1 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
6 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
9 define_insn with define_expand and new define_insn
10 *altivec_lve<VI_char>x_internal.
11 (altivec_stve<VI_char>x): Replace define_insn with define_expand
12 and new define_insn *altivec_stve<VI_char>x_internal.
13 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
15 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
17 (altivec_expand_stvex_be): New function.
19 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
21 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
23 to STACK_POINTER_REGNUM if !frame_pointer_needed.
24 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
25 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
27 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
30 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
31 instead of emit_move_insn.
33 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
37 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
40 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
42 * config/rs6000/altivec.md (altivec_lvxl): Rename as
43 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
45 (altivec_lvxl_<mode>): New define_expand incorporating
46 -maltivec=be semantics where needed.
47 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
48 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
49 semantics where needed.
50 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
51 (altivec_stvx_<mode>): New define_expand incorporating
52 -maltivec=be semantics where needed.
53 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
54 VM2 iterator instead of V4SI.
55 (altivec_stvxl_<mode>): New define_expand incorporating
56 -maltivec=be semantics where needed.
57 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
58 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
59 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
60 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
61 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
63 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
64 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
65 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
66 ALTIVEC_BUILTIN_STVXL.
67 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
69 (altivec_expand_stvx_be): Likewise.
70 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
71 (altivec_expand_lvx_be): Likewise.
72 (altivec_expand_stvx_be): Likewise.
73 (altivec_expand_builtin): Add cases for
74 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
75 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
76 (altivec_init_builtins): Add definitions for
77 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
78 __builtin_altivec_stvx_<mode>, and
79 __builtin_altivec_stvxl_<mode>.
81 2014-02-21 Catherine Moore <clm@codesourcery.com>
83 * doc/invoke.texi (mvirt, mno-virt): Document.
84 * config/mips/mips.opt (mvirt): New option.
85 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
87 2014-02-21 Richard Biener <rguenther@suse.de>
89 PR tree-optimization/60276
90 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
91 (STMT_VINFO_MIN_NEG_DIST): New macro.
92 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
93 STMT_VINFO_MIN_NEG_DIST.
94 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
95 made for negative dependence distances still hold.
97 2014-02-21 Richard Biener <rguenther@suse.de>
100 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
101 DECL_INITIAL for globals not in the current function context.
103 2014-02-21 Jakub Jelinek <jakub@redhat.com>
105 PR tree-optimization/56490
106 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
107 * tree-ssa-uninit.c: Include params.h.
108 (compute_control_dep_chain): Add num_calls argument, return false
109 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
110 num_calls to recursive call.
111 (find_predicates): Change dep_chain into normal array,
112 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
113 variable and adjust compute_control_dep_chain caller.
114 (find_def_preds): Likewise.
116 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
118 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
119 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
121 2014-02-21 Nick Clifton <nickc@redhat.com>
123 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
125 (popqi1): Add mode to pre_dec.
128 2014-02-21 Jakub Jelinek <jakub@redhat.com>
130 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
131 mode for mask of V8SFmode permutation.
133 2014-02-20 Richard Henderson <rth@redhat.com>
136 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
137 a new pseudo for OLDVAL.
139 2014-02-20 Jakub Jelinek <jakub@redhat.com>
142 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
143 gen_reg_rtx if d->testing_p.
144 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
145 if d->testing_p and we will certainly return true.
146 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
149 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
151 * emit-rtl.c (gen_reg_rtx): Assert that
152 crtl->emit.regno_pointer_align_length is non-zero.
154 2014-02-20 Richard Henderson <rth@redhat.com>
157 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
158 on failure the store back into EXPECT.
160 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
161 Sandra Loosemore <sandra@codesourcery.com>
163 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
164 * config/nios2/nios2.c (nios2_function_profiler):
165 Add -fPIC (flag_pic == 2) support.
166 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
167 (nios2_large_offset_p): New function.
168 (nios2_unspec_reloc_p): Move up position, update to use
169 nios2_large_offset_p.
170 (nios2_unspec_address): Remove function.
171 (nios2_unspec_offset): New function.
172 (nios2_large_got_address): New function.
173 (nios2_got_address): Add large offset support.
174 (nios2_legitimize_tls_address): Update usage of removed and new
176 (nios2_symbol_binds_local_p): New function.
177 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
178 (nios2_legitimize_address): Update to use nios2_large_offset_p.
179 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
180 (nios2_print_operand): Merge H/L processing, add hiadj/lo
181 processing for (const (unspec ...)).
182 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
184 2014-02-20 Richard Biener <rguenther@suse.de>
186 * tree-cfg.c (replace_uses_by): Mark altered BBs before
187 doing the substitution.
188 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
190 2014-02-20 Martin Jambor <mjambor@suse.cz>
193 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
194 info when checking whether lattices are bottom.
196 2014-02-20 Richard Biener <rguenther@suse.de>
199 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
202 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
205 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
206 parameter specifying the scaling.
207 (inline_call): Update.
208 (want_inline_recursively): Guard division by zero.
209 (recursive_inlining): Update.
210 * ipa-inline.h (clone_inlined_nodes): Update.
212 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
215 * config/i386/i386.c (classify_argument): Pass structures of size
216 64 bytes or less in register.
218 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
219 Kirill Yukhin <kirill.yukhin@intel.com>
221 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
222 (_mm_rcp28_round_ss): Ditto.
223 (_mm_rsqrt28_round_sd): Ditto.
224 (_mm_rsqrt28_round_ss): Ditto.
225 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
226 (_mm_rcp14_round_ss): Ditto.
227 (_mm_rsqrt14_round_sd): Ditto.
228 (_mm_rsqrt14_round_ss): Ditto.
229 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
230 the first input operand, get rid of match_dup.
231 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
233 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
235 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
236 operand as the first input operand, set type attribute.
237 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
239 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
240 operand as the first input operand, set type attribute.
242 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
244 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
247 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
250 * config/i386/i386.c (construct_container): Remove TFmode check
251 for X86_64_INTEGER_CLASS.
253 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
256 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
257 only when -Wpsabi is enabled.
259 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
262 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
263 passing arrays in registers are the same as for structs, so remove the
264 special case for them.
266 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
268 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
269 destination type, extract only the valid bits if the source type is not
270 integral and has a different mode.
272 2014-02-19 Richard Biener <rguenther@suse.de>
275 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
278 2014-02-19 Richard Biener <rguenther@suse.de>
281 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
282 (ipa_modify_call_arguments): Emit an argument load explicitely and
283 preserve virtual SSA form there and for the replacement call.
284 Do not update SSA form nor free dominance info.
286 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
288 * ipa.c (function_and_variable_visibility): Also clear WEAK
289 flag when disolving COMDAT_GROUP.
291 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
293 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
294 * ipa-prop.c (ipa_set_jf_known_type): Return early when
296 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
297 do more sanity checks.
298 (detect_type_change): Return true when giving up early.
299 (compute_complex_assign_jump_func): Fix type parameter of
301 (compute_complex_ancestor_jump_func): Likewise.
302 (update_jump_functions_after_inlining): Fix updating of
304 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
306 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
308 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
309 inline clones when edge disappears.
311 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
314 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
315 Split 64-bit moves into 2 patterns. Do not allow the use of
316 direct move for TDmode in little endian, since the decimal value
317 has little endian bytes within a word, but the 64-bit pieces are
318 ordered in a big endian fashion, and normal subreg's of TDmode are
320 (mov<mode>_64bit_dm): Likewise.
321 (movtd_64bit_nodm): Likewise.
323 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
325 PR tree-optimization/60174
326 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
327 statement of an SSA_NAME that occurs in an abnormal PHI node.
329 2014-02-18 Jakub Jelinek <jakub@redhat.com>
332 * final.c (SEEN_BB): Remove.
333 (SEEN_NOTE, SEEN_EMITTED): Renumber.
334 (final_scan_insn): Don't force_source_line on second
335 NOTE_INSN_BASIC_BLOCK.
337 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
340 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
341 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
342 (type_natural_mode): Warn ABI change when %zmm register is not
343 available for AVX512F vector value passing.
345 2014-02-18 Kai Tietz <ktietz@redhat.com>
348 * config/i386/i386.c (ix86_expand_prologue): Use value in
349 rax register as displacement when restoring %r10 or %rax.
350 Fix wrong offset when restoring both registers.
352 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
354 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
355 assertion with conditional return.
357 2014-02-18 Jakub Jelinek <jakub@redhat.com>
358 Uros Bizjak <ubizjak@gmail.com>
361 * config/i386/driver-i386.c (host_detect_local_cpu): If
362 YMM state is not saved by the OS, also clear has_f16c. Move
363 CPUID 0x80000001 handling before YMM state saving checking.
365 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
367 PR rtl-optimization/58960
368 * haifa-sched.c (alloc_global_sched_pressure_data): New,
369 factored out from ...
370 (sched_init): ... here.
371 (free_global_sched_pressure_data): New, factored out from ...
372 (sched_finish): ... here.
373 * sched-int.h (free_global_sched_pressure_data): Declare.
374 * sched-rgn.c (nr_regions_initial): New static global.
375 (haifa_find_rgns): Initialize it.
376 (schedule_region): Disable sched-pressure for the newly
379 2014-02-17 Richard Biener <rguenther@suse.de>
381 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
382 release SSA defs of pattern stmts.
384 2014-02-17 Richard Biener <rguenther@suse.de>
386 * tree-inline.c (expand_call_inline): Release the virtual
387 operand defined by the call we are about to inline.
389 2014-02-17 Richard Biener <rguenther@suse.de>
391 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
393 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
394 Ilya Tocar <ilya.tocar@intel.com>
396 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
397 arguments order in builtin.
398 (_mm512_permutexvar_epi64): Ditto.
399 (_mm512_mask_permutexvar_epi64): Ditto
400 (_mm512_maskz_permutexvar_epi32): Ditto
401 (_mm512_permutexvar_epi32): Ditto
402 (_mm512_mask_permutexvar_epi32): Ditto
404 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
406 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
407 (p8_vmrgow): Likewise.
409 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
411 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
414 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
417 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
418 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
419 into 64-bit and 32-bit moves. On 64-bit moves, add support for
420 using direct move instructions on ISA 2.07. Also adjust
421 instruction length for 64-bit.
422 (mov<mode>_64bit, TFmode/TDmode): Likewise.
423 (mov<mode>_32bit, TFmode/TDmode): Likewise.
425 2014-02-15 Alan Modra <amodra@gmail.com>
429 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
430 find_replacement on parts of insn rtl that might be reloaded.
432 2014-02-15 Richard Biener <rguenther@suse.de>
434 PR tree-optimization/60183
435 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
436 (tree_ssa_phiprop): Calculate and free post-dominators.
438 2014-02-14 Jeff Law <law@redhat.com>
440 PR rtl-optimization/60131
441 * ree.c (get_extended_src_reg): New function.
442 (combine_reaching_defs): Use it rather than assuming location of REG.
443 (find_and_remove_re): Verify first operand of extension is
444 a REG before adding the insns to the copy list.
446 2014-02-14 Roland McGrath <mcgrathr@google.com>
448 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
449 * configure: Regenerated.
450 * config.in: Regenerated.
451 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
452 instead of ASM_SHORT.
454 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
455 Richard Earnshaw <rearnsha@arm.com>
457 PR rtl-optimization/59535
458 * lra-constraints.c (process_alt_operands): Encourage alternative
459 when unassigned pseudo class is superset of the alternative class.
460 (inherit_reload_reg): Don't inherit when optimizing for code size.
461 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
462 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
463 modes not less than 4 for Thumb1.
465 2014-02-14 Kyle McMartin <kyle@redhat.com>
468 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
470 2014-02-14 Richard Biener <rguenther@suse.de>
472 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
473 (get_frame_arg): Drop the assert with langhook types_compatible_p.
474 Do not strip INDIRECT_REFs.
476 2014-02-14 Richard Biener <rguenther@suse.de>
479 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
480 DECL_FUNCTION_SPECIFIC_TARGET.
481 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
482 * tree-streamer-out.c (pack_ts_target_option): Remove.
483 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
484 (write_ts_function_decl_tree_pointers): Do not stream
485 DECL_FUNCTION_SPECIFIC_TARGET.
486 * tree-streamer-in.c (unpack_ts_target_option): Remove.
487 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
488 (lto_input_ts_function_decl_tree_pointers): Do not stream
489 DECL_FUNCTION_SPECIFIC_TARGET.
491 2014-02-14 Jakub Jelinek <jakub@redhat.com>
493 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
494 (get_initial_def_for_induction, vectorizable_induction): Ignore
495 debug stmts when looking for exit_phi.
496 (vectorizable_live_operation): Fix up condition.
498 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
500 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
501 nreverse() because it changes the content of original tree list.
503 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
505 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
506 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
508 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
510 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
511 GNU coding standards.
513 2014-02-13 Jakub Jelinek <jakub@redhat.com>
516 * dwarf2out.c (gen_subprogram_die): Don't call
517 add_calling_convention_attribute if subr_die is old_die.
519 2014-02-13 Sharad Singhai <singhai@google.com>
521 * doc/optinfo.texi: Fix order of nodes.
523 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
525 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
526 operands[2], not operands[3].
528 2014-02-13 Richard Biener <rguenther@suse.de>
531 * doc/install.texi (ISL): Update recommended version to 0.12.2,
532 mention the possibility of an in-tree build.
533 (CLooG): Update recommended version to 0.18.1, mention the
534 possibility of an in-tree build and clarify that the ISL
535 bundled with CLooG does not work.
537 2014-02-13 Jakub Jelinek <jakub@redhat.com>
540 * expr.c (compress_float_constant): If x is a hard register,
541 extend into a pseudo and then move to x.
543 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
545 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
546 caused by bad second argument to warning_at() with -mhotpatch and
547 nested functions (e.g. with gfortran).
549 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
551 * opts.c (option_name): Remove "enabled by default" rider.
553 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
555 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
557 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
558 Uros Bizjak <ubizjak@gmail.com>
561 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
562 * configure: Regenerated.
564 2014-02-12 Richard Biener <rguenther@suse.de>
566 * vec.c (vec_prefix::calculate_allocation): Move as
567 inline variant to vec.h.
568 (vec_prefix::calculate_allocation_1): New out-of-line version.
569 * vec.h (vec_prefix::calculate_allocation_1): Declare.
570 (vec_prefix::m_has_auto_buf): Rename to ...
571 (vec_prefix::m_using_auto_storage): ... this.
572 (vec_prefix::calculate_allocation): Inline the easy cases
573 and dispatch to calculate_allocation_1 which doesn't need the
575 (va_heap::reserve): Use gcc_checking_assert.
576 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
577 m_using_auto_storage.
578 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
580 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
581 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
582 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
584 2014-02-12 Richard Biener <rguenther@suse.de>
586 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
587 when we found a dependence.
589 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
591 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
593 (maybe_fold_stmt): ... into this new function.
594 * omp-low.c (lower_omp): Update comment.
596 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
599 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
602 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
604 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
605 identifiers in comments.
606 (cortexa53_extra_costs): Likewise.
607 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
608 (cortexa7_extra_costs): Likewise.
609 (cortexa12_extra_costs): Likewise.
610 (cortexa15_extra_costs): Likewise.
611 (v7m_extra_costs): Likewise.
613 2014-02-12 Richard Biener <rguenther@suse.de>
616 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
617 of posix_memalign being successful.
618 (lower_stmt): Restrict lowering of posix_memalign to when
619 -ftree-bit-ccp is enabled.
621 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
623 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
625 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
627 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
629 PR rtl-optimization/60116
630 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
631 other_insn once the combination has been validated.
633 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
636 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
638 * ipa-devirt.c: Include demangle.h
639 (odr_violation_reported): New static variable.
640 (add_type_duplicate): Update odr_violations.
641 (maybe_record_node): Add completep parameter; update it.
642 (record_target_from_binfo): Add COMPLETEP parameter;
644 (possible_polymorphic_call_targets_1): Likewise.
645 (struct polymorphic_call_target_d): Add nonconstruction_targets;
646 rename FINAL to COMPLETE.
647 (record_targets_from_bases): Sanity check we found the binfo;
648 fix COMPLETEP updating.
649 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
650 parameter, fix computing of COMPLETEP.
651 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
652 at LTO time do demangling.
653 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
654 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
656 (gimple_get_virt_method_for_binfo): Likewise.
657 * gimple-fold.h (gimple_get_virt_method_for_binfo,
658 gimple_get_virt_method_for_vtable): Update prototypes.
660 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
663 * genautomata.c (add_presence_absence): Fix typo with
664 {final_}presence_list.
666 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
669 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
670 for VSX/Altivec vectors that land in GPR registers.
672 2014-02-11 Richard Henderson <rth@redhat.com>
673 Jakub Jelinek <jakub@redhat.com>
676 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
677 around drhs if type conversion to lacc->type is not useless.
679 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
681 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
683 (cortex-a57.cortex-a53): Likewise.
684 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
686 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
688 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
691 2014-02-11 Renlin Li <Renlin.Li@arm.com>
693 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
694 add_options_for_arm_vfp3.
696 2014-02-11 Jeff Law <law@redhat.com>
699 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
700 object with an undesirable mode.
702 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
705 * config/i386/sol2-9.h: New file.
706 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
707 *-*-solaris2.9*): Use it.
709 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
711 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
712 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
714 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
716 * config/microblaze/microblaze.c: Extend mcpu version format
718 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
720 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
722 2014-02-10 Richard Henderson <rth@redhat.com>
725 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
726 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
727 ms-abi vs -mno-accumulate-outgoing-args.
728 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
729 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
732 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
735 * cfgexpand.c (expand_asm_operands): Attach source location to
736 ASM_INPUT rtx objects.
737 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
739 2014-02-10 Nick Clifton <nickc@redhat.com>
741 * config/mn10300/mn10300.c (popcount): New function.
742 (mn10300_expand_prologue): Include saved registers in stack usage
745 2014-02-10 Jeff Law <law@redhat.com>
748 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
749 when changing the SET_DEST of a prior insn to avoid an input reload.
751 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
753 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
754 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
755 -mcall-openbsd, or -mcall-linux.
756 (CC1_ENDIAN_BIG_SPEC): Remove.
757 (CC1_ENDIAN_LITTLE_SPEC): Remove.
758 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
759 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
760 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
761 and %cc1_endian_default.
762 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
764 2014-02-10 Richard Biener <rguenther@suse.de>
766 PR tree-optimization/60115
767 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
768 MEM_REF handling. Properly verify that the accesses are not
769 out of the objects bound.
771 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
773 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
776 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
778 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
779 proper constants and fix formatting.
780 (possible_polymorphic_call_targets): Fix formatting.
782 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
783 Ilya Tocar <ilya.tocar@intel.com>
785 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
786 (_mm512_loadu_epi32): Renamed into...
787 (_mm512_loadu_si512): This.
788 (_mm512_storeu_epi32): Renamed into...
789 (_mm512_storeu_si512): This.
790 (_mm512_maskz_ceil_ps): Removed.
791 (_mm512_maskz_ceil_pd): Ditto.
792 (_mm512_maskz_floor_ps): Ditto.
793 (_mm512_maskz_floor_pd): Ditto.
794 (_mm512_floor_round_ps): Ditto.
795 (_mm512_floor_round_pd): Ditto.
796 (_mm512_ceil_round_ps): Ditto.
797 (_mm512_ceil_round_pd): Ditto.
798 (_mm512_mask_floor_round_ps): Ditto.
799 (_mm512_mask_floor_round_pd): Ditto.
800 (_mm512_mask_ceil_round_ps): Ditto.
801 (_mm512_mask_ceil_round_pd): Ditto.
802 (_mm512_maskz_floor_round_ps): Ditto.
803 (_mm512_maskz_floor_round_pd): Ditto.
804 (_mm512_maskz_ceil_round_ps): Ditto.
805 (_mm512_maskz_ceil_round_pd): Ditto.
806 (_mm512_expand_pd): Ditto.
807 (_mm512_expand_ps): Ditto.
808 * config/i386/i386.c (ix86_builtins): Remove
809 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
811 * config/i386/predicates.md (const1256_operand): New.
812 (const_1_to_2_operand): Ditto.
813 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
814 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
815 (*avx512pf_gatherpf<mode>sf): Ditto.
816 (avx512pf_gatherpf<mode>df): Ditto.
817 (*avx512pf_gatherpf<mode>df_mask): Ditto.
818 (*avx512pf_gatherpf<mode>df): Ditto.
819 (avx512pf_scatterpf<mode>sf): Ditto.
820 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
821 (*avx512pf_scatterpf<mode>sf): Ditto.
822 (avx512pf_scatterpf<mode>df): Ditto.
823 (*avx512pf_scatterpf<mode>df_mask): Ditto.
824 (*avx512pf_scatterpf<mode>df): Ditto.
825 (avx512f_expand<mode>): Removed.
826 (<shift_insn><mode>3<mask_name>): Change predicate type.
828 2014-02-08 Jakub Jelinek <jakub@redhat.com>
830 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
831 not at the end of datarefs vector use ordered_remove to avoid
832 reordering datarefs vector.
835 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
836 mark local addressable non-static vars as GOVD_PRIVATE
837 instead of GOVD_LOCAL.
838 * omp-low.c (lower_omp_for): Move gimple_bind_vars
839 and BLOCK_VARS of gimple_bind_block to new_stmt rather
843 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
844 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
845 assume_aligned or alloc_align attributes.
846 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
847 arguments. Handle also assume_aligned and alloc_align attributes.
848 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
849 calls to functions with assume_aligned or alloc_align attributes.
850 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
852 2014-02-08 Terry Guo <terry.guo@arm.com>
854 * doc/invoke.texi: Document ARM -march=armv7e-m.
856 2014-02-08 Jakub Jelinek <jakub@redhat.com>
858 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
859 flag on __cilkrts_rethrow builtin.
862 * ipa-cp.c (determine_versionability): Fail at -O0
863 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
864 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
867 2014-02-04 Jakub Jelinek <jakub@redhat.com>
870 * tree-inline.c (copy_forbidden): Fail for
871 __attribute__((optimize (0))) functions.
873 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
875 * varpool.c: Include pointer-set.h.
876 (varpool_remove_unreferenced_decls): Variables in other partitions
877 will not be output; be however careful to not lose information
880 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
882 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
883 lookup in the vtable constructor.
885 2014-02-07 Jeff Law <law@redhat.com>
888 * config/m68k/m68k.md (ashldi_extsi): Turn into a
889 define_insn_and_split.
891 * ipa-inline.c (inline_small_functions): Fix typos.
893 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
895 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
896 (s390_can_use_return_insn): Declare.
897 * config/s390/s390.h (EPILOGUE_USES): Define.
898 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
900 (s390_chunkify_start): Handle return JUMP_LABELs.
901 (s390_early_mach): Emit a main_pool instruction on the entry edge.
902 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
903 (s390_can_use_return_insn): New functions.
904 (s390_fix_long_loop_prediction): Handle conditional returns.
905 (TARGET_SET_UP_BY_PROLOGUE): Define.
906 * config/s390/s390.md (ANY_RETURN): New code iterator.
907 (*creturn, *csimple_return, return, simple_return): New patterns.
909 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
911 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
912 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
913 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
914 REG_CFA_RESTORE list when deciding not to restore a register.
916 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
918 * config/s390/s390.c: Include tree-pass.h and context.h.
919 (s390_early_mach): New function, split out from...
920 (s390_emit_prologue): ...here.
921 (pass_data_s390_early_mach): New pass structure.
922 (pass_s390_early_mach): New class.
923 (s390_option_override): Create and register early_mach pass.
926 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
928 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
929 to match for the exit block.
931 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
933 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
934 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
935 Reject misaligned operands.
937 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
939 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
941 2014-02-07 Richard Biener <rguenther@suse.de>
944 * gimple-low.c (lower_builtin_posix_memalign): New function.
945 (lower_stmt): Call it to lower posix_memalign in a way
946 to make alignment info accessible.
948 2014-02-07 Jakub Jelinek <jakub@redhat.com>
951 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
952 __builtin_setjmp_receiver.
954 2014-02-07 Richard Biener <rguenther@suse.de>
957 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
958 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
959 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
960 Handle BUILT_IN_POSIX_MEMALIGN.
961 (find_func_clobbers): Likewise.
962 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
963 (call_may_clobber_ref_p_1): Likewise.
965 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
968 * ipa-devirt.c (record_target_from_binfo): Remove overactive
971 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
974 * lto-cgraph.c (lto_output_node): Use
975 symtab_get_symbol_partitioning_class.
976 (lto_output_varpool_node): likewise.
977 (symtab_get_symbol_partitioning_class): Move here from
979 * cgraph.h (symbol_partitioning_class): Likewise.
980 (symtab_get_symbol_partitioning_class): Declare.
982 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
984 * ggc.h (ggc_internal_cleared_alloc): New macro.
985 * vec.h (vec_safe_copy): Handle memory stats.
986 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
987 * target-globals.c (save_target_globals): Likewise.
989 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
992 * expr.c (emit_move_resolve_push): Export; be bit more selective
993 on when to clear alias set.
994 * expr.h (emit_move_resolve_push): Declare.
995 * function.h (struct function): Add tail_call_marked.
996 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
997 * config/i386/i386-protos.h (ix86_expand_push): Remove.
998 * config/i386/i386.md (TImode move expander): De not call
1000 (FP push expanders): Preserve memory attributes.
1001 * config/i386/sse.md (push<mode>1): Remove.
1002 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1003 (ix86_expand_push): Remove.
1004 * config/i386/mmx.md (push<mode>1): Remove.
1006 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1008 PR rtl-optimization/60030
1009 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1010 lopart with paradoxical subreg before shifting it up by hprec.
1012 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1014 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1015 Remove extra newline at end of file.
1016 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1017 (arm_issue_rate): Handle cortexa57.
1018 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1019 (cortex-a57.cortex-a53): Likewise.
1021 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1024 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1025 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1027 (arm_expand_prologue): Adjust all callers.
1028 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1029 info, registers also at the lowest numbered registers side. Use
1030 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1034 * var-tracking.c (adjust_mems): Before adding a SET to
1035 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1037 2014-02-06 Alan Modra <amodra@gmail.com>
1040 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1041 change SDmode to DDmode when lra_in_progress.
1043 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1046 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1047 free_data_ref on the dr first, and before goto again also set dr
1048 to the next dr. For simd_lane_access, free old datarefs[i] before
1049 overwriting it. For get_vectype_for_scalar_type failure, don't
1050 free_data_ref if simd_lane_access.
1052 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1055 * tree.h (opts_for_fn): New inline function.
1056 (opt_for_fn): Define.
1057 * config/i386/i386.c (ix86_function_regparm): Use
1058 opt_for_fn (decl, optimize) instead of optimize.
1060 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1062 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1063 for SYMBOL_REF in large memory model.
1065 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1067 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1069 (cortex-a57): Likewise.
1070 (cortex-a57.cortex-a53): Likewise.
1072 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1073 Kugan Vivekanandarajah <kuganv@linaro.org>
1075 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1077 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1079 2014-02-06 Richard Biener <rguenther@suse.de>
1081 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1082 set_loop_copy and initialize_original_copy_tables.
1084 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1086 * config/aarch64/aarch64-simd.md
1087 (aarch64_ashr_simddi): Change QI to SI.
1089 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1090 Jakub Jelinek <jakub@redhat.com>
1093 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1096 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1098 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1099 CODE_FOR_altivec_vpku[hw]um to
1100 CODE_FOR_altivec_vpku[hw]um_direct.
1101 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1102 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1103 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1104 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1106 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1108 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1109 generation for -maltivec=be.
1110 (altivec_vsumsws): Simplify redundant test.
1112 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1114 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1115 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1116 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1117 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1118 gen_altivec_vpkuwum.
1119 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1121 (altivec_vpks<VI_char>ss): Likewise.
1122 (altivec_vpks<VI_char>us): Likewise.
1123 (altivec_vpku<VI_char>us): Likewise.
1124 (altivec_vpku<VI_char>um): Likewise.
1125 (altivec_vpku<VI_char>um_direct): New (copy of
1126 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1128 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1129 target is little endian and -maltivec=be is not specified.
1130 (*altivec_vupkhs<VU_char>_direct): New (copy of
1131 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1132 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1133 target is little endian and -maltivec=be is not specified.
1134 (*altivec_vupkls<VU_char>_direct): New (copy of
1135 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1136 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1137 little endian and -maltivec=be is not specified.
1138 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1139 little endian and -maltivec=be is not specified.
1141 2014-02-05 Richard Henderson <rth@redhat.com>
1144 * combine-stack-adj.c: Revert r206943.
1145 * sched-int.h (struct deps_desc): Add last_args_size.
1146 * sched-deps.c (init_deps): Initialize it.
1147 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1148 contain REG_ARGS_SIZE notes.
1150 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1152 * lto-cgraph.c (asm_nodes_output): Make global.
1153 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1154 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1155 (driver_handle_option): Handle OPT_fwpa.
1157 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1160 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1161 a comment typo and formatting issue. If odr_hash hasn't been
1162 created, return vNULL and set *completep to false.
1165 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1166 bb with no successors.
1168 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1171 * doc/invoke.texi (-march): Clarify documentation for ARM.
1175 2014-02-05 Richard Biener <rguenther@suse.de>
1177 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1178 when not vectorizing because of too many alias checks.
1179 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1180 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1182 2014-02-05 Nick Clifton <nickc@redhat.com>
1184 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1185 accept extended registers in any mode when compiling for the MN10300.
1187 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1189 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1190 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1191 sanitization attributes.
1192 (can_inline_edge_p): Likewise.
1193 (sanitize_attrs_match_for_inline_p): New function.
1195 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1197 * ipa-prop.c (detect_type_change): Shor circuit testing of
1198 type changes on THIS pointer.
1200 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1203 * config/pa/pa.c (legitimize_tls_address): Return original address
1204 if not passed a SYMBOL_REF rtx.
1205 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1207 (pa_emit_move_sequence): Simplify TLS source operands.
1208 (pa_legitimate_constant_p): Reject all TLS constants.
1209 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1210 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1212 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1214 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1215 groups when we know they are controlled by LTO.
1216 * varasm.c (default_binds_local_p_1): If object is in other partition,
1217 it will be resolved locally.
1219 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1221 * config/host-linux.c (linux_gt_pch_use_address): Don't
1222 use SSIZE_MAX because it is not always defined.
1224 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1227 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1228 threshold for pseudo splitting.
1229 (update_ebb_live_info): Process call argument hard registers and
1230 hard registers from insn definition too.
1231 (max_small_class_regs_num): New constant.
1232 (inherit_in_ebb): Update live hard regs through EBBs. Update
1233 reloads_num only for small register classes. Don't split for
1236 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1239 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1242 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1244 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1247 2014-02-04 Marek Polacek <polacek@redhat.com>
1249 * gdbinit.in (pel): Define.
1251 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1253 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1256 2014-02-04 Richard Biener <rguenther@suse.de>
1259 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1260 in function context local.
1261 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1262 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1263 similar to LTO_imported_decl_ref.
1265 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1267 PR tree-optimization/60002
1268 * cgraphclones.c (build_function_decl_skip_args): Clear
1271 PR tree-optimization/60023
1272 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1273 false to gsi_replace.
1274 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1275 has been in some EH region and vec_stmt could throw, add
1276 vec_stmt into the same EH region.
1277 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1278 has no lhs, ignore it.
1279 * internal-fn.c (expand_MASK_LOAD): Likewise.
1282 * tree-inline.c (copy_forbidden): Fail for
1283 __attribute__((optimize (0))) functions.
1286 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1287 is set, copy one less argument.
1288 (expand_simd_clones): Don't subtract clone_info->inbranch
1289 from simd_clone_struct_alloc argument.
1291 PR rtl-optimization/57915
1292 * recog.c (simplify_while_replacing): If all unary/binary/relational
1293 operation arguments are constant, attempt to simplify those.
1296 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1297 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1299 2014-02-04 Richard Biener <rguenther@suse.de>
1301 PR tree-optimization/60012
1302 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1303 TBAA disambiguation to all DDRs.
1305 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1308 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1309 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1311 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1314 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1316 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1318 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1319 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1321 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1324 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1325 to figure out targets of polymorphic calls with known decl.
1326 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1327 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1328 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1329 (get_polymorphic_call_info): ... here.
1330 (get_polymorphic_call_info_from_invariant): New function.
1332 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1334 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1335 lookup via vtable pointer; check for type consistency
1336 and turn inconsitent facts into UNREACHABLE.
1337 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1338 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1339 type inconsistent querries; return UNREACHABLE instead.
1341 2014-02-03 Richard Henderson <rth@twiddle.net>
1344 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1345 already processed this node.
1346 (normalize_one_pred_1): Pass along mark_set.
1347 (normalize_one_pred): Create and destroy a pointer_set_t.
1348 (normalize_one_pred_chain): Likewise.
1350 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1352 PR gcov-profile/58602
1353 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1355 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1358 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1359 -fno-devirtualize; try to devirtualize by the knowledge of
1360 virtual table pointer given by aggregate propagation.
1361 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1362 (ipa_print_node_jump_functions): Dump also offset that
1363 is relevant for polymorphic calls.
1364 (determine_known_aggregate_parts): Add arg_type parameter; use it
1365 instead of determining the type from pointer type.
1366 (ipa_compute_jump_functions_for_edge): Update call of
1367 determine_known_aggregate_parts.
1368 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1369 (gimple_get_virt_method_for_binfo): ... here; simplify using
1370 vtable_pointer_value_to_vtable.
1371 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1372 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1373 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1374 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1376 (vtable_pointer_value_to_binfo): ... here.
1377 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1379 2014-02-03 Teresa Johnson <tejohnson@google.com>
1381 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1382 redef of outer loop index variable.
1384 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1388 * doc/extend.texi (Function Attributes): Typo.
1390 2014-02-03 Cong Hou <congh@google.com>
1392 PR tree-optimization/60000
1393 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1394 if the vectorized statement is a store. A store statement can only
1395 appear at the end of pattern statements.
1397 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1399 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1400 (ix86_option_override_internal): Default long double to 64-bit for
1401 32-bit Bionic and to 128-bit for 64-bit Bionic.
1403 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1404 TARGET_LONG_DOUBLE_128 is true.
1405 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1407 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1408 (mlong-double-64): Negate -mlong-double-128.
1409 (mlong-double-128): New option.
1411 * config/i386/i386-c.c (ix86_target_macros): Define
1412 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1414 * doc/invoke.texi: Document -mlong-double-128.
1416 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1418 PR rtl-optimization/60024
1419 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1421 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1423 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1425 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1427 PR rtl-optimization/57662
1428 * sel-sched.c (code_motion_path_driver): Do not mark already not
1429 existing blocks in the visiting bitmap.
1431 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1433 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1434 on the insn being emitted.
1436 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1437 Will Deacon <will.deacon@arm.com>
1439 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1441 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1443 * config/arm/arm-tables.opt: Regenerate.
1445 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1447 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1448 for vector types other than V16QImode.
1449 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1450 define_expand, and call altivec_expand_vec_perm_le when producing
1451 code with little endian element order.
1452 (*altivec_vperm_<mode>_internal): New insn having previous
1453 behavior of altivec_vperm_<mode>.
1454 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1455 altivec_expand_vec_perm_le when producing code with little endian
1457 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1458 behavior of altivec_vperm_<mode>_uns.
1460 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1462 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1463 (altivec_vsumsws): Add handling for -maltivec=be with a little
1465 (altivec_vsumsws_direct): New.
1466 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1467 gen_altivec_vsumsws.
1469 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1471 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1472 vtable_pointer_value_to_binfo): New functions.
1473 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1474 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1476 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1478 * config/nios2/nios2.md (load_got_register): Initialize GOT
1479 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1480 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1482 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1484 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1485 preserverd by passthrough, do not propagate the type.
1487 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1489 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1490 (mips_atomic_assign_expand_fenv): New function.
1491 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1493 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1495 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1496 (__builtin_mips_set_fcsr): Likewise.
1497 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1498 MIPS_USI_FTYPE_VOID.
1499 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1500 (mips16_expand_set_fcsr): Likewise.
1501 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1502 (mips16_set_fcsr_stub): Likewise.
1503 (mips16_get_fcsr_one_only_stub): New class.
1504 (mips16_set_fcsr_one_only_stub): Likewise.
1505 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1506 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1507 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1508 (hard_float): New availability predicate.
1509 (mips_builtins): Add get_fcsr and set_fcsr.
1510 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1511 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1512 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1513 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1514 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1517 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1519 * config/mips/mips.c (mips_one_only_stub): New class.
1520 (mips_need_mips16_rdhwr_p): Replace with...
1521 (mips16_rdhwr_stub): ...this new variable.
1522 (mips16_stub_call_address): New function.
1523 (mips16_rdhwr_one_only_stub): New class.
1524 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1525 (mips_output_mips16_rdhwr): Delete.
1526 (mips_finish_stub): New function.
1527 (mips_code_end): Use it to handle rdhwr stubs.
1529 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1532 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1533 when calculating size of integer atomic types.
1535 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1537 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1539 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1541 PR tree-optimization/60003
1542 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1543 * profile.c (branch_prob): Use gimple_call_builtin_p
1544 to check for BUILT_IN_SETJMP_RECEIVER.
1545 * tree-inline.c (copy_bb): Call notice_special_calls.
1547 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1550 * lra-constraints.c (process_alt_operands): Update reload_sum only
1553 2014-01-31 Richard Henderson <rth@redhat.com>
1556 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1557 until after else_eh is processed.
1559 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1561 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1562 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1563 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1564 in smmintrin.h, remove them.
1565 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1566 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1567 * config/i386/i386.md (ROUND_SAE): Fix value.
1568 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1569 (const48_operand): New.
1570 * config/i386/subst.md (round), (round_expand): Use
1571 const_4_or_8_to_11_operand.
1572 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1574 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1576 * config/i386/constraints.md (Yk): Swap meaning with k.
1577 * config/i386/i386.md (movhi_internal): Change Yk to k.
1578 (movqi_internal): Ditto.
1579 (*k<logic><mode>): Ditto.
1582 (kandn<mode>): Ditto.
1583 (*<code>hi_1): Ditto.
1584 (*<code>qi_1): Ditto.
1585 (kxnor<mode>): Ditto.
1586 (kortestzhi): Ditto.
1587 (kortestchi): Ditto.
1589 (*one_cmplhi2_1): Ditto.
1590 (*one_cmplqi2_1): Ditto.
1591 * config/i386/sse.md (): Change k to Yk.
1592 (avx512f_load<mode>_mask): Ditto.
1593 (avx512f_blendm<mode>): Ditto.
1594 (avx512f_store<mode>_mask): Ditto.
1595 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1596 (avx512f_storedqu<mode>_mask): Ditto.
1597 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1599 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1600 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1601 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1602 (avx512f_maskcmp<mode>3): Ditto.
1603 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1604 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1605 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1606 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1607 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1608 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1609 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1610 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1611 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1612 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1613 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1614 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1615 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1616 (vec_extract_lo_<mode>_maskm): Ditto.
1617 (vec_extract_hi_<mode>_maskm): Ditto.
1618 (avx512f_vternlog<mode>_mask): Ditto.
1619 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1620 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1621 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1622 (avx512f_<code>v8div16qi2_mask): Ditto.
1623 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1624 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1625 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1626 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1627 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1628 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1629 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1630 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1631 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1632 (avx512cd_maskb_vec_dupv8di): Ditto.
1633 (avx512cd_maskw_vec_dupv16si): Ditto.
1634 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1635 (avx512f_vpermi2var<mode>3_mask): Ditto.
1636 (avx512f_vpermi2var<mode>3_mask): Ditto.
1637 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1638 (*avx512f_gathersi<mode>): Ditto.
1639 (*avx512f_gathersi<mode>_2): Ditto.
1640 (*avx512f_gatherdi<mode>): Ditto.
1641 (*avx512f_gatherdi<mode>_2): Ditto.
1642 (*avx512f_scattersi<mode>): Ditto.
1643 (*avx512f_scatterdi<mode>): Ditto.
1644 (avx512f_compress<mode>_mask): Ditto.
1645 (avx512f_compressstore<mode>_mask): Ditto.
1646 (avx512f_expand<mode>_mask): Ditto.
1647 * config/i386/subst.md (mask): Change k to Yk.
1648 (mask_scalar_merge): Ditto.
1651 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1653 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1655 2014-01-31 Richard Biener <rguenther@suse.de>
1658 * builtins.c (fold_builtin_memory_op): Make sure to not
1659 use a floating-point mode or a boolean or enumeral type for
1662 2014-01-30 DJ Delorie <dj@redhat.com>
1664 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1665 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1666 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1667 whenever main() has an epilogue.
1669 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1671 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1672 unused variable "field".
1673 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1674 (vsx_mergeh_<mode>): Likewise.
1675 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1676 (altivec_vmrghh): Likewise.
1677 (altivec_vmrghw): Likewise.
1678 (altivec_vmrglb): Likewise.
1679 (altivec_vmrglh): Likewise.
1680 (altivec_vmrglw): Likewise.
1681 (altivec_vspltb): Add missing uses.
1682 (altivec_vsplth): Likewise.
1683 (altivec_vspltw): Likewise.
1684 (altivec_vspltsf): Likewise.
1686 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1689 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1690 frame related instructions.
1692 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1694 PR rtl-optimization/59959
1695 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1696 any reload of register whose subreg is invalid.
1698 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1700 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1701 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1702 Add missing return type - void.
1704 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1706 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1707 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1708 remove element index adjustment for endian (now handled in vsx.md
1710 (altivec_expand_vec_perm_const): Use
1711 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1712 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1713 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1714 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1715 define_expand and a new define_insn *altivec_vspltb_internal;
1716 adjust for -maltivec=be on a little endian target.
1717 (altivec_vspltb_direct): New.
1718 (altivec_vsplth): Divide into a define_expand and a new
1719 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1720 little endian target.
1721 (altivec_vsplth_direct): New.
1722 (altivec_vspltw): Divide into a define_expand and a new
1723 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1724 little endian target.
1725 (altivec_vspltw_direct): New.
1726 (altivec_vspltsf): Divide into a define_expand and a new
1727 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1728 a little endian target.
1730 2014-01-30 Richard Biener <rguenther@suse.de>
1732 PR tree-optimization/59993
1733 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1734 can propagate form the earlier stmt and avoid the transform
1735 when the intermediate result is needed.
1737 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1739 * README.Portability: Fix typo.
1741 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1743 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1744 comparison_operator with ordered_comparison_operator.
1746 2014-01-30 Nick Clifton <nickc@redhat.com>
1748 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1749 Rename to mn10300_store_multiple_regs.
1750 * config/mn10300/mn10300.c: Likewise.
1751 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1752 store_multiple_regs.
1753 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1754 Call mn10300_store_multiple_regs.
1756 2014-01-30 Nick Clifton <nickc@redhat.com>
1757 DJ Delorie <dj@redhat.com>
1759 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1760 %fp 2 to keep registers after it properly word-aligned.
1761 (rl78_alloc_physical_registers_umul): Handle the case where both
1762 input operands are the same.
1764 2014-01-30 Richard Biener <rguenther@suse.de>
1766 PR tree-optimization/59903
1767 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1770 2014-01-30 Jason Merrill <jason@redhat.com>
1773 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1776 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1778 2014-01-30 Richard Biener <rguenther@suse.de>
1780 PR tree-optimization/59951
1781 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1783 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1786 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1787 SFmode to DFmode case.
1789 2014-01-29 DJ Delorie <dj@redhat.com>
1791 * config/msp430/msp430.opt (-minrt): New.
1792 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1794 (ENDFILE_SPEC): Likewise.
1796 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1798 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1799 (estimate_function_body_sizes): Use it.
1801 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1804 * dwarf2out.c (is_cxx_auto): New.
1805 (is_base_type): Use it.
1806 (gen_type_die_with_usage): Likewise.
1808 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1810 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1811 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1812 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1813 -maltivec=be with LE targets.
1814 (vsx_mergeh_<mode>): Likewise.
1815 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1816 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1817 (altivec_vmrghb): Replace with define_expand and new
1818 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1819 (altivec_vmrghb_direct): New define_insn.
1820 (altivec_vmrghh): Replace with define_expand and new
1821 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1822 (altivec_vmrghh_direct): New define_insn.
1823 (altivec_vmrghw): Replace with define_expand and new
1824 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1825 (altivec_vmrghw_direct): New define_insn.
1826 (*altivec_vmrghsf): Adjust for endianness.
1827 (altivec_vmrglb): Replace with define_expand and new
1828 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1829 (altivec_vmrglb_direct): New define_insn.
1830 (altivec_vmrglh): Replace with define_expand and new
1831 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1832 (altivec_vmrglh_direct): New define_insn.
1833 (altivec_vmrglw): Replace with define_expand and new
1834 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1835 (altivec_vmrglw_direct): New define_insn.
1836 (*altivec_vmrglsf): Adjust for endianness.
1837 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1838 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1839 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1840 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1841 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1842 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1843 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1844 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1846 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1848 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1849 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1852 2014-01-29 Richard Biener <rguenther@suse.de>
1854 PR tree-optimization/58742
1855 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1856 associate_pointerplus_align.
1857 (associate_pointerplus_diff): New function.
1858 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1859 and associate_pointerplus_diff.
1861 2014-01-29 Richard Biener <rguenther@suse.de>
1863 * lto-streamer.h (LTO_major_version): Bump to 3.
1864 (LTO_minor_version): Reset to 0.
1866 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1868 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1869 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1870 (arm_file_start): Generate correct asm header for armv7ve.
1871 * config/arm/bpabi.h: Add multilib support for armv7ve.
1872 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1873 and cortex-a15 to armv7ve.
1874 * config/arm/t-aprofile: Add multilib support for armv7ve.
1875 * doc/invoke.texi: Document -march=armv7ve.
1877 2014-01-29 Richard Biener <rguenther@suse.de>
1879 PR tree-optimization/58742
1880 * tree-ssa-forwprop.c (associate_plusminus): Return true
1881 if we changed sth, defer EH cleanup to ...
1882 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1883 (simplify_mult): New function.
1885 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1888 PR tree-optimization/59920
1889 * tree.c (build_common_builtin_nodes): Remove
1890 __builtin_setjmp_dispatcher initialization.
1891 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1892 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1893 instead of gsi_after_labels + manually skipping debug stmts.
1894 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1895 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1896 * tree-inline.c (copy_edges_for_bb): Remove
1897 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1898 argument. Ignore computed_goto_p stmts. Don't call
1899 make_abnormal_goto_edges. If a call might need abnormal edges
1900 for non-local gotos, see if it already has an edge to
1901 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1902 with true argument, don't do anything then, otherwise add
1903 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1904 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1906 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1907 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1908 (lower_stmt): Don't set data->calls_builtin_setjmp.
1909 (lower_builtin_setjmp): Adjust comment.
1910 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1911 * tree-cfg.c (found_computed_goto): Remove.
1912 (factor_computed_gotos): Remove.
1913 (make_goto_expr_edges): Return bool, true for computed gotos.
1914 Don't call make_abnormal_goto_edges.
1915 (build_gimple_cfg): Don't set found_computed_goto, don't call
1916 factor_computed_gotos.
1917 (computed_goto_p): No longer static.
1918 (make_blocks): Don't set found_computed_goto.
1919 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1920 (make_edges): If make_goto_expr_edges returns true, push bb
1921 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1922 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1923 vector. Record mapping between bbs and OpenMP regions if there
1924 are any, adjust make_gimple_omp_edges caller. Call
1925 handle_abnormal_edges.
1926 (make_abnormal_goto_edges): Remove.
1927 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1928 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1929 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1930 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1931 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1932 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1933 filling *region also set *region_idx to (*region)->entry->index.
1936 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1937 For REGs set ORIGINAL_REGNO.
1939 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1941 * doc/md.texi: Mention that a target shouldn't implement
1942 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1945 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1947 PR tree-optimization/59594
1948 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1949 a copy of the datarefs vector rather than the vector itself.
1951 2014-01-28 Jason Merrill <jason@redhat.com>
1954 * dwarf2out.c (auto_die): New static.
1955 (gen_type_die_with_usage): Handle C++1y 'auto'.
1956 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1959 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1962 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1963 (SPEC_X32): Likewise.
1964 (SPEC_64): Likewise.
1965 * config/i386/i386.c (ix86_option_override_internal): Turn off
1966 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1968 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1969 * config/i386/i386.h (TARGET_16BIT): New macro.
1970 (TARGET_16BIT_P): Likewise.
1971 * config/i386/i386.opt: Add m16.
1972 * doc/invoke.texi: Document -m16.
1974 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1976 PR preprocessor/59935
1977 * input.c (location_get_source_line): Bail out on when line number
1978 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1980 2014-01-28 Richard Biener <rguenther@suse.de>
1982 PR tree-optimization/58742
1983 * tree-ssa-forwprop.c (associate_plusminus): Handle
1984 pointer subtraction of the form (T)(P + A) - (T)P.
1986 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1988 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1991 2014-01-28 Richard Biener <rguenther@suse.de>
1994 2014-01-28 Richard Biener <rguenther@suse.de>
1996 PR rtl-optimization/45364
1997 PR rtl-optimization/59890
1998 * var-tracking.c (local_get_addr_clear_given_value): Handle
1999 already cleared slot.
2000 (val_reset): Handle not allocated local_get_addr_cache.
2001 (vt_find_locations): Use post-order on the inverted CFG.
2003 2014-01-28 Richard Biener <rguenther@suse.de>
2005 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2007 2014-01-28 Richard Biener <rguenther@suse.de>
2009 PR rtl-optimization/45364
2010 PR rtl-optimization/59890
2011 * var-tracking.c (local_get_addr_clear_given_value): Handle
2012 already cleared slot.
2013 (val_reset): Handle not allocated local_get_addr_cache.
2014 (vt_find_locations): Use post-order on the inverted CFG.
2016 2014-01-28 Alan Modra <amodra@gmail.com>
2018 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2019 * configure.ac <recursive call for build != host>: Define
2020 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2021 and LD_FOR_BUILD too.
2022 * configure: Regenerate.
2024 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2026 * config/i386/i386.c (get_builtin_code_for_version): Separate
2027 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2028 Broadwell from Haswell.
2030 2014-01-27 Steve Ellcey <sellcey@mips.com>
2032 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2033 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2034 * config/mips/mips.c (mips_option_override): Change setting
2036 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2037 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2038 Change from Mask to Var.
2040 2014-01-27 Jeff Law <law@redhat.com>
2042 * ipa-inline.c (inline_small_functions): Fix typo.
2044 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2046 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2047 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2048 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2049 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2050 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2051 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2052 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2053 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2054 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2055 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2056 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2057 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2058 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2059 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2060 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2061 (_mm512_storeu_epi64): Ditto.
2062 (_mm512_cmpge_epi32_mask): Ditto.
2063 (_mm512_cmpge_epu32_mask): Ditto.
2064 (_mm512_cmpge_epi64_mask): Ditto.
2065 (_mm512_cmpge_epu64_mask): Ditto.
2066 (_mm512_cmple_epi32_mask): Ditto.
2067 (_mm512_cmple_epu32_mask): Ditto.
2068 (_mm512_cmple_epi64_mask): Ditto.
2069 (_mm512_cmple_epu64_mask): Ditto.
2070 (_mm512_cmplt_epi32_mask): Ditto.
2071 (_mm512_cmplt_epu32_mask): Ditto.
2072 (_mm512_cmplt_epi64_mask): Ditto.
2073 (_mm512_cmplt_epu64_mask): Ditto.
2074 (_mm512_cmpneq_epi32_mask): Ditto.
2075 (_mm512_cmpneq_epu32_mask): Ditto.
2076 (_mm512_cmpneq_epi64_mask): Ditto.
2077 (_mm512_cmpneq_epu64_mask): Ditto.
2078 (_mm512_expand_pd): Ditto.
2079 (_mm512_expand_ps): Ditto.
2080 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2081 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2082 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2083 * config/i386/i386.c (ix86_builtins): Add
2084 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2085 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2086 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2087 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2088 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2089 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2090 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2091 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2092 IX86_BUILTIN_PMOVUSQW512_MEM.
2093 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2094 __builtin_ia32_pmovsqd512mem_mask,
2095 __builtin_ia32_pmovqd512mem_mask,
2096 __builtin_ia32_pmovusqw512mem_mask,
2097 __builtin_ia32_pmovsqw512mem_mask,
2098 __builtin_ia32_pmovqw512mem_mask,
2099 __builtin_ia32_pmovusdw512mem_mask,
2100 __builtin_ia32_pmovsdw512mem_mask,
2101 __builtin_ia32_pmovdw512mem_mask,
2102 __builtin_ia32_pmovqb512mem_mask,
2103 __builtin_ia32_pmovusqb512mem_mask,
2104 __builtin_ia32_pmovsqb512mem_mask,
2105 __builtin_ia32_pmovusdb512mem_mask,
2106 __builtin_ia32_pmovsdb512mem_mask,
2107 __builtin_ia32_pmovdb512mem_mask.
2108 (bdesc_args): Add __builtin_ia32_expanddf512,
2109 __builtin_ia32_expandsf512.
2110 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2111 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2112 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2113 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2114 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2115 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2116 (avx512f_<code>v8div16qi2_mask_store): This.
2117 (avx512f_expand<mode>): New.
2119 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2121 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2123 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2124 (_mm512_prefetch_i32scatter_pd): Ditto.
2125 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2126 (_mm512_prefetch_i64scatter_pd): Ditto.
2127 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2128 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2129 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2130 (_mm512_prefetch_i32scatter_ps): Ditto.
2131 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2132 (_mm512_prefetch_i64scatter_ps): Ditto.
2133 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2134 * config/i386/i386-builtin-types.def: Define
2135 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2136 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2137 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2138 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2139 IX86_BUILTIN_SCATTERPFQPD.
2140 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2141 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2142 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2143 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2144 __builtin_ia32_scatterpfqps.
2145 (ix86_expand_builtin): Expand new built-ins.
2146 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2147 fix memory access data type.
2148 (*avx512pf_gatherpf<mode>_mask): Ditto.
2149 (*avx512pf_gatherpf<mode>): Ditto.
2150 (avx512pf_scatterpf<mode>): Ditto.
2151 (*avx512pf_scatterpf<mode>_mask): Ditto.
2152 (*avx512pf_scatterpf<mode>): Ditto.
2153 (GATHER_SCATTER_SF_MEM_MODE): New.
2154 (avx512pf_gatherpf<mode>df): Ditto.
2155 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2156 (*avx512pf_scatterpf<mode>df): Ditto.
2158 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2161 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2162 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2165 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2167 * common/config/arm/arm-common.c
2168 (arm_rewrite_mcpu): Handle multiple names.
2170 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2172 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2174 * gimple-builder.h (create_gimple_tmp): Delete.
2176 2014-01-27 Christian Bruel <christian.bruel@st.com>
2178 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2181 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2183 * config/pa/pa.md (call): Generate indirect long calls to non-local
2184 functions when outputing 32-bit code.
2185 (call_value): Likewise except for special call to buggy powf function.
2187 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2188 portable runtime and PIC indirect calls.
2189 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2190 and PIC call sequences. Use ldo instead of blr to set return register
2191 in PIC call sequence.
2193 2014-01-25 Walter Lee <walt@tilera.com>
2195 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2196 avoid clobbering a live register.
2198 2014-01-25 Walter Lee <walt@tilera.com>
2200 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2201 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2202 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2203 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2205 2014-01-25 Walter Lee <walt@tilera.com>
2207 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2208 arguments on even registers.
2209 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2211 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2212 (BIGGEST_ALIGNMENT): Ditto.
2213 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2215 2014-01-25 Walter Lee <walt@tilera.com>
2217 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2218 insns before bundling.
2219 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2221 2014-01-25 Walter Lee <walt@tilera.com>
2223 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2224 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2225 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2227 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2229 * config/mips/constraints.md (kl): Delete.
2230 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2231 define expands, using...
2232 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2233 instructions for MIPS16.
2234 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2235 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2237 2014-01-25 Walter Lee <walt@tilera.com>
2239 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2243 2014-01-25 Walter Lee <walt@tilera.com>
2245 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2246 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2248 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2250 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2253 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2255 * print-rtl.c (in_call_function_usage): New var.
2256 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2257 EXPR_LIST mode as mode and not as reg note name.
2260 * cfgloopmanip.c (copy_loop_info): If
2261 loop->warned_aggressive_loop_optimizations, make sure
2262 the flag is set in target loop too.
2264 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2266 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2268 * builtins.def: Likewise.
2269 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2270 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2271 * ira.c (ira_setup_eliminable_regset): Likewise.
2272 * omp-low.c (gate_expand_omp): Likewise.
2273 (execute_lower_omp): Likewise.
2274 (diagnose_sb_0): Likewise.
2275 (gate_diagnose_omp_blocks): Likewise.
2276 (simd_clone_clauses_extract): Likewise.
2279 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2281 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2282 correction for little endian...
2283 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2286 2014-01-24 Jeff Law <law@redhat.com>
2288 PR tree-optimization/59919
2289 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2290 for non-returning calls.
2292 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2294 * common/config/aarch64/aarch64-common.c
2295 (aarch64_rewrite_mcpu): Handle multiple names.
2296 * config/aarch64/aarch64.h
2297 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2299 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2301 * input.c (add_file_to_cache_tab): Handle the case where fopen
2304 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2307 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2308 from push operand if code of push isn't PRE_DEC.
2310 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2313 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2314 -mquad-memory-atomic. Update -mquad-memory documentation to say
2315 it is only used for non-atomic loads/stores.
2317 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2318 -mquad-memory or -mquad-memory-atomic switches.
2320 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2321 -mquad-memory-atomic to ISA 2.07 support.
2323 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2324 to separate support of normal quad word memory operations (ldq, stq)
2325 from the atomic quad word memory operations.
2327 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2328 support to separate non-atomic quad word operations from atomic
2329 quad word operations. Disable non-atomic quad word operations in
2330 little endian mode so that we don't have to swap words after the
2331 load and before the store.
2332 (quad_load_store_p): Add comment about atomic quad word support.
2333 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2334 options printed with -mdebug=reg.
2336 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2337 -mquad-memory-atomic as the test for whether we have quad word
2338 atomic instructions.
2339 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2340 or -mp8-vector are used, allow byte/half-word atomic operations.
2342 * config/rs6000/sync.md (load_lockedti): Insure that the address
2343 is a proper indexed or indirect address for the lqarx instruction.
2344 On little endian systems, swap the hi/lo registers after the lqarx
2346 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2347 insure the address is valid for the lqarx instruction.
2348 (store_conditionalti): Insure that the address is a proper indexed
2349 or indirect address for the stqcrx. instruction. On little endian
2350 systems, swap the hi/lo registers before doing the stqcrx.
2352 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2353 insure the address is valid for the stqcrx. instruction.
2355 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2356 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2357 type of quad memory support is available.
2359 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2362 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2363 there is a danger of looping.
2365 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2368 force flag_ira_loop_pressure if set via command line.
2370 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2372 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2373 (ashr_simd): New builtin handling DI mode.
2374 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2375 (aarch64_sshr_simddi): New match pattern.
2376 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2377 (vshrd_n_s64): Likewise.
2378 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2380 2014-01-23 Nick Clifton <nickc@redhat.com>
2382 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2383 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2384 favour of mcu specific scripts.
2385 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2388 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2389 Alex Velenko <Alex.Velenko@arm.com>
2391 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2392 (vaddv_s16): Likewise.
2393 (vaddv_s32): Likewise.
2394 (vaddv_u8): Likewise.
2395 (vaddv_u16): Likewise.
2396 (vaddv_u32): Likewise.
2397 (vaddvq_s8): Likewise.
2398 (vaddvq_s16): Likewise.
2399 (vaddvq_s32): Likewise.
2400 (vaddvq_s64): Likewise.
2401 (vaddvq_u8): Likewise.
2402 (vaddvq_u16): Likewise.
2403 (vaddvq_u32): Likewise.
2404 (vaddvq_u64): Likewise.
2405 (vaddv_f32): Likewise.
2406 (vaddvq_f32): Likewise.
2407 (vaddvq_f64): Likewise.
2408 (vmaxv_f32): Likewise.
2409 (vmaxv_s8): Likewise.
2410 (vmaxv_s16): Likewise.
2411 (vmaxv_s32): Likewise.
2412 (vmaxv_u8): Likewise.
2413 (vmaxv_u16): Likewise.
2414 (vmaxv_u32): Likewise.
2415 (vmaxvq_f32): Likewise.
2416 (vmaxvq_f64): Likewise.
2417 (vmaxvq_s8): Likewise.
2418 (vmaxvq_s16): Likewise.
2419 (vmaxvq_s32): Likewise.
2420 (vmaxvq_u8): Likewise.
2421 (vmaxvq_u16): Likewise.
2422 (vmaxvq_u32): Likewise.
2423 (vmaxnmv_f32): Likewise.
2424 (vmaxnmvq_f32): Likewise.
2425 (vmaxnmvq_f64): Likewise.
2426 (vminv_f32): Likewise.
2427 (vminv_s8): Likewise.
2428 (vminv_s16): Likewise.
2429 (vminv_s32): Likewise.
2430 (vminv_u8): Likewise.
2431 (vminv_u16): Likewise.
2432 (vminv_u32): Likewise.
2433 (vminvq_f32): Likewise.
2434 (vminvq_f64): Likewise.
2435 (vminvq_s8): Likewise.
2436 (vminvq_s16): Likewise.
2437 (vminvq_s32): Likewise.
2438 (vminvq_u8): Likewise.
2439 (vminvq_u16): Likewise.
2440 (vminvq_u32): Likewise.
2441 (vminnmv_f32): Likewise.
2442 (vminnmvq_f32): Likewise.
2443 (vminnmvq_f64): Likewise.
2445 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2447 * config/aarch64/aarch64-simd.md
2448 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2449 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2450 (*aarch64_mul3_elt<mode>): Likewise.
2451 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2452 (*aarch64_mul3_elt_to_64v2df): Likewise.
2453 (*aarch64_mla_elt<mode>): Likewise.
2454 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2455 (*aarch64_mls_elt<mode>): Likewise.
2456 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2457 (*aarch64_fma4_elt<mode>): Likewise.
2458 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2459 (*aarch64_fma4_elt_to_64v2df): Likewise.
2460 (*aarch64_fnma4_elt<mode>): Likewise.
2461 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2462 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2463 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2464 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2465 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2466 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2467 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2468 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2469 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2471 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2473 * config/aarch64/aarch64-simd.md
2474 (aarch64_be_checked_get_lane<mode>): New define_expand.
2475 * config/aarch64/aarch64-simd-builtins.def
2476 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2477 New builtin definition.
2478 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2479 Use new safe be builtin.
2481 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2483 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2485 (aarch64_be_st1<mode>): Likewise.
2486 (aarch_ld1<VALL:mode>): Define_expand modified.
2487 (aarch_st1<VALL:mode>): Likewise.
2488 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2489 (UNSPEC_ST1): Likewise.
2491 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2493 * config/microblaze/microblaze.md: Add trap insn and attribute
2495 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2497 PR preprocessor/58580
2498 * input.h (location_get_source_line): Take an additional line_size
2500 (void diagnostics_file_cache_fini): Declare new function.
2501 * input.c (struct fcache): New type.
2502 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2503 New static constants.
2504 (diagnostic_file_cache_init, total_lines_num)
2505 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2506 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2507 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2508 (get_next_line, read_next_line, goto_next_line, read_line_num):
2509 New static function definitions.
2510 (diagnostic_file_cache_fini): New function.
2511 (location_get_source_line): Take an additional output line_len
2512 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2514 * diagnostic.c (diagnostic_finish): Call
2515 diagnostic_file_cache_fini.
2516 (adjust_line): Take an additional input parameter for the length
2517 of the line, rather than calculating it with strlen.
2518 (diagnostic_show_locus): Adjust the use of
2519 location_get_source_line and adjust_line with respect to their new
2520 signature. While displaying a line now, do not stop at the first
2521 null byte. Rather, display the zero byte as a space and keep
2522 going until we reach the size of the line.
2523 * Makefile.in: Add vec.o to OBJS-libcommon
2525 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2526 Ilya Tocar <ilya.tocar@intel.com>
2528 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2529 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2530 (__builtin_ia32_kmov16): Ditto.
2531 * config/i386/i386.md (UNSPEC_KMOV): New.
2534 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2536 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2537 (_mm512_storeu_si512): Ditto.
2539 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2542 * rtl.h (get_referenced_operands): Declare.
2543 * recog.c (get_referenced_operands): New function.
2544 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2545 operands have been referenced when recording LO_SUM references.
2547 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2549 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2551 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2553 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2554 Enable for generic and recent AMD targets.
2556 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2558 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2559 ARG_SIZE note when adjustment was eliminated.
2561 2014-01-22 Jeff Law <law@redhat.com>
2563 PR tree-optimization/59597
2564 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2565 in file. Accept new argument REGISTERING and use it to modify
2566 dump output appropriately.
2567 (register_jump_thread): Corresponding changes.
2568 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2569 thread paths involving joiner blocks. Add code to dump cancelled
2570 jump threading paths.
2572 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2574 PR rtl-optimization/59477
2575 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2576 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2578 2014-01-22 Tom Tromey <tromey@redhat.com>
2580 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2582 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2584 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2586 PR rtl-optimization/59896
2587 * lra-constraints.c (process_alt_operands): Check unused note for
2588 matched operands of insn with no output reloads.
2590 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2592 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2593 (mips_move_from_gpr_cost): Likewise.
2595 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2597 PR rtl-optimization/59858
2598 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2599 ira_class_hard_regs_num.
2600 (process_alt_operands): Increase reject for dying matched operand.
2602 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2605 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2606 smaller than size, perform several stores or loads and stores
2607 at dst + count - size to store or copy all of size bytes, rather
2608 than just last modesize bytes.
2610 2014-01-20 DJ Delorie <dj@redhat.com>
2612 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2613 that CLOBBERs are REGs before propogating their values.
2615 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2618 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2619 (cgraph_inline_failed_type): New function.
2620 * cgraph.h (DEFCIFCODE): Add type.
2621 (cgraph_inline_failed_type_t): New enum.
2622 (cgraph_inline_failed_type): New prototype.
2623 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2624 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2625 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2626 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2627 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2628 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2629 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2630 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2631 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2632 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2633 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2634 OPTIMIZATION_MISMATCH.
2635 * tree-inline.c (expand_call_inline): Emit errors during
2636 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2638 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2641 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2642 mode attribute in insn output.
2644 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2646 * output.h (output_constant): Delete.
2647 * varasm.c (output_constant): Make private.
2649 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2651 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2653 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2656 * tree.h (fold_builtin_strcat): New prototype.
2657 * builtins.c (fold_builtin_strcat): No longer static. Add len
2658 argument, if non-NULL, don't call c_strlen. Optimize
2659 directly into __builtin_memcpy instead of __builtin_strcpy.
2660 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2661 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2663 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2665 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2666 for SImode_address_operand operands, having only a REG argument.
2668 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2670 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2671 loader name using mbig-endian.
2672 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2674 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2676 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2680 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2682 * config/aarch64/aarch64-protos.h
2683 (aarch64_cannot_change_mode_class_ptr): Declare.
2684 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2685 aarch64_cannot_change_mode_class_ptr): New.
2686 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2687 backend hook aarch64_cannot_change_mode_class.
2689 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2691 * common/config/aarch64/aarch64-common.c
2692 (aarch64_handle_option): Don't handle any option order logic here.
2693 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2694 selected_cpu, warn on architecture version mismatch.
2695 (aarch64_override_options): Fix parsing order for option strings.
2697 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2698 Iain Sandoe <iain@codesourcery.com>
2701 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2702 warning. Amend comment to reflect current functionality.
2704 2014-01-20 Richard Biener <rguenther@suse.de>
2707 * builtins.c (fold_builtin_strcat): Remove case better handled
2708 by tree-ssa-strlen.c.
2710 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2712 * config/aarch64/aarch64.opt
2713 (mcpu, march, mtune): Make case-insensitive.
2715 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2718 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2719 if operands[1] is a REG or ZERO_EXTEND of a REG.
2721 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2723 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2725 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2727 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2728 long non-pic millicode calls.
2730 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2732 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2734 2014-01-19 Kito Cheng <kito@0xlab.org>
2736 * builtins.c (expand_movstr): Check movstr expand done or fail.
2738 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2739 H.J. Lu <hongjiu.lu@intel.com>
2742 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2743 to DImode for zero-extended addresses.
2745 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2747 PR rtl-optimization/57763
2748 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2749 on the new indirect jump_insn and increment LABEL_NUSES (label).
2751 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2755 * config.gcc (x86_archs): New variable.
2756 (x86_64_archs): Likewise.
2757 (x86_cpus): Likewise.
2758 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2759 --with-arch/--with-cpu= options.
2760 Support --with-arch=/--with-cpu={nehalem,westmere,
2761 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2763 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2765 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2766 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2768 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2770 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2772 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2775 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2776 clear cpp_get_options (parse_in)->warn_unused_macros for
2777 ix86_target_macros_internal with cpp_define.
2779 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2781 * jump.c (delete_related_insns): Keep (use (insn))s.
2782 * reorg.c (redundant_insn): Check for barriers too.
2784 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2786 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2788 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2790 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2791 call to $$dyncall when TARGET_LONG_CALLS is true.
2793 2014-01-17 Jeff Law <law@redhat.com>
2795 * ree.c (combine_set_extension): Temporarily disable test for
2796 changing number of hard registers.
2798 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2801 * ipa-inline-analysis.c (inline_free_summary):
2802 Do not free summary of aliases.
2804 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2807 * gimplify.c (gimplify_expr): Use create_tmp_var
2808 instead of create_tmp_var_raw. If cond doesn't have
2809 integral type, don't add the IFN_ANNOTATE builtin at all.
2811 2014-01-17 Martin Jambor <mjambor@suse.cz>
2814 * ipa-cp.c (prev_edge_clone): New variable.
2815 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2816 Also resize prev_edge_clone vector.
2817 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2818 (ipcp_edge_removal_hook): New function.
2819 (ipcp_driver): Register ipcp_edge_removal_hook.
2821 2014-01-17 Andrew Pinski <apinski@cavium.com>
2822 Steve Ellcey <sellcey@mips.com>
2825 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2828 2014-01-17 Jeff Law <law@redhat.com>
2831 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2832 so that pass_ccp runs first.
2834 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2836 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2837 (ix86_adjust_cost): Use !TARGET_XXX.
2838 (do_reorder_for_imul): Likewise.
2839 (swap_top_of_ready_list): Likewise.
2840 (ix86_sched_reorder): Likewise.
2842 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2844 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2845 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2846 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2847 (intel_memset): New. Duplicate slm_memset.
2848 (intel_cost): New. Duplicate slm_cost.
2849 (m_INTEL): New macro.
2850 (processor_target_table): Add "intel".
2851 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2852 with PROCESSOR_INTEL for "intel".
2853 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2854 PROCESSOR_SILVERMONT.
2855 (ix86_issue_rate): Likewise.
2856 (ix86_adjust_cost): Likewise.
2857 (ia32_multipass_dfa_lookahead): Likewise.
2858 (swap_top_of_ready_list): Likewise.
2859 (ix86_sched_reorder): Likewise.
2860 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2861 instead of TARGET_OPT_AGU.
2862 * config/i386/i386.h (TARGET_INTEL): New.
2863 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2864 (processor_type): Add PROCESSOR_INTEL.
2865 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2866 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2868 2014-01-17 Marek Polacek <polacek@redhat.com>
2871 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2874 2014-01-17 Richard Biener <rguenther@suse.de>
2876 PR tree-optimization/46590
2877 * opts.c (default_options_table): Add entries for
2878 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2879 all enabled at -O1 but not for -Og.
2880 * common.opt (fbranch-count-reg): Remove Init(1).
2881 (fmove-loop-invariants): Likewise.
2882 (ftree-pta): Likewise.
2884 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2886 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2887 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2888 decls to at least the GCC 4.8 used alignments.
2891 * tree-nested.c (convert_nonlocal_reference_stmt,
2892 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2893 of GIMPLE_BIND stmts, adjust associated decls.
2895 2014-01-17 Richard Biener <rguenther@suse.de>
2897 PR tree-optimization/46590
2898 * vec.h (vec<>::bseach): New member function implementing
2899 binary search according to C89 bsearch.
2900 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2901 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2902 bitmap pointer again. Make accesses_in_loop a flat array.
2903 (mem_ref_obstack): New global.
2904 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2905 (mark_ref_stored): Likewise.
2906 (ref_indep_loop_p_2): Likewise.
2907 (set_ref_stored_in_loop): New helper function.
2908 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2909 (memref_free): Adjust.
2910 (record_mem_ref_loc): Simplify.
2911 (gather_mem_refs_stmt): Adjust.
2912 (sort_locs_in_loop_postorder_cmp): New function.
2913 (analyze_memory_references): Sort accesses_in_loop after
2914 loop postorder number.
2915 (find_ref_loc_in_loop_cmp): New function.
2916 (for_all_locs_in_loop): Find relevant cluster of locs in
2917 accesses_in_loop and iterate without recursion.
2918 (execute_sm): Avoid uninit warning.
2919 (struct ref_always_accessed): Simplify.
2920 (ref_always_accessed::operator ()): Likewise.
2921 (ref_always_accessed_p): Likewise.
2922 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2923 loop postorder numbers here.
2924 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2927 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2930 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2931 on decls for which assemble_alias has been called.
2933 2014-01-17 Nick Clifton <nickc@redhat.com>
2935 * config/msp430/msp430.opt: (mcpu): New option.
2936 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2937 (msp430_option_override): Parse target_cpu. If the MCU name
2938 matches a generic string, clear target_mcu.
2939 (msp430_attr): Allow numeric interrupt values up to 63.
2940 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2941 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2943 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2945 * config/msp430/msp430.md (popm): Use %J rather than %I.
2946 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2947 (addhi_cy_i): Use immediate_operand for operand 2.
2948 * doc/invoke.texi: Document -mcpu option.
2950 2014-01-17 Richard Biener <rguenther@suse.de>
2952 PR rtl-optimization/38518
2953 * df.h (df_analyze_loop): Declare.
2954 * df-core.c: Include cfgloop.h.
2955 (df_analyze_1): Split out main part of df_analyze.
2956 (df_analyze): Adjust.
2957 (loop_inverted_post_order_compute): New function.
2958 (loop_post_order_compute): Likewise.
2959 (df_analyze_loop): New function avoiding whole-function
2961 * loop-invariant.c (find_defs): Use df_analyze_loop.
2962 (find_invariants): Adjust.
2963 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2965 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2967 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2968 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2970 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2972 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2973 traversal when removing references.
2975 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2978 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2980 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2983 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2984 pushing a reload for an autoinc when we had previously reloaded an
2985 inner part of the address.
2987 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2989 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2991 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2992 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2993 when not giving up or versioning for alias only because of
2995 (vect_analyze_data_ref_dependences): Set to true.
2996 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2998 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2999 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3003 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3006 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3007 operand 0 predicate for gathers, use a new pseudo as subtarget.
3009 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3012 * lra-constraints.c (process_alt_operands): Add printing debug info.
3013 Check absence of input/output reloads for matched operands too.
3015 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3017 PR rtl-optimization/59835
3018 * ira.c (ira_init_register_move_cost): Increase cost for
3021 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3023 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3025 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3028 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3029 non-register objects. Use gen_(high/low)part more consistently.
3032 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3035 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3036 endian support, remove tests for WORDS_BIG_ENDIAN.
3037 (p8_mfvsrd_3_<mode>): Likewise.
3038 (reload_gpr_from_vsx<mode>): Likewise.
3039 (reload_gpr_from_vsxsf): Likewise.
3040 (p8_mfvsrd_4_disf): Likewise.
3042 2014-01-16 Richard Biener <rguenther@suse.de>
3044 PR rtl-optimization/46590
3045 * lcm.c (compute_antinout_edge): Use postorder iteration.
3046 (compute_laterin): Use inverted postorder iteration.
3048 2014-01-16 Nick Clifton <nickc@redhat.com>
3051 * varasm.c (output_constant): Return the number of bytes actually
3053 (output_constructor_array_range): Update the field size with the
3054 number of bytes emitted by output_constant.
3055 (output_constructor_regular_field): Likewise. Also do not
3056 complain if the total number of bytes emitted is now greater
3057 than the expected fieldpos.
3058 * output.h (output_constant): Update prototype and descriptive comment.
3060 2014-01-16 Marek Polacek <polacek@redhat.com>
3063 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3064 it is error_mark_node.
3066 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3068 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3069 VALID_AVX256_REG_OR_OI_MODE.
3071 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3073 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3074 current procedure should be profiled.
3076 2014-01-15 Andrew Pinski <apinski@cavium.com>
3078 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3079 of moving from/to the STACK_REG register class.
3081 2014-01-15 Richard Henderson <rth@redhat.com>
3084 * reginfo.c (global_regs_decl): Globalize.
3085 * rtl.h (global_regs_decl): Declare.
3086 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3087 reserved via global_regs.
3089 2014-01-15 Teresa Johnson <tejohnson@google.com>
3091 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3093 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3095 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3096 and vmulosh rather than call gen_vec_widen_smult_*.
3097 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3098 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3099 (vec_widen_smult_even_v16qi): Likewise.
3100 (vec_widen_umult_even_v8hi): Likewise.
3101 (vec_widen_smult_even_v8hi): Likewise.
3102 (vec_widen_umult_odd_v16qi): Likewise.
3103 (vec_widen_smult_odd_v16qi): Likewise.
3104 (vec_widen_umult_odd_v8hi): Likewise.
3105 (vec_widen_smult_odd_v8hi): Likewise.
3106 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3107 vmuloub rather than call gen_vec_widen_umult_*.
3108 (vec_widen_umult_lo_v16qi): Likewise.
3109 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3110 vmulosb rather than call gen_vec_widen_smult_*.
3111 (vec_widen_smult_lo_v16qi): Likewise.
3112 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3113 rather than call gen_vec_widen_umult_*.
3114 (vec_widen_umult_lo_v8hi): Likewise.
3115 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3116 rather than call gen_vec_widen_smult_*.
3117 (vec_widen_smult_lo_v8hi): Likewise.
3119 2014-01-15 Jeff Law <law@redhat.com>
3121 PR tree-optimization/59747
3122 * ree.c (find_and_remove_re): Properly handle case where a second
3123 eliminated extension requires widening a copy created for elimination
3124 of a prior extension.
3125 (combine_set_extension): Ensure that the number of hard regs needed
3126 for a destination register does not change when we widen it.
3128 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3130 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3131 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3132 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3133 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3134 (avr-*-rtems*): Likewise.
3135 (bfin*-rtems*): Likewise.
3136 (moxie-*-rtems*): Likewise.
3137 (h8300-*-rtems*): Likewise.
3138 (i[34567]86-*-rtems*): Likewise.
3139 (lm32-*-rtems*): Likewise.
3140 (m32r-*-rtems*): Likewise.
3141 (m68k-*-rtems*): Likewise.
3142 (microblaze*-*-rtems*): Likewise.
3143 (mips*-*-rtems*): Likewise.
3144 (powerpc-*-rtems*): Likewise.
3145 (sh-*-rtems*): Likewise.
3146 (sparc-*-rtems*): Likewise.
3147 (sparc64-*-rtems*): Likewise.
3148 (v850-*-rtems*): Likewise.
3149 (m32c-*-rtems*): Likewise.
3151 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3153 PR rtl-optimization/59511
3154 * ira.c (ira_init_register_move_cost): Use memory costs for some
3155 cases of register move cost calculations.
3156 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3157 instead of BB frequency.
3158 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3159 * lra-assigns.c (find_hard_regno_for): Ditto.
3161 2014-01-15 Richard Biener <rguenther@suse.de>
3163 PR tree-optimization/59822
3164 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3165 (vectorizable_load): Use it to hoist defs of uses of invariant
3166 loads out of the loop.
3168 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3169 Kugan Vivekanandarajah <kuganv@linaro.org>
3172 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3175 2014-01-15 Richard Biener <rguenther@suse.de>
3177 PR rtl-optimization/59802
3178 * lcm.c (compute_available): Use inverted postorder to seed
3179 the initial worklist.
3181 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3184 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3185 ADDR_REGS for invalid symrefs in non-PIC code.
3187 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3190 * builtins.c (determine_block_size): Initialize *probable_max_size
3191 even if len_rtx is CONST_INT.
3193 2014-01-14 Andrew Pinski <apinski@cavium.com>
3195 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3196 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3197 (cortexa53_tunings): Likewise.
3198 (aarch64_sched_issue_rate): New function.
3199 (TARGET_SCHED_ISSUE_RATE): Define.
3201 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3203 * ira-costs.c (find_costs_and_classes): Add missed
3204 ira_init_register_move_cost_if_necessary.
3206 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3209 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3211 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3214 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3215 to indicate if type is used for function return value. Warn ABI
3216 change if the vector mode isn't available for function return value.
3217 (ix86_function_arg_advance): Pass false to type_natural_mode.
3218 (ix86_function_arg): Likewise.
3219 (ix86_gimplify_va_arg): Likewise.
3220 (function_arg_32): Don't warn ABI change.
3221 (ix86_function_value): Pass true to type_natural_mode.
3222 (ix86_return_in_memory): Likewise.
3223 (ix86_struct_value_rtx): Removed.
3224 (TARGET_STRUCT_VALUE_RTX): Likewise.
3226 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3228 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3229 converting a conditional jump into a conditional return.
3231 2014-01-14 Richard Biener <rguenther@suse.de>
3233 PR tree-optimization/58921
3234 PR tree-optimization/59006
3235 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3236 hoisting invariant stmts.
3237 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3238 invariant loads on the preheader edge if possible.
3240 2014-01-14 Joey Ye <joey.ye@arm.com>
3242 * doc/plugin.texi (Building GCC plugins): Update to C++.
3244 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3246 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3247 (_mm_rcp28_round_ss): Ditto.
3248 (_mm_rsqrt28_round_sd): Ditto.
3249 (_mm_rsqrt28_round_ss): Ditto.
3250 (_mm_rcp28_sd): Ditto.
3251 (_mm_rcp28_ss): Ditto.
3252 (_mm_rsqrt28_sd): Ditto.
3253 (_mm_rsqrt28_ss): Ditto.
3254 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3255 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3256 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3257 (IX86_BUILTIN_RCP28SD): Ditto.
3258 (IX86_BUILTIN_RCP28SS): Ditto.
3259 (IX86_BUILTIN_RSQRT28SD): Ditto.
3260 (IX86_BUILTIN_RSQRT28SS): Ditto.
3261 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3262 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3263 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3264 (ix86_expand_special_args_builtin): Expand new FTYPE.
3265 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3266 (srcp14<mode>): Make insn unary.
3267 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3268 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3269 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3270 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3271 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3272 Fix rounding: make it SAE only.
3273 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3275 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3277 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3278 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3279 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3280 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3281 (round_saeonly_mask_scalar_operand4): Ditto.
3282 (round_saeonly_mask_scalar_op3): Ditto.
3283 (round_saeonly_mask_scalar_op4): Ditto.
3285 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3287 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3288 Implement -maltivec=be for vec_insert and vec_extract.
3290 2014-01-10 DJ Delorie <dj@redhat.com>
3292 * config/msp430/msp430.md (call_internal): Don't allow memory
3293 references with SP as the base register.
3294 (call_value_internal): Likewise.
3295 * config/msp430/constraints.md (Yc): New. For memory references
3296 that don't use SP as a base register.
3298 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3299 "an integer without a # prefix"
3300 * config/msp430/msp430.md (epilogue_helper): Use it.
3302 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3305 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3306 AVX512F gather builtins.
3307 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3308 on gather decls with INTEGER_TYPE masktype.
3309 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3310 directly into the builtin rather than hoisting it before loop.
3312 PR tree-optimization/59387
3313 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3314 (scev_const_prop): If folded_casts and type has undefined overflow,
3315 use force_gimple_operand instead of force_gimple_operand_gsi and
3316 for each added stmt if it is assign with
3317 arith_code_with_undefined_signed_overflow, call
3318 rewrite_to_defined_overflow.
3319 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3320 gimple-fold.h instead.
3321 (arith_code_with_undefined_signed_overflow,
3322 rewrite_to_defined_overflow): Moved to ...
3323 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3324 rewrite_to_defined_overflow): ... here. No longer static.
3325 Include gimplify-me.h.
3326 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3327 rewrite_to_defined_overflow): New prototypes.
3329 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3331 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3333 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3335 * builtins.c (get_object_alignment_2): Minor tweak.
3336 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3338 2014-01-13 Christian Bruel <christian.bruel@st.com>
3340 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3341 optimized non constant lengths.
3343 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3346 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3347 load as __atomic_load_N if possible.
3349 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3351 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3353 (rs6000_expand_builtin): Adjust call.
3355 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3358 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3359 * config/rs6000/rs6000.c: Include target-globals.h.
3360 (rs6000_set_current_function): Instead of doing target_reinit
3361 unconditionally, use save_target_globals_default_opts and
3362 restore_target_globals.
3364 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3366 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3367 (rs6000_expand_builtin): Handle mffs and mtfsf.
3368 (rs6000_init_builtins): Define mffs and mtfsf.
3369 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3370 (rs6000_mffs): New pattern.
3371 (rs6000_mtfsf): New pattern.
3373 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3375 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3376 Start narrowing with START. Apply candidate-use pair
3377 and check overall cost in narrowing.
3378 (iv_ca_prune): Pass new argument.
3380 2014-01-10 Jeff Law <law@redhat.com>
3383 * ree.c (combine_reaching_defs): Ensure the defining statement
3384 occurs before the extension when optimizing extensions with
3385 different source and destination hard registers.
3387 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3390 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3391 vtables into the type inheritance graph.
3393 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3395 PR rtl-optimization/59754
3396 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3397 modes in the REGNO != REGNO case.
3399 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3401 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3403 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3405 PR tree-optimization/59745
3406 * tree-predcom.c (tree_predictive_commoning_loop): Call
3407 free_affine_expand_cache if giving up because components is NULL.
3409 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3410 GC in payload of target_globals struct instead of allocating them on
3411 the heap and the larger structs separately using GC.
3412 * target-globals.h (struct target_globals): Make regs, hard_regs,
3413 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3414 of GTY((skip)) and change type to void *.
3415 (reset_target_globals): Cast loads from those fields to corresponding
3418 2014-01-10 Steve Ellcey <sellcey@mips.com>
3421 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3422 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3423 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3425 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3428 * aarch64-modes.def (CC_Zmode): New flags mode.
3429 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3430 represents an equality.
3431 (aarch64_get_condition_code): Handle CC_Zmode.
3432 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3434 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3436 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3437 extraction in good case.
3439 2014-01-10 Richard Biener <rguenther@suse.de>
3441 PR tree-optimization/59374
3442 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3443 checking after SLP discovery. Mark stmts not participating
3444 in any SLP instance properly.
3446 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3448 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3449 when handling a SET rtx.
3451 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3453 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3454 (cortex-a57): Likewise.
3455 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3457 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3459 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3460 non-iwmmxt builtins.
3462 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3466 * ipa-devirt.c record_target_from_binfo): Take as argument
3467 stack of binfos and lookup matching one for virtual inheritance.
3468 (possible_polymorphic_call_targets_1): Update.
3470 2014-01-10 Huacai Chen <chenhc@lemote.com>
3472 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3473 kernel strings for Loongson-2E/2F/3A.
3475 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3478 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3479 is_gimple_call before calling gimple_call_internal_p.
3481 2014-01-09 Steve Ellcey <sellcey@mips.com>
3483 * Makefile.in (TREE_FLOW_H): Remove.
3484 (TREE_SSA_H): Add file names from tree-flow.h.
3485 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3486 * tree.h: Remove tree-flow.h reference.
3487 * hash-table.h: Remove tree-flow.h reference.
3488 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3489 reference with tree-ssa-loop.h.
3491 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3493 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3494 default element-order behavior for -maltivec.
3495 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3496 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3497 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3498 when targeting big endian, at least for now.
3499 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3501 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3504 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3505 var satisfies use_register_for_decl, just take into account type
3506 alignment, rather than decl alignment.
3508 PR tree-optimization/59622
3509 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3510 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3511 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3512 Don't devirtualize for inplace at all. For targets.length () == 1,
3513 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3515 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3517 * config/i386/i386.md (cpu): Remove the unused btver1.
3519 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3521 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3523 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3526 * tree-core.h (struct target_globals): New forward declaration.
3527 (struct tree_target_option): Add globals field.
3528 * tree.h (TREE_TARGET_GLOBALS): Define.
3529 (prepare_target_option_nodes_for_pch): New prototype.
3530 * target-globals.h (struct target_globals): Define even if
3532 * tree.c (prepare_target_option_node_for_pch,
3533 prepare_target_option_nodes_for_pch): New functions.
3534 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3535 * config/i386/i386.c: Include target-globals.h.
3536 (ix86_set_current_function): Instead of doing target_reinit
3537 unconditionally, use save_target_globals_default_opts and
3538 restore_target_globals.
3540 2014-01-09 Richard Biener <rguenther@suse.de>
3542 PR tree-optimization/59715
3543 * tree-cfg.h (split_critical_edges): Declare.
3544 * tree-cfg.c (split_critical_edges): Export.
3545 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3547 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3549 * cfgexpand.c (expand_stack_vars): Optionally disable
3550 asan stack protection.
3551 (expand_used_vars): Likewise.
3552 (partition_stack_vars): Likewise.
3553 * asan.c (asan_emit_stack_protection): Optionally disable
3554 after return stack usage.
3555 (instrument_derefs): Optionally disable memory access instrumentation.
3556 (instrument_builtin_call): Likewise.
3557 (instrument_strlen_call): Likewise.
3558 (asan_protect_global): Optionally disable global variables protection.
3559 * doc/invoke.texi: Added doc for new options.
3560 * params.def: Added new options.
3561 * params.h: Likewise.
3563 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3565 PR rtl-optimization/59724
3566 * ifcvt.c (cond_exec_process_if_block): Don't call
3567 flow_find_head_matching_sequence with 0 longest_match.
3568 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3569 non-active insns if !stop_after.
3570 (try_head_merge_bb): Revert 2014-01-07 changes.
3572 2014-01-08 Jeff Law <law@redhat.com>
3574 * ree.c (get_sub_rtx): New function, extracted from...
3575 (merge_def_and_ext): Here.
3576 (combine_reaching_defs): Use get_sub_rtx.
3578 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3580 * cgraph.h (varpool_variable_node): Do not choke on null node.
3582 2014-01-08 Catherine Moore <clm@codesourcery.com>
3584 * config/mips/mips.md (simple_return): Attempt to use JRC
3586 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3588 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3590 PR rtl-optimization/59137
3591 * reorg.c (steal_delay_list_from_target): Call update_block for
3593 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3595 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3597 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3598 two duplicate entries.
3600 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3603 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3605 * config/mips/mips.c (mips_truncated_op_cost): New function.
3606 (mips_rtx_costs): Adjust test for BADDU.
3607 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3609 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3611 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3612 (*baddu_si): ...this new pattern.
3614 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3617 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3619 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3622 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3624 (expand_expr, expand_normal): Adjust.
3625 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3626 inner_reference_p. Use inner_reference_p to expand inner references.
3627 (store_expr): Adjust.
3628 * cfgexpand.c (expand_call_stmt): Adjust.
3630 2014-01-08 Rong Xu <xur@google.com>
3632 * gcov-io.c (gcov_var): Move from gcov-io.h.
3633 (gcov_position): Ditto.
3634 (gcov_is_error): Ditto.
3635 (gcov_rewrite): Ditto.
3636 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3637 only part to libgcc/libgcov.h.
3639 2014-01-08 Marek Polacek <polacek@redhat.com>
3642 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3644 2014-01-08 Marek Polacek <polacek@redhat.com>
3647 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3649 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3651 PR rtl-optimization/59649
3652 * stor-layout.c (get_mode_bounds): For BImode return
3653 0 and STORE_FLAG_VALUE.
3655 2014-01-08 Richard Biener <rguenther@suse.de>
3658 * gimple.h (is_gimple_builtin_call): Remove.
3659 (gimple_builtin_call_types_compatible_p): New.
3660 (gimple_call_builtin_p): New overload.
3661 * gimple.c (is_gimple_builtin_call): Remove.
3662 (validate_call): Rename to ...
3663 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3665 (validate_type): New static function.
3666 (gimple_call_builtin_p): New overload and adjust.
3667 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3668 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3669 (gimple_fold_stmt_to_constant_1): Likewise.
3670 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3672 2014-01-08 Richard Biener <rguenther@suse.de>
3675 * gimplify.c (gimplify_expr): Gimplify register-register type
3676 VIEW_CONVERT_EXPRs to separate stmts.
3678 2014-01-07 Jeff Law <law@redhat.com>
3681 * ree.c (combine_set_extension): Handle case where source
3682 and destination registers in an extension insn are different.
3683 (combine_reaching_defs): Allow source and destination registers
3684 in extension to be different under limited circumstances.
3685 (add_removable_extension): Remove restriction that the
3686 source and destination registers in the extension are the same.
3687 (find_and_remove_re): Emit a copy from the extension's
3688 destination to its source after the defining insn if
3689 the source and destination registers are different.
3692 * ifcvt.c (merge_if_block): If we are merging a block with more than
3693 one successor with a block with no successors, remove any BARRIER
3694 after the second block.
3696 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3698 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3700 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3703 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3704 for 14-bit register offsets when INT14_OK_STRICT is false.
3706 2014-01-07 Roland Stigge <stigge@antcom.de>
3707 Michael Meissner <meissner@linux.vnet.ibm.com>
3710 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3711 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3713 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3715 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3718 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3720 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3721 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3722 rtx is const0_rtx or not.
3724 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3727 * target-globals.c (save_target_globals): Remove this_fn_optab
3729 * toplev.c: Include optabs.h.
3730 (target_reinit): Temporarily restore the global options if another
3731 set of options are in force.
3733 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3735 PR rtl-optimization/58668
3736 * cfgcleanup.c (flow_find_cross_jump): Don't count
3737 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3738 to determine what is counted.
3739 (flow_find_head_matching_sequence): Use active_insn_p to determine
3741 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3743 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3744 determine what is counted.
3746 PR tree-optimization/59643
3747 * tree-predcom.c (split_data_refs_to_components): If one dr is
3748 read and one write, determine_offset fails and the write isn't
3749 in the bad component, just put the read into the bad component.
3751 2014-01-07 Mike Stump <mikestump@comcast.net>
3752 Jakub Jelinek <jakub@redhat.com>
3755 * tree-core.h (struct tree_optimization_option): Change optabs
3756 type from unsigned char * to void *.
3757 * optabs.c (init_tree_optimization_optabs): Adjust
3758 TREE_OPTIMIZATION_OPTABS initialization.
3760 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3763 * config/i386/i386.h (struct machine_function): Add
3764 no_drap_save_restore field.
3765 * config/i386/i386.c (ix86_save_reg): Use
3766 !cfun->machine->no_drap_save_restore instead of
3767 crtl->stack_realign_needed.
3768 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3769 this function clears frame_pointer_needed. Set
3770 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3771 and DRAP reg is needed.
3773 2014-01-06 Marek Polacek <polacek@redhat.com>
3776 * doc/implement-c.texi: Mention that other integer types are
3777 permitted as bit-field types in strictly conforming mode.
3779 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3781 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3784 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3786 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3788 2014-01-06 Martin Jambor <mjambor@suse.cz>
3791 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3793 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3795 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3799 * var-tracking.c (add_stores): Preserve the value of the source even if
3800 we don't record the store.
3802 2014-01-06 Terry Guo <terry.guo@arm.com>
3804 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3806 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3809 * config/darwin.c (darwin_function_section): Adjust return values to
3810 correspond to optimisation changes made in r206070.
3812 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3814 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3815 from prefetch_block tune setting.
3816 (nocona_cost): Correct size of prefetch block to 64.
3818 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3820 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3821 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3822 used to save the static chain register in the computation of the offset
3823 from which the FP registers need to be restored.
3825 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3827 PR tree-optimization/59519
3828 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3829 ICE if get_current_def (current_new_name) is already non-NULL, as long
3830 as it is a phi result of some other phi in *new_exit_bb that has
3833 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3834 or vmovdqu* for misaligned_operand.
3835 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3836 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3837 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3838 aligned_mem for AVX512F masked aligned load and store builtins and for
3841 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3843 PR tree-optimization/59651
3844 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3845 Address range for negative step should be added by TYPE_SIZE_UNIT.
3847 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3849 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3850 overlapping registers also for registers other than the stack pointer.
3852 2014-01-03 Marek Polacek <polacek@redhat.com>
3855 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3858 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3861 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3864 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3865 (push splitter): Use <P:MODE_SIZE> instead of
3866 GET_MODE_SIZE (<P:MODE>mode).
3867 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3868 (mov -1, reg peephole2): Likewise.
3869 * config/i386/sse.md (*mov<mode>_internal,
3870 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3871 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3872 *<code><mode>3, *andnot<mode>3<mask_name>,
3873 <mask_codefor><code><mode>3<mask_name>): Likewise.
3874 * config/i386/subst.md (mask_mode512bit_condition,
3875 sd_mask_mode512bit_condition): Likewise.
3877 2014-01-02 Xinliang David Li <davidxl@google.com>
3879 PR tree-optimization/59303
3880 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3881 (dump_predicates): Better output format.
3882 (pred_equal_p): New function.
3883 (is_neq_relop_p): Ditto.
3884 (is_neq_zero_form_p): Ditto.
3885 (pred_expr_equal_p): Ditto.
3886 (pred_neg_p): Ditto.
3887 (simplify_pred): Ditto.
3888 (simplify_preds_2): Ditto.
3889 (simplify_preds_3): Ditto.
3890 (simplify_preds_4): Ditto.
3891 (simplify_preds): Ditto.
3893 (push_to_worklist): Ditto.
3894 (get_pred_info_from_cmp): Ditto.
3895 (is_degenerated_phi): Ditto.
3896 (normalize_one_pred_1): Ditto.
3897 (normalize_one_pred): Ditto.
3898 (normalize_one_pred_chain): Ditto.
3899 (normalize_preds): Ditto.
3900 (normalize_cond_1): Remove function.
3901 (normalize_cond): Ditto.
3902 (is_gcond_subset_of): Ditto.
3903 (is_subset_of_any): Ditto.
3904 (is_or_set_subset_of): Ditto.
3905 (is_and_set_subset_of): Ditto.
3906 (is_norm_cond_subset_of): Ditto.
3907 (pred_chain_length_cmp): Ditto.
3908 (convert_control_dep_chain_into_preds): Type change.
3909 (find_predicates): Ditto.
3910 (find_def_preds): Ditto.
3911 (destroy_predicates_vecs): Ditto.
3912 (find_matching_predicates_in_rest_chains): Ditto.
3913 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3914 (is_pred_expr_subset): Ditto.
3915 (is_pred_chain_subset_of): Ditto.
3916 (is_included_in): Ditto.
3917 (is_superset_of): Ditto.
3919 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3921 Update copyright years.
3923 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3925 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3926 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3927 config/arc/arc.md, config/arc/arc.opt,
3928 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3929 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3930 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3931 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3932 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3933 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3934 vtable-verify.c, vtable-verify.h: Use the standard form for the
3937 2014-01-02 Tobias Burnus <burnus@net-b.de>
3939 * gcc.c (process_command): Update copyright notice dates.
3940 * gcov-dump.c: Ditto.
3942 * doc/cpp.texi: Bump @copying's copyright year.
3943 * doc/cppinternals.texi: Ditto.
3944 * doc/gcc.texi: Ditto.
3945 * doc/gccint.texi: Ditto.
3946 * doc/gcov.texi: Ditto.
3947 * doc/install.texi: Ditto.
3948 * doc/invoke.texi: Ditto.
3950 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3952 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3954 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3956 * config/i386/sse.md (*mov<mode>_internal): Guard
3957 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3959 PR rtl-optimization/59647
3960 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3961 new_rtx into UNSIGNED_FLOAT rtxes.
3963 Copyright (C) 2014 Free Software Foundation, Inc.
3965 Copying and distribution of this file, with or without modification,
3966 are permitted in any medium without royalty provided the copyright
3967 notice and this notice are preserved.