1 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
6 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
8 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk
9 and define TARGET_ASM_OUTPUT_MI_THUNK and
10 TARGET_ASM_CAN_OUTPUT_MI_THUNK.
12 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
14 * config/microblaze/predicates.md: Add cmp_op predicate.
15 * config/microblaze/microblaze.md: Add branch_compare instruction
16 which uses cmp_op predicate and emits cmp insn before branch.
17 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
18 to microblaze_expand_conditional_branch and consolidate logic.
19 (microblaze_expand_conditional_branch): emit branch_compare
20 insn instead of handling cmp op separate from branch insn.
22 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
27 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
30 define_insn with define_expand and new define_insn
31 *altivec_lve<VI_char>x_internal.
32 (altivec_stve<VI_char>x): Replace define_insn with define_expand
33 and new define_insn *altivec_stve<VI_char>x_internal.
34 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
36 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
38 (altivec_expand_stvex_be): New function.
40 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
42 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
44 to STACK_POINTER_REGNUM if !frame_pointer_needed.
45 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
46 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
48 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
51 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
52 instead of emit_move_insn.
54 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
56 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
58 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
61 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
63 * config/rs6000/altivec.md (altivec_lvxl): Rename as
64 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
66 (altivec_lvxl_<mode>): New define_expand incorporating
67 -maltivec=be semantics where needed.
68 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
69 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
70 semantics where needed.
71 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
72 (altivec_stvx_<mode>): New define_expand incorporating
73 -maltivec=be semantics where needed.
74 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
75 VM2 iterator instead of V4SI.
76 (altivec_stvxl_<mode>): New define_expand incorporating
77 -maltivec=be semantics where needed.
78 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
79 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
80 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
81 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
82 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
84 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
85 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
86 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
87 ALTIVEC_BUILTIN_STVXL.
88 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
90 (altivec_expand_stvx_be): Likewise.
91 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
92 (altivec_expand_lvx_be): Likewise.
93 (altivec_expand_stvx_be): Likewise.
94 (altivec_expand_builtin): Add cases for
95 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
96 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
97 (altivec_init_builtins): Add definitions for
98 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
99 __builtin_altivec_stvx_<mode>, and
100 __builtin_altivec_stvxl_<mode>.
102 2014-02-21 Catherine Moore <clm@codesourcery.com>
104 * doc/invoke.texi (mvirt, mno-virt): Document.
105 * config/mips/mips.opt (mvirt): New option.
106 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
108 2014-02-21 Richard Biener <rguenther@suse.de>
110 PR tree-optimization/60276
111 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
112 (STMT_VINFO_MIN_NEG_DIST): New macro.
113 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
114 STMT_VINFO_MIN_NEG_DIST.
115 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
116 made for negative dependence distances still hold.
118 2014-02-21 Richard Biener <rguenther@suse.de>
121 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
122 DECL_INITIAL for globals not in the current function context.
124 2014-02-21 Jakub Jelinek <jakub@redhat.com>
126 PR tree-optimization/56490
127 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
128 * tree-ssa-uninit.c: Include params.h.
129 (compute_control_dep_chain): Add num_calls argument, return false
130 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
131 num_calls to recursive call.
132 (find_predicates): Change dep_chain into normal array,
133 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
134 variable and adjust compute_control_dep_chain caller.
135 (find_def_preds): Likewise.
137 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
139 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
140 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
142 2014-02-21 Nick Clifton <nickc@redhat.com>
144 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
146 (popqi1): Add mode to pre_dec.
149 2014-02-21 Jakub Jelinek <jakub@redhat.com>
151 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
152 mode for mask of V8SFmode permutation.
154 2014-02-20 Richard Henderson <rth@redhat.com>
157 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
158 a new pseudo for OLDVAL.
160 2014-02-20 Jakub Jelinek <jakub@redhat.com>
163 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
164 gen_reg_rtx if d->testing_p.
165 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
166 if d->testing_p and we will certainly return true.
167 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
170 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
172 * emit-rtl.c (gen_reg_rtx): Assert that
173 crtl->emit.regno_pointer_align_length is non-zero.
175 2014-02-20 Richard Henderson <rth@redhat.com>
178 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
179 on failure the store back into EXPECT.
181 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
182 Sandra Loosemore <sandra@codesourcery.com>
184 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
185 * config/nios2/nios2.c (nios2_function_profiler):
186 Add -fPIC (flag_pic == 2) support.
187 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
188 (nios2_large_offset_p): New function.
189 (nios2_unspec_reloc_p): Move up position, update to use
190 nios2_large_offset_p.
191 (nios2_unspec_address): Remove function.
192 (nios2_unspec_offset): New function.
193 (nios2_large_got_address): New function.
194 (nios2_got_address): Add large offset support.
195 (nios2_legitimize_tls_address): Update usage of removed and new
197 (nios2_symbol_binds_local_p): New function.
198 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
199 (nios2_legitimize_address): Update to use nios2_large_offset_p.
200 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
201 (nios2_print_operand): Merge H/L processing, add hiadj/lo
202 processing for (const (unspec ...)).
203 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
205 2014-02-20 Richard Biener <rguenther@suse.de>
207 * tree-cfg.c (replace_uses_by): Mark altered BBs before
208 doing the substitution.
209 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
211 2014-02-20 Martin Jambor <mjambor@suse.cz>
214 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
215 info when checking whether lattices are bottom.
217 2014-02-20 Richard Biener <rguenther@suse.de>
220 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
223 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
226 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
227 parameter specifying the scaling.
228 (inline_call): Update.
229 (want_inline_recursively): Guard division by zero.
230 (recursive_inlining): Update.
231 * ipa-inline.h (clone_inlined_nodes): Update.
233 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
236 * config/i386/i386.c (classify_argument): Pass structures of size
237 64 bytes or less in register.
239 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
240 Kirill Yukhin <kirill.yukhin@intel.com>
242 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
243 (_mm_rcp28_round_ss): Ditto.
244 (_mm_rsqrt28_round_sd): Ditto.
245 (_mm_rsqrt28_round_ss): Ditto.
246 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
247 (_mm_rcp14_round_ss): Ditto.
248 (_mm_rsqrt14_round_sd): Ditto.
249 (_mm_rsqrt14_round_ss): Ditto.
250 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
251 the first input operand, get rid of match_dup.
252 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
254 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
256 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
257 operand as the first input operand, set type attribute.
258 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
260 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
261 operand as the first input operand, set type attribute.
263 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
265 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
268 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
271 * config/i386/i386.c (construct_container): Remove TFmode check
272 for X86_64_INTEGER_CLASS.
274 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
277 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
278 only when -Wpsabi is enabled.
280 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
283 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
284 passing arrays in registers are the same as for structs, so remove the
285 special case for them.
287 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
289 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
290 destination type, extract only the valid bits if the source type is not
291 integral and has a different mode.
293 2014-02-19 Richard Biener <rguenther@suse.de>
296 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
299 2014-02-19 Richard Biener <rguenther@suse.de>
302 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
303 (ipa_modify_call_arguments): Emit an argument load explicitely and
304 preserve virtual SSA form there and for the replacement call.
305 Do not update SSA form nor free dominance info.
307 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
309 * ipa.c (function_and_variable_visibility): Also clear WEAK
310 flag when disolving COMDAT_GROUP.
312 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
314 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
315 * ipa-prop.c (ipa_set_jf_known_type): Return early when
317 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
318 do more sanity checks.
319 (detect_type_change): Return true when giving up early.
320 (compute_complex_assign_jump_func): Fix type parameter of
322 (compute_complex_ancestor_jump_func): Likewise.
323 (update_jump_functions_after_inlining): Fix updating of
325 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
327 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
329 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
330 inline clones when edge disappears.
332 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
335 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
336 Split 64-bit moves into 2 patterns. Do not allow the use of
337 direct move for TDmode in little endian, since the decimal value
338 has little endian bytes within a word, but the 64-bit pieces are
339 ordered in a big endian fashion, and normal subreg's of TDmode are
341 (mov<mode>_64bit_dm): Likewise.
342 (movtd_64bit_nodm): Likewise.
344 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
346 PR tree-optimization/60174
347 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
348 statement of an SSA_NAME that occurs in an abnormal PHI node.
350 2014-02-18 Jakub Jelinek <jakub@redhat.com>
353 * final.c (SEEN_BB): Remove.
354 (SEEN_NOTE, SEEN_EMITTED): Renumber.
355 (final_scan_insn): Don't force_source_line on second
356 NOTE_INSN_BASIC_BLOCK.
358 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
361 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
362 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
363 (type_natural_mode): Warn ABI change when %zmm register is not
364 available for AVX512F vector value passing.
366 2014-02-18 Kai Tietz <ktietz@redhat.com>
369 * config/i386/i386.c (ix86_expand_prologue): Use value in
370 rax register as displacement when restoring %r10 or %rax.
371 Fix wrong offset when restoring both registers.
373 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
375 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
376 assertion with conditional return.
378 2014-02-18 Jakub Jelinek <jakub@redhat.com>
379 Uros Bizjak <ubizjak@gmail.com>
382 * config/i386/driver-i386.c (host_detect_local_cpu): If
383 YMM state is not saved by the OS, also clear has_f16c. Move
384 CPUID 0x80000001 handling before YMM state saving checking.
386 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
388 PR rtl-optimization/58960
389 * haifa-sched.c (alloc_global_sched_pressure_data): New,
390 factored out from ...
391 (sched_init): ... here.
392 (free_global_sched_pressure_data): New, factored out from ...
393 (sched_finish): ... here.
394 * sched-int.h (free_global_sched_pressure_data): Declare.
395 * sched-rgn.c (nr_regions_initial): New static global.
396 (haifa_find_rgns): Initialize it.
397 (schedule_region): Disable sched-pressure for the newly
400 2014-02-17 Richard Biener <rguenther@suse.de>
402 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
403 release SSA defs of pattern stmts.
405 2014-02-17 Richard Biener <rguenther@suse.de>
407 * tree-inline.c (expand_call_inline): Release the virtual
408 operand defined by the call we are about to inline.
410 2014-02-17 Richard Biener <rguenther@suse.de>
412 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
414 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
415 Ilya Tocar <ilya.tocar@intel.com>
417 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
418 arguments order in builtin.
419 (_mm512_permutexvar_epi64): Ditto.
420 (_mm512_mask_permutexvar_epi64): Ditto
421 (_mm512_maskz_permutexvar_epi32): Ditto
422 (_mm512_permutexvar_epi32): Ditto
423 (_mm512_mask_permutexvar_epi32): Ditto
425 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
427 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
428 (p8_vmrgow): Likewise.
430 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
432 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
435 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
438 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
439 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
440 into 64-bit and 32-bit moves. On 64-bit moves, add support for
441 using direct move instructions on ISA 2.07. Also adjust
442 instruction length for 64-bit.
443 (mov<mode>_64bit, TFmode/TDmode): Likewise.
444 (mov<mode>_32bit, TFmode/TDmode): Likewise.
446 2014-02-15 Alan Modra <amodra@gmail.com>
450 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
451 find_replacement on parts of insn rtl that might be reloaded.
453 2014-02-15 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/60183
456 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
457 (tree_ssa_phiprop): Calculate and free post-dominators.
459 2014-02-14 Jeff Law <law@redhat.com>
461 PR rtl-optimization/60131
462 * ree.c (get_extended_src_reg): New function.
463 (combine_reaching_defs): Use it rather than assuming location of REG.
464 (find_and_remove_re): Verify first operand of extension is
465 a REG before adding the insns to the copy list.
467 2014-02-14 Roland McGrath <mcgrathr@google.com>
469 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
470 * configure: Regenerated.
471 * config.in: Regenerated.
472 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
473 instead of ASM_SHORT.
475 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
476 Richard Earnshaw <rearnsha@arm.com>
478 PR rtl-optimization/59535
479 * lra-constraints.c (process_alt_operands): Encourage alternative
480 when unassigned pseudo class is superset of the alternative class.
481 (inherit_reload_reg): Don't inherit when optimizing for code size.
482 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
483 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
484 modes not less than 4 for Thumb1.
486 2014-02-14 Kyle McMartin <kyle@redhat.com>
489 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
491 2014-02-14 Richard Biener <rguenther@suse.de>
493 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
494 (get_frame_arg): Drop the assert with langhook types_compatible_p.
495 Do not strip INDIRECT_REFs.
497 2014-02-14 Richard Biener <rguenther@suse.de>
500 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
501 DECL_FUNCTION_SPECIFIC_TARGET.
502 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
503 * tree-streamer-out.c (pack_ts_target_option): Remove.
504 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
505 (write_ts_function_decl_tree_pointers): Do not stream
506 DECL_FUNCTION_SPECIFIC_TARGET.
507 * tree-streamer-in.c (unpack_ts_target_option): Remove.
508 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
509 (lto_input_ts_function_decl_tree_pointers): Do not stream
510 DECL_FUNCTION_SPECIFIC_TARGET.
512 2014-02-14 Jakub Jelinek <jakub@redhat.com>
514 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
515 (get_initial_def_for_induction, vectorizable_induction): Ignore
516 debug stmts when looking for exit_phi.
517 (vectorizable_live_operation): Fix up condition.
519 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
521 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
522 nreverse() because it changes the content of original tree list.
524 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
526 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
527 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
529 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
531 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
532 GNU coding standards.
534 2014-02-13 Jakub Jelinek <jakub@redhat.com>
537 * dwarf2out.c (gen_subprogram_die): Don't call
538 add_calling_convention_attribute if subr_die is old_die.
540 2014-02-13 Sharad Singhai <singhai@google.com>
542 * doc/optinfo.texi: Fix order of nodes.
544 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
546 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
547 operands[2], not operands[3].
549 2014-02-13 Richard Biener <rguenther@suse.de>
552 * doc/install.texi (ISL): Update recommended version to 0.12.2,
553 mention the possibility of an in-tree build.
554 (CLooG): Update recommended version to 0.18.1, mention the
555 possibility of an in-tree build and clarify that the ISL
556 bundled with CLooG does not work.
558 2014-02-13 Jakub Jelinek <jakub@redhat.com>
561 * expr.c (compress_float_constant): If x is a hard register,
562 extend into a pseudo and then move to x.
564 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
566 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
567 caused by bad second argument to warning_at() with -mhotpatch and
568 nested functions (e.g. with gfortran).
570 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
572 * opts.c (option_name): Remove "enabled by default" rider.
574 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
576 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
578 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
579 Uros Bizjak <ubizjak@gmail.com>
582 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
583 * configure: Regenerated.
585 2014-02-12 Richard Biener <rguenther@suse.de>
587 * vec.c (vec_prefix::calculate_allocation): Move as
588 inline variant to vec.h.
589 (vec_prefix::calculate_allocation_1): New out-of-line version.
590 * vec.h (vec_prefix::calculate_allocation_1): Declare.
591 (vec_prefix::m_has_auto_buf): Rename to ...
592 (vec_prefix::m_using_auto_storage): ... this.
593 (vec_prefix::calculate_allocation): Inline the easy cases
594 and dispatch to calculate_allocation_1 which doesn't need the
596 (va_heap::reserve): Use gcc_checking_assert.
597 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
598 m_using_auto_storage.
599 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
601 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
602 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
603 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
605 2014-02-12 Richard Biener <rguenther@suse.de>
607 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
608 when we found a dependence.
610 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
612 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
614 (maybe_fold_stmt): ... into this new function.
615 * omp-low.c (lower_omp): Update comment.
617 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
620 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
623 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
625 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
626 identifiers in comments.
627 (cortexa53_extra_costs): Likewise.
628 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
629 (cortexa7_extra_costs): Likewise.
630 (cortexa12_extra_costs): Likewise.
631 (cortexa15_extra_costs): Likewise.
632 (v7m_extra_costs): Likewise.
634 2014-02-12 Richard Biener <rguenther@suse.de>
637 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
638 of posix_memalign being successful.
639 (lower_stmt): Restrict lowering of posix_memalign to when
640 -ftree-bit-ccp is enabled.
642 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
644 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
646 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
648 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
650 PR rtl-optimization/60116
651 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
652 other_insn once the combination has been validated.
654 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
657 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
659 * ipa-devirt.c: Include demangle.h
660 (odr_violation_reported): New static variable.
661 (add_type_duplicate): Update odr_violations.
662 (maybe_record_node): Add completep parameter; update it.
663 (record_target_from_binfo): Add COMPLETEP parameter;
665 (possible_polymorphic_call_targets_1): Likewise.
666 (struct polymorphic_call_target_d): Add nonconstruction_targets;
667 rename FINAL to COMPLETE.
668 (record_targets_from_bases): Sanity check we found the binfo;
669 fix COMPLETEP updating.
670 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
671 parameter, fix computing of COMPLETEP.
672 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
673 at LTO time do demangling.
674 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
675 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
677 (gimple_get_virt_method_for_binfo): Likewise.
678 * gimple-fold.h (gimple_get_virt_method_for_binfo,
679 gimple_get_virt_method_for_vtable): Update prototypes.
681 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
684 * genautomata.c (add_presence_absence): Fix typo with
685 {final_}presence_list.
687 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
690 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
691 for VSX/Altivec vectors that land in GPR registers.
693 2014-02-11 Richard Henderson <rth@redhat.com>
694 Jakub Jelinek <jakub@redhat.com>
697 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
698 around drhs if type conversion to lacc->type is not useless.
700 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
702 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
704 (cortex-a57.cortex-a53): Likewise.
705 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
707 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
709 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
712 2014-02-11 Renlin Li <Renlin.Li@arm.com>
714 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
715 add_options_for_arm_vfp3.
717 2014-02-11 Jeff Law <law@redhat.com>
720 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
721 object with an undesirable mode.
723 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
726 * config/i386/sol2-9.h: New file.
727 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
728 *-*-solaris2.9*): Use it.
730 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
732 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
733 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
735 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
737 * config/microblaze/microblaze.c: Extend mcpu version format
739 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
741 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
743 2014-02-10 Richard Henderson <rth@redhat.com>
746 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
747 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
748 ms-abi vs -mno-accumulate-outgoing-args.
749 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
750 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
753 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
756 * cfgexpand.c (expand_asm_operands): Attach source location to
757 ASM_INPUT rtx objects.
758 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
760 2014-02-10 Nick Clifton <nickc@redhat.com>
762 * config/mn10300/mn10300.c (popcount): New function.
763 (mn10300_expand_prologue): Include saved registers in stack usage
766 2014-02-10 Jeff Law <law@redhat.com>
769 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
770 when changing the SET_DEST of a prior insn to avoid an input reload.
772 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
774 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
775 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
776 -mcall-openbsd, or -mcall-linux.
777 (CC1_ENDIAN_BIG_SPEC): Remove.
778 (CC1_ENDIAN_LITTLE_SPEC): Remove.
779 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
780 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
781 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
782 and %cc1_endian_default.
783 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
785 2014-02-10 Richard Biener <rguenther@suse.de>
787 PR tree-optimization/60115
788 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
789 MEM_REF handling. Properly verify that the accesses are not
790 out of the objects bound.
792 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
794 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
797 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
799 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
800 proper constants and fix formatting.
801 (possible_polymorphic_call_targets): Fix formatting.
803 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
804 Ilya Tocar <ilya.tocar@intel.com>
806 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
807 (_mm512_loadu_epi32): Renamed into...
808 (_mm512_loadu_si512): This.
809 (_mm512_storeu_epi32): Renamed into...
810 (_mm512_storeu_si512): This.
811 (_mm512_maskz_ceil_ps): Removed.
812 (_mm512_maskz_ceil_pd): Ditto.
813 (_mm512_maskz_floor_ps): Ditto.
814 (_mm512_maskz_floor_pd): Ditto.
815 (_mm512_floor_round_ps): Ditto.
816 (_mm512_floor_round_pd): Ditto.
817 (_mm512_ceil_round_ps): Ditto.
818 (_mm512_ceil_round_pd): Ditto.
819 (_mm512_mask_floor_round_ps): Ditto.
820 (_mm512_mask_floor_round_pd): Ditto.
821 (_mm512_mask_ceil_round_ps): Ditto.
822 (_mm512_mask_ceil_round_pd): Ditto.
823 (_mm512_maskz_floor_round_ps): Ditto.
824 (_mm512_maskz_floor_round_pd): Ditto.
825 (_mm512_maskz_ceil_round_ps): Ditto.
826 (_mm512_maskz_ceil_round_pd): Ditto.
827 (_mm512_expand_pd): Ditto.
828 (_mm512_expand_ps): Ditto.
829 * config/i386/i386.c (ix86_builtins): Remove
830 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
832 * config/i386/predicates.md (const1256_operand): New.
833 (const_1_to_2_operand): Ditto.
834 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
835 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
836 (*avx512pf_gatherpf<mode>sf): Ditto.
837 (avx512pf_gatherpf<mode>df): Ditto.
838 (*avx512pf_gatherpf<mode>df_mask): Ditto.
839 (*avx512pf_gatherpf<mode>df): Ditto.
840 (avx512pf_scatterpf<mode>sf): Ditto.
841 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
842 (*avx512pf_scatterpf<mode>sf): Ditto.
843 (avx512pf_scatterpf<mode>df): Ditto.
844 (*avx512pf_scatterpf<mode>df_mask): Ditto.
845 (*avx512pf_scatterpf<mode>df): Ditto.
846 (avx512f_expand<mode>): Removed.
847 (<shift_insn><mode>3<mask_name>): Change predicate type.
849 2014-02-08 Jakub Jelinek <jakub@redhat.com>
851 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
852 not at the end of datarefs vector use ordered_remove to avoid
853 reordering datarefs vector.
856 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
857 mark local addressable non-static vars as GOVD_PRIVATE
858 instead of GOVD_LOCAL.
859 * omp-low.c (lower_omp_for): Move gimple_bind_vars
860 and BLOCK_VARS of gimple_bind_block to new_stmt rather
864 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
865 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
866 assume_aligned or alloc_align attributes.
867 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
868 arguments. Handle also assume_aligned and alloc_align attributes.
869 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
870 calls to functions with assume_aligned or alloc_align attributes.
871 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
873 2014-02-08 Terry Guo <terry.guo@arm.com>
875 * doc/invoke.texi: Document ARM -march=armv7e-m.
877 2014-02-08 Jakub Jelinek <jakub@redhat.com>
879 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
880 flag on __cilkrts_rethrow builtin.
883 * ipa-cp.c (determine_versionability): Fail at -O0
884 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
885 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
888 2014-02-04 Jakub Jelinek <jakub@redhat.com>
891 * tree-inline.c (copy_forbidden): Fail for
892 __attribute__((optimize (0))) functions.
894 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
896 * varpool.c: Include pointer-set.h.
897 (varpool_remove_unreferenced_decls): Variables in other partitions
898 will not be output; be however careful to not lose information
901 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
903 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
904 lookup in the vtable constructor.
906 2014-02-07 Jeff Law <law@redhat.com>
909 * config/m68k/m68k.md (ashldi_extsi): Turn into a
910 define_insn_and_split.
912 * ipa-inline.c (inline_small_functions): Fix typos.
914 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
916 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
917 (s390_can_use_return_insn): Declare.
918 * config/s390/s390.h (EPILOGUE_USES): Define.
919 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
921 (s390_chunkify_start): Handle return JUMP_LABELs.
922 (s390_early_mach): Emit a main_pool instruction on the entry edge.
923 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
924 (s390_can_use_return_insn): New functions.
925 (s390_fix_long_loop_prediction): Handle conditional returns.
926 (TARGET_SET_UP_BY_PROLOGUE): Define.
927 * config/s390/s390.md (ANY_RETURN): New code iterator.
928 (*creturn, *csimple_return, return, simple_return): New patterns.
930 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
932 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
933 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
934 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
935 REG_CFA_RESTORE list when deciding not to restore a register.
937 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
939 * config/s390/s390.c: Include tree-pass.h and context.h.
940 (s390_early_mach): New function, split out from...
941 (s390_emit_prologue): ...here.
942 (pass_data_s390_early_mach): New pass structure.
943 (pass_s390_early_mach): New class.
944 (s390_option_override): Create and register early_mach pass.
947 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
949 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
950 to match for the exit block.
952 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
954 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
955 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
956 Reject misaligned operands.
958 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
960 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
962 2014-02-07 Richard Biener <rguenther@suse.de>
965 * gimple-low.c (lower_builtin_posix_memalign): New function.
966 (lower_stmt): Call it to lower posix_memalign in a way
967 to make alignment info accessible.
969 2014-02-07 Jakub Jelinek <jakub@redhat.com>
972 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
973 __builtin_setjmp_receiver.
975 2014-02-07 Richard Biener <rguenther@suse.de>
978 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
979 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
980 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
981 Handle BUILT_IN_POSIX_MEMALIGN.
982 (find_func_clobbers): Likewise.
983 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
984 (call_may_clobber_ref_p_1): Likewise.
986 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
989 * ipa-devirt.c (record_target_from_binfo): Remove overactive
992 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
995 * lto-cgraph.c (lto_output_node): Use
996 symtab_get_symbol_partitioning_class.
997 (lto_output_varpool_node): likewise.
998 (symtab_get_symbol_partitioning_class): Move here from
1000 * cgraph.h (symbol_partitioning_class): Likewise.
1001 (symtab_get_symbol_partitioning_class): Declare.
1003 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1005 * ggc.h (ggc_internal_cleared_alloc): New macro.
1006 * vec.h (vec_safe_copy): Handle memory stats.
1007 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1008 * target-globals.c (save_target_globals): Likewise.
1010 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1013 * expr.c (emit_move_resolve_push): Export; be bit more selective
1014 on when to clear alias set.
1015 * expr.h (emit_move_resolve_push): Declare.
1016 * function.h (struct function): Add tail_call_marked.
1017 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1018 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1019 * config/i386/i386.md (TImode move expander): De not call
1021 (FP push expanders): Preserve memory attributes.
1022 * config/i386/sse.md (push<mode>1): Remove.
1023 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1024 (ix86_expand_push): Remove.
1025 * config/i386/mmx.md (push<mode>1): Remove.
1027 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1029 PR rtl-optimization/60030
1030 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1031 lopart with paradoxical subreg before shifting it up by hprec.
1033 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1035 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1036 Remove extra newline at end of file.
1037 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1038 (arm_issue_rate): Handle cortexa57.
1039 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1040 (cortex-a57.cortex-a53): Likewise.
1042 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1045 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1046 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1048 (arm_expand_prologue): Adjust all callers.
1049 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1050 info, registers also at the lowest numbered registers side. Use
1051 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1055 * var-tracking.c (adjust_mems): Before adding a SET to
1056 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1058 2014-02-06 Alan Modra <amodra@gmail.com>
1061 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1062 change SDmode to DDmode when lra_in_progress.
1064 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1067 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1068 free_data_ref on the dr first, and before goto again also set dr
1069 to the next dr. For simd_lane_access, free old datarefs[i] before
1070 overwriting it. For get_vectype_for_scalar_type failure, don't
1071 free_data_ref if simd_lane_access.
1073 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1076 * tree.h (opts_for_fn): New inline function.
1077 (opt_for_fn): Define.
1078 * config/i386/i386.c (ix86_function_regparm): Use
1079 opt_for_fn (decl, optimize) instead of optimize.
1081 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1083 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1084 for SYMBOL_REF in large memory model.
1086 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1088 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1090 (cortex-a57): Likewise.
1091 (cortex-a57.cortex-a53): Likewise.
1093 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1094 Kugan Vivekanandarajah <kuganv@linaro.org>
1096 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1098 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1100 2014-02-06 Richard Biener <rguenther@suse.de>
1102 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1103 set_loop_copy and initialize_original_copy_tables.
1105 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1107 * config/aarch64/aarch64-simd.md
1108 (aarch64_ashr_simddi): Change QI to SI.
1110 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1111 Jakub Jelinek <jakub@redhat.com>
1114 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1117 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1119 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1120 CODE_FOR_altivec_vpku[hw]um to
1121 CODE_FOR_altivec_vpku[hw]um_direct.
1122 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1123 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1124 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1125 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1127 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1129 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1130 generation for -maltivec=be.
1131 (altivec_vsumsws): Simplify redundant test.
1133 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1135 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1136 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1137 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1138 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1139 gen_altivec_vpkuwum.
1140 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1142 (altivec_vpks<VI_char>ss): Likewise.
1143 (altivec_vpks<VI_char>us): Likewise.
1144 (altivec_vpku<VI_char>us): Likewise.
1145 (altivec_vpku<VI_char>um): Likewise.
1146 (altivec_vpku<VI_char>um_direct): New (copy of
1147 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1149 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1150 target is little endian and -maltivec=be is not specified.
1151 (*altivec_vupkhs<VU_char>_direct): New (copy of
1152 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1153 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1154 target is little endian and -maltivec=be is not specified.
1155 (*altivec_vupkls<VU_char>_direct): New (copy of
1156 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1157 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1158 little endian and -maltivec=be is not specified.
1159 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1160 little endian and -maltivec=be is not specified.
1162 2014-02-05 Richard Henderson <rth@redhat.com>
1165 * combine-stack-adj.c: Revert r206943.
1166 * sched-int.h (struct deps_desc): Add last_args_size.
1167 * sched-deps.c (init_deps): Initialize it.
1168 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1169 contain REG_ARGS_SIZE notes.
1171 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1173 * lto-cgraph.c (asm_nodes_output): Make global.
1174 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1175 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1176 (driver_handle_option): Handle OPT_fwpa.
1178 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1181 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1182 a comment typo and formatting issue. If odr_hash hasn't been
1183 created, return vNULL and set *completep to false.
1186 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1187 bb with no successors.
1189 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1192 * doc/invoke.texi (-march): Clarify documentation for ARM.
1196 2014-02-05 Richard Biener <rguenther@suse.de>
1198 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1199 when not vectorizing because of too many alias checks.
1200 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1201 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1203 2014-02-05 Nick Clifton <nickc@redhat.com>
1205 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1206 accept extended registers in any mode when compiling for the MN10300.
1208 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1210 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1211 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1212 sanitization attributes.
1213 (can_inline_edge_p): Likewise.
1214 (sanitize_attrs_match_for_inline_p): New function.
1216 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1218 * ipa-prop.c (detect_type_change): Shor circuit testing of
1219 type changes on THIS pointer.
1221 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1224 * config/pa/pa.c (legitimize_tls_address): Return original address
1225 if not passed a SYMBOL_REF rtx.
1226 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1228 (pa_emit_move_sequence): Simplify TLS source operands.
1229 (pa_legitimate_constant_p): Reject all TLS constants.
1230 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1231 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1233 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1235 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1236 groups when we know they are controlled by LTO.
1237 * varasm.c (default_binds_local_p_1): If object is in other partition,
1238 it will be resolved locally.
1240 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1242 * config/host-linux.c (linux_gt_pch_use_address): Don't
1243 use SSIZE_MAX because it is not always defined.
1245 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1248 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1249 threshold for pseudo splitting.
1250 (update_ebb_live_info): Process call argument hard registers and
1251 hard registers from insn definition too.
1252 (max_small_class_regs_num): New constant.
1253 (inherit_in_ebb): Update live hard regs through EBBs. Update
1254 reloads_num only for small register classes. Don't split for
1257 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1260 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1263 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1265 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1268 2014-02-04 Marek Polacek <polacek@redhat.com>
1270 * gdbinit.in (pel): Define.
1272 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1274 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1277 2014-02-04 Richard Biener <rguenther@suse.de>
1280 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1281 in function context local.
1282 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1283 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1284 similar to LTO_imported_decl_ref.
1286 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1288 PR tree-optimization/60002
1289 * cgraphclones.c (build_function_decl_skip_args): Clear
1292 PR tree-optimization/60023
1293 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1294 false to gsi_replace.
1295 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1296 has been in some EH region and vec_stmt could throw, add
1297 vec_stmt into the same EH region.
1298 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1299 has no lhs, ignore it.
1300 * internal-fn.c (expand_MASK_LOAD): Likewise.
1303 * tree-inline.c (copy_forbidden): Fail for
1304 __attribute__((optimize (0))) functions.
1307 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1308 is set, copy one less argument.
1309 (expand_simd_clones): Don't subtract clone_info->inbranch
1310 from simd_clone_struct_alloc argument.
1312 PR rtl-optimization/57915
1313 * recog.c (simplify_while_replacing): If all unary/binary/relational
1314 operation arguments are constant, attempt to simplify those.
1317 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1318 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1320 2014-02-04 Richard Biener <rguenther@suse.de>
1322 PR tree-optimization/60012
1323 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1324 TBAA disambiguation to all DDRs.
1326 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1329 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1330 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1332 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1335 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1337 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1339 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1340 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1342 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1345 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1346 to figure out targets of polymorphic calls with known decl.
1347 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1348 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1349 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1350 (get_polymorphic_call_info): ... here.
1351 (get_polymorphic_call_info_from_invariant): New function.
1353 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1355 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1356 lookup via vtable pointer; check for type consistency
1357 and turn inconsitent facts into UNREACHABLE.
1358 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1359 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1360 type inconsistent querries; return UNREACHABLE instead.
1362 2014-02-03 Richard Henderson <rth@twiddle.net>
1365 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1366 already processed this node.
1367 (normalize_one_pred_1): Pass along mark_set.
1368 (normalize_one_pred): Create and destroy a pointer_set_t.
1369 (normalize_one_pred_chain): Likewise.
1371 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1373 PR gcov-profile/58602
1374 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1376 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1379 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1380 -fno-devirtualize; try to devirtualize by the knowledge of
1381 virtual table pointer given by aggregate propagation.
1382 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1383 (ipa_print_node_jump_functions): Dump also offset that
1384 is relevant for polymorphic calls.
1385 (determine_known_aggregate_parts): Add arg_type parameter; use it
1386 instead of determining the type from pointer type.
1387 (ipa_compute_jump_functions_for_edge): Update call of
1388 determine_known_aggregate_parts.
1389 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1390 (gimple_get_virt_method_for_binfo): ... here; simplify using
1391 vtable_pointer_value_to_vtable.
1392 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1393 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1394 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1395 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1397 (vtable_pointer_value_to_binfo): ... here.
1398 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1400 2014-02-03 Teresa Johnson <tejohnson@google.com>
1402 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1403 redef of outer loop index variable.
1405 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1409 * doc/extend.texi (Function Attributes): Typo.
1411 2014-02-03 Cong Hou <congh@google.com>
1413 PR tree-optimization/60000
1414 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1415 if the vectorized statement is a store. A store statement can only
1416 appear at the end of pattern statements.
1418 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1420 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1421 (ix86_option_override_internal): Default long double to 64-bit for
1422 32-bit Bionic and to 128-bit for 64-bit Bionic.
1424 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1425 TARGET_LONG_DOUBLE_128 is true.
1426 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1428 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1429 (mlong-double-64): Negate -mlong-double-128.
1430 (mlong-double-128): New option.
1432 * config/i386/i386-c.c (ix86_target_macros): Define
1433 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1435 * doc/invoke.texi: Document -mlong-double-128.
1437 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1439 PR rtl-optimization/60024
1440 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1442 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1444 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1446 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1448 PR rtl-optimization/57662
1449 * sel-sched.c (code_motion_path_driver): Do not mark already not
1450 existing blocks in the visiting bitmap.
1452 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1454 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1455 on the insn being emitted.
1457 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1458 Will Deacon <will.deacon@arm.com>
1460 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1462 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1464 * config/arm/arm-tables.opt: Regenerate.
1466 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1468 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1469 for vector types other than V16QImode.
1470 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1471 define_expand, and call altivec_expand_vec_perm_le when producing
1472 code with little endian element order.
1473 (*altivec_vperm_<mode>_internal): New insn having previous
1474 behavior of altivec_vperm_<mode>.
1475 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1476 altivec_expand_vec_perm_le when producing code with little endian
1478 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1479 behavior of altivec_vperm_<mode>_uns.
1481 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1483 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1484 (altivec_vsumsws): Add handling for -maltivec=be with a little
1486 (altivec_vsumsws_direct): New.
1487 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1488 gen_altivec_vsumsws.
1490 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1492 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1493 vtable_pointer_value_to_binfo): New functions.
1494 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1495 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1497 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1499 * config/nios2/nios2.md (load_got_register): Initialize GOT
1500 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1501 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1503 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1505 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1506 preserverd by passthrough, do not propagate the type.
1508 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1510 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1511 (mips_atomic_assign_expand_fenv): New function.
1512 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1514 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1516 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1517 (__builtin_mips_set_fcsr): Likewise.
1518 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1519 MIPS_USI_FTYPE_VOID.
1520 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1521 (mips16_expand_set_fcsr): Likewise.
1522 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1523 (mips16_set_fcsr_stub): Likewise.
1524 (mips16_get_fcsr_one_only_stub): New class.
1525 (mips16_set_fcsr_one_only_stub): Likewise.
1526 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1527 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1528 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1529 (hard_float): New availability predicate.
1530 (mips_builtins): Add get_fcsr and set_fcsr.
1531 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1532 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1533 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1534 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1535 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1538 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1540 * config/mips/mips.c (mips_one_only_stub): New class.
1541 (mips_need_mips16_rdhwr_p): Replace with...
1542 (mips16_rdhwr_stub): ...this new variable.
1543 (mips16_stub_call_address): New function.
1544 (mips16_rdhwr_one_only_stub): New class.
1545 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1546 (mips_output_mips16_rdhwr): Delete.
1547 (mips_finish_stub): New function.
1548 (mips_code_end): Use it to handle rdhwr stubs.
1550 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1553 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1554 when calculating size of integer atomic types.
1556 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1558 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1560 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1562 PR tree-optimization/60003
1563 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1564 * profile.c (branch_prob): Use gimple_call_builtin_p
1565 to check for BUILT_IN_SETJMP_RECEIVER.
1566 * tree-inline.c (copy_bb): Call notice_special_calls.
1568 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1571 * lra-constraints.c (process_alt_operands): Update reload_sum only
1574 2014-01-31 Richard Henderson <rth@redhat.com>
1577 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1578 until after else_eh is processed.
1580 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1582 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1583 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1584 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1585 in smmintrin.h, remove them.
1586 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1587 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1588 * config/i386/i386.md (ROUND_SAE): Fix value.
1589 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1590 (const48_operand): New.
1591 * config/i386/subst.md (round), (round_expand): Use
1592 const_4_or_8_to_11_operand.
1593 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1595 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1597 * config/i386/constraints.md (Yk): Swap meaning with k.
1598 * config/i386/i386.md (movhi_internal): Change Yk to k.
1599 (movqi_internal): Ditto.
1600 (*k<logic><mode>): Ditto.
1603 (kandn<mode>): Ditto.
1604 (*<code>hi_1): Ditto.
1605 (*<code>qi_1): Ditto.
1606 (kxnor<mode>): Ditto.
1607 (kortestzhi): Ditto.
1608 (kortestchi): Ditto.
1610 (*one_cmplhi2_1): Ditto.
1611 (*one_cmplqi2_1): Ditto.
1612 * config/i386/sse.md (): Change k to Yk.
1613 (avx512f_load<mode>_mask): Ditto.
1614 (avx512f_blendm<mode>): Ditto.
1615 (avx512f_store<mode>_mask): Ditto.
1616 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1617 (avx512f_storedqu<mode>_mask): Ditto.
1618 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1620 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1621 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1622 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1623 (avx512f_maskcmp<mode>3): Ditto.
1624 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1625 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1626 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1627 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1628 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1629 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1630 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1631 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1632 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1633 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1634 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1635 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1636 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1637 (vec_extract_lo_<mode>_maskm): Ditto.
1638 (vec_extract_hi_<mode>_maskm): Ditto.
1639 (avx512f_vternlog<mode>_mask): Ditto.
1640 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1641 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1642 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1643 (avx512f_<code>v8div16qi2_mask): Ditto.
1644 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1645 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1646 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1647 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1648 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1649 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1650 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1651 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1652 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1653 (avx512cd_maskb_vec_dupv8di): Ditto.
1654 (avx512cd_maskw_vec_dupv16si): Ditto.
1655 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1656 (avx512f_vpermi2var<mode>3_mask): Ditto.
1657 (avx512f_vpermi2var<mode>3_mask): Ditto.
1658 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1659 (*avx512f_gathersi<mode>): Ditto.
1660 (*avx512f_gathersi<mode>_2): Ditto.
1661 (*avx512f_gatherdi<mode>): Ditto.
1662 (*avx512f_gatherdi<mode>_2): Ditto.
1663 (*avx512f_scattersi<mode>): Ditto.
1664 (*avx512f_scatterdi<mode>): Ditto.
1665 (avx512f_compress<mode>_mask): Ditto.
1666 (avx512f_compressstore<mode>_mask): Ditto.
1667 (avx512f_expand<mode>_mask): Ditto.
1668 * config/i386/subst.md (mask): Change k to Yk.
1669 (mask_scalar_merge): Ditto.
1672 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1674 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1676 2014-01-31 Richard Biener <rguenther@suse.de>
1679 * builtins.c (fold_builtin_memory_op): Make sure to not
1680 use a floating-point mode or a boolean or enumeral type for
1683 2014-01-30 DJ Delorie <dj@redhat.com>
1685 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1686 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1687 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1688 whenever main() has an epilogue.
1690 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1692 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1693 unused variable "field".
1694 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1695 (vsx_mergeh_<mode>): Likewise.
1696 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1697 (altivec_vmrghh): Likewise.
1698 (altivec_vmrghw): Likewise.
1699 (altivec_vmrglb): Likewise.
1700 (altivec_vmrglh): Likewise.
1701 (altivec_vmrglw): Likewise.
1702 (altivec_vspltb): Add missing uses.
1703 (altivec_vsplth): Likewise.
1704 (altivec_vspltw): Likewise.
1705 (altivec_vspltsf): Likewise.
1707 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1710 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1711 frame related instructions.
1713 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1715 PR rtl-optimization/59959
1716 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1717 any reload of register whose subreg is invalid.
1719 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1721 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1722 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1723 Add missing return type - void.
1725 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1727 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1728 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1729 remove element index adjustment for endian (now handled in vsx.md
1731 (altivec_expand_vec_perm_const): Use
1732 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1733 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1734 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1735 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1736 define_expand and a new define_insn *altivec_vspltb_internal;
1737 adjust for -maltivec=be on a little endian target.
1738 (altivec_vspltb_direct): New.
1739 (altivec_vsplth): Divide into a define_expand and a new
1740 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1741 little endian target.
1742 (altivec_vsplth_direct): New.
1743 (altivec_vspltw): Divide into a define_expand and a new
1744 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1745 little endian target.
1746 (altivec_vspltw_direct): New.
1747 (altivec_vspltsf): Divide into a define_expand and a new
1748 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1749 a little endian target.
1751 2014-01-30 Richard Biener <rguenther@suse.de>
1753 PR tree-optimization/59993
1754 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1755 can propagate form the earlier stmt and avoid the transform
1756 when the intermediate result is needed.
1758 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1760 * README.Portability: Fix typo.
1762 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1764 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1765 comparison_operator with ordered_comparison_operator.
1767 2014-01-30 Nick Clifton <nickc@redhat.com>
1769 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1770 Rename to mn10300_store_multiple_regs.
1771 * config/mn10300/mn10300.c: Likewise.
1772 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1773 store_multiple_regs.
1774 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1775 Call mn10300_store_multiple_regs.
1777 2014-01-30 Nick Clifton <nickc@redhat.com>
1778 DJ Delorie <dj@redhat.com>
1780 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1781 %fp 2 to keep registers after it properly word-aligned.
1782 (rl78_alloc_physical_registers_umul): Handle the case where both
1783 input operands are the same.
1785 2014-01-30 Richard Biener <rguenther@suse.de>
1787 PR tree-optimization/59903
1788 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1791 2014-01-30 Jason Merrill <jason@redhat.com>
1794 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1797 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1799 2014-01-30 Richard Biener <rguenther@suse.de>
1801 PR tree-optimization/59951
1802 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1804 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1807 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1808 SFmode to DFmode case.
1810 2014-01-29 DJ Delorie <dj@redhat.com>
1812 * config/msp430/msp430.opt (-minrt): New.
1813 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1815 (ENDFILE_SPEC): Likewise.
1817 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1819 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1820 (estimate_function_body_sizes): Use it.
1822 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1825 * dwarf2out.c (is_cxx_auto): New.
1826 (is_base_type): Use it.
1827 (gen_type_die_with_usage): Likewise.
1829 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1831 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1832 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1833 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1834 -maltivec=be with LE targets.
1835 (vsx_mergeh_<mode>): Likewise.
1836 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1837 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1838 (altivec_vmrghb): Replace with define_expand and new
1839 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1840 (altivec_vmrghb_direct): New define_insn.
1841 (altivec_vmrghh): Replace with define_expand and new
1842 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1843 (altivec_vmrghh_direct): New define_insn.
1844 (altivec_vmrghw): Replace with define_expand and new
1845 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1846 (altivec_vmrghw_direct): New define_insn.
1847 (*altivec_vmrghsf): Adjust for endianness.
1848 (altivec_vmrglb): Replace with define_expand and new
1849 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1850 (altivec_vmrglb_direct): New define_insn.
1851 (altivec_vmrglh): Replace with define_expand and new
1852 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1853 (altivec_vmrglh_direct): New define_insn.
1854 (altivec_vmrglw): Replace with define_expand and new
1855 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1856 (altivec_vmrglw_direct): New define_insn.
1857 (*altivec_vmrglsf): Adjust for endianness.
1858 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1859 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1860 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1861 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1862 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1863 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1864 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1865 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1867 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1869 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1870 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1873 2014-01-29 Richard Biener <rguenther@suse.de>
1875 PR tree-optimization/58742
1876 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1877 associate_pointerplus_align.
1878 (associate_pointerplus_diff): New function.
1879 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1880 and associate_pointerplus_diff.
1882 2014-01-29 Richard Biener <rguenther@suse.de>
1884 * lto-streamer.h (LTO_major_version): Bump to 3.
1885 (LTO_minor_version): Reset to 0.
1887 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1889 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1890 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1891 (arm_file_start): Generate correct asm header for armv7ve.
1892 * config/arm/bpabi.h: Add multilib support for armv7ve.
1893 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1894 and cortex-a15 to armv7ve.
1895 * config/arm/t-aprofile: Add multilib support for armv7ve.
1896 * doc/invoke.texi: Document -march=armv7ve.
1898 2014-01-29 Richard Biener <rguenther@suse.de>
1900 PR tree-optimization/58742
1901 * tree-ssa-forwprop.c (associate_plusminus): Return true
1902 if we changed sth, defer EH cleanup to ...
1903 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1904 (simplify_mult): New function.
1906 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1909 PR tree-optimization/59920
1910 * tree.c (build_common_builtin_nodes): Remove
1911 __builtin_setjmp_dispatcher initialization.
1912 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1913 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1914 instead of gsi_after_labels + manually skipping debug stmts.
1915 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1916 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1917 * tree-inline.c (copy_edges_for_bb): Remove
1918 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1919 argument. Ignore computed_goto_p stmts. Don't call
1920 make_abnormal_goto_edges. If a call might need abnormal edges
1921 for non-local gotos, see if it already has an edge to
1922 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1923 with true argument, don't do anything then, otherwise add
1924 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1925 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1927 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1928 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1929 (lower_stmt): Don't set data->calls_builtin_setjmp.
1930 (lower_builtin_setjmp): Adjust comment.
1931 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1932 * tree-cfg.c (found_computed_goto): Remove.
1933 (factor_computed_gotos): Remove.
1934 (make_goto_expr_edges): Return bool, true for computed gotos.
1935 Don't call make_abnormal_goto_edges.
1936 (build_gimple_cfg): Don't set found_computed_goto, don't call
1937 factor_computed_gotos.
1938 (computed_goto_p): No longer static.
1939 (make_blocks): Don't set found_computed_goto.
1940 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1941 (make_edges): If make_goto_expr_edges returns true, push bb
1942 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1943 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1944 vector. Record mapping between bbs and OpenMP regions if there
1945 are any, adjust make_gimple_omp_edges caller. Call
1946 handle_abnormal_edges.
1947 (make_abnormal_goto_edges): Remove.
1948 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1949 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1950 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1951 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1952 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1953 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1954 filling *region also set *region_idx to (*region)->entry->index.
1957 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1958 For REGs set ORIGINAL_REGNO.
1960 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1962 * doc/md.texi: Mention that a target shouldn't implement
1963 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1966 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1968 PR tree-optimization/59594
1969 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1970 a copy of the datarefs vector rather than the vector itself.
1972 2014-01-28 Jason Merrill <jason@redhat.com>
1975 * dwarf2out.c (auto_die): New static.
1976 (gen_type_die_with_usage): Handle C++1y 'auto'.
1977 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1980 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1983 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1984 (SPEC_X32): Likewise.
1985 (SPEC_64): Likewise.
1986 * config/i386/i386.c (ix86_option_override_internal): Turn off
1987 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1989 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1990 * config/i386/i386.h (TARGET_16BIT): New macro.
1991 (TARGET_16BIT_P): Likewise.
1992 * config/i386/i386.opt: Add m16.
1993 * doc/invoke.texi: Document -m16.
1995 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1997 PR preprocessor/59935
1998 * input.c (location_get_source_line): Bail out on when line number
1999 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2001 2014-01-28 Richard Biener <rguenther@suse.de>
2003 PR tree-optimization/58742
2004 * tree-ssa-forwprop.c (associate_plusminus): Handle
2005 pointer subtraction of the form (T)(P + A) - (T)P.
2007 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2009 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2012 2014-01-28 Richard Biener <rguenther@suse.de>
2015 2014-01-28 Richard Biener <rguenther@suse.de>
2017 PR rtl-optimization/45364
2018 PR rtl-optimization/59890
2019 * var-tracking.c (local_get_addr_clear_given_value): Handle
2020 already cleared slot.
2021 (val_reset): Handle not allocated local_get_addr_cache.
2022 (vt_find_locations): Use post-order on the inverted CFG.
2024 2014-01-28 Richard Biener <rguenther@suse.de>
2026 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2028 2014-01-28 Richard Biener <rguenther@suse.de>
2030 PR rtl-optimization/45364
2031 PR rtl-optimization/59890
2032 * var-tracking.c (local_get_addr_clear_given_value): Handle
2033 already cleared slot.
2034 (val_reset): Handle not allocated local_get_addr_cache.
2035 (vt_find_locations): Use post-order on the inverted CFG.
2037 2014-01-28 Alan Modra <amodra@gmail.com>
2039 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2040 * configure.ac <recursive call for build != host>: Define
2041 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2042 and LD_FOR_BUILD too.
2043 * configure: Regenerate.
2045 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2047 * config/i386/i386.c (get_builtin_code_for_version): Separate
2048 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2049 Broadwell from Haswell.
2051 2014-01-27 Steve Ellcey <sellcey@mips.com>
2053 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2054 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2055 * config/mips/mips.c (mips_option_override): Change setting
2057 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2058 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2059 Change from Mask to Var.
2061 2014-01-27 Jeff Law <law@redhat.com>
2063 * ipa-inline.c (inline_small_functions): Fix typo.
2065 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2067 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2068 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2069 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2070 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2071 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2072 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2073 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2074 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2075 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2076 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2077 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2078 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2079 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2080 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2081 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2082 (_mm512_storeu_epi64): Ditto.
2083 (_mm512_cmpge_epi32_mask): Ditto.
2084 (_mm512_cmpge_epu32_mask): Ditto.
2085 (_mm512_cmpge_epi64_mask): Ditto.
2086 (_mm512_cmpge_epu64_mask): Ditto.
2087 (_mm512_cmple_epi32_mask): Ditto.
2088 (_mm512_cmple_epu32_mask): Ditto.
2089 (_mm512_cmple_epi64_mask): Ditto.
2090 (_mm512_cmple_epu64_mask): Ditto.
2091 (_mm512_cmplt_epi32_mask): Ditto.
2092 (_mm512_cmplt_epu32_mask): Ditto.
2093 (_mm512_cmplt_epi64_mask): Ditto.
2094 (_mm512_cmplt_epu64_mask): Ditto.
2095 (_mm512_cmpneq_epi32_mask): Ditto.
2096 (_mm512_cmpneq_epu32_mask): Ditto.
2097 (_mm512_cmpneq_epi64_mask): Ditto.
2098 (_mm512_cmpneq_epu64_mask): Ditto.
2099 (_mm512_expand_pd): Ditto.
2100 (_mm512_expand_ps): Ditto.
2101 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2102 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2103 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2104 * config/i386/i386.c (ix86_builtins): Add
2105 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2106 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2107 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2108 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2109 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2110 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2111 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2112 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2113 IX86_BUILTIN_PMOVUSQW512_MEM.
2114 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2115 __builtin_ia32_pmovsqd512mem_mask,
2116 __builtin_ia32_pmovqd512mem_mask,
2117 __builtin_ia32_pmovusqw512mem_mask,
2118 __builtin_ia32_pmovsqw512mem_mask,
2119 __builtin_ia32_pmovqw512mem_mask,
2120 __builtin_ia32_pmovusdw512mem_mask,
2121 __builtin_ia32_pmovsdw512mem_mask,
2122 __builtin_ia32_pmovdw512mem_mask,
2123 __builtin_ia32_pmovqb512mem_mask,
2124 __builtin_ia32_pmovusqb512mem_mask,
2125 __builtin_ia32_pmovsqb512mem_mask,
2126 __builtin_ia32_pmovusdb512mem_mask,
2127 __builtin_ia32_pmovsdb512mem_mask,
2128 __builtin_ia32_pmovdb512mem_mask.
2129 (bdesc_args): Add __builtin_ia32_expanddf512,
2130 __builtin_ia32_expandsf512.
2131 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2132 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2133 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2134 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2135 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2136 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2137 (avx512f_<code>v8div16qi2_mask_store): This.
2138 (avx512f_expand<mode>): New.
2140 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2142 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2144 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2145 (_mm512_prefetch_i32scatter_pd): Ditto.
2146 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2147 (_mm512_prefetch_i64scatter_pd): Ditto.
2148 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2149 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2150 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2151 (_mm512_prefetch_i32scatter_ps): Ditto.
2152 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2153 (_mm512_prefetch_i64scatter_ps): Ditto.
2154 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2155 * config/i386/i386-builtin-types.def: Define
2156 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2157 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2158 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2159 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2160 IX86_BUILTIN_SCATTERPFQPD.
2161 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2162 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2163 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2164 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2165 __builtin_ia32_scatterpfqps.
2166 (ix86_expand_builtin): Expand new built-ins.
2167 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2168 fix memory access data type.
2169 (*avx512pf_gatherpf<mode>_mask): Ditto.
2170 (*avx512pf_gatherpf<mode>): Ditto.
2171 (avx512pf_scatterpf<mode>): Ditto.
2172 (*avx512pf_scatterpf<mode>_mask): Ditto.
2173 (*avx512pf_scatterpf<mode>): Ditto.
2174 (GATHER_SCATTER_SF_MEM_MODE): New.
2175 (avx512pf_gatherpf<mode>df): Ditto.
2176 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2177 (*avx512pf_scatterpf<mode>df): Ditto.
2179 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2182 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2183 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2186 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2188 * common/config/arm/arm-common.c
2189 (arm_rewrite_mcpu): Handle multiple names.
2191 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2193 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2195 * gimple-builder.h (create_gimple_tmp): Delete.
2197 2014-01-27 Christian Bruel <christian.bruel@st.com>
2199 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2202 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2204 * config/pa/pa.md (call): Generate indirect long calls to non-local
2205 functions when outputing 32-bit code.
2206 (call_value): Likewise except for special call to buggy powf function.
2208 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2209 portable runtime and PIC indirect calls.
2210 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2211 and PIC call sequences. Use ldo instead of blr to set return register
2212 in PIC call sequence.
2214 2014-01-25 Walter Lee <walt@tilera.com>
2216 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2217 avoid clobbering a live register.
2219 2014-01-25 Walter Lee <walt@tilera.com>
2221 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2222 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2223 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2224 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2226 2014-01-25 Walter Lee <walt@tilera.com>
2228 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2229 arguments on even registers.
2230 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2232 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2233 (BIGGEST_ALIGNMENT): Ditto.
2234 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2236 2014-01-25 Walter Lee <walt@tilera.com>
2238 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2239 insns before bundling.
2240 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2242 2014-01-25 Walter Lee <walt@tilera.com>
2244 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2245 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2246 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2248 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2250 * config/mips/constraints.md (kl): Delete.
2251 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2252 define expands, using...
2253 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2254 instructions for MIPS16.
2255 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2256 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2258 2014-01-25 Walter Lee <walt@tilera.com>
2260 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2264 2014-01-25 Walter Lee <walt@tilera.com>
2266 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2267 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2269 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2271 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2274 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2276 * print-rtl.c (in_call_function_usage): New var.
2277 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2278 EXPR_LIST mode as mode and not as reg note name.
2281 * cfgloopmanip.c (copy_loop_info): If
2282 loop->warned_aggressive_loop_optimizations, make sure
2283 the flag is set in target loop too.
2285 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2287 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2289 * builtins.def: Likewise.
2290 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2291 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2292 * ira.c (ira_setup_eliminable_regset): Likewise.
2293 * omp-low.c (gate_expand_omp): Likewise.
2294 (execute_lower_omp): Likewise.
2295 (diagnose_sb_0): Likewise.
2296 (gate_diagnose_omp_blocks): Likewise.
2297 (simd_clone_clauses_extract): Likewise.
2300 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2302 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2303 correction for little endian...
2304 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2307 2014-01-24 Jeff Law <law@redhat.com>
2309 PR tree-optimization/59919
2310 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2311 for non-returning calls.
2313 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2315 * common/config/aarch64/aarch64-common.c
2316 (aarch64_rewrite_mcpu): Handle multiple names.
2317 * config/aarch64/aarch64.h
2318 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2320 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2322 * input.c (add_file_to_cache_tab): Handle the case where fopen
2325 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2328 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2329 from push operand if code of push isn't PRE_DEC.
2331 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2334 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2335 -mquad-memory-atomic. Update -mquad-memory documentation to say
2336 it is only used for non-atomic loads/stores.
2338 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2339 -mquad-memory or -mquad-memory-atomic switches.
2341 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2342 -mquad-memory-atomic to ISA 2.07 support.
2344 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2345 to separate support of normal quad word memory operations (ldq, stq)
2346 from the atomic quad word memory operations.
2348 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2349 support to separate non-atomic quad word operations from atomic
2350 quad word operations. Disable non-atomic quad word operations in
2351 little endian mode so that we don't have to swap words after the
2352 load and before the store.
2353 (quad_load_store_p): Add comment about atomic quad word support.
2354 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2355 options printed with -mdebug=reg.
2357 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2358 -mquad-memory-atomic as the test for whether we have quad word
2359 atomic instructions.
2360 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2361 or -mp8-vector are used, allow byte/half-word atomic operations.
2363 * config/rs6000/sync.md (load_lockedti): Insure that the address
2364 is a proper indexed or indirect address for the lqarx instruction.
2365 On little endian systems, swap the hi/lo registers after the lqarx
2367 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2368 insure the address is valid for the lqarx instruction.
2369 (store_conditionalti): Insure that the address is a proper indexed
2370 or indirect address for the stqcrx. instruction. On little endian
2371 systems, swap the hi/lo registers before doing the stqcrx.
2373 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2374 insure the address is valid for the stqcrx. instruction.
2376 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2377 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2378 type of quad memory support is available.
2380 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2383 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2384 there is a danger of looping.
2386 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2388 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2389 force flag_ira_loop_pressure if set via command line.
2391 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2393 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2394 (ashr_simd): New builtin handling DI mode.
2395 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2396 (aarch64_sshr_simddi): New match pattern.
2397 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2398 (vshrd_n_s64): Likewise.
2399 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2401 2014-01-23 Nick Clifton <nickc@redhat.com>
2403 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2404 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2405 favour of mcu specific scripts.
2406 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2409 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2410 Alex Velenko <Alex.Velenko@arm.com>
2412 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2413 (vaddv_s16): Likewise.
2414 (vaddv_s32): Likewise.
2415 (vaddv_u8): Likewise.
2416 (vaddv_u16): Likewise.
2417 (vaddv_u32): Likewise.
2418 (vaddvq_s8): Likewise.
2419 (vaddvq_s16): Likewise.
2420 (vaddvq_s32): Likewise.
2421 (vaddvq_s64): Likewise.
2422 (vaddvq_u8): Likewise.
2423 (vaddvq_u16): Likewise.
2424 (vaddvq_u32): Likewise.
2425 (vaddvq_u64): Likewise.
2426 (vaddv_f32): Likewise.
2427 (vaddvq_f32): Likewise.
2428 (vaddvq_f64): Likewise.
2429 (vmaxv_f32): Likewise.
2430 (vmaxv_s8): Likewise.
2431 (vmaxv_s16): Likewise.
2432 (vmaxv_s32): Likewise.
2433 (vmaxv_u8): Likewise.
2434 (vmaxv_u16): Likewise.
2435 (vmaxv_u32): Likewise.
2436 (vmaxvq_f32): Likewise.
2437 (vmaxvq_f64): Likewise.
2438 (vmaxvq_s8): Likewise.
2439 (vmaxvq_s16): Likewise.
2440 (vmaxvq_s32): Likewise.
2441 (vmaxvq_u8): Likewise.
2442 (vmaxvq_u16): Likewise.
2443 (vmaxvq_u32): Likewise.
2444 (vmaxnmv_f32): Likewise.
2445 (vmaxnmvq_f32): Likewise.
2446 (vmaxnmvq_f64): Likewise.
2447 (vminv_f32): Likewise.
2448 (vminv_s8): Likewise.
2449 (vminv_s16): Likewise.
2450 (vminv_s32): Likewise.
2451 (vminv_u8): Likewise.
2452 (vminv_u16): Likewise.
2453 (vminv_u32): Likewise.
2454 (vminvq_f32): Likewise.
2455 (vminvq_f64): Likewise.
2456 (vminvq_s8): Likewise.
2457 (vminvq_s16): Likewise.
2458 (vminvq_s32): Likewise.
2459 (vminvq_u8): Likewise.
2460 (vminvq_u16): Likewise.
2461 (vminvq_u32): Likewise.
2462 (vminnmv_f32): Likewise.
2463 (vminnmvq_f32): Likewise.
2464 (vminnmvq_f64): Likewise.
2466 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2468 * config/aarch64/aarch64-simd.md
2469 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2470 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2471 (*aarch64_mul3_elt<mode>): Likewise.
2472 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2473 (*aarch64_mul3_elt_to_64v2df): Likewise.
2474 (*aarch64_mla_elt<mode>): Likewise.
2475 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2476 (*aarch64_mls_elt<mode>): Likewise.
2477 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2478 (*aarch64_fma4_elt<mode>): Likewise.
2479 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2480 (*aarch64_fma4_elt_to_64v2df): Likewise.
2481 (*aarch64_fnma4_elt<mode>): Likewise.
2482 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2483 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2484 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2485 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2486 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2487 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2488 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2489 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2490 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2492 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2494 * config/aarch64/aarch64-simd.md
2495 (aarch64_be_checked_get_lane<mode>): New define_expand.
2496 * config/aarch64/aarch64-simd-builtins.def
2497 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2498 New builtin definition.
2499 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2500 Use new safe be builtin.
2502 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2504 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2506 (aarch64_be_st1<mode>): Likewise.
2507 (aarch_ld1<VALL:mode>): Define_expand modified.
2508 (aarch_st1<VALL:mode>): Likewise.
2509 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2510 (UNSPEC_ST1): Likewise.
2512 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2514 * config/microblaze/microblaze.md: Add trap insn and attribute
2516 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2518 PR preprocessor/58580
2519 * input.h (location_get_source_line): Take an additional line_size
2521 (void diagnostics_file_cache_fini): Declare new function.
2522 * input.c (struct fcache): New type.
2523 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2524 New static constants.
2525 (diagnostic_file_cache_init, total_lines_num)
2526 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2527 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2528 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2529 (get_next_line, read_next_line, goto_next_line, read_line_num):
2530 New static function definitions.
2531 (diagnostic_file_cache_fini): New function.
2532 (location_get_source_line): Take an additional output line_len
2533 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2535 * diagnostic.c (diagnostic_finish): Call
2536 diagnostic_file_cache_fini.
2537 (adjust_line): Take an additional input parameter for the length
2538 of the line, rather than calculating it with strlen.
2539 (diagnostic_show_locus): Adjust the use of
2540 location_get_source_line and adjust_line with respect to their new
2541 signature. While displaying a line now, do not stop at the first
2542 null byte. Rather, display the zero byte as a space and keep
2543 going until we reach the size of the line.
2544 * Makefile.in: Add vec.o to OBJS-libcommon
2546 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2547 Ilya Tocar <ilya.tocar@intel.com>
2549 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2550 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2551 (__builtin_ia32_kmov16): Ditto.
2552 * config/i386/i386.md (UNSPEC_KMOV): New.
2555 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2557 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2558 (_mm512_storeu_si512): Ditto.
2560 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2563 * rtl.h (get_referenced_operands): Declare.
2564 * recog.c (get_referenced_operands): New function.
2565 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2566 operands have been referenced when recording LO_SUM references.
2568 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2570 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2572 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2574 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2575 Enable for generic and recent AMD targets.
2577 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2579 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2580 ARG_SIZE note when adjustment was eliminated.
2582 2014-01-22 Jeff Law <law@redhat.com>
2584 PR tree-optimization/59597
2585 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2586 in file. Accept new argument REGISTERING and use it to modify
2587 dump output appropriately.
2588 (register_jump_thread): Corresponding changes.
2589 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2590 thread paths involving joiner blocks. Add code to dump cancelled
2591 jump threading paths.
2593 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2595 PR rtl-optimization/59477
2596 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2597 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2599 2014-01-22 Tom Tromey <tromey@redhat.com>
2601 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2603 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2605 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2607 PR rtl-optimization/59896
2608 * lra-constraints.c (process_alt_operands): Check unused note for
2609 matched operands of insn with no output reloads.
2611 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2613 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2614 (mips_move_from_gpr_cost): Likewise.
2616 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2618 PR rtl-optimization/59858
2619 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2620 ira_class_hard_regs_num.
2621 (process_alt_operands): Increase reject for dying matched operand.
2623 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2626 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2627 smaller than size, perform several stores or loads and stores
2628 at dst + count - size to store or copy all of size bytes, rather
2629 than just last modesize bytes.
2631 2014-01-20 DJ Delorie <dj@redhat.com>
2633 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2634 that CLOBBERs are REGs before propogating their values.
2636 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2639 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2640 (cgraph_inline_failed_type): New function.
2641 * cgraph.h (DEFCIFCODE): Add type.
2642 (cgraph_inline_failed_type_t): New enum.
2643 (cgraph_inline_failed_type): New prototype.
2644 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2645 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2646 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2647 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2648 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2649 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2650 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2651 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2652 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2653 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2654 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2655 OPTIMIZATION_MISMATCH.
2656 * tree-inline.c (expand_call_inline): Emit errors during
2657 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2659 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2662 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2663 mode attribute in insn output.
2665 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2667 * output.h (output_constant): Delete.
2668 * varasm.c (output_constant): Make private.
2670 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2672 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2674 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2677 * tree.h (fold_builtin_strcat): New prototype.
2678 * builtins.c (fold_builtin_strcat): No longer static. Add len
2679 argument, if non-NULL, don't call c_strlen. Optimize
2680 directly into __builtin_memcpy instead of __builtin_strcpy.
2681 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2682 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2684 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2686 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2687 for SImode_address_operand operands, having only a REG argument.
2689 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2691 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2692 loader name using mbig-endian.
2693 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2695 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2697 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2701 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2703 * config/aarch64/aarch64-protos.h
2704 (aarch64_cannot_change_mode_class_ptr): Declare.
2705 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2706 aarch64_cannot_change_mode_class_ptr): New.
2707 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2708 backend hook aarch64_cannot_change_mode_class.
2710 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2712 * common/config/aarch64/aarch64-common.c
2713 (aarch64_handle_option): Don't handle any option order logic here.
2714 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2715 selected_cpu, warn on architecture version mismatch.
2716 (aarch64_override_options): Fix parsing order for option strings.
2718 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2719 Iain Sandoe <iain@codesourcery.com>
2722 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2723 warning. Amend comment to reflect current functionality.
2725 2014-01-20 Richard Biener <rguenther@suse.de>
2728 * builtins.c (fold_builtin_strcat): Remove case better handled
2729 by tree-ssa-strlen.c.
2731 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2733 * config/aarch64/aarch64.opt
2734 (mcpu, march, mtune): Make case-insensitive.
2736 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2739 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2740 if operands[1] is a REG or ZERO_EXTEND of a REG.
2742 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2744 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2746 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2748 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2749 long non-pic millicode calls.
2751 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2753 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2755 2014-01-19 Kito Cheng <kito@0xlab.org>
2757 * builtins.c (expand_movstr): Check movstr expand done or fail.
2759 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2760 H.J. Lu <hongjiu.lu@intel.com>
2763 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2764 to DImode for zero-extended addresses.
2766 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2768 PR rtl-optimization/57763
2769 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2770 on the new indirect jump_insn and increment LABEL_NUSES (label).
2772 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2776 * config.gcc (x86_archs): New variable.
2777 (x86_64_archs): Likewise.
2778 (x86_cpus): Likewise.
2779 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2780 --with-arch/--with-cpu= options.
2781 Support --with-arch=/--with-cpu={nehalem,westmere,
2782 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2784 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2786 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2787 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2789 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2791 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2793 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2796 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2797 clear cpp_get_options (parse_in)->warn_unused_macros for
2798 ix86_target_macros_internal with cpp_define.
2800 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2802 * jump.c (delete_related_insns): Keep (use (insn))s.
2803 * reorg.c (redundant_insn): Check for barriers too.
2805 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2807 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2809 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2811 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2812 call to $$dyncall when TARGET_LONG_CALLS is true.
2814 2014-01-17 Jeff Law <law@redhat.com>
2816 * ree.c (combine_set_extension): Temporarily disable test for
2817 changing number of hard registers.
2819 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2822 * ipa-inline-analysis.c (inline_free_summary):
2823 Do not free summary of aliases.
2825 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2828 * gimplify.c (gimplify_expr): Use create_tmp_var
2829 instead of create_tmp_var_raw. If cond doesn't have
2830 integral type, don't add the IFN_ANNOTATE builtin at all.
2832 2014-01-17 Martin Jambor <mjambor@suse.cz>
2835 * ipa-cp.c (prev_edge_clone): New variable.
2836 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2837 Also resize prev_edge_clone vector.
2838 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2839 (ipcp_edge_removal_hook): New function.
2840 (ipcp_driver): Register ipcp_edge_removal_hook.
2842 2014-01-17 Andrew Pinski <apinski@cavium.com>
2843 Steve Ellcey <sellcey@mips.com>
2846 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2849 2014-01-17 Jeff Law <law@redhat.com>
2852 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2853 so that pass_ccp runs first.
2855 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2857 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2858 (ix86_adjust_cost): Use !TARGET_XXX.
2859 (do_reorder_for_imul): Likewise.
2860 (swap_top_of_ready_list): Likewise.
2861 (ix86_sched_reorder): Likewise.
2863 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2865 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2866 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2867 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2868 (intel_memset): New. Duplicate slm_memset.
2869 (intel_cost): New. Duplicate slm_cost.
2870 (m_INTEL): New macro.
2871 (processor_target_table): Add "intel".
2872 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2873 with PROCESSOR_INTEL for "intel".
2874 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2875 PROCESSOR_SILVERMONT.
2876 (ix86_issue_rate): Likewise.
2877 (ix86_adjust_cost): Likewise.
2878 (ia32_multipass_dfa_lookahead): Likewise.
2879 (swap_top_of_ready_list): Likewise.
2880 (ix86_sched_reorder): Likewise.
2881 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2882 instead of TARGET_OPT_AGU.
2883 * config/i386/i386.h (TARGET_INTEL): New.
2884 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2885 (processor_type): Add PROCESSOR_INTEL.
2886 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2887 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2889 2014-01-17 Marek Polacek <polacek@redhat.com>
2892 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2895 2014-01-17 Richard Biener <rguenther@suse.de>
2897 PR tree-optimization/46590
2898 * opts.c (default_options_table): Add entries for
2899 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2900 all enabled at -O1 but not for -Og.
2901 * common.opt (fbranch-count-reg): Remove Init(1).
2902 (fmove-loop-invariants): Likewise.
2903 (ftree-pta): Likewise.
2905 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2907 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2908 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2909 decls to at least the GCC 4.8 used alignments.
2912 * tree-nested.c (convert_nonlocal_reference_stmt,
2913 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2914 of GIMPLE_BIND stmts, adjust associated decls.
2916 2014-01-17 Richard Biener <rguenther@suse.de>
2918 PR tree-optimization/46590
2919 * vec.h (vec<>::bseach): New member function implementing
2920 binary search according to C89 bsearch.
2921 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2922 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2923 bitmap pointer again. Make accesses_in_loop a flat array.
2924 (mem_ref_obstack): New global.
2925 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2926 (mark_ref_stored): Likewise.
2927 (ref_indep_loop_p_2): Likewise.
2928 (set_ref_stored_in_loop): New helper function.
2929 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2930 (memref_free): Adjust.
2931 (record_mem_ref_loc): Simplify.
2932 (gather_mem_refs_stmt): Adjust.
2933 (sort_locs_in_loop_postorder_cmp): New function.
2934 (analyze_memory_references): Sort accesses_in_loop after
2935 loop postorder number.
2936 (find_ref_loc_in_loop_cmp): New function.
2937 (for_all_locs_in_loop): Find relevant cluster of locs in
2938 accesses_in_loop and iterate without recursion.
2939 (execute_sm): Avoid uninit warning.
2940 (struct ref_always_accessed): Simplify.
2941 (ref_always_accessed::operator ()): Likewise.
2942 (ref_always_accessed_p): Likewise.
2943 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2944 loop postorder numbers here.
2945 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2948 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2951 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2952 on decls for which assemble_alias has been called.
2954 2014-01-17 Nick Clifton <nickc@redhat.com>
2956 * config/msp430/msp430.opt: (mcpu): New option.
2957 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2958 (msp430_option_override): Parse target_cpu. If the MCU name
2959 matches a generic string, clear target_mcu.
2960 (msp430_attr): Allow numeric interrupt values up to 63.
2961 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2962 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2964 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2966 * config/msp430/msp430.md (popm): Use %J rather than %I.
2967 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2968 (addhi_cy_i): Use immediate_operand for operand 2.
2969 * doc/invoke.texi: Document -mcpu option.
2971 2014-01-17 Richard Biener <rguenther@suse.de>
2973 PR rtl-optimization/38518
2974 * df.h (df_analyze_loop): Declare.
2975 * df-core.c: Include cfgloop.h.
2976 (df_analyze_1): Split out main part of df_analyze.
2977 (df_analyze): Adjust.
2978 (loop_inverted_post_order_compute): New function.
2979 (loop_post_order_compute): Likewise.
2980 (df_analyze_loop): New function avoiding whole-function
2982 * loop-invariant.c (find_defs): Use df_analyze_loop.
2983 (find_invariants): Adjust.
2984 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2986 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2988 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2989 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2991 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2993 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2994 traversal when removing references.
2996 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2999 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3001 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3004 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3005 pushing a reload for an autoinc when we had previously reloaded an
3006 inner part of the address.
3008 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3010 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3012 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3013 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3014 when not giving up or versioning for alias only because of
3016 (vect_analyze_data_ref_dependences): Set to true.
3017 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3019 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3020 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3024 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3027 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3028 operand 0 predicate for gathers, use a new pseudo as subtarget.
3030 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3033 * lra-constraints.c (process_alt_operands): Add printing debug info.
3034 Check absence of input/output reloads for matched operands too.
3036 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3038 PR rtl-optimization/59835
3039 * ira.c (ira_init_register_move_cost): Increase cost for
3042 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3044 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3046 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3049 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3050 non-register objects. Use gen_(high/low)part more consistently.
3053 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3056 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3057 endian support, remove tests for WORDS_BIG_ENDIAN.
3058 (p8_mfvsrd_3_<mode>): Likewise.
3059 (reload_gpr_from_vsx<mode>): Likewise.
3060 (reload_gpr_from_vsxsf): Likewise.
3061 (p8_mfvsrd_4_disf): Likewise.
3063 2014-01-16 Richard Biener <rguenther@suse.de>
3065 PR rtl-optimization/46590
3066 * lcm.c (compute_antinout_edge): Use postorder iteration.
3067 (compute_laterin): Use inverted postorder iteration.
3069 2014-01-16 Nick Clifton <nickc@redhat.com>
3072 * varasm.c (output_constant): Return the number of bytes actually
3074 (output_constructor_array_range): Update the field size with the
3075 number of bytes emitted by output_constant.
3076 (output_constructor_regular_field): Likewise. Also do not
3077 complain if the total number of bytes emitted is now greater
3078 than the expected fieldpos.
3079 * output.h (output_constant): Update prototype and descriptive comment.
3081 2014-01-16 Marek Polacek <polacek@redhat.com>
3084 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3085 it is error_mark_node.
3087 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3089 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3090 VALID_AVX256_REG_OR_OI_MODE.
3092 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3094 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3095 current procedure should be profiled.
3097 2014-01-15 Andrew Pinski <apinski@cavium.com>
3099 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3100 of moving from/to the STACK_REG register class.
3102 2014-01-15 Richard Henderson <rth@redhat.com>
3105 * reginfo.c (global_regs_decl): Globalize.
3106 * rtl.h (global_regs_decl): Declare.
3107 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3108 reserved via global_regs.
3110 2014-01-15 Teresa Johnson <tejohnson@google.com>
3112 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3114 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3116 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3117 and vmulosh rather than call gen_vec_widen_smult_*.
3118 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3119 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3120 (vec_widen_smult_even_v16qi): Likewise.
3121 (vec_widen_umult_even_v8hi): Likewise.
3122 (vec_widen_smult_even_v8hi): Likewise.
3123 (vec_widen_umult_odd_v16qi): Likewise.
3124 (vec_widen_smult_odd_v16qi): Likewise.
3125 (vec_widen_umult_odd_v8hi): Likewise.
3126 (vec_widen_smult_odd_v8hi): Likewise.
3127 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3128 vmuloub rather than call gen_vec_widen_umult_*.
3129 (vec_widen_umult_lo_v16qi): Likewise.
3130 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3131 vmulosb rather than call gen_vec_widen_smult_*.
3132 (vec_widen_smult_lo_v16qi): Likewise.
3133 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3134 rather than call gen_vec_widen_umult_*.
3135 (vec_widen_umult_lo_v8hi): Likewise.
3136 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3137 rather than call gen_vec_widen_smult_*.
3138 (vec_widen_smult_lo_v8hi): Likewise.
3140 2014-01-15 Jeff Law <law@redhat.com>
3142 PR tree-optimization/59747
3143 * ree.c (find_and_remove_re): Properly handle case where a second
3144 eliminated extension requires widening a copy created for elimination
3145 of a prior extension.
3146 (combine_set_extension): Ensure that the number of hard regs needed
3147 for a destination register does not change when we widen it.
3149 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3151 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3152 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3153 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3154 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3155 (avr-*-rtems*): Likewise.
3156 (bfin*-rtems*): Likewise.
3157 (moxie-*-rtems*): Likewise.
3158 (h8300-*-rtems*): Likewise.
3159 (i[34567]86-*-rtems*): Likewise.
3160 (lm32-*-rtems*): Likewise.
3161 (m32r-*-rtems*): Likewise.
3162 (m68k-*-rtems*): Likewise.
3163 (microblaze*-*-rtems*): Likewise.
3164 (mips*-*-rtems*): Likewise.
3165 (powerpc-*-rtems*): Likewise.
3166 (sh-*-rtems*): Likewise.
3167 (sparc-*-rtems*): Likewise.
3168 (sparc64-*-rtems*): Likewise.
3169 (v850-*-rtems*): Likewise.
3170 (m32c-*-rtems*): Likewise.
3172 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3174 PR rtl-optimization/59511
3175 * ira.c (ira_init_register_move_cost): Use memory costs for some
3176 cases of register move cost calculations.
3177 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3178 instead of BB frequency.
3179 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3180 * lra-assigns.c (find_hard_regno_for): Ditto.
3182 2014-01-15 Richard Biener <rguenther@suse.de>
3184 PR tree-optimization/59822
3185 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3186 (vectorizable_load): Use it to hoist defs of uses of invariant
3187 loads out of the loop.
3189 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3190 Kugan Vivekanandarajah <kuganv@linaro.org>
3193 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3196 2014-01-15 Richard Biener <rguenther@suse.de>
3198 PR rtl-optimization/59802
3199 * lcm.c (compute_available): Use inverted postorder to seed
3200 the initial worklist.
3202 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3205 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3206 ADDR_REGS for invalid symrefs in non-PIC code.
3208 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3211 * builtins.c (determine_block_size): Initialize *probable_max_size
3212 even if len_rtx is CONST_INT.
3214 2014-01-14 Andrew Pinski <apinski@cavium.com>
3216 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3217 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3218 (cortexa53_tunings): Likewise.
3219 (aarch64_sched_issue_rate): New function.
3220 (TARGET_SCHED_ISSUE_RATE): Define.
3222 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3224 * ira-costs.c (find_costs_and_classes): Add missed
3225 ira_init_register_move_cost_if_necessary.
3227 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3230 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3232 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3235 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3236 to indicate if type is used for function return value. Warn ABI
3237 change if the vector mode isn't available for function return value.
3238 (ix86_function_arg_advance): Pass false to type_natural_mode.
3239 (ix86_function_arg): Likewise.
3240 (ix86_gimplify_va_arg): Likewise.
3241 (function_arg_32): Don't warn ABI change.
3242 (ix86_function_value): Pass true to type_natural_mode.
3243 (ix86_return_in_memory): Likewise.
3244 (ix86_struct_value_rtx): Removed.
3245 (TARGET_STRUCT_VALUE_RTX): Likewise.
3247 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3249 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3250 converting a conditional jump into a conditional return.
3252 2014-01-14 Richard Biener <rguenther@suse.de>
3254 PR tree-optimization/58921
3255 PR tree-optimization/59006
3256 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3257 hoisting invariant stmts.
3258 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3259 invariant loads on the preheader edge if possible.
3261 2014-01-14 Joey Ye <joey.ye@arm.com>
3263 * doc/plugin.texi (Building GCC plugins): Update to C++.
3265 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3267 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3268 (_mm_rcp28_round_ss): Ditto.
3269 (_mm_rsqrt28_round_sd): Ditto.
3270 (_mm_rsqrt28_round_ss): Ditto.
3271 (_mm_rcp28_sd): Ditto.
3272 (_mm_rcp28_ss): Ditto.
3273 (_mm_rsqrt28_sd): Ditto.
3274 (_mm_rsqrt28_ss): Ditto.
3275 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3276 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3277 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3278 (IX86_BUILTIN_RCP28SD): Ditto.
3279 (IX86_BUILTIN_RCP28SS): Ditto.
3280 (IX86_BUILTIN_RSQRT28SD): Ditto.
3281 (IX86_BUILTIN_RSQRT28SS): Ditto.
3282 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3283 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3284 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3285 (ix86_expand_special_args_builtin): Expand new FTYPE.
3286 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3287 (srcp14<mode>): Make insn unary.
3288 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3289 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3290 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3291 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3292 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3293 Fix rounding: make it SAE only.
3294 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3296 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3298 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3299 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3300 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3301 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3302 (round_saeonly_mask_scalar_operand4): Ditto.
3303 (round_saeonly_mask_scalar_op3): Ditto.
3304 (round_saeonly_mask_scalar_op4): Ditto.
3306 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3308 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3309 Implement -maltivec=be for vec_insert and vec_extract.
3311 2014-01-10 DJ Delorie <dj@redhat.com>
3313 * config/msp430/msp430.md (call_internal): Don't allow memory
3314 references with SP as the base register.
3315 (call_value_internal): Likewise.
3316 * config/msp430/constraints.md (Yc): New. For memory references
3317 that don't use SP as a base register.
3319 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3320 "an integer without a # prefix"
3321 * config/msp430/msp430.md (epilogue_helper): Use it.
3323 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3326 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3327 AVX512F gather builtins.
3328 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3329 on gather decls with INTEGER_TYPE masktype.
3330 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3331 directly into the builtin rather than hoisting it before loop.
3333 PR tree-optimization/59387
3334 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3335 (scev_const_prop): If folded_casts and type has undefined overflow,
3336 use force_gimple_operand instead of force_gimple_operand_gsi and
3337 for each added stmt if it is assign with
3338 arith_code_with_undefined_signed_overflow, call
3339 rewrite_to_defined_overflow.
3340 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3341 gimple-fold.h instead.
3342 (arith_code_with_undefined_signed_overflow,
3343 rewrite_to_defined_overflow): Moved to ...
3344 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3345 rewrite_to_defined_overflow): ... here. No longer static.
3346 Include gimplify-me.h.
3347 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3348 rewrite_to_defined_overflow): New prototypes.
3350 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3352 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3354 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3356 * builtins.c (get_object_alignment_2): Minor tweak.
3357 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3359 2014-01-13 Christian Bruel <christian.bruel@st.com>
3361 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3362 optimized non constant lengths.
3364 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3367 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3368 load as __atomic_load_N if possible.
3370 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3372 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3374 (rs6000_expand_builtin): Adjust call.
3376 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3379 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3380 * config/rs6000/rs6000.c: Include target-globals.h.
3381 (rs6000_set_current_function): Instead of doing target_reinit
3382 unconditionally, use save_target_globals_default_opts and
3383 restore_target_globals.
3385 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3387 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3388 (rs6000_expand_builtin): Handle mffs and mtfsf.
3389 (rs6000_init_builtins): Define mffs and mtfsf.
3390 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3391 (rs6000_mffs): New pattern.
3392 (rs6000_mtfsf): New pattern.
3394 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3396 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3397 Start narrowing with START. Apply candidate-use pair
3398 and check overall cost in narrowing.
3399 (iv_ca_prune): Pass new argument.
3401 2014-01-10 Jeff Law <law@redhat.com>
3404 * ree.c (combine_reaching_defs): Ensure the defining statement
3405 occurs before the extension when optimizing extensions with
3406 different source and destination hard registers.
3408 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3411 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3412 vtables into the type inheritance graph.
3414 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3416 PR rtl-optimization/59754
3417 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3418 modes in the REGNO != REGNO case.
3420 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3422 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3424 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3426 PR tree-optimization/59745
3427 * tree-predcom.c (tree_predictive_commoning_loop): Call
3428 free_affine_expand_cache if giving up because components is NULL.
3430 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3431 GC in payload of target_globals struct instead of allocating them on
3432 the heap and the larger structs separately using GC.
3433 * target-globals.h (struct target_globals): Make regs, hard_regs,
3434 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3435 of GTY((skip)) and change type to void *.
3436 (reset_target_globals): Cast loads from those fields to corresponding
3439 2014-01-10 Steve Ellcey <sellcey@mips.com>
3442 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3443 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3444 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3446 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3449 * aarch64-modes.def (CC_Zmode): New flags mode.
3450 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3451 represents an equality.
3452 (aarch64_get_condition_code): Handle CC_Zmode.
3453 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3455 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3457 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3458 extraction in good case.
3460 2014-01-10 Richard Biener <rguenther@suse.de>
3462 PR tree-optimization/59374
3463 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3464 checking after SLP discovery. Mark stmts not participating
3465 in any SLP instance properly.
3467 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3469 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3470 when handling a SET rtx.
3472 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3474 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3475 (cortex-a57): Likewise.
3476 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3478 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3480 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3481 non-iwmmxt builtins.
3483 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3487 * ipa-devirt.c record_target_from_binfo): Take as argument
3488 stack of binfos and lookup matching one for virtual inheritance.
3489 (possible_polymorphic_call_targets_1): Update.
3491 2014-01-10 Huacai Chen <chenhc@lemote.com>
3493 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3494 kernel strings for Loongson-2E/2F/3A.
3496 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3499 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3500 is_gimple_call before calling gimple_call_internal_p.
3502 2014-01-09 Steve Ellcey <sellcey@mips.com>
3504 * Makefile.in (TREE_FLOW_H): Remove.
3505 (TREE_SSA_H): Add file names from tree-flow.h.
3506 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3507 * tree.h: Remove tree-flow.h reference.
3508 * hash-table.h: Remove tree-flow.h reference.
3509 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3510 reference with tree-ssa-loop.h.
3512 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3514 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3515 default element-order behavior for -maltivec.
3516 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3517 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3518 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3519 when targeting big endian, at least for now.
3520 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3522 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3525 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3526 var satisfies use_register_for_decl, just take into account type
3527 alignment, rather than decl alignment.
3529 PR tree-optimization/59622
3530 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3531 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3532 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3533 Don't devirtualize for inplace at all. For targets.length () == 1,
3534 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3536 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3538 * config/i386/i386.md (cpu): Remove the unused btver1.
3540 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3542 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3544 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3547 * tree-core.h (struct target_globals): New forward declaration.
3548 (struct tree_target_option): Add globals field.
3549 * tree.h (TREE_TARGET_GLOBALS): Define.
3550 (prepare_target_option_nodes_for_pch): New prototype.
3551 * target-globals.h (struct target_globals): Define even if
3553 * tree.c (prepare_target_option_node_for_pch,
3554 prepare_target_option_nodes_for_pch): New functions.
3555 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3556 * config/i386/i386.c: Include target-globals.h.
3557 (ix86_set_current_function): Instead of doing target_reinit
3558 unconditionally, use save_target_globals_default_opts and
3559 restore_target_globals.
3561 2014-01-09 Richard Biener <rguenther@suse.de>
3563 PR tree-optimization/59715
3564 * tree-cfg.h (split_critical_edges): Declare.
3565 * tree-cfg.c (split_critical_edges): Export.
3566 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3568 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3570 * cfgexpand.c (expand_stack_vars): Optionally disable
3571 asan stack protection.
3572 (expand_used_vars): Likewise.
3573 (partition_stack_vars): Likewise.
3574 * asan.c (asan_emit_stack_protection): Optionally disable
3575 after return stack usage.
3576 (instrument_derefs): Optionally disable memory access instrumentation.
3577 (instrument_builtin_call): Likewise.
3578 (instrument_strlen_call): Likewise.
3579 (asan_protect_global): Optionally disable global variables protection.
3580 * doc/invoke.texi: Added doc for new options.
3581 * params.def: Added new options.
3582 * params.h: Likewise.
3584 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3586 PR rtl-optimization/59724
3587 * ifcvt.c (cond_exec_process_if_block): Don't call
3588 flow_find_head_matching_sequence with 0 longest_match.
3589 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3590 non-active insns if !stop_after.
3591 (try_head_merge_bb): Revert 2014-01-07 changes.
3593 2014-01-08 Jeff Law <law@redhat.com>
3595 * ree.c (get_sub_rtx): New function, extracted from...
3596 (merge_def_and_ext): Here.
3597 (combine_reaching_defs): Use get_sub_rtx.
3599 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3601 * cgraph.h (varpool_variable_node): Do not choke on null node.
3603 2014-01-08 Catherine Moore <clm@codesourcery.com>
3605 * config/mips/mips.md (simple_return): Attempt to use JRC
3607 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3609 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3611 PR rtl-optimization/59137
3612 * reorg.c (steal_delay_list_from_target): Call update_block for
3614 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3616 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3618 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3619 two duplicate entries.
3621 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3624 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3626 * config/mips/mips.c (mips_truncated_op_cost): New function.
3627 (mips_rtx_costs): Adjust test for BADDU.
3628 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3630 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3632 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3633 (*baddu_si): ...this new pattern.
3635 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3638 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3640 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3643 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3645 (expand_expr, expand_normal): Adjust.
3646 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3647 inner_reference_p. Use inner_reference_p to expand inner references.
3648 (store_expr): Adjust.
3649 * cfgexpand.c (expand_call_stmt): Adjust.
3651 2014-01-08 Rong Xu <xur@google.com>
3653 * gcov-io.c (gcov_var): Move from gcov-io.h.
3654 (gcov_position): Ditto.
3655 (gcov_is_error): Ditto.
3656 (gcov_rewrite): Ditto.
3657 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3658 only part to libgcc/libgcov.h.
3660 2014-01-08 Marek Polacek <polacek@redhat.com>
3663 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3665 2014-01-08 Marek Polacek <polacek@redhat.com>
3668 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3670 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3672 PR rtl-optimization/59649
3673 * stor-layout.c (get_mode_bounds): For BImode return
3674 0 and STORE_FLAG_VALUE.
3676 2014-01-08 Richard Biener <rguenther@suse.de>
3679 * gimple.h (is_gimple_builtin_call): Remove.
3680 (gimple_builtin_call_types_compatible_p): New.
3681 (gimple_call_builtin_p): New overload.
3682 * gimple.c (is_gimple_builtin_call): Remove.
3683 (validate_call): Rename to ...
3684 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3686 (validate_type): New static function.
3687 (gimple_call_builtin_p): New overload and adjust.
3688 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3689 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3690 (gimple_fold_stmt_to_constant_1): Likewise.
3691 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3693 2014-01-08 Richard Biener <rguenther@suse.de>
3696 * gimplify.c (gimplify_expr): Gimplify register-register type
3697 VIEW_CONVERT_EXPRs to separate stmts.
3699 2014-01-07 Jeff Law <law@redhat.com>
3702 * ree.c (combine_set_extension): Handle case where source
3703 and destination registers in an extension insn are different.
3704 (combine_reaching_defs): Allow source and destination registers
3705 in extension to be different under limited circumstances.
3706 (add_removable_extension): Remove restriction that the
3707 source and destination registers in the extension are the same.
3708 (find_and_remove_re): Emit a copy from the extension's
3709 destination to its source after the defining insn if
3710 the source and destination registers are different.
3713 * ifcvt.c (merge_if_block): If we are merging a block with more than
3714 one successor with a block with no successors, remove any BARRIER
3715 after the second block.
3717 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3719 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3721 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3724 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3725 for 14-bit register offsets when INT14_OK_STRICT is false.
3727 2014-01-07 Roland Stigge <stigge@antcom.de>
3728 Michael Meissner <meissner@linux.vnet.ibm.com>
3731 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3732 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3734 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3736 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3739 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3741 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3742 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3743 rtx is const0_rtx or not.
3745 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3748 * target-globals.c (save_target_globals): Remove this_fn_optab
3750 * toplev.c: Include optabs.h.
3751 (target_reinit): Temporarily restore the global options if another
3752 set of options are in force.
3754 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3756 PR rtl-optimization/58668
3757 * cfgcleanup.c (flow_find_cross_jump): Don't count
3758 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3759 to determine what is counted.
3760 (flow_find_head_matching_sequence): Use active_insn_p to determine
3762 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3764 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3765 determine what is counted.
3767 PR tree-optimization/59643
3768 * tree-predcom.c (split_data_refs_to_components): If one dr is
3769 read and one write, determine_offset fails and the write isn't
3770 in the bad component, just put the read into the bad component.
3772 2014-01-07 Mike Stump <mikestump@comcast.net>
3773 Jakub Jelinek <jakub@redhat.com>
3776 * tree-core.h (struct tree_optimization_option): Change optabs
3777 type from unsigned char * to void *.
3778 * optabs.c (init_tree_optimization_optabs): Adjust
3779 TREE_OPTIMIZATION_OPTABS initialization.
3781 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3784 * config/i386/i386.h (struct machine_function): Add
3785 no_drap_save_restore field.
3786 * config/i386/i386.c (ix86_save_reg): Use
3787 !cfun->machine->no_drap_save_restore instead of
3788 crtl->stack_realign_needed.
3789 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3790 this function clears frame_pointer_needed. Set
3791 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3792 and DRAP reg is needed.
3794 2014-01-06 Marek Polacek <polacek@redhat.com>
3797 * doc/implement-c.texi: Mention that other integer types are
3798 permitted as bit-field types in strictly conforming mode.
3800 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3802 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3805 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3807 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3809 2014-01-06 Martin Jambor <mjambor@suse.cz>
3812 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3814 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3816 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3820 * var-tracking.c (add_stores): Preserve the value of the source even if
3821 we don't record the store.
3823 2014-01-06 Terry Guo <terry.guo@arm.com>
3825 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3827 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3830 * config/darwin.c (darwin_function_section): Adjust return values to
3831 correspond to optimisation changes made in r206070.
3833 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3835 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3836 from prefetch_block tune setting.
3837 (nocona_cost): Correct size of prefetch block to 64.
3839 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3841 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3842 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3843 used to save the static chain register in the computation of the offset
3844 from which the FP registers need to be restored.
3846 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3848 PR tree-optimization/59519
3849 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3850 ICE if get_current_def (current_new_name) is already non-NULL, as long
3851 as it is a phi result of some other phi in *new_exit_bb that has
3854 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3855 or vmovdqu* for misaligned_operand.
3856 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3857 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3858 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3859 aligned_mem for AVX512F masked aligned load and store builtins and for
3862 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3864 PR tree-optimization/59651
3865 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3866 Address range for negative step should be added by TYPE_SIZE_UNIT.
3868 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3870 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3871 overlapping registers also for registers other than the stack pointer.
3873 2014-01-03 Marek Polacek <polacek@redhat.com>
3876 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3879 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3882 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3885 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3886 (push splitter): Use <P:MODE_SIZE> instead of
3887 GET_MODE_SIZE (<P:MODE>mode).
3888 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3889 (mov -1, reg peephole2): Likewise.
3890 * config/i386/sse.md (*mov<mode>_internal,
3891 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3892 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3893 *<code><mode>3, *andnot<mode>3<mask_name>,
3894 <mask_codefor><code><mode>3<mask_name>): Likewise.
3895 * config/i386/subst.md (mask_mode512bit_condition,
3896 sd_mask_mode512bit_condition): Likewise.
3898 2014-01-02 Xinliang David Li <davidxl@google.com>
3900 PR tree-optimization/59303
3901 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3902 (dump_predicates): Better output format.
3903 (pred_equal_p): New function.
3904 (is_neq_relop_p): Ditto.
3905 (is_neq_zero_form_p): Ditto.
3906 (pred_expr_equal_p): Ditto.
3907 (pred_neg_p): Ditto.
3908 (simplify_pred): Ditto.
3909 (simplify_preds_2): Ditto.
3910 (simplify_preds_3): Ditto.
3911 (simplify_preds_4): Ditto.
3912 (simplify_preds): Ditto.
3914 (push_to_worklist): Ditto.
3915 (get_pred_info_from_cmp): Ditto.
3916 (is_degenerated_phi): Ditto.
3917 (normalize_one_pred_1): Ditto.
3918 (normalize_one_pred): Ditto.
3919 (normalize_one_pred_chain): Ditto.
3920 (normalize_preds): Ditto.
3921 (normalize_cond_1): Remove function.
3922 (normalize_cond): Ditto.
3923 (is_gcond_subset_of): Ditto.
3924 (is_subset_of_any): Ditto.
3925 (is_or_set_subset_of): Ditto.
3926 (is_and_set_subset_of): Ditto.
3927 (is_norm_cond_subset_of): Ditto.
3928 (pred_chain_length_cmp): Ditto.
3929 (convert_control_dep_chain_into_preds): Type change.
3930 (find_predicates): Ditto.
3931 (find_def_preds): Ditto.
3932 (destroy_predicates_vecs): Ditto.
3933 (find_matching_predicates_in_rest_chains): Ditto.
3934 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3935 (is_pred_expr_subset): Ditto.
3936 (is_pred_chain_subset_of): Ditto.
3937 (is_included_in): Ditto.
3938 (is_superset_of): Ditto.
3940 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3942 Update copyright years.
3944 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3946 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3947 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3948 config/arc/arc.md, config/arc/arc.opt,
3949 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3950 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3951 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3952 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3953 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3954 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3955 vtable-verify.c, vtable-verify.h: Use the standard form for the
3958 2014-01-02 Tobias Burnus <burnus@net-b.de>
3960 * gcc.c (process_command): Update copyright notice dates.
3961 * gcov-dump.c: Ditto.
3963 * doc/cpp.texi: Bump @copying's copyright year.
3964 * doc/cppinternals.texi: Ditto.
3965 * doc/gcc.texi: Ditto.
3966 * doc/gccint.texi: Ditto.
3967 * doc/gcov.texi: Ditto.
3968 * doc/install.texi: Ditto.
3969 * doc/invoke.texi: Ditto.
3971 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3973 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3975 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3977 * config/i386/sse.md (*mov<mode>_internal): Guard
3978 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3980 PR rtl-optimization/59647
3981 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3982 new_rtx into UNSIGNED_FLOAT rtxes.
3984 Copyright (C) 2014 Free Software Foundation, Inc.
3986 Copying and distribution of this file, with or without modification,
3987 are permitted in any medium without royalty provided the copyright
3988 notice and this notice are preserved.