post-order on the inverted CFG.
2019 2014-01-28 Richard Biener <rguenther@suse.de>
2021 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2023 2014-01-28 Richard Biener <rguenther@suse.de>
2025 PR rtl-optimization/45364
2026 PR rtl-optimization/59890
2027 * var-tracking.c (local_get_addr_clear_given_value): Handle
2028 already cleared slot.
2029 (val_reset): Handle not allocated local_get_addr_cache.
2030 (vt_find_locations): Use post-order on the inverted CFG.
2032 2014-01-28 Alan Modra <amodra@gmail.com>
2034 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2035 * configure.ac <recursive call for build != host>: Define
2036 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2037 and LD_FOR_BUILD too.
2038 * configure: Regenerate.
2040 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2042 * config/i386/i386.c (get_builtin_code_for_version): Separate
2043 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2044 Broadwell from Haswell.
2046 2014-01-27 Steve Ellcey <sellcey@mips.com>
2048 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2049 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2050 * config/mips/mips.c (mips_option_override): Change setting
2052 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2053 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2054 Change from Mask to Var.
2056 2014-01-27 Jeff Law <law@redhat.com>
2058 * ipa-inline.c (inline_small_functions): Fix typo.
2060 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2062 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2063 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2064 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2065 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2066 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2067 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2068 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2069 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2070 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2071 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2072 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2073 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2074 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2075 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2076 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2077 (_mm512_storeu_epi64): Ditto.
2078 (_mm512_cmpge_epi32_mask): Ditto.
2079 (_mm512_cmpge_epu32_mask): Ditto.
2080 (_mm512_cmpge_epi64_mask): Ditto.
2081 (_mm512_cmpge_epu64_mask): Ditto.
2082 (_mm512_cmple_epi32_mask): Ditto.
2083 (_mm512_cmple_epu32_mask): Ditto.
2084 (_mm512_cmple_epi64_mask): Ditto.
2085 (_mm512_cmple_epu64_mask): Ditto.
2086 (_mm512_cmplt_epi32_mask): Ditto.
2087 (_mm512_cmplt_epu32_mask): Ditto.
2088 (_mm512_cmplt_epi64_mask): Ditto.
2089 (_mm512_cmplt_epu64_mask): Ditto.
2090 (_mm512_cmpneq_epi32_mask): Ditto.
2091 (_mm512_cmpneq_epu32_mask): Ditto.
2092 (_mm512_cmpneq_epi64_mask): Ditto.
2093 (_mm512_cmpneq_epu64_mask): Ditto.
2094 (_mm512_expand_pd): Ditto.
2095 (_mm512_expand_ps): Ditto.
2096 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2097 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2098 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2099 * config/i386/i386.c (ix86_builtins): Add
2100 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2101 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2102 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2103 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2104 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2105 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2106 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2107 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2108 IX86_BUILTIN_PMOVUSQW512_MEM.
2109 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2110 __builtin_ia32_pmovsqd512mem_mask,
2111 __builtin_ia32_pmovqd512mem_mask,
2112 __builtin_ia32_pmovusqw512mem_mask,
2113 __builtin_ia32_pmovsqw512mem_mask,
2114 __builtin_ia32_pmovqw512mem_mask,
2115 __builtin_ia32_pmovusdw512mem_mask,
2116 __builtin_ia32_pmovsdw512mem_mask,
2117 __builtin_ia32_pmovdw512mem_mask,
2118 __builtin_ia32_pmovqb512mem_mask,
2119 __builtin_ia32_pmovusqb512mem_mask,
2120 __builtin_ia32_pmovsqb512mem_mask,
2121 __builtin_ia32_pmovusdb512mem_mask,
2122 __builtin_ia32_pmovsdb512mem_mask,
2123 __builtin_ia32_pmovdb512mem_mask.
2124 (bdesc_args): Add __builtin_ia32_expanddf512,
2125 __builtin_ia32_expandsf512.
2126 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2127 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2128 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2129 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2130 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2131 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2132 (avx512f_<code>v8div16qi2_mask_store): This.
2133 (avx512f_expand<mode>): New.
2135 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2137 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2139 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2140 (_mm512_prefetch_i32scatter_pd): Ditto.
2141 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2142 (_mm512_prefetch_i64scatter_pd): Ditto.
2143 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2144 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2145 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2146 (_mm512_prefetch_i32scatter_ps): Ditto.
2147 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2148 (_mm512_prefetch_i64scatter_ps): Ditto.
2149 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2150 * config/i386/i386-builtin-types.def: Define
2151 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2152 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2153 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2154 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2155 IX86_BUILTIN_SCATTERPFQPD.
2156 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2157 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2158 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2159 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2160 __builtin_ia32_scatterpfqps.
2161 (ix86_expand_builtin): Expand new built-ins.
2162 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2163 fix memory access data type.
2164 (*avx512pf_gatherpf<mode>_mask): Ditto.
2165 (*avx512pf_gatherpf<mode>): Ditto.
2166 (avx512pf_scatterpf<mode>): Ditto.
2167 (*avx512pf_scatterpf<mode>_mask): Ditto.
2168 (*avx512pf_scatterpf<mode>): Ditto.
2169 (GATHER_SCATTER_SF_MEM_MODE): New.
2170 (avx512pf_gatherpf<mode>df): Ditto.
2171 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2172 (*avx512pf_scatterpf<mode>df): Ditto.
2174 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2177 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2178 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2181 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2183 * common/config/arm/arm-common.c
2184 (arm_rewrite_mcpu): Handle multiple names.
2186 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2188 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2190 * gimple-builder.h (create_gimple_tmp): Delete.
2192 2014-01-27 Christian Bruel <christian.bruel@st.com>
2194 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2197 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2199 * config/pa/pa.md (call): Generate indirect long calls to non-local
2200 functions when outputing 32-bit code.
2201 (call_value): Likewise except for special call to buggy powf function.
2203 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2204 portable runtime and PIC indirect calls.
2205 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2206 and PIC call sequences. Use ldo instead of blr to set return register
2207 in PIC call sequence.
2209 2014-01-25 Walter Lee <walt@tilera.com>
2211 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2212 avoid clobbering a live register.
2214 2014-01-25 Walter Lee <walt@tilera.com>
2216 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2217 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2218 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2219 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2221 2014-01-25 Walter Lee <walt@tilera.com>
2223 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2224 arguments on even registers.
2225 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2227 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2228 (BIGGEST_ALIGNMENT): Ditto.
2229 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2231 2014-01-25 Walter Lee <walt@tilera.com>
2233 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2234 insns before bundling.
2235 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2237 2014-01-25 Walter Lee <walt@tilera.com>
2239 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2240 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2241 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2243 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2245 * config/mips/constraints.md (kl): Delete.
2246 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2247 define expands, using...
2248 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2249 instructions for MIPS16.
2250 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2251 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2253 2014-01-25 Walter Lee <walt@tilera.com>
2255 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2259 2014-01-25 Walter Lee <walt@tilera.com>
2261 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2262 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2264 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2266 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2269 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2271 * print-rtl.c (in_call_function_usage): New var.
2272 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2273 EXPR_LIST mode as mode and not as reg note name.
2276 * cfgloopmanip.c (copy_loop_info): If
2277 loop->warned_aggressive_loop_optimizations, make sure
2278 the flag is set in target loop too.
2280 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2282 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2284 * builtins.def: Likewise.
2285 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2286 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2287 * ira.c (ira_setup_eliminable_regset): Likewise.
2288 * omp-low.c (gate_expand_omp): Likewise.
2289 (execute_lower_omp): Likewise.
2290 (diagnose_sb_0): Likewise.
2291 (gate_diagnose_omp_blocks): Likewise.
2292 (simd_clone_clauses_extract): Likewise.
2295 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2297 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2298 correction for little endian...
2299 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2302 2014-01-24 Jeff Law <law@redhat.com>
2304 PR tree-optimization/59919
2305 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2306 for non-returning calls.
2308 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2310 * common/config/aarch64/aarch64-common.c
2311 (aarch64_rewrite_mcpu): Handle multiple names.
2312 * config/aarch64/aarch64.h
2313 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2315 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2317 * input.c (add_file_to_cache_tab): Handle the case where fopen
2320 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2323 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2324 from push operand if code of push isn't PRE_DEC.
2326 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2329 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2330 -mquad-memory-atomic. Update -mquad-memory documentation to say
2331 it is only used for non-atomic loads/stores.
2333 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2334 -mquad-memory or -mquad-memory-atomic switches.
2336 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2337 -mquad-memory-atomic to ISA 2.07 support.
2339 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2340 to separate support of normal quad word memory operations (ldq, stq)
2341 from the atomic quad word memory operations.
2343 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2344 support to separate non-atomic quad word operations from atomic
2345 quad word operations. Disable non-atomic quad word operations in
2346 little endian mode so that we don't have to swap words after the
2347 load and before the store.
2348 (quad_load_store_p): Add comment about atomic quad word support.
2349 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2350 options printed with -mdebug=reg.
2352 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2353 -mquad-memory-atomic as the test for whether we have quad word
2354 atomic instructions.
2355 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2356 or -mp8-vector are used, allow byte/half-word atomic operations.
2358 * config/rs6000/sync.md (load_lockedti): Insure that the address
2359 is a proper indexed or indirect address for the lqarx instruction.
2360 On little endian systems, swap the hi/lo registers after the lqarx
2362 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2363 insure the address is valid for the lqarx instruction.
2364 (store_conditionalti): Insure that the address is a proper indexed
2365 or indirect address for the stqcrx. instruction. On little endian
2366 systems, swap the hi/lo registers before doing the stqcrx.
2368 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2369 insure the address is valid for the stqcrx. instruction.
2371 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2372 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2373 type of quad memory support is available.
2375 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2378 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2379 there is a danger of looping.
2381 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2383 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2384 force flag_ira_loop_pressure if set via command line.
2386 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2388 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2389 (ashr_simd): New builtin handling DI mode.
2390 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2391 (aarch64_sshr_simddi): New match pattern.
2392 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2393 (vshrd_n_s64): Likewise.
2394 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2396 2014-01-23 Nick Clifton <nickc@redhat.com>
2398 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2399 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2400 favour of mcu specific scripts.
2401 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2404 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2405 Alex Velenko <Alex.Velenko@arm.com>
2407 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2408 (vaddv_s16): Likewise.
2409 (vaddv_s32): Likewise.
2410 (vaddv_u8): Likewise.
2411 (vaddv_u16): Likewise.
2412 (vaddv_u32): Likewise.
2413 (vaddvq_s8): Likewise.
2414 (vaddvq_s16): Likewise.
2415 (vaddvq_s32): Likewise.
2416 (vaddvq_s64): Likewise.
2417 (vaddvq_u8): Likewise.
2418 (vaddvq_u16): Likewise.
2419 (vaddvq_u32): Likewise.
2420 (vaddvq_u64): Likewise.
2421 (vaddv_f32): Likewise.
2422 (vaddvq_f32): Likewise.
2423 (vaddvq_f64): Likewise.
2424 (vmaxv_f32): Likewise.
2425 (vmaxv_s8): Likewise.
2426 (vmaxv_s16): Likewise.
2427 (vmaxv_s32): Likewise.
2428 (vmaxv_u8): Likewise.
2429 (vmaxv_u16): Likewise.
2430 (vmaxv_u32): Likewise.
2431 (vmaxvq_f32): Likewise.
2432 (vmaxvq_f64): Likewise.
2433 (vmaxvq_s8): Likewise.
2434 (vmaxvq_s16): Likewise.
2435 (vmaxvq_s32): Likewise.
2436 (vmaxvq_u8): Likewise.
2437 (vmaxvq_u16): Likewise.
2438 (vmaxvq_u32): Likewise.
2439 (vmaxnmv_f32): Likewise.
2440 (vmaxnmvq_f32): Likewise.
2441 (vmaxnmvq_f64): Likewise.
2442 (vminv_f32): Likewise.
2443 (vminv_s8): Likewise.
2444 (vminv_s16): Likewise.
2445 (vminv_s32): Likewise.
2446 (vminv_u8): Likewise.
2447 (vminv_u16): Likewise.
2448 (vminv_u32): Likewise.
2449 (vminvq_f32): Likewise.
2450 (vminvq_f64): Likewise.
2451 (vminvq_s8): Likewise.
2452 (vminvq_s16): Likewise.
2453 (vminvq_s32): Likewise.
2454 (vminvq_u8): Likewise.
2455 (vminvq_u16): Likewise.
2456 (vminvq_u32): Likewise.
2457 (vminnmv_f32): Likewise.
2458 (vminnmvq_f32): Likewise.
2459 (vminnmvq_f64): Likewise.
2461 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2463 * config/aarch64/aarch64-simd.md
2464 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2465 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2466 (*aarch64_mul3_elt<mode>): Likewise.
2467 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2468 (*aarch64_mul3_elt_to_64v2df): Likewise.
2469 (*aarch64_mla_elt<mode>): Likewise.
2470 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2471 (*aarch64_mls_elt<mode>): Likewise.
2472 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2473 (*aarch64_fma4_elt<mode>): Likewise.
2474 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2475 (*aarch64_fma4_elt_to_64v2df): Likewise.
2476 (*aarch64_fnma4_elt<mode>): Likewise.
2477 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2478 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2479 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2480 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2481 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2482 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2483 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2484 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2485 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2487 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2489 * config/aarch64/aarch64-simd.md
2490 (aarch64_be_checked_get_lane<mode>): New define_expand.
2491 * config/aarch64/aarch64-simd-builtins.def
2492 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2493 New builtin definition.
2494 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2495 Use new safe be builtin.
2497 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2499 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2501 (aarch64_be_st1<mode>): Likewise.
2502 (aarch_ld1<VALL:mode>): Define_expand modified.
2503 (aarch_st1<VALL:mode>): Likewise.
2504 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2505 (UNSPEC_ST1): Likewise.
2507 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2509 * config/microblaze/microblaze.md: Add trap insn and attribute
2511 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2513 PR preprocessor/58580
2514 * input.h (location_get_source_line): Take an additional line_size
2516 (void diagnostics_file_cache_fini): Declare new function.
2517 * input.c (struct fcache): New type.
2518 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2519 New static constants.
2520 (diagnostic_file_cache_init, total_lines_num)
2521 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2522 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2523 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2524 (get_next_line, read_next_line, goto_next_line, read_line_num):
2525 New static function definitions.
2526 (diagnostic_file_cache_fini): New function.
2527 (location_get_source_line): Take an additional output line_len
2528 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2530 * diagnostic.c (diagnostic_finish): Call
2531 diagnostic_file_cache_fini.
2532 (adjust_line): Take an additional input parameter for the length
2533 of the line, rather than calculating it with strlen.
2534 (diagnostic_show_locus): Adjust the use of
2535 location_get_source_line and adjust_line with respect to their new
2536 signature. While displaying a line now, do not stop at the first
2537 null byte. Rather, display the zero byte as a space and keep
2538 going until we reach the size of the line.
2539 * Makefile.in: Add vec.o to OBJS-libcommon
2541 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2542 Ilya Tocar <ilya.tocar@intel.com>
2544 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2545 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2546 (__builtin_ia32_kmov16): Ditto.
2547 * config/i386/i386.md (UNSPEC_KMOV): New.
2550 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2552 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2553 (_mm512_storeu_si512): Ditto.
2555 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2558 * rtl.h (get_referenced_operands): Declare.
2559 * recog.c (get_referenced_operands): New function.
2560 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2561 operands have been referenced when recording LO_SUM references.
2563 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2565 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2567 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2569 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2570 Enable for generic and recent AMD targets.
2572 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2574 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2575 ARG_SIZE note when adjustment was eliminated.
2577 2014-01-22 Jeff Law <law@redhat.com>
2579 PR tree-optimization/59597
2580 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2581 in file. Accept new argument REGISTERING and use it to modify
2582 dump output appropriately.
2583 (register_jump_thread): Corresponding changes.
2584 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2585 thread paths involving joiner blocks. Add code to dump cancelled
2586 jump threading paths.
2588 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2590 PR rtl-optimization/59477
2591 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2592 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2594 2014-01-22 Tom Tromey <tromey@redhat.com>
2596 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2598 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2600 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2602 PR rtl-optimization/59896
2603 * lra-constraints.c (process_alt_operands): Check unused note for
2604 matched operands of insn with no output reloads.
2606 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2608 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2609 (mips_move_from_gpr_cost): Likewise.
2611 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2613 PR rtl-optimization/59858
2614 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2615 ira_class_hard_regs_num.
2616 (process_alt_operands): Increase reject for dying matched operand.
2618 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2621 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2622 smaller than size, perform several stores or loads and stores
2623 at dst + count - size to store or copy all of size bytes, rather
2624 than just last modesize bytes.
2626 2014-01-20 DJ Delorie <dj@redhat.com>
2628 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2629 that CLOBBERs are REGs before propogating their values.
2631 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2634 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2635 (cgraph_inline_failed_type): New function.
2636 * cgraph.h (DEFCIFCODE): Add type.
2637 (cgraph_inline_failed_type_t): New enum.
2638 (cgraph_inline_failed_type): New prototype.
2639 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2640 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2641 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2642 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2643 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2644 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2645 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2646 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2647 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2648 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2649 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2650 OPTIMIZATION_MISMATCH.
2651 * tree-inline.c (expand_call_inline): Emit errors during
2652 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2654 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2657 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2658 mode attribute in insn output.
2660 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2662 * output.h (output_constant): Delete.
2663 * varasm.c (output_constant): Make private.
2665 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2667 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2669 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2672 * tree.h (fold_builtin_strcat): New prototype.
2673 * builtins.c (fold_builtin_strcat): No longer static. Add len
2674 argument, if non-NULL, don't call c_strlen. Optimize
2675 directly into __builtin_memcpy instead of __builtin_strcpy.
2676 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2677 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2679 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2681 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2682 for SImode_address_operand operands, having only a REG argument.
2684 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2686 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2687 loader name using mbig-endian.
2688 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2690 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2692 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2696 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2698 * config/aarch64/aarch64-protos.h
2699 (aarch64_cannot_change_mode_class_ptr): Declare.
2700 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2701 aarch64_cannot_change_mode_class_ptr): New.
2702 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2703 backend hook aarch64_cannot_change_mode_class.
2705 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2707 * common/config/aarch64/aarch64-common.c
2708 (aarch64_handle_option): Don't handle any option order logic here.
2709 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2710 selected_cpu, warn on architecture version mismatch.
2711 (aarch64_override_options): Fix parsing order for option strings.
2713 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2714 Iain Sandoe <iain@codesourcery.com>
2717 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2718 warning. Amend comment to reflect current functionality.
2720 2014-01-20 Richard Biener <rguenther@suse.de>
2723 * builtins.c (fold_builtin_strcat): Remove case better handled
2724 by tree-ssa-strlen.c.
2726 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2728 * config/aarch64/aarch64.opt
2729 (mcpu, march, mtune): Make case-insensitive.
2731 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2734 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2735 if operands[1] is a REG or ZERO_EXTEND of a REG.
2737 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2739 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2741 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2743 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2744 long non-pic millicode calls.
2746 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2748 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2750 2014-01-19 Kito Cheng <kito@0xlab.org>
2752 * builtins.c (expand_movstr): Check movstr expand done or fail.
2754 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2755 H.J. Lu <hongjiu.lu@intel.com>
2758 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2759 to DImode for zero-extended addresses.
2761 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2763 PR rtl-optimization/57763
2764 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2765 on the new indirect jump_insn and increment LABEL_NUSES (label).
2767 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2771 * config.gcc (x86_archs): New variable.
2772 (x86_64_archs): Likewise.
2773 (x86_cpus): Likewise.
2774 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2775 --with-arch/--with-cpu= options.
2776 Support --with-arch=/--with-cpu={nehalem,westmere,
2777 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2779 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2781 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2782 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2784 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2786 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2788 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2791 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2792 clear cpp_get_options (parse_in)->warn_unused_macros for
2793 ix86_target_macros_internal with cpp_define.
2795 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2797 * jump.c (delete_related_insns): Keep (use (insn))s.
2798 * reorg.c (redundant_insn): Check for barriers too.
2800 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2802 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2804 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2806 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2807 call to $$dyncall when TARGET_LONG_CALLS is true.
2809 2014-01-17 Jeff Law <law@redhat.com>
2811 * ree.c (combine_set_extension): Temporarily disable test for
2812 changing number of hard registers.
2814 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2817 * ipa-inline-analysis.c (inline_free_summary):
2818 Do not free summary of aliases.
2820 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2823 * gimplify.c (gimplify_expr): Use create_tmp_var
2824 instead of create_tmp_var_raw. If cond doesn't have
2825 integral type, don't add the IFN_ANNOTATE builtin at all.
2827 2014-01-17 Martin Jambor <mjambor@suse.cz>
2830 * ipa-cp.c (prev_edge_clone): New variable.
2831 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2832 Also resize prev_edge_clone vector.
2833 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2834 (ipcp_edge_removal_hook): New function.
2835 (ipcp_driver): Register ipcp_edge_removal_hook.
2837 2014-01-17 Andrew Pinski <apinski@cavium.com>
2838 Steve Ellcey <sellcey@mips.com>
2841 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2844 2014-01-17 Jeff Law <law@redhat.com>
2847 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2848 so that pass_ccp runs first.
2850 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2852 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2853 (ix86_adjust_cost): Use !TARGET_XXX.
2854 (do_reorder_for_imul): Likewise.
2855 (swap_top_of_ready_list): Likewise.
2856 (ix86_sched_reorder): Likewise.
2858 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2860 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2861 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2862 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2863 (intel_memset): New. Duplicate slm_memset.
2864 (intel_cost): New. Duplicate slm_cost.
2865 (m_INTEL): New macro.
2866 (processor_target_table): Add "intel".
2867 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2868 with PROCESSOR_INTEL for "intel".
2869 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2870 PROCESSOR_SILVERMONT.
2871 (ix86_issue_rate): Likewise.
2872 (ix86_adjust_cost): Likewise.
2873 (ia32_multipass_dfa_lookahead): Likewise.
2874 (swap_top_of_ready_list): Likewise.
2875 (ix86_sched_reorder): Likewise.
2876 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2877 instead of TARGET_OPT_AGU.
2878 * config/i386/i386.h (TARGET_INTEL): New.
2879 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2880 (processor_type): Add PROCESSOR_INTEL.
2881 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2882 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2884 2014-01-17 Marek Polacek <polacek@redhat.com>
2887 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2890 2014-01-17 Richard Biener <rguenther@suse.de>
2892 PR tree-optimization/46590
2893 * opts.c (default_options_table): Add entries for
2894 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2895 all enabled at -O1 but not for -Og.
2896 * common.opt (fbranch-count-reg): Remove Init(1).
2897 (fmove-loop-invariants): Likewise.
2898 (ftree-pta): Likewise.
2900 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2902 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2903 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2904 decls to at least the GCC 4.8 used alignments.
2907 * tree-nested.c (convert_nonlocal_reference_stmt,
2908 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2909 of GIMPLE_BIND stmts, adjust associated decls.
2911 2014-01-17 Richard Biener <rguenther@suse.de>
2913 PR tree-optimization/46590
2914 * vec.h (vec<>::bseach): New member function implementing
2915 binary search according to C89 bsearch.
2916 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2917 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2918 bitmap pointer again. Make accesses_in_loop a flat array.
2919 (mem_ref_obstack): New global.
2920 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2921 (mark_ref_stored): Likewise.
2922 (ref_indep_loop_p_2): Likewise.
2923 (set_ref_stored_in_loop): New helper function.
2924 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2925 (memref_free): Adjust.
2926 (record_mem_ref_loc): Simplify.
2927 (gather_mem_refs_stmt): Adjust.
2928 (sort_locs_in_loop_postorder_cmp): New function.
2929 (analyze_memory_references): Sort accesses_in_loop after
2930 loop postorder number.
2931 (find_ref_loc_in_loop_cmp): New function.
2932 (for_all_locs_in_loop): Find relevant cluster of locs in
2933 accesses_in_loop and iterate without recursion.
2934 (execute_sm): Avoid uninit warning.
2935 (struct ref_always_accessed): Simplify.
2936 (ref_always_accessed::operator ()): Likewise.
2937 (ref_always_accessed_p): Likewise.
2938 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2939 loop postorder numbers here.
2940 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2943 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2946 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2947 on decls for which assemble_alias has been called.
2949 2014-01-17 Nick Clifton <nickc@redhat.com>
2951 * config/msp430/msp430.opt: (mcpu): New option.
2952 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2953 (msp430_option_override): Parse target_cpu. If the MCU name
2954 matches a generic string, clear target_mcu.
2955 (msp430_attr): Allow numeric interrupt values up to 63.
2956 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2957 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2959 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2961 * config/msp430/msp430.md (popm): Use %J rather than %I.
2962 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2963 (addhi_cy_i): Use immediate_operand for operand 2.
2964 * doc/invoke.texi: Document -mcpu option.
2966 2014-01-17 Richard Biener <rguenther@suse.de>
2968 PR rtl-optimization/38518
2969 * df.h (df_analyze_loop): Declare.
2970 * df-core.c: Include cfgloop.h.
2971 (df_analyze_1): Split out main part of df_analyze.
2972 (df_analyze): Adjust.
2973 (loop_inverted_post_order_compute): New function.
2974 (loop_post_order_compute): Likewise.
2975 (df_analyze_loop): New function avoiding whole-function
2977 * loop-invariant.c (find_defs): Use df_analyze_loop.
2978 (find_invariants): Adjust.
2979 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2981 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2983 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2984 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2986 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2988 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2989 traversal when removing references.
2991 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2994 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2996 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2999 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3000 pushing a reload for an autoinc when we had previously reloaded an
3001 inner part of the address.
3003 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3005 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3007 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3008 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3009 when not giving up or versioning for alias only because of
3011 (vect_analyze_data_ref_dependences): Set to true.
3012 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3014 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3015 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3019 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3022 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3023 operand 0 predicate for gathers, use a new pseudo as subtarget.
3025 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3028 * lra-constraints.c (process_alt_operands): Add printing debug info.
3029 Check absence of input/output reloads for matched operands too.
3031 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3033 PR rtl-optimization/59835
3034 * ira.c (ira_init_register_move_cost): Increase cost for
3037 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3039 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3041 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3044 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3045 non-register objects. Use gen_(high/low)part more consistently.
3048 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3051 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3052 endian support, remove tests for WORDS_BIG_ENDIAN.
3053 (p8_mfvsrd_3_<mode>): Likewise.
3054 (reload_gpr_from_vsx<mode>): Likewise.
3055 (reload_gpr_from_vsxsf): Likewise.
3056 (p8_mfvsrd_4_disf): Likewise.
3058 2014-01-16 Richard Biener <rguenther@suse.de>
3060 PR rtl-optimization/46590
3061 * lcm.c (compute_antinout_edge): Use postorder iteration.
3062 (compute_laterin): Use inverted postorder iteration.
3064 2014-01-16 Nick Clifton <nickc@redhat.com>
3067 * varasm.c (output_constant): Return the number of bytes actually
3069 (output_constructor_array_range): Update the field size with the
3070 number of bytes emitted by output_constant.
3071 (output_constructor_regular_field): Likewise. Also do not
3072 complain if the total number of bytes emitted is now greater
3073 than the expected fieldpos.
3074 * output.h (output_constant): Update prototype and descriptive comment.
3076 2014-01-16 Marek Polacek <polacek@redhat.com>
3079 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3080 it is error_mark_node.
3082 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3084 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3085 VALID_AVX256_REG_OR_OI_MODE.
3087 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3089 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3090 current procedure should be profiled.
3092 2014-01-15 Andrew Pinski <apinski@cavium.com>
3094 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3095 of moving from/to the STACK_REG register class.
3097 2014-01-15 Richard Henderson <rth@redhat.com>
3100 * reginfo.c (global_regs_decl): Globalize.
3101 * rtl.h (global_regs_decl): Declare.
3102 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3103 reserved via global_regs.
3105 2014-01-15 Teresa Johnson <tejohnson@google.com>
3107 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3109 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3111 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3112 and vmulosh rather than call gen_vec_widen_smult_*.
3113 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3114 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3115 (vec_widen_smult_even_v16qi): Likewise.
3116 (vec_widen_umult_even_v8hi): Likewise.
3117 (vec_widen_smult_even_v8hi): Likewise.
3118 (vec_widen_umult_odd_v16qi): Likewise.
3119 (vec_widen_smult_odd_v16qi): Likewise.
3120 (vec_widen_umult_odd_v8hi): Likewise.
3121 (vec_widen_smult_odd_v8hi): Likewise.
3122 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3123 vmuloub rather than call gen_vec_widen_umult_*.
3124 (vec_widen_umult_lo_v16qi): Likewise.
3125 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3126 vmulosb rather than call gen_vec_widen_smult_*.
3127 (vec_widen_smult_lo_v16qi): Likewise.
3128 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3129 rather than call gen_vec_widen_umult_*.
3130 (vec_widen_umult_lo_v8hi): Likewise.
3131 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3132 rather than call gen_vec_widen_smult_*.
3133 (vec_widen_smult_lo_v8hi): Likewise.
3135 2014-01-15 Jeff Law <law@redhat.com>
3137 PR tree-optimization/59747
3138 * ree.c (find_and_remove_re): Properly handle case where a second
3139 eliminated extension requires widening a copy created for elimination
3140 of a prior extension.
3141 (combine_set_extension): Ensure that the number of hard regs needed
3142 for a destination register does not change when we widen it.
3144 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3146 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3147 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3148 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3149 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3150 (avr-*-rtems*): Likewise.
3151 (bfin*-rtems*): Likewise.
3152 (moxie-*-rtems*): Likewise.
3153 (h8300-*-rtems*): Likewise.
3154 (i[34567]86-*-rtems*): Likewise.
3155 (lm32-*-rtems*): Likewise.
3156 (m32r-*-rtems*): Likewise.
3157 (m68k-*-rtems*): Likewise.
3158 (microblaze*-*-rtems*): Likewise.
3159 (mips*-*-rtems*): Likewise.
3160 (powerpc-*-rtems*): Likewise.
3161 (sh-*-rtems*): Likewise.
3162 (sparc-*-rtems*): Likewise.
3163 (sparc64-*-rtems*): Likewise.
3164 (v850-*-rtems*): Likewise.
3165 (m32c-*-rtems*): Likewise.
3167 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3169 PR rtl-optimization/59511
3170 * ira.c (ira_init_register_move_cost): Use memory costs for some
3171 cases of register move cost calculations.
3172 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3173 instead of BB frequency.
3174 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3175 * lra-assigns.c (find_hard_regno_for): Ditto.
3177 2014-01-15 Richard Biener <rguenther@suse.de>
3179 PR tree-optimization/59822
3180 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3181 (vectorizable_load): Use it to hoist defs of uses of invariant
3182 loads out of the loop.
3184 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3185 Kugan Vivekanandarajah <kuganv@linaro.org>
3188 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3191 2014-01-15 Richard Biener <rguenther@suse.de>
3193 PR rtl-optimization/59802
3194 * lcm.c (compute_available): Use inverted postorder to seed
3195 the initial worklist.
3197 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3200 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3201 ADDR_REGS for invalid symrefs in non-PIC code.
3203 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3206 * builtins.c (determine_block_size): Initialize *probable_max_size
3207 even if len_rtx is CONST_INT.
3209 2014-01-14 Andrew Pinski <apinski@cavium.com>
3211 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3212 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3213 (cortexa53_tunings): Likewise.
3214 (aarch64_sched_issue_rate): New function.
3215 (TARGET_SCHED_ISSUE_RATE): Define.
3217 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3219 * ira-costs.c (find_costs_and_classes): Add missed
3220 ira_init_register_move_cost_if_necessary.
3222 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3225 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3227 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3230 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3231 to indicate if type is used for function return value. Warn ABI
3232 change if the vector mode isn't available for function return value.
3233 (ix86_function_arg_advance): Pass false to type_natural_mode.
3234 (ix86_function_arg): Likewise.
3235 (ix86_gimplify_va_arg): Likewise.
3236 (function_arg_32): Don't warn ABI change.
3237 (ix86_function_value): Pass true to type_natural_mode.
3238 (ix86_return_in_memory): Likewise.
3239 (ix86_struct_value_rtx): Removed.
3240 (TARGET_STRUCT_VALUE_RTX): Likewise.
3242 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3244 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3245 converting a conditional jump into a conditional return.
3247 2014-01-14 Richard Biener <rguenther@suse.de>
3249 PR tree-optimization/58921
3250 PR tree-optimization/59006
3251 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3252 hoisting invariant stmts.
3253 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3254 invariant loads on the preheader edge if possible.
3256 2014-01-14 Joey Ye <joey.ye@arm.com>
3258 * doc/plugin.texi (Building GCC plugins): Update to C++.
3260 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3262 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3263 (_mm_rcp28_round_ss): Ditto.
3264 (_mm_rsqrt28_round_sd): Ditto.
3265 (_mm_rsqrt28_round_ss): Ditto.
3266 (_mm_rcp28_sd): Ditto.
3267 (_mm_rcp28_ss): Ditto.
3268 (_mm_rsqrt28_sd): Ditto.
3269 (_mm_rsqrt28_ss): Ditto.
3270 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3271 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3272 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3273 (IX86_BUILTIN_RCP28SD): Ditto.
3274 (IX86_BUILTIN_RCP28SS): Ditto.
3275 (IX86_BUILTIN_RSQRT28SD): Ditto.
3276 (IX86_BUILTIN_RSQRT28SS): Ditto.
3277 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3278 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3279 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3280 (ix86_expand_special_args_builtin): Expand new FTYPE.
3281 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3282 (srcp14<mode>): Make insn unary.
3283 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3284 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3285 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3286 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3287 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3288 Fix rounding: make it SAE only.
3289 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3291 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3293 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3294 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3295 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3296 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3297 (round_saeonly_mask_scalar_operand4): Ditto.
3298 (round_saeonly_mask_scalar_op3): Ditto.
3299 (round_saeonly_mask_scalar_op4): Ditto.
3301 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3303 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3304 Implement -maltivec=be for vec_insert and vec_extract.
3306 2014-01-10 DJ Delorie <dj@redhat.com>
3308 * config/msp430/msp430.md (call_internal): Don't allow memory
3309 references with SP as the base register.
3310 (call_value_internal): Likewise.
3311 * config/msp430/constraints.md (Yc): New. For memory references
3312 that don't use SP as a base register.
3314 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3315 "an integer without a # prefix"
3316 * config/msp430/msp430.md (epilogue_helper): Use it.
3318 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3321 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3322 AVX512F gather builtins.
3323 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3324 on gather decls with INTEGER_TYPE masktype.
3325 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3326 directly into the builtin rather than hoisting it before loop.
3328 PR tree-optimization/59387
3329 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3330 (scev_const_prop): If folded_casts and type has undefined overflow,
3331 use force_gimple_operand instead of force_gimple_operand_gsi and
3332 for each added stmt if it is assign with
3333 arith_code_with_undefined_signed_overflow, call
3334 rewrite_to_defined_overflow.
3335 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3336 gimple-fold.h instead.
3337 (arith_code_with_undefined_signed_overflow,
3338 rewrite_to_defined_overflow): Moved to ...
3339 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3340 rewrite_to_defined_overflow): ... here. No longer static.
3341 Include gimplify-me.h.
3342 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3343 rewrite_to_defined_overflow): New prototypes.
3345 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3347 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3349 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3351 * builtins.c (get_object_alignment_2): Minor tweak.
3352 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3354 2014-01-13 Christian Bruel <christian.bruel@st.com>
3356 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3357 optimized non constant lengths.
3359 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3362 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3363 load as __atomic_load_N if possible.
3365 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3367 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3369 (rs6000_expand_builtin): Adjust call.
3371 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3374 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3375 * config/rs6000/rs6000.c: Include target-globals.h.
3376 (rs6000_set_current_function): Instead of doing target_reinit
3377 unconditionally, use save_target_globals_default_opts and
3378 restore_target_globals.
3380 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3382 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3383 (rs6000_expand_builtin): Handle mffs and mtfsf.
3384 (rs6000_init_builtins): Define mffs and mtfsf.
3385 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3386 (rs6000_mffs): New pattern.
3387 (rs6000_mtfsf): New pattern.
3389 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3391 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3392 Start narrowing with START. Apply candidate-use pair
3393 and check overall cost in narrowing.
3394 (iv_ca_prune): Pass new argument.
3396 2014-01-10 Jeff Law <law@redhat.com>
3399 * ree.c (combine_reaching_defs): Ensure the defining statement
3400 occurs before the extension when optimizing extensions with
3401 different source and destination hard registers.
3403 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3406 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3407 vtables into the type inheritance graph.
3409 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3411 PR rtl-optimization/59754
3412 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3413 modes in the REGNO != REGNO case.
3415 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3417 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3419 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3421 PR tree-optimization/59745
3422 * tree-predcom.c (tree_predictive_commoning_loop): Call
3423 free_affine_expand_cache if giving up because components is NULL.
3425 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3426 GC in payload of target_globals struct instead of allocating them on
3427 the heap and the larger structs separately using GC.
3428 * target-globals.h (struct target_globals): Make regs, hard_regs,
3429 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3430 of GTY((skip)) and change type to void *.
3431 (reset_target_globals): Cast loads from those fields to corresponding
3434 2014-01-10 Steve Ellcey <sellcey@mips.com>
3437 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3438 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3439 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3441 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3444 * aarch64-modes.def (CC_Zmode): New flags mode.
3445 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3446 represents an equality.
3447 (aarch64_get_condition_code): Handle CC_Zmode.
3448 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3450 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3452 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3453 extraction in good case.
3455 2014-01-10 Richard Biener <rguenther@suse.de>
3457 PR tree-optimization/59374
3458 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3459 checking after SLP discovery. Mark stmts not participating
3460 in any SLP instance properly.
3462 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3464 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3465 when handling a SET rtx.
3467 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3469 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3470 (cortex-a57): Likewise.
3471 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3473 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3475 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3476 non-iwmmxt builtins.
3478 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3482 * ipa-devirt.c record_target_from_binfo): Take as argument
3483 stack of binfos and lookup matching one for virtual inheritance.
3484 (possible_polymorphic_call_targets_1): Update.
3486 2014-01-10 Huacai Chen <chenhc@lemote.com>
3488 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3489 kernel strings for Loongson-2E/2F/3A.
3491 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3494 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3495 is_gimple_call before calling gimple_call_internal_p.
3497 2014-01-09 Steve Ellcey <sellcey@mips.com>
3499 * Makefile.in (TREE_FLOW_H): Remove.
3500 (TREE_SSA_H): Add file names from tree-flow.h.
3501 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3502 * tree.h: Remove tree-flow.h reference.
3503 * hash-table.h: Remove tree-flow.h reference.
3504 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3505 reference with tree-ssa-loop.h.
3507 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3509 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3510 default element-order behavior for -maltivec.
3511 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3512 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3513 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3514 when targeting big endian, at least for now.
3515 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3517 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3520 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3521 var satisfies use_register_for_decl, just take into account type
3522 alignment, rather than decl alignment.
3524 PR tree-optimization/59622
3525 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3526 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3527 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3528 Don't devirtualize for inplace at all. For targets.length () == 1,
3529 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3531 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3533 * config/i386/i386.md (cpu): Remove the unused btver1.
3535 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3537 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3539 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3542 * tree-core.h (struct target_globals): New forward declaration.
3543 (struct tree_target_option): Add globals field.
3544 * tree.h (TREE_TARGET_GLOBALS): Define.
3545 (prepare_target_option_nodes_for_pch): New prototype.
3546 * target-globals.h (struct target_globals): Define even if
3548 * tree.c (prepare_target_option_node_for_pch,
3549 prepare_target_option_nodes_for_pch): New functions.
3550 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3551 * config/i386/i386.c: Include target-globals.h.
3552 (ix86_set_current_function): Instead of doing target_reinit
3553 unconditionally, use save_target_globals_default_opts and
3554 restore_target_globals.
3556 2014-01-09 Richard Biener <rguenther@suse.de>
3558 PR tree-optimization/59715
3559 * tree-cfg.h (split_critical_edges): Declare.
3560 * tree-cfg.c (split_critical_edges): Export.
3561 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3563 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3565 * cfgexpand.c (expand_stack_vars): Optionally disable
3566 asan stack protection.
3567 (expand_used_vars): Likewise.
3568 (partition_stack_vars): Likewise.
3569 * asan.c (asan_emit_stack_protection): Optionally disable
3570 after return stack usage.
3571 (instrument_derefs): Optionally disable memory access instrumentation.
3572 (instrument_builtin_call): Likewise.
3573 (instrument_strlen_call): Likewise.
3574 (asan_protect_global): Optionally disable global variables protection.
3575 * doc/invoke.texi: Added doc for new options.
3576 * params.def: Added new options.
3577 * params.h: Likewise.
3579 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3581 PR rtl-optimization/59724
3582 * ifcvt.c (cond_exec_process_if_block): Don't call
3583 flow_find_head_matching_sequence with 0 longest_match.
3584 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3585 non-active insns if !stop_after.
3586 (try_head_merge_bb): Revert 2014-01-07 changes.
3588 2014-01-08 Jeff Law <law@redhat.com>
3590 * ree.c (get_sub_rtx): New function, extracted from...
3591 (merge_def_and_ext): Here.
3592 (combine_reaching_defs): Use get_sub_rtx.
3594 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3596 * cgraph.h (varpool_variable_node): Do not choke on null node.
3598 2014-01-08 Catherine Moore <clm@codesourcery.com>
3600 * config/mips/mips.md (simple_return): Attempt to use JRC
3602 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3604 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3606 PR rtl-optimization/59137
3607 * reorg.c (steal_delay_list_from_target): Call update_block for
3609 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3611 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3613 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3614 two duplicate entries.
3616 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3619 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3621 * config/mips/mips.c (mips_truncated_op_cost): New function.
3622 (mips_rtx_costs): Adjust test for BADDU.
3623 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3625 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3627 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3628 (*baddu_si): ...this new pattern.
3630 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3633 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3635 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3638 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3640 (expand_expr, expand_normal): Adjust.
3641 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3642 inner_reference_p. Use inner_reference_p to expand inner references.
3643 (store_expr): Adjust.
3644 * cfgexpand.c (expand_call_stmt): Adjust.
3646 2014-01-08 Rong Xu <xur@google.com>
3648 * gcov-io.c (gcov_var): Move from gcov-io.h.
3649 (gcov_position): Ditto.
3650 (gcov_is_error): Ditto.
3651 (gcov_rewrite): Ditto.
3652 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3653 only part to libgcc/libgcov.h.
3655 2014-01-08 Marek Polacek <polacek@redhat.com>
3658 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3660 2014-01-08 Marek Polacek <polacek@redhat.com>
3663 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3665 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3667 PR rtl-optimization/59649
3668 * stor-layout.c (get_mode_bounds): For BImode return
3669 0 and STORE_FLAG_VALUE.
3671 2014-01-08 Richard Biener <rguenther@suse.de>
3674 * gimple.h (is_gimple_builtin_call): Remove.
3675 (gimple_builtin_call_types_compatible_p): New.
3676 (gimple_call_builtin_p): New overload.
3677 * gimple.c (is_gimple_builtin_call): Remove.
3678 (validate_call): Rename to ...
3679 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3681 (validate_type): New static function.
3682 (gimple_call_builtin_p): New overload and adjust.
3683 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3684 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3685 (gimple_fold_stmt_to_constant_1): Likewise.
3686 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3688 2014-01-08 Richard Biener <rguenther@suse.de>
3691 * gimplify.c (gimplify_expr): Gimplify register-register type
3692 VIEW_CONVERT_EXPRs to separate stmts.
3694 2014-01-07 Jeff Law <law@redhat.com>
3697 * ree.c (combine_set_extension): Handle case where source
3698 and destination registers in an extension insn are different.
3699 (combine_reaching_defs): Allow source and destination registers
3700 in extension to be different under limited circumstances.
3701 (add_removable_extension): Remove restriction that the
3702 source and destination registers in the extension are the same.
3703 (find_and_remove_re): Emit a copy from the extension's
3704 destination to its source after the defining insn if
3705 the source and destination registers are different.
3708 * ifcvt.c (merge_if_block): If we are merging a block with more than
3709 one successor with a block with no successors, remove any BARRIER
3710 after the second block.
3712 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3714 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3716 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3719 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3720 for 14-bit register offsets when INT14_OK_STRICT is false.
3722 2014-01-07 Roland Stigge <stigge@antcom.de>
3723 Michael Meissner <meissner@linux.vnet.ibm.com>
3726 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3727 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3729 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3731 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3734 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3736 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3737 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3738 rtx is const0_rtx or not.
3740 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3743 * target-globals.c (save_target_globals): Remove this_fn_optab
3745 * toplev.c: Include optabs.h.
3746 (target_reinit): Temporarily restore the global options if another
3747 set of options are in force.
3749 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3751 PR rtl-optimization/58668
3752 * cfgcleanup.c (flow_find_cross_jump): Don't count
3753 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3754 to determine what is counted.
3755 (flow_find_head_matching_sequence): Use active_insn_p to determine
3757 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3759 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3760 determine what is counted.
3762 PR tree-optimization/59643
3763 * tree-predcom.c (split_data_refs_to_components): If one dr is
3764 read and one write, determine_offset fails and the write isn't
3765 in the bad component, just put the read into the bad component.
3767 2014-01-07 Mike Stump <mikestump@comcast.net>
3768 Jakub Jelinek <jakub@redhat.com>
3771 * tree-core.h (struct tree_optimization_option): Change optabs
3772 type from unsigned char * to void *.
3773 * optabs.c (init_tree_optimization_optabs): Adjust
3774 TREE_OPTIMIZATION_OPTABS initialization.
3776 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3779 * config/i386/i386.h (struct machine_function): Add
3780 no_drap_save_restore field.
3781 * config/i386/i386.c (ix86_save_reg): Use
3782 !cfun->machine->no_drap_save_restore instead of
3783 crtl->stack_realign_needed.
3784 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3785 this function clears frame_pointer_needed. Set
3786 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3787 and DRAP reg is needed.
3789 2014-01-06 Marek Polacek <polacek@redhat.com>
3792 * doc/implement-c.texi: Mention that other integer types are
3793 permitted as bit-field types in strictly conforming mode.
3795 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3797 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3800 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3802 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3804 2014-01-06 Martin Jambor <mjambor@suse.cz>
3807 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3809 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3811 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3815 * var-tracking.c (add_stores): Preserve the value of the source even if
3816 we don't record the store.
3818 2014-01-06 Terry Guo <terry.guo@arm.com>
3820 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3822 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3825 * config/darwin.c (darwin_function_section): Adjust return values to
3826 correspond to optimisation changes made in r206070.
3828 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3830 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3831 from prefetch_block tune setting.
3832 (nocona_cost): Correct size of prefetch block to 64.
3834 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3836 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3837 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3838 used to save the static chain register in the computation of the offset
3839 from which the FP registers need to be restored.
3841 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3843 PR tree-optimization/59519
3844 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3845 ICE if get_current_def (current_new_name) is already non-NULL, as long
3846 as it is a phi result of some other phi in *new_exit_bb that has
3849 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3850 or vmovdqu* for misaligned_operand.
3851 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3852 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3853 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3854 aligned_mem for AVX512F masked aligned load and store builtins and for
3857 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3859 PR tree-optimization/59651
3860 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3861 Address range for negative step should be added by TYPE_SIZE_UNIT.
3863 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3865 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3866 overlapping registers also for registers other than the stack pointer.
3868 2014-01-03 Marek Polacek <polacek@redhat.com>
3871 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3874 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3877 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3880 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3881 (push splitter): Use <P:MODE_SIZE> instead of
3882 GET_MODE_SIZE (<P:MODE>mode).
3883 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3884 (mov -1, reg peephole2): Likewise.
3885 * config/i386/sse.md (*mov<mode>_internal,
3886 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3887 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3888 *<code><mode>3, *andnot<mode>3<mask_name>,
3889 <mask_codefor><code><mode>3<mask_name>): Likewise.
3890 * config/i386/subst.md (mask_mode512bit_condition,
3891 sd_mask_mode512bit_condition): Likewise.
3893 2014-01-02 Xinliang David Li <davidxl@google.com>
3895 PR tree-optimization/59303
3896 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3897 (dump_predicates): Better output format.
3898 (pred_equal_p): New function.
3899 (is_neq_relop_p): Ditto.
3900 (is_neq_zero_form_p): Ditto.
3901 (pred_expr_equal_p): Ditto.
3902 (pred_neg_p): Ditto.
3903 (simplify_pred): Ditto.
3904 (simplify_preds_2): Ditto.
3905 (simplify_preds_3): Ditto.
3906 (simplify_preds_4): Ditto.
3907 (simplify_preds): Ditto.
3909 (push_to_worklist): Ditto.
3910 (get_pred_info_from_cmp): Ditto.
3911 (is_degenerated_phi): Ditto.
3912 (normalize_one_pred_1): Ditto.
3913 (normalize_one_pred): Ditto.
3914 (normalize_one_pred_chain): Ditto.
3915 (normalize_preds): Ditto.
3916 (normalize_cond_1): Remove function.
3917 (normalize_cond): Ditto.
3918 (is_gcond_subset_of): Ditto.
3919 (is_subset_of_any): Ditto.
3920 (is_or_set_subset_of): Ditto.
3921 (is_and_set_subset_of): Ditto.
3922 (is_norm_cond_subset_of): Ditto.
3923 (pred_chain_length_cmp): Ditto.
3924 (convert_control_dep_chain_into_preds): Type change.
3925 (find_predicates): Ditto.
3926 (find_def_preds): Ditto.
3927 (destroy_predicates_vecs): Ditto.
3928 (find_matching_predicates_in_rest_chains): Ditto.
3929 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3930 (is_pred_expr_subset): Ditto.
3931 (is_pred_chain_subset_of): Ditto.
3932 (is_included_in): Ditto.
3933 (is_superset_of): Ditto.
3935 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3937 Update copyright years.
3939 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3941 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3942 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3943 config/arc/arc.md, config/arc/arc.opt,
3944 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3945 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3946 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3947 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3948 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3949 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3950 vtable-verify.c, vtable-verify.h: Use the standard form for the
3953 2014-01-02 Tobias Burnus <burnus@net-b.de>
3955 * gcc.c (process_command): Update copyright notice dates.
3956 * gcov-dump.c: Ditto.
3958 * doc/cpp.texi: Bump @copying's copyright year.
3959 * doc/cppinternals.texi: Ditto.
3960 * doc/gcc.texi: Ditto.
3961 * doc/gccint.texi: Ditto.
3962 * doc/gcov.texi: Ditto.
3963 * doc/install.texi: Ditto.
3964 * doc/invoke.texi: Ditto.
3966 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3968 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3970 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3972 * config/i386/sse.md (*mov<mode>_internal): Guard
3973 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3975 PR rtl-optimization/59647
3976 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3977 new_rtx into UNSIGNED_FLOAT rtxes.
3979 Copyright (C) 2014 Free Software Foundation, Inc.
3981 Copying and distribution of this file, with or without modification,
3982 are permitted in any medium without royalty provided the copyright
3983 notice and this notice are preserved.