1 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
4 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
5 instead of emit_move_insn.
7 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
11 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
14 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16 * config/rs6000/altivec.md (altivec_lvxl): Rename as
17 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of
19 (altivec_lvxl_<mode>): New define_expand incorporating
20 -maltivec=be semantics where needed.
21 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
22 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
23 semantics where needed.
24 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
25 (altivec_stvx_<mode>): New define_expand incorporating
26 -maltivec=be semantics where needed.
27 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
28 VM2 iterator instead of V4SI.
29 (altivec_stvxl_<mode>): New define_expand incorporating
30 -maltivec=be semantics where needed.
31 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
32 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
33 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI,
34 STVX_V2DF, STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI,
35 STVXL_V2DF, STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI,
37 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
38 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
39 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
40 ALTIVEC_BUILTIN_STVXL.
41 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New
43 (altivec_expand_stvx_be): Likewise.
44 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
45 (altivec_expand_lvx_be): Likewise.
46 (altivec_expand_stvx_be): Likewise.
47 (altivec_expand_builtin): Add cases for
48 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
49 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
50 (altivec_init_builtins): Add definitions for
51 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
52 __builtin_altivec_stvx_<mode>, and
53 __builtin_altivec_stvxl_<mode>.
55 2014-02-21 Catherine Moore <clm@codesourcery.com>
57 * doc/invoke.texi (mvirt, mno-virt): Document.
58 * config/mips/mips.opt (mvirt): New option.
59 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
61 2014-02-21 Richard Biener <rguenther@suse.de>
63 PR tree-optimization/60276
64 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
65 (STMT_VINFO_MIN_NEG_DIST): New macro.
66 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
67 STMT_VINFO_MIN_NEG_DIST.
68 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
69 made for negative dependence distances still hold.
71 2014-02-21 Richard Biener <rguenther@suse.de>
74 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
75 DECL_INITIAL for globals not in the current function context.
77 2014-02-21 Jakub Jelinek <jakub@redhat.com>
79 PR tree-optimization/56490
80 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
81 * tree-ssa-uninit.c: Include params.h.
82 (compute_control_dep_chain): Add num_calls argument, return false
83 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
84 num_calls to recursive call.
85 (find_predicates): Change dep_chain into normal array,
86 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
87 variable and adjust compute_control_dep_chain caller.
88 (find_def_preds): Likewise.
90 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
92 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
93 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
95 2014-02-21 Nick Clifton <nickc@redhat.com>
97 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
99 (popqi1): Add mode to pre_dec.
102 2014-02-21 Jakub Jelinek <jakub@redhat.com>
104 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
105 mode for mask of V8SFmode permutation.
107 2014-02-20 Richard Henderson <rth@redhat.com>
110 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
111 a new pseudo for OLDVAL.
113 2014-02-20 Jakub Jelinek <jakub@redhat.com>
116 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
117 gen_reg_rtx if d->testing_p.
118 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
119 if d->testing_p and we will certainly return true.
120 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
123 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
125 * emit-rtl.c (gen_reg_rtx): Assert that
126 crtl->emit.regno_pointer_align_length is non-zero.
128 2014-02-20 Richard Henderson <rth@redhat.com>
131 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
132 on failure the store back into EXPECT.
134 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
135 Sandra Loosemore <sandra@codesourcery.com>
137 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
138 * config/nios2/nios2.c (nios2_function_profiler):
139 Add -fPIC (flag_pic == 2) support.
140 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
141 (nios2_large_offset_p): New function.
142 (nios2_unspec_reloc_p): Move up position, update to use
143 nios2_large_offset_p.
144 (nios2_unspec_address): Remove function.
145 (nios2_unspec_offset): New function.
146 (nios2_large_got_address): New function.
147 (nios2_got_address): Add large offset support.
148 (nios2_legitimize_tls_address): Update usage of removed and new
150 (nios2_symbol_binds_local_p): New function.
151 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
152 (nios2_legitimize_address): Update to use nios2_large_offset_p.
153 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
154 (nios2_print_operand): Merge H/L processing, add hiadj/lo
155 processing for (const (unspec ...)).
156 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
158 2014-02-20 Richard Biener <rguenther@suse.de>
160 * tree-cfg.c (replace_uses_by): Mark altered BBs before
161 doing the substitution.
162 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
164 2014-02-20 Martin Jambor <mjambor@suse.cz>
167 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
168 info when checking whether lattices are bottom.
170 2014-02-20 Richard Biener <rguenther@suse.de>
173 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
176 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
179 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
180 parameter specifying the scaling.
181 (inline_call): Update.
182 (want_inline_recursively): Guard division by zero.
183 (recursive_inlining): Update.
184 * ipa-inline.h (clone_inlined_nodes): Update.
186 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
189 * config/i386/i386.c (classify_argument): Pass structures of size
190 64 bytes or less in register.
192 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
193 Kirill Yukhin <kirill.yukhin@intel.com>
195 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
196 (_mm_rcp28_round_ss): Ditto.
197 (_mm_rsqrt28_round_sd): Ditto.
198 (_mm_rsqrt28_round_ss): Ditto.
199 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
200 (_mm_rcp14_round_ss): Ditto.
201 (_mm_rsqrt14_round_sd): Ditto.
202 (_mm_rsqrt14_round_ss): Ditto.
203 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
204 the first input operand, get rid of match_dup.
205 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
207 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
209 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
210 operand as the first input operand, set type attribute.
211 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
213 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
214 operand as the first input operand, set type attribute.
216 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
218 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
221 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
224 * config/i386/i386.c (construct_container): Remove TFmode check
225 for X86_64_INTEGER_CLASS.
227 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
230 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
231 only when -Wpsabi is enabled.
233 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
236 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
237 passing arrays in registers are the same as for structs, so remove the
238 special case for them.
240 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
242 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
243 destination type, extract only the valid bits if the source type is not
244 integral and has a different mode.
246 2014-02-19 Richard Biener <rguenther@suse.de>
249 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
252 2014-02-19 Richard Biener <rguenther@suse.de>
255 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
256 (ipa_modify_call_arguments): Emit an argument load explicitely and
257 preserve virtual SSA form there and for the replacement call.
258 Do not update SSA form nor free dominance info.
260 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
262 * ipa.c (function_and_variable_visibility): Also clear WEAK
263 flag when disolving COMDAT_GROUP.
265 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
267 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
268 * ipa-prop.c (ipa_set_jf_known_type): Return early when
270 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
271 do more sanity checks.
272 (detect_type_change): Return true when giving up early.
273 (compute_complex_assign_jump_func): Fix type parameter of
275 (compute_complex_ancestor_jump_func): Likewise.
276 (update_jump_functions_after_inlining): Fix updating of
278 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
280 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
282 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
283 inline clones when edge disappears.
285 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
288 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
289 Split 64-bit moves into 2 patterns. Do not allow the use of
290 direct move for TDmode in little endian, since the decimal value
291 has little endian bytes within a word, but the 64-bit pieces are
292 ordered in a big endian fashion, and normal subreg's of TDmode are
294 (mov<mode>_64bit_dm): Likewise.
295 (movtd_64bit_nodm): Likewise.
297 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
299 PR tree-optimization/60174
300 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
301 statement of an SSA_NAME that occurs in an abnormal PHI node.
303 2014-02-18 Jakub Jelinek <jakub@redhat.com>
306 * final.c (SEEN_BB): Remove.
307 (SEEN_NOTE, SEEN_EMITTED): Renumber.
308 (final_scan_insn): Don't force_source_line on second
309 NOTE_INSN_BASIC_BLOCK.
311 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
314 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
315 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
316 (type_natural_mode): Warn ABI change when %zmm register is not
317 available for AVX512F vector value passing.
319 2014-02-18 Kai Tietz <ktietz@redhat.com>
322 * config/i386/i386.c (ix86_expand_prologue): Use value in
323 rax register as displacement when restoring %r10 or %rax.
324 Fix wrong offset when restoring both registers.
326 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
328 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
329 assertion with conditional return.
331 2014-02-18 Jakub Jelinek <jakub@redhat.com>
332 Uros Bizjak <ubizjak@gmail.com>
335 * config/i386/driver-i386.c (host_detect_local_cpu): If
336 YMM state is not saved by the OS, also clear has_f16c. Move
337 CPUID 0x80000001 handling before YMM state saving checking.
339 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
341 PR rtl-optimization/58960
342 * haifa-sched.c (alloc_global_sched_pressure_data): New,
343 factored out from ...
344 (sched_init): ... here.
345 (free_global_sched_pressure_data): New, factored out from ...
346 (sched_finish): ... here.
347 * sched-int.h (free_global_sched_pressure_data): Declare.
348 * sched-rgn.c (nr_regions_initial): New static global.
349 (haifa_find_rgns): Initialize it.
350 (schedule_region): Disable sched-pressure for the newly
353 2014-02-17 Richard Biener <rguenther@suse.de>
355 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
356 release SSA defs of pattern stmts.
358 2014-02-17 Richard Biener <rguenther@suse.de>
360 * tree-inline.c (expand_call_inline): Release the virtual
361 operand defined by the call we are about to inline.
363 2014-02-17 Richard Biener <rguenther@suse.de>
365 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
367 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
368 Ilya Tocar <ilya.tocar@intel.com>
370 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
371 arguments order in builtin.
372 (_mm512_permutexvar_epi64): Ditto.
373 (_mm512_mask_permutexvar_epi64): Ditto
374 (_mm512_maskz_permutexvar_epi32): Ditto
375 (_mm512_permutexvar_epi32): Ditto
376 (_mm512_mask_permutexvar_epi32): Ditto
378 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
380 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
381 (p8_vmrgow): Likewise.
383 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
385 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
388 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
391 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
392 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
393 into 64-bit and 32-bit moves. On 64-bit moves, add support for
394 using direct move instructions on ISA 2.07. Also adjust
395 instruction length for 64-bit.
396 (mov<mode>_64bit, TFmode/TDmode): Likewise.
397 (mov<mode>_32bit, TFmode/TDmode): Likewise.
399 2014-02-15 Alan Modra <amodra@gmail.com>
403 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
404 find_replacement on parts of insn rtl that might be reloaded.
406 2014-02-15 Richard Biener <rguenther@suse.de>
408 PR tree-optimization/60183
409 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
410 (tree_ssa_phiprop): Calculate and free post-dominators.
412 2014-02-14 Jeff Law <law@redhat.com>
414 PR rtl-optimization/60131
415 * ree.c (get_extended_src_reg): New function.
416 (combine_reaching_defs): Use it rather than assuming location of REG.
417 (find_and_remove_re): Verify first operand of extension is
418 a REG before adding the insns to the copy list.
420 2014-02-14 Roland McGrath <mcgrathr@google.com>
422 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
423 * configure: Regenerated.
424 * config.in: Regenerated.
425 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
426 instead of ASM_SHORT.
428 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
429 Richard Earnshaw <rearnsha@arm.com>
431 PR rtl-optimization/59535
432 * lra-constraints.c (process_alt_operands): Encourage alternative
433 when unassigned pseudo class is superset of the alternative class.
434 (inherit_reload_reg): Don't inherit when optimizing for code size.
435 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
436 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
437 modes not less than 4 for Thumb1.
439 2014-02-14 Kyle McMartin <kyle@redhat.com>
442 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
444 2014-02-14 Richard Biener <rguenther@suse.de>
446 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
447 (get_frame_arg): Drop the assert with langhook types_compatible_p.
448 Do not strip INDIRECT_REFs.
450 2014-02-14 Richard Biener <rguenther@suse.de>
453 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
454 DECL_FUNCTION_SPECIFIC_TARGET.
455 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
456 * tree-streamer-out.c (pack_ts_target_option): Remove.
457 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
458 (write_ts_function_decl_tree_pointers): Do not stream
459 DECL_FUNCTION_SPECIFIC_TARGET.
460 * tree-streamer-in.c (unpack_ts_target_option): Remove.
461 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
462 (lto_input_ts_function_decl_tree_pointers): Do not stream
463 DECL_FUNCTION_SPECIFIC_TARGET.
465 2014-02-14 Jakub Jelinek <jakub@redhat.com>
467 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
468 (get_initial_def_for_induction, vectorizable_induction): Ignore
469 debug stmts when looking for exit_phi.
470 (vectorizable_live_operation): Fix up condition.
472 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
474 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
475 nreverse() because it changes the content of original tree list.
477 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
479 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
480 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
482 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
484 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
485 GNU coding standards.
487 2014-02-13 Jakub Jelinek <jakub@redhat.com>
490 * dwarf2out.c (gen_subprogram_die): Don't call
491 add_calling_convention_attribute if subr_die is old_die.
493 2014-02-13 Sharad Singhai <singhai@google.com>
495 * doc/optinfo.texi: Fix order of nodes.
497 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
499 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
500 operands[2], not operands[3].
502 2014-02-13 Richard Biener <rguenther@suse.de>
505 * doc/install.texi (ISL): Update recommended version to 0.12.2,
506 mention the possibility of an in-tree build.
507 (CLooG): Update recommended version to 0.18.1, mention the
508 possibility of an in-tree build and clarify that the ISL
509 bundled with CLooG does not work.
511 2014-02-13 Jakub Jelinek <jakub@redhat.com>
514 * expr.c (compress_float_constant): If x is a hard register,
515 extend into a pseudo and then move to x.
517 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
519 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
520 caused by bad second argument to warning_at() with -mhotpatch and
521 nested functions (e.g. with gfortran).
523 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
525 * opts.c (option_name): Remove "enabled by default" rider.
527 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
529 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
531 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
532 Uros Bizjak <ubizjak@gmail.com>
535 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
536 * configure: Regenerated.
538 2014-02-12 Richard Biener <rguenther@suse.de>
540 * vec.c (vec_prefix::calculate_allocation): Move as
541 inline variant to vec.h.
542 (vec_prefix::calculate_allocation_1): New out-of-line version.
543 * vec.h (vec_prefix::calculate_allocation_1): Declare.
544 (vec_prefix::m_has_auto_buf): Rename to ...
545 (vec_prefix::m_using_auto_storage): ... this.
546 (vec_prefix::calculate_allocation): Inline the easy cases
547 and dispatch to calculate_allocation_1 which doesn't need the
549 (va_heap::reserve): Use gcc_checking_assert.
550 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
551 m_using_auto_storage.
552 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
554 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
555 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
556 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
558 2014-02-12 Richard Biener <rguenther@suse.de>
560 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
561 when we found a dependence.
563 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
565 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
567 (maybe_fold_stmt): ... into this new function.
568 * omp-low.c (lower_omp): Update comment.
570 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
573 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
576 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
578 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
579 identifiers in comments.
580 (cortexa53_extra_costs): Likewise.
581 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
582 (cortexa7_extra_costs): Likewise.
583 (cortexa12_extra_costs): Likewise.
584 (cortexa15_extra_costs): Likewise.
585 (v7m_extra_costs): Likewise.
587 2014-02-12 Richard Biener <rguenther@suse.de>
590 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
591 of posix_memalign being successful.
592 (lower_stmt): Restrict lowering of posix_memalign to when
593 -ftree-bit-ccp is enabled.
595 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
597 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
599 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
601 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
603 PR rtl-optimization/60116
604 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
605 other_insn once the combination has been validated.
607 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
610 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
612 * ipa-devirt.c: Include demangle.h
613 (odr_violation_reported): New static variable.
614 (add_type_duplicate): Update odr_violations.
615 (maybe_record_node): Add completep parameter; update it.
616 (record_target_from_binfo): Add COMPLETEP parameter;
618 (possible_polymorphic_call_targets_1): Likewise.
619 (struct polymorphic_call_target_d): Add nonconstruction_targets;
620 rename FINAL to COMPLETE.
621 (record_targets_from_bases): Sanity check we found the binfo;
622 fix COMPLETEP updating.
623 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
624 parameter, fix computing of COMPLETEP.
625 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
626 at LTO time do demangling.
627 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
628 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
630 (gimple_get_virt_method_for_binfo): Likewise.
631 * gimple-fold.h (gimple_get_virt_method_for_binfo,
632 gimple_get_virt_method_for_vtable): Update prototypes.
634 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
637 * genautomata.c (add_presence_absence): Fix typo with
638 {final_}presence_list.
640 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
643 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
644 for VSX/Altivec vectors that land in GPR registers.
646 2014-02-11 Richard Henderson <rth@redhat.com>
647 Jakub Jelinek <jakub@redhat.com>
650 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
651 around drhs if type conversion to lacc->type is not useless.
653 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
655 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
657 (cortex-a57.cortex-a53): Likewise.
658 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
660 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
662 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
665 2014-02-11 Renlin Li <Renlin.Li@arm.com>
667 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
668 add_options_for_arm_vfp3.
670 2014-02-11 Jeff Law <law@redhat.com>
673 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
674 object with an undesirable mode.
676 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
679 * config/i386/sol2-9.h: New file.
680 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
681 *-*-solaris2.9*): Use it.
683 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
685 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
686 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
688 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
690 * config/microblaze/microblaze.c: Extend mcpu version format
692 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
694 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
696 2014-02-10 Richard Henderson <rth@redhat.com>
699 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
700 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
701 ms-abi vs -mno-accumulate-outgoing-args.
702 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
703 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
706 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
709 * cfgexpand.c (expand_asm_operands): Attach source location to
710 ASM_INPUT rtx objects.
711 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
713 2014-02-10 Nick Clifton <nickc@redhat.com>
715 * config/mn10300/mn10300.c (popcount): New function.
716 (mn10300_expand_prologue): Include saved registers in stack usage
719 2014-02-10 Jeff Law <law@redhat.com>
722 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
723 when changing the SET_DEST of a prior insn to avoid an input reload.
725 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
727 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
728 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
729 -mcall-openbsd, or -mcall-linux.
730 (CC1_ENDIAN_BIG_SPEC): Remove.
731 (CC1_ENDIAN_LITTLE_SPEC): Remove.
732 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
733 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
734 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
735 and %cc1_endian_default.
736 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
738 2014-02-10 Richard Biener <rguenther@suse.de>
740 PR tree-optimization/60115
741 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
742 MEM_REF handling. Properly verify that the accesses are not
743 out of the objects bound.
745 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
747 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
750 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
752 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
753 proper constants and fix formatting.
754 (possible_polymorphic_call_targets): Fix formatting.
756 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
757 Ilya Tocar <ilya.tocar@intel.com>
759 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
760 (_mm512_loadu_epi32): Renamed into...
761 (_mm512_loadu_si512): This.
762 (_mm512_storeu_epi32): Renamed into...
763 (_mm512_storeu_si512): This.
764 (_mm512_maskz_ceil_ps): Removed.
765 (_mm512_maskz_ceil_pd): Ditto.
766 (_mm512_maskz_floor_ps): Ditto.
767 (_mm512_maskz_floor_pd): Ditto.
768 (_mm512_floor_round_ps): Ditto.
769 (_mm512_floor_round_pd): Ditto.
770 (_mm512_ceil_round_ps): Ditto.
771 (_mm512_ceil_round_pd): Ditto.
772 (_mm512_mask_floor_round_ps): Ditto.
773 (_mm512_mask_floor_round_pd): Ditto.
774 (_mm512_mask_ceil_round_ps): Ditto.
775 (_mm512_mask_ceil_round_pd): Ditto.
776 (_mm512_maskz_floor_round_ps): Ditto.
777 (_mm512_maskz_floor_round_pd): Ditto.
778 (_mm512_maskz_ceil_round_ps): Ditto.
779 (_mm512_maskz_ceil_round_pd): Ditto.
780 (_mm512_expand_pd): Ditto.
781 (_mm512_expand_ps): Ditto.
782 * config/i386/i386.c (ix86_builtins): Remove
783 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
785 * config/i386/predicates.md (const1256_operand): New.
786 (const_1_to_2_operand): Ditto.
787 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
788 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
789 (*avx512pf_gatherpf<mode>sf): Ditto.
790 (avx512pf_gatherpf<mode>df): Ditto.
791 (*avx512pf_gatherpf<mode>df_mask): Ditto.
792 (*avx512pf_gatherpf<mode>df): Ditto.
793 (avx512pf_scatterpf<mode>sf): Ditto.
794 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
795 (*avx512pf_scatterpf<mode>sf): Ditto.
796 (avx512pf_scatterpf<mode>df): Ditto.
797 (*avx512pf_scatterpf<mode>df_mask): Ditto.
798 (*avx512pf_scatterpf<mode>df): Ditto.
799 (avx512f_expand<mode>): Removed.
800 (<shift_insn><mode>3<mask_name>): Change predicate type.
802 2014-02-08 Jakub Jelinek <jakub@redhat.com>
804 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
805 not at the end of datarefs vector use ordered_remove to avoid
806 reordering datarefs vector.
809 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
810 mark local addressable non-static vars as GOVD_PRIVATE
811 instead of GOVD_LOCAL.
812 * omp-low.c (lower_omp_for): Move gimple_bind_vars
813 and BLOCK_VARS of gimple_bind_block to new_stmt rather
817 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
818 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
819 assume_aligned or alloc_align attributes.
820 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
821 arguments. Handle also assume_aligned and alloc_align attributes.
822 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
823 calls to functions with assume_aligned or alloc_align attributes.
824 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
826 2014-02-08 Terry Guo <terry.guo@arm.com>
828 * doc/invoke.texi: Document ARM -march=armv7e-m.
830 2014-02-08 Jakub Jelinek <jakub@redhat.com>
832 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
833 flag on __cilkrts_rethrow builtin.
836 * ipa-cp.c (determine_versionability): Fail at -O0
837 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
838 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
841 2014-02-04 Jakub Jelinek <jakub@redhat.com>
844 * tree-inline.c (copy_forbidden): Fail for
845 __attribute__((optimize (0))) functions.
847 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
849 * varpool.c: Include pointer-set.h.
850 (varpool_remove_unreferenced_decls): Variables in other partitions
851 will not be output; be however careful to not lose information
854 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
856 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
857 lookup in the vtable constructor.
859 2014-02-07 Jeff Law <law@redhat.com>
862 * config/m68k/m68k.md (ashldi_extsi): Turn into a
863 define_insn_and_split.
865 * ipa-inline.c (inline_small_functions): Fix typos.
867 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
869 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
870 (s390_can_use_return_insn): Declare.
871 * config/s390/s390.h (EPILOGUE_USES): Define.
872 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
874 (s390_chunkify_start): Handle return JUMP_LABELs.
875 (s390_early_mach): Emit a main_pool instruction on the entry edge.
876 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
877 (s390_can_use_return_insn): New functions.
878 (s390_fix_long_loop_prediction): Handle conditional returns.
879 (TARGET_SET_UP_BY_PROLOGUE): Define.
880 * config/s390/s390.md (ANY_RETURN): New code iterator.
881 (*creturn, *csimple_return, return, simple_return): New patterns.
883 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
885 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
886 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
887 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
888 REG_CFA_RESTORE list when deciding not to restore a register.
890 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
892 * config/s390/s390.c: Include tree-pass.h and context.h.
893 (s390_early_mach): New function, split out from...
894 (s390_emit_prologue): ...here.
895 (pass_data_s390_early_mach): New pass structure.
896 (pass_s390_early_mach): New class.
897 (s390_option_override): Create and register early_mach pass.
900 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
902 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
903 to match for the exit block.
905 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
907 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
908 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
909 Reject misaligned operands.
911 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
913 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
915 2014-02-07 Richard Biener <rguenther@suse.de>
918 * gimple-low.c (lower_builtin_posix_memalign): New function.
919 (lower_stmt): Call it to lower posix_memalign in a way
920 to make alignment info accessible.
922 2014-02-07 Jakub Jelinek <jakub@redhat.com>
925 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
926 __builtin_setjmp_receiver.
928 2014-02-07 Richard Biener <rguenther@suse.de>
931 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
932 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
933 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
934 Handle BUILT_IN_POSIX_MEMALIGN.
935 (find_func_clobbers): Likewise.
936 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
937 (call_may_clobber_ref_p_1): Likewise.
939 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
942 * ipa-devirt.c (record_target_from_binfo): Remove overactive
945 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
948 * lto-cgraph.c (lto_output_node): Use
949 symtab_get_symbol_partitioning_class.
950 (lto_output_varpool_node): likewise.
951 (symtab_get_symbol_partitioning_class): Move here from
953 * cgraph.h (symbol_partitioning_class): Likewise.
954 (symtab_get_symbol_partitioning_class): Declare.
956 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
958 * ggc.h (ggc_internal_cleared_alloc): New macro.
959 * vec.h (vec_safe_copy): Handle memory stats.
960 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
961 * target-globals.c (save_target_globals): Likewise.
963 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
966 * expr.c (emit_move_resolve_push): Export; be bit more selective
967 on when to clear alias set.
968 * expr.h (emit_move_resolve_push): Declare.
969 * function.h (struct function): Add tail_call_marked.
970 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
971 * config/i386/i386-protos.h (ix86_expand_push): Remove.
972 * config/i386/i386.md (TImode move expander): De not call
974 (FP push expanders): Preserve memory attributes.
975 * config/i386/sse.md (push<mode>1): Remove.
976 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
977 (ix86_expand_push): Remove.
978 * config/i386/mmx.md (push<mode>1): Remove.
980 2014-02-06 Jakub Jelinek <jakub@redhat.com>
982 PR rtl-optimization/60030
983 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
984 lopart with paradoxical subreg before shifting it up by hprec.
986 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
988 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
989 Remove extra newline at end of file.
990 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
991 (arm_issue_rate): Handle cortexa57.
992 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
993 (cortex-a57.cortex-a53): Likewise.
995 2014-02-06 Jakub Jelinek <jakub@redhat.com>
998 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
999 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1001 (arm_expand_prologue): Adjust all callers.
1002 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1003 info, registers also at the lowest numbered registers side. Use
1004 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1008 * var-tracking.c (adjust_mems): Before adding a SET to
1009 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1011 2014-02-06 Alan Modra <amodra@gmail.com>
1014 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1015 change SDmode to DDmode when lra_in_progress.
1017 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1020 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1021 free_data_ref on the dr first, and before goto again also set dr
1022 to the next dr. For simd_lane_access, free old datarefs[i] before
1023 overwriting it. For get_vectype_for_scalar_type failure, don't
1024 free_data_ref if simd_lane_access.
1026 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1029 * tree.h (opts_for_fn): New inline function.
1030 (opt_for_fn): Define.
1031 * config/i386/i386.c (ix86_function_regparm): Use
1032 opt_for_fn (decl, optimize) instead of optimize.
1034 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1036 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1037 for SYMBOL_REF in large memory model.
1039 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1041 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1043 (cortex-a57): Likewise.
1044 (cortex-a57.cortex-a53): Likewise.
1046 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1047 Kugan Vivekanandarajah <kuganv@linaro.org>
1049 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1051 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1053 2014-02-06 Richard Biener <rguenther@suse.de>
1055 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1056 set_loop_copy and initialize_original_copy_tables.
1058 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1060 * config/aarch64/aarch64-simd.md
1061 (aarch64_ashr_simddi): Change QI to SI.
1063 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1064 Jakub Jelinek <jakub@redhat.com>
1067 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1070 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1072 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1073 CODE_FOR_altivec_vpku[hw]um to
1074 CODE_FOR_altivec_vpku[hw]um_direct.
1075 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1076 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1077 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1078 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1080 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1082 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1083 generation for -maltivec=be.
1084 (altivec_vsumsws): Simplify redundant test.
1086 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1088 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1089 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1090 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1091 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1092 gen_altivec_vpkuwum.
1093 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1095 (altivec_vpks<VI_char>ss): Likewise.
1096 (altivec_vpks<VI_char>us): Likewise.
1097 (altivec_vpku<VI_char>us): Likewise.
1098 (altivec_vpku<VI_char>um): Likewise.
1099 (altivec_vpku<VI_char>um_direct): New (copy of
1100 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1102 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1103 target is little endian and -maltivec=be is not specified.
1104 (*altivec_vupkhs<VU_char>_direct): New (copy of
1105 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1106 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1107 target is little endian and -maltivec=be is not specified.
1108 (*altivec_vupkls<VU_char>_direct): New (copy of
1109 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1110 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1111 little endian and -maltivec=be is not specified.
1112 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1113 little endian and -maltivec=be is not specified.
1115 2014-02-05 Richard Henderson <rth@redhat.com>
1118 * combine-stack-adj.c: Revert r206943.
1119 * sched-int.h (struct deps_desc): Add last_args_size.
1120 * sched-deps.c (init_deps): Initialize it.
1121 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1122 contain REG_ARGS_SIZE notes.
1124 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1126 * lto-cgraph.c (asm_nodes_output): Make global.
1127 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1128 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1129 (driver_handle_option): Handle OPT_fwpa.
1131 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1134 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1135 a comment typo and formatting issue. If odr_hash hasn't been
1136 created, return vNULL and set *completep to false.
1139 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1140 bb with no successors.
1142 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1145 * doc/invoke.texi (-march): Clarify documentation for ARM.
1149 2014-02-05 Richard Biener <rguenther@suse.de>
1151 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1152 when not vectorizing because of too many alias checks.
1153 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1154 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1156 2014-02-05 Nick Clifton <nickc@redhat.com>
1158 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1159 accept extended registers in any mode when compiling for the MN10300.
1161 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1163 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1164 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1165 sanitization attributes.
1166 (can_inline_edge_p): Likewise.
1167 (sanitize_attrs_match_for_inline_p): New function.
1169 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1171 * ipa-prop.c (detect_type_change): Shor circuit testing of
1172 type changes on THIS pointer.
1174 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1177 * config/pa/pa.c (legitimize_tls_address): Return original address
1178 if not passed a SYMBOL_REF rtx.
1179 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1181 (pa_emit_move_sequence): Simplify TLS source operands.
1182 (pa_legitimate_constant_p): Reject all TLS constants.
1183 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1184 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1186 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1188 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1189 groups when we know they are controlled by LTO.
1190 * varasm.c (default_binds_local_p_1): If object is in other partition,
1191 it will be resolved locally.
1193 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1195 * config/host-linux.c (linux_gt_pch_use_address): Don't
1196 use SSIZE_MAX because it is not always defined.
1198 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1201 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1202 threshold for pseudo splitting.
1203 (update_ebb_live_info): Process call argument hard registers and
1204 hard registers from insn definition too.
1205 (max_small_class_regs_num): New constant.
1206 (inherit_in_ebb): Update live hard regs through EBBs. Update
1207 reloads_num only for small register classes. Don't split for
1210 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1213 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1216 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1218 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1221 2014-02-04 Marek Polacek <polacek@redhat.com>
1223 * gdbinit.in (pel): Define.
1225 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1227 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1230 2014-02-04 Richard Biener <rguenther@suse.de>
1233 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1234 in function context local.
1235 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1236 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1237 similar to LTO_imported_decl_ref.
1239 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1241 PR tree-optimization/60002
1242 * cgraphclones.c (build_function_decl_skip_args): Clear
1245 PR tree-optimization/60023
1246 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1247 false to gsi_replace.
1248 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1249 has been in some EH region and vec_stmt could throw, add
1250 vec_stmt into the same EH region.
1251 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1252 has no lhs, ignore it.
1253 * internal-fn.c (expand_MASK_LOAD): Likewise.
1256 * tree-inline.c (copy_forbidden): Fail for
1257 __attribute__((optimize (0))) functions.
1260 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1261 is set, copy one less argument.
1262 (expand_simd_clones): Don't subtract clone_info->inbranch
1263 from simd_clone_struct_alloc argument.
1265 PR rtl-optimization/57915
1266 * recog.c (simplify_while_replacing): If all unary/binary/relational
1267 operation arguments are constant, attempt to simplify those.
1270 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1271 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1273 2014-02-04 Richard Biener <rguenther@suse.de>
1275 PR tree-optimization/60012
1276 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1277 TBAA disambiguation to all DDRs.
1279 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1282 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1283 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1285 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1288 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1290 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1292 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1293 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1295 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1298 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1299 to figure out targets of polymorphic calls with known decl.
1300 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1301 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1302 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1303 (get_polymorphic_call_info): ... here.
1304 (get_polymorphic_call_info_from_invariant): New function.
1306 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1308 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1309 lookup via vtable pointer; check for type consistency
1310 and turn inconsitent facts into UNREACHABLE.
1311 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1312 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1313 type inconsistent querries; return UNREACHABLE instead.
1315 2014-02-03 Richard Henderson <rth@twiddle.net>
1318 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1319 already processed this node.
1320 (normalize_one_pred_1): Pass along mark_set.
1321 (normalize_one_pred): Create and destroy a pointer_set_t.
1322 (normalize_one_pred_chain): Likewise.
1324 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1326 PR gcov-profile/58602
1327 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1329 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1332 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1333 -fno-devirtualize; try to devirtualize by the knowledge of
1334 virtual table pointer given by aggregate propagation.
1335 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1336 (ipa_print_node_jump_functions): Dump also offset that
1337 is relevant for polymorphic calls.
1338 (determine_known_aggregate_parts): Add arg_type parameter; use it
1339 instead of determining the type from pointer type.
1340 (ipa_compute_jump_functions_for_edge): Update call of
1341 determine_known_aggregate_parts.
1342 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1343 (gimple_get_virt_method_for_binfo): ... here; simplify using
1344 vtable_pointer_value_to_vtable.
1345 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1346 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1347 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1348 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1350 (vtable_pointer_value_to_binfo): ... here.
1351 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1353 2014-02-03 Teresa Johnson <tejohnson@google.com>
1355 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1356 redef of outer loop index variable.
1358 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1362 * doc/extend.texi (Function Attributes): Typo.
1364 2014-02-03 Cong Hou <congh@google.com>
1366 PR tree-optimization/60000
1367 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1368 if the vectorized statement is a store. A store statement can only
1369 appear at the end of pattern statements.
1371 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1373 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1374 (ix86_option_override_internal): Default long double to 64-bit for
1375 32-bit Bionic and to 128-bit for 64-bit Bionic.
1377 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1378 TARGET_LONG_DOUBLE_128 is true.
1379 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1381 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1382 (mlong-double-64): Negate -mlong-double-128.
1383 (mlong-double-128): New option.
1385 * config/i386/i386-c.c (ix86_target_macros): Define
1386 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1388 * doc/invoke.texi: Document -mlong-double-128.
1390 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1392 PR rtl-optimization/60024
1393 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1395 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1397 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1399 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1401 PR rtl-optimization/57662
1402 * sel-sched.c (code_motion_path_driver): Do not mark already not
1403 existing blocks in the visiting bitmap.
1405 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1407 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1408 on the insn being emitted.
1410 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1411 Will Deacon <will.deacon@arm.com>
1413 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1415 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1417 * config/arm/arm-tables.opt: Regenerate.
1419 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1421 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1422 for vector types other than V16QImode.
1423 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1424 define_expand, and call altivec_expand_vec_perm_le when producing
1425 code with little endian element order.
1426 (*altivec_vperm_<mode>_internal): New insn having previous
1427 behavior of altivec_vperm_<mode>.
1428 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1429 altivec_expand_vec_perm_le when producing code with little endian
1431 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1432 behavior of altivec_vperm_<mode>_uns.
1434 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1436 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1437 (altivec_vsumsws): Add handling for -maltivec=be with a little
1439 (altivec_vsumsws_direct): New.
1440 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1441 gen_altivec_vsumsws.
1443 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1445 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1446 vtable_pointer_value_to_binfo): New functions.
1447 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1448 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1450 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1452 * config/nios2/nios2.md (load_got_register): Initialize GOT
1453 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1454 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1456 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1458 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1459 preserverd by passthrough, do not propagate the type.
1461 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1463 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1464 (mips_atomic_assign_expand_fenv): New function.
1465 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1467 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1469 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1470 (__builtin_mips_set_fcsr): Likewise.
1471 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1472 MIPS_USI_FTYPE_VOID.
1473 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1474 (mips16_expand_set_fcsr): Likewise.
1475 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1476 (mips16_set_fcsr_stub): Likewise.
1477 (mips16_get_fcsr_one_only_stub): New class.
1478 (mips16_set_fcsr_one_only_stub): Likewise.
1479 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1480 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1481 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1482 (hard_float): New availability predicate.
1483 (mips_builtins): Add get_fcsr and set_fcsr.
1484 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1485 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1486 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1487 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1488 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1491 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1493 * config/mips/mips.c (mips_one_only_stub): New class.
1494 (mips_need_mips16_rdhwr_p): Replace with...
1495 (mips16_rdhwr_stub): ...this new variable.
1496 (mips16_stub_call_address): New function.
1497 (mips16_rdhwr_one_only_stub): New class.
1498 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1499 (mips_output_mips16_rdhwr): Delete.
1500 (mips_finish_stub): New function.
1501 (mips_code_end): Use it to handle rdhwr stubs.
1503 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1506 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1507 when calculating size of integer atomic types.
1509 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1511 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1513 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1515 PR tree-optimization/60003
1516 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1517 * profile.c (branch_prob): Use gimple_call_builtin_p
1518 to check for BUILT_IN_SETJMP_RECEIVER.
1519 * tree-inline.c (copy_bb): Call notice_special_calls.
1521 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1524 * lra-constraints.c (process_alt_operands): Update reload_sum only
1527 2014-01-31 Richard Henderson <rth@redhat.com>
1530 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1531 until after else_eh is processed.
1533 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1535 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1536 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1537 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1538 in smmintrin.h, remove them.
1539 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1540 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1541 * config/i386/i386.md (ROUND_SAE): Fix value.
1542 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1543 (const48_operand): New.
1544 * config/i386/subst.md (round), (round_expand): Use
1545 const_4_or_8_to_11_operand.
1546 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1548 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1550 * config/i386/constraints.md (Yk): Swap meaning with k.
1551 * config/i386/i386.md (movhi_internal): Change Yk to k.
1552 (movqi_internal): Ditto.
1553 (*k<logic><mode>): Ditto.
1556 (kandn<mode>): Ditto.
1557 (*<code>hi_1): Ditto.
1558 (*<code>qi_1): Ditto.
1559 (kxnor<mode>): Ditto.
1560 (kortestzhi): Ditto.
1561 (kortestchi): Ditto.
1563 (*one_cmplhi2_1): Ditto.
1564 (*one_cmplqi2_1): Ditto.
1565 * config/i386/sse.md (): Change k to Yk.
1566 (avx512f_load<mode>_mask): Ditto.
1567 (avx512f_blendm<mode>): Ditto.
1568 (avx512f_store<mode>_mask): Ditto.
1569 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1570 (avx512f_storedqu<mode>_mask): Ditto.
1571 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1573 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1574 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1575 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1576 (avx512f_maskcmp<mode>3): Ditto.
1577 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1578 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1579 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1580 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1581 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1582 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1583 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1584 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1585 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1586 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1587 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1588 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1589 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1590 (vec_extract_lo_<mode>_maskm): Ditto.
1591 (vec_extract_hi_<mode>_maskm): Ditto.
1592 (avx512f_vternlog<mode>_mask): Ditto.
1593 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1594 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1595 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1596 (avx512f_<code>v8div16qi2_mask): Ditto.
1597 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1598 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1599 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1600 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1601 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1602 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1603 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1604 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1605 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1606 (avx512cd_maskb_vec_dupv8di): Ditto.
1607 (avx512cd_maskw_vec_dupv16si): Ditto.
1608 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1609 (avx512f_vpermi2var<mode>3_mask): Ditto.
1610 (avx512f_vpermi2var<mode>3_mask): Ditto.
1611 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1612 (*avx512f_gathersi<mode>): Ditto.
1613 (*avx512f_gathersi<mode>_2): Ditto.
1614 (*avx512f_gatherdi<mode>): Ditto.
1615 (*avx512f_gatherdi<mode>_2): Ditto.
1616 (*avx512f_scattersi<mode>): Ditto.
1617 (*avx512f_scatterdi<mode>): Ditto.
1618 (avx512f_compress<mode>_mask): Ditto.
1619 (avx512f_compressstore<mode>_mask): Ditto.
1620 (avx512f_expand<mode>_mask): Ditto.
1621 * config/i386/subst.md (mask): Change k to Yk.
1622 (mask_scalar_merge): Ditto.
1625 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1627 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1629 2014-01-31 Richard Biener <rguenther@suse.de>
1632 * builtins.c (fold_builtin_memory_op): Make sure to not
1633 use a floating-point mode or a boolean or enumeral type for
1636 2014-01-30 DJ Delorie <dj@redhat.com>
1638 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1639 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1640 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1641 whenever main() has an epilogue.
1643 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1645 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1646 unused variable "field".
1647 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1648 (vsx_mergeh_<mode>): Likewise.
1649 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1650 (altivec_vmrghh): Likewise.
1651 (altivec_vmrghw): Likewise.
1652 (altivec_vmrglb): Likewise.
1653 (altivec_vmrglh): Likewise.
1654 (altivec_vmrglw): Likewise.
1655 (altivec_vspltb): Add missing uses.
1656 (altivec_vsplth): Likewise.
1657 (altivec_vspltw): Likewise.
1658 (altivec_vspltsf): Likewise.
1660 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1663 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1664 frame related instructions.
1666 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1668 PR rtl-optimization/59959
1669 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1670 any reload of register whose subreg is invalid.
1672 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1674 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1675 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1676 Add missing return type - void.
1678 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1680 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1681 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1682 remove element index adjustment for endian (now handled in vsx.md
1684 (altivec_expand_vec_perm_const): Use
1685 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1686 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1687 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1688 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1689 define_expand and a new define_insn *altivec_vspltb_internal;
1690 adjust for -maltivec=be on a little endian target.
1691 (altivec_vspltb_direct): New.
1692 (altivec_vsplth): Divide into a define_expand and a new
1693 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1694 little endian target.
1695 (altivec_vsplth_direct): New.
1696 (altivec_vspltw): Divide into a define_expand and a new
1697 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1698 little endian target.
1699 (altivec_vspltw_direct): New.
1700 (altivec_vspltsf): Divide into a define_expand and a new
1701 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1702 a little endian target.
1704 2014-01-30 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/59993
1707 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1708 can propagate form the earlier stmt and avoid the transform
1709 when the intermediate result is needed.
1711 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1713 * README.Portability: Fix typo.
1715 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1717 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1718 comparison_operator with ordered_comparison_operator.
1720 2014-01-30 Nick Clifton <nickc@redhat.com>
1722 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1723 Rename to mn10300_store_multiple_regs.
1724 * config/mn10300/mn10300.c: Likewise.
1725 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1726 store_multiple_regs.
1727 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1728 Call mn10300_store_multiple_regs.
1730 2014-01-30 Nick Clifton <nickc@redhat.com>
1731 DJ Delorie <dj@redhat.com>
1733 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1734 %fp 2 to keep registers after it properly word-aligned.
1735 (rl78_alloc_physical_registers_umul): Handle the case where both
1736 input operands are the same.
1738 2014-01-30 Richard Biener <rguenther@suse.de>
1740 PR tree-optimization/59903
1741 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1744 2014-01-30 Jason Merrill <jason@redhat.com>
1747 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1750 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1752 2014-01-30 Richard Biener <rguenther@suse.de>
1754 PR tree-optimization/59951
1755 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1757 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1760 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1761 SFmode to DFmode case.
1763 2014-01-29 DJ Delorie <dj@redhat.com>
1765 * config/msp430/msp430.opt (-minrt): New.
1766 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1768 (ENDFILE_SPEC): Likewise.
1770 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1772 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1773 (estimate_function_body_sizes): Use it.
1775 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1778 * dwarf2out.c (is_cxx_auto): New.
1779 (is_base_type): Use it.
1780 (gen_type_die_with_usage): Likewise.
1782 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1784 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1785 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1786 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1787 -maltivec=be with LE targets.
1788 (vsx_mergeh_<mode>): Likewise.
1789 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1790 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1791 (altivec_vmrghb): Replace with define_expand and new
1792 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1793 (altivec_vmrghb_direct): New define_insn.
1794 (altivec_vmrghh): Replace with define_expand and new
1795 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1796 (altivec_vmrghh_direct): New define_insn.
1797 (altivec_vmrghw): Replace with define_expand and new
1798 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1799 (altivec_vmrghw_direct): New define_insn.
1800 (*altivec_vmrghsf): Adjust for endianness.
1801 (altivec_vmrglb): Replace with define_expand and new
1802 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1803 (altivec_vmrglb_direct): New define_insn.
1804 (altivec_vmrglh): Replace with define_expand and new
1805 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1806 (altivec_vmrglh_direct): New define_insn.
1807 (altivec_vmrglw): Replace with define_expand and new
1808 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1809 (altivec_vmrglw_direct): New define_insn.
1810 (*altivec_vmrglsf): Adjust for endianness.
1811 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1812 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1813 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1814 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1815 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1816 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1817 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1818 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1820 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1822 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1823 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1826 2014-01-29 Richard Biener <rguenther@suse.de>
1828 PR tree-optimization/58742
1829 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1830 associate_pointerplus_align.
1831 (associate_pointerplus_diff): New function.
1832 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1833 and associate_pointerplus_diff.
1835 2014-01-29 Richard Biener <rguenther@suse.de>
1837 * lto-streamer.h (LTO_major_version): Bump to 3.
1838 (LTO_minor_version): Reset to 0.
1840 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1842 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1843 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1844 (arm_file_start): Generate correct asm header for armv7ve.
1845 * config/arm/bpabi.h: Add multilib support for armv7ve.
1846 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1847 and cortex-a15 to armv7ve.
1848 * config/arm/t-aprofile: Add multilib support for armv7ve.
1849 * doc/invoke.texi: Document -march=armv7ve.
1851 2014-01-29 Richard Biener <rguenther@suse.de>
1853 PR tree-optimization/58742
1854 * tree-ssa-forwprop.c (associate_plusminus): Return true
1855 if we changed sth, defer EH cleanup to ...
1856 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1857 (simplify_mult): New function.
1859 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1862 PR tree-optimization/59920
1863 * tree.c (build_common_builtin_nodes): Remove
1864 __builtin_setjmp_dispatcher initialization.
1865 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1866 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1867 instead of gsi_after_labels + manually skipping debug stmts.
1868 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1869 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1870 * tree-inline.c (copy_edges_for_bb): Remove
1871 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1872 argument. Ignore computed_goto_p stmts. Don't call
1873 make_abnormal_goto_edges. If a call might need abnormal edges
1874 for non-local gotos, see if it already has an edge to
1875 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1876 with true argument, don't do anything then, otherwise add
1877 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1878 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1880 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1881 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1882 (lower_stmt): Don't set data->calls_builtin_setjmp.
1883 (lower_builtin_setjmp): Adjust comment.
1884 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1885 * tree-cfg.c (found_computed_goto): Remove.
1886 (factor_computed_gotos): Remove.
1887 (make_goto_expr_edges): Return bool, true for computed gotos.
1888 Don't call make_abnormal_goto_edges.
1889 (build_gimple_cfg): Don't set found_computed_goto, don't call
1890 factor_computed_gotos.
1891 (computed_goto_p): No longer static.
1892 (make_blocks): Don't set found_computed_goto.
1893 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1894 (make_edges): If make_goto_expr_edges returns true, push bb
1895 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1896 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1897 vector. Record mapping between bbs and OpenMP regions if there
1898 are any, adjust make_gimple_omp_edges caller. Call
1899 handle_abnormal_edges.
1900 (make_abnormal_goto_edges): Remove.
1901 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1902 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1903 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1904 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1905 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1906 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1907 filling *region also set *region_idx to (*region)->entry->index.
1910 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1911 For REGs set ORIGINAL_REGNO.
1913 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1915 * doc/md.texi: Mention that a target shouldn't implement
1916 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1919 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1921 PR tree-optimization/59594
1922 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1923 a copy of the datarefs vector rather than the vector itself.
1925 2014-01-28 Jason Merrill <jason@redhat.com>
1928 * dwarf2out.c (auto_die): New static.
1929 (gen_type_die_with_usage): Handle C++1y 'auto'.
1930 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1933 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1936 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1937 (SPEC_X32): Likewise.
1938 (SPEC_64): Likewise.
1939 * config/i386/i386.c (ix86_option_override_internal): Turn off
1940 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1942 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1943 * config/i386/i386.h (TARGET_16BIT): New macro.
1944 (TARGET_16BIT_P): Likewise.
1945 * config/i386/i386.opt: Add m16.
1946 * doc/invoke.texi: Document -m16.
1948 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1950 PR preprocessor/59935
1951 * input.c (location_get_source_line): Bail out on when line number
1952 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1954 2014-01-28 Richard Biener <rguenther@suse.de>
1956 PR tree-optimization/58742
1957 * tree-ssa-forwprop.c (associate_plusminus): Handle
1958 pointer subtraction of the form (T)(P + A) - (T)P.
1960 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1962 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1965 2014-01-28 Richard Biener <rguenther@suse.de>
1968 2014-01-28 Richard Biener <rguenther@suse.de>
1970 PR rtl-optimization/45364
1971 PR rtl-optimization/59890
1972 * var-tracking.c (local_get_addr_clear_given_value): Handle
1973 already cleared slot.
1974 (val_reset): Handle not allocated local_get_addr_cache.
1975 (vt_find_locations): Use post-order on the inverted CFG.
1977 2014-01-28 Richard Biener <rguenther@suse.de>
1979 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1981 2014-01-28 Richard Biener <rguenther@suse.de>
1983 PR rtl-optimization/45364
1984 PR rtl-optimization/59890
1985 * var-tracking.c (local_get_addr_clear_given_value): Handle
1986 already cleared slot.
1987 (val_reset): Handle not allocated local_get_addr_cache.
1988 (vt_find_locations): Use post-order on the inverted CFG.
1990 2014-01-28 Alan Modra <amodra@gmail.com>
1992 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1993 * configure.ac <recursive call for build != host>: Define
1994 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1995 and LD_FOR_BUILD too.
1996 * configure: Regenerate.
1998 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2000 * config/i386/i386.c (get_builtin_code_for_version): Separate
2001 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2002 Broadwell from Haswell.
2004 2014-01-27 Steve Ellcey <sellcey@mips.com>
2006 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2007 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2008 * config/mips/mips.c (mips_option_override): Change setting
2010 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2011 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2012 Change from Mask to Var.
2014 2014-01-27 Jeff Law <law@redhat.com>
2016 * ipa-inline.c (inline_small_functions): Fix typo.
2018 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2020 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2021 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2022 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2023 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2024 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2025 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2026 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2027 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2028 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2029 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2030 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2031 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2032 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2033 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2034 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2035 (_mm512_storeu_epi64): Ditto.
2036 (_mm512_cmpge_epi32_mask): Ditto.
2037 (_mm512_cmpge_epu32_mask): Ditto.
2038 (_mm512_cmpge_epi64_mask): Ditto.
2039 (_mm512_cmpge_epu64_mask): Ditto.
2040 (_mm512_cmple_epi32_mask): Ditto.
2041 (_mm512_cmple_epu32_mask): Ditto.
2042 (_mm512_cmple_epi64_mask): Ditto.
2043 (_mm512_cmple_epu64_mask): Ditto.
2044 (_mm512_cmplt_epi32_mask): Ditto.
2045 (_mm512_cmplt_epu32_mask): Ditto.
2046 (_mm512_cmplt_epi64_mask): Ditto.
2047 (_mm512_cmplt_epu64_mask): Ditto.
2048 (_mm512_cmpneq_epi32_mask): Ditto.
2049 (_mm512_cmpneq_epu32_mask): Ditto.
2050 (_mm512_cmpneq_epi64_mask): Ditto.
2051 (_mm512_cmpneq_epu64_mask): Ditto.
2052 (_mm512_expand_pd): Ditto.
2053 (_mm512_expand_ps): Ditto.
2054 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2055 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2056 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2057 * config/i386/i386.c (ix86_builtins): Add
2058 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2059 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2060 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2061 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2062 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2063 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2064 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2065 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2066 IX86_BUILTIN_PMOVUSQW512_MEM.
2067 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2068 __builtin_ia32_pmovsqd512mem_mask,
2069 __builtin_ia32_pmovqd512mem_mask,
2070 __builtin_ia32_pmovusqw512mem_mask,
2071 __builtin_ia32_pmovsqw512mem_mask,
2072 __builtin_ia32_pmovqw512mem_mask,
2073 __builtin_ia32_pmovusdw512mem_mask,
2074 __builtin_ia32_pmovsdw512mem_mask,
2075 __builtin_ia32_pmovdw512mem_mask,
2076 __builtin_ia32_pmovqb512mem_mask,
2077 __builtin_ia32_pmovusqb512mem_mask,
2078 __builtin_ia32_pmovsqb512mem_mask,
2079 __builtin_ia32_pmovusdb512mem_mask,
2080 __builtin_ia32_pmovsdb512mem_mask,
2081 __builtin_ia32_pmovdb512mem_mask.
2082 (bdesc_args): Add __builtin_ia32_expanddf512,
2083 __builtin_ia32_expandsf512.
2084 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2085 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2086 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2087 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2088 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2089 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2090 (avx512f_<code>v8div16qi2_mask_store): This.
2091 (avx512f_expand<mode>): New.
2093 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2095 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2097 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2098 (_mm512_prefetch_i32scatter_pd): Ditto.
2099 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2100 (_mm512_prefetch_i64scatter_pd): Ditto.
2101 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2102 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2103 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2104 (_mm512_prefetch_i32scatter_ps): Ditto.
2105 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2106 (_mm512_prefetch_i64scatter_ps): Ditto.
2107 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2108 * config/i386/i386-builtin-types.def: Define
2109 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2110 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2111 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2112 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2113 IX86_BUILTIN_SCATTERPFQPD.
2114 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2115 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2116 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2117 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2118 __builtin_ia32_scatterpfqps.
2119 (ix86_expand_builtin): Expand new built-ins.
2120 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2121 fix memory access data type.
2122 (*avx512pf_gatherpf<mode>_mask): Ditto.
2123 (*avx512pf_gatherpf<mode>): Ditto.
2124 (avx512pf_scatterpf<mode>): Ditto.
2125 (*avx512pf_scatterpf<mode>_mask): Ditto.
2126 (*avx512pf_scatterpf<mode>): Ditto.
2127 (GATHER_SCATTER_SF_MEM_MODE): New.
2128 (avx512pf_gatherpf<mode>df): Ditto.
2129 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2130 (*avx512pf_scatterpf<mode>df): Ditto.
2132 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2135 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2136 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2139 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2141 * common/config/arm/arm-common.c
2142 (arm_rewrite_mcpu): Handle multiple names.
2144 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2146 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2148 * gimple-builder.h (create_gimple_tmp): Delete.
2150 2014-01-27 Christian Bruel <christian.bruel@st.com>
2152 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2155 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2157 * config/pa/pa.md (call): Generate indirect long calls to non-local
2158 functions when outputing 32-bit code.
2159 (call_value): Likewise except for special call to buggy powf function.
2161 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2162 portable runtime and PIC indirect calls.
2163 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2164 and PIC call sequences. Use ldo instead of blr to set return register
2165 in PIC call sequence.
2167 2014-01-25 Walter Lee <walt@tilera.com>
2169 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2170 avoid clobbering a live register.
2172 2014-01-25 Walter Lee <walt@tilera.com>
2174 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2175 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2176 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2177 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2179 2014-01-25 Walter Lee <walt@tilera.com>
2181 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2182 arguments on even registers.
2183 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2185 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2186 (BIGGEST_ALIGNMENT): Ditto.
2187 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2189 2014-01-25 Walter Lee <walt@tilera.com>
2191 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2192 insns before bundling.
2193 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2195 2014-01-25 Walter Lee <walt@tilera.com>
2197 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2198 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2199 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2201 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2203 * config/mips/constraints.md (kl): Delete.
2204 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2205 define expands, using...
2206 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2207 instructions for MIPS16.
2208 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2209 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2211 2014-01-25 Walter Lee <walt@tilera.com>
2213 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2217 2014-01-25 Walter Lee <walt@tilera.com>
2219 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2220 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2222 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2224 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2227 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2229 * print-rtl.c (in_call_function_usage): New var.
2230 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2231 EXPR_LIST mode as mode and not as reg note name.
2234 * cfgloopmanip.c (copy_loop_info): If
2235 loop->warned_aggressive_loop_optimizations, make sure
2236 the flag is set in target loop too.
2238 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2240 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2242 * builtins.def: Likewise.
2243 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2244 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2245 * ira.c (ira_setup_eliminable_regset): Likewise.
2246 * omp-low.c (gate_expand_omp): Likewise.
2247 (execute_lower_omp): Likewise.
2248 (diagnose_sb_0): Likewise.
2249 (gate_diagnose_omp_blocks): Likewise.
2250 (simd_clone_clauses_extract): Likewise.
2253 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2255 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2256 correction for little endian...
2257 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2260 2014-01-24 Jeff Law <law@redhat.com>
2262 PR tree-optimization/59919
2263 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2264 for non-returning calls.
2266 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2268 * common/config/aarch64/aarch64-common.c
2269 (aarch64_rewrite_mcpu): Handle multiple names.
2270 * config/aarch64/aarch64.h
2271 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2273 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2275 * input.c (add_file_to_cache_tab): Handle the case where fopen
2278 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2281 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2282 from push operand if code of push isn't PRE_DEC.
2284 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2287 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2288 -mquad-memory-atomic. Update -mquad-memory documentation to say
2289 it is only used for non-atomic loads/stores.
2291 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2292 -mquad-memory or -mquad-memory-atomic switches.
2294 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2295 -mquad-memory-atomic to ISA 2.07 support.
2297 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2298 to separate support of normal quad word memory operations (ldq, stq)
2299 from the atomic quad word memory operations.
2301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2302 support to separate non-atomic quad word operations from atomic
2303 quad word operations. Disable non-atomic quad word operations in
2304 little endian mode so that we don't have to swap words after the
2305 load and before the store.
2306 (quad_load_store_p): Add comment about atomic quad word support.
2307 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2308 options printed with -mdebug=reg.
2310 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2311 -mquad-memory-atomic as the test for whether we have quad word
2312 atomic instructions.
2313 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2314 or -mp8-vector are used, allow byte/half-word atomic operations.
2316 * config/rs6000/sync.md (load_lockedti): Insure that the address
2317 is a proper indexed or indirect address for the lqarx instruction.
2318 On little endian systems, swap the hi/lo registers after the lqarx
2320 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2321 insure the address is valid for the lqarx instruction.
2322 (store_conditionalti): Insure that the address is a proper indexed
2323 or indirect address for the stqcrx. instruction. On little endian
2324 systems, swap the hi/lo registers before doing the stqcrx.
2326 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2327 insure the address is valid for the stqcrx. instruction.
2329 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2330 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2331 type of quad memory support is available.
2333 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2336 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2337 there is a danger of looping.
2339 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2342 force flag_ira_loop_pressure if set via command line.
2344 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2346 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2347 (ashr_simd): New builtin handling DI mode.
2348 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2349 (aarch64_sshr_simddi): New match pattern.
2350 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2351 (vshrd_n_s64): Likewise.
2352 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2354 2014-01-23 Nick Clifton <nickc@redhat.com>
2356 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2357 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2358 favour of mcu specific scripts.
2359 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2362 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2363 Alex Velenko <Alex.Velenko@arm.com>
2365 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2366 (vaddv_s16): Likewise.
2367 (vaddv_s32): Likewise.
2368 (vaddv_u8): Likewise.
2369 (vaddv_u16): Likewise.
2370 (vaddv_u32): Likewise.
2371 (vaddvq_s8): Likewise.
2372 (vaddvq_s16): Likewise.
2373 (vaddvq_s32): Likewise.
2374 (vaddvq_s64): Likewise.
2375 (vaddvq_u8): Likewise.
2376 (vaddvq_u16): Likewise.
2377 (vaddvq_u32): Likewise.
2378 (vaddvq_u64): Likewise.
2379 (vaddv_f32): Likewise.
2380 (vaddvq_f32): Likewise.
2381 (vaddvq_f64): Likewise.
2382 (vmaxv_f32): Likewise.
2383 (vmaxv_s8): Likewise.
2384 (vmaxv_s16): Likewise.
2385 (vmaxv_s32): Likewise.
2386 (vmaxv_u8): Likewise.
2387 (vmaxv_u16): Likewise.
2388 (vmaxv_u32): Likewise.
2389 (vmaxvq_f32): Likewise.
2390 (vmaxvq_f64): Likewise.
2391 (vmaxvq_s8): Likewise.
2392 (vmaxvq_s16): Likewise.
2393 (vmaxvq_s32): Likewise.
2394 (vmaxvq_u8): Likewise.
2395 (vmaxvq_u16): Likewise.
2396 (vmaxvq_u32): Likewise.
2397 (vmaxnmv_f32): Likewise.
2398 (vmaxnmvq_f32): Likewise.
2399 (vmaxnmvq_f64): Likewise.
2400 (vminv_f32): Likewise.
2401 (vminv_s8): Likewise.
2402 (vminv_s16): Likewise.
2403 (vminv_s32): Likewise.
2404 (vminv_u8): Likewise.
2405 (vminv_u16): Likewise.
2406 (vminv_u32): Likewise.
2407 (vminvq_f32): Likewise.
2408 (vminvq_f64): Likewise.
2409 (vminvq_s8): Likewise.
2410 (vminvq_s16): Likewise.
2411 (vminvq_s32): Likewise.
2412 (vminvq_u8): Likewise.
2413 (vminvq_u16): Likewise.
2414 (vminvq_u32): Likewise.
2415 (vminnmv_f32): Likewise.
2416 (vminnmvq_f32): Likewise.
2417 (vminnmvq_f64): Likewise.
2419 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2421 * config/aarch64/aarch64-simd.md
2422 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2423 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2424 (*aarch64_mul3_elt<mode>): Likewise.
2425 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2426 (*aarch64_mul3_elt_to_64v2df): Likewise.
2427 (*aarch64_mla_elt<mode>): Likewise.
2428 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2429 (*aarch64_mls_elt<mode>): Likewise.
2430 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2431 (*aarch64_fma4_elt<mode>): Likewise.
2432 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2433 (*aarch64_fma4_elt_to_64v2df): Likewise.
2434 (*aarch64_fnma4_elt<mode>): Likewise.
2435 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2436 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2437 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2438 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2439 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2440 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2441 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2442 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2443 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2445 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2447 * config/aarch64/aarch64-simd.md
2448 (aarch64_be_checked_get_lane<mode>): New define_expand.
2449 * config/aarch64/aarch64-simd-builtins.def
2450 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2451 New builtin definition.
2452 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2453 Use new safe be builtin.
2455 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2457 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2459 (aarch64_be_st1<mode>): Likewise.
2460 (aarch_ld1<VALL:mode>): Define_expand modified.
2461 (aarch_st1<VALL:mode>): Likewise.
2462 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2463 (UNSPEC_ST1): Likewise.
2465 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2467 * config/microblaze/microblaze.md: Add trap insn and attribute
2469 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2471 PR preprocessor/58580
2472 * input.h (location_get_source_line): Take an additional line_size
2474 (void diagnostics_file_cache_fini): Declare new function.
2475 * input.c (struct fcache): New type.
2476 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2477 New static constants.
2478 (diagnostic_file_cache_init, total_lines_num)
2479 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2480 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2481 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2482 (get_next_line, read_next_line, goto_next_line, read_line_num):
2483 New static function definitions.
2484 (diagnostic_file_cache_fini): New function.
2485 (location_get_source_line): Take an additional output line_len
2486 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2488 * diagnostic.c (diagnostic_finish): Call
2489 diagnostic_file_cache_fini.
2490 (adjust_line): Take an additional input parameter for the length
2491 of the line, rather than calculating it with strlen.
2492 (diagnostic_show_locus): Adjust the use of
2493 location_get_source_line and adjust_line with respect to their new
2494 signature. While displaying a line now, do not stop at the first
2495 null byte. Rather, display the zero byte as a space and keep
2496 going until we reach the size of the line.
2497 * Makefile.in: Add vec.o to OBJS-libcommon
2499 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2500 Ilya Tocar <ilya.tocar@intel.com>
2502 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2503 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2504 (__builtin_ia32_kmov16): Ditto.
2505 * config/i386/i386.md (UNSPEC_KMOV): New.
2508 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2510 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2511 (_mm512_storeu_si512): Ditto.
2513 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2516 * rtl.h (get_referenced_operands): Declare.
2517 * recog.c (get_referenced_operands): New function.
2518 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2519 operands have been referenced when recording LO_SUM references.
2521 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2523 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2525 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2527 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2528 Enable for generic and recent AMD targets.
2530 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2532 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2533 ARG_SIZE note when adjustment was eliminated.
2535 2014-01-22 Jeff Law <law@redhat.com>
2537 PR tree-optimization/59597
2538 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2539 in file. Accept new argument REGISTERING and use it to modify
2540 dump output appropriately.
2541 (register_jump_thread): Corresponding changes.
2542 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2543 thread paths involving joiner blocks. Add code to dump cancelled
2544 jump threading paths.
2546 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2548 PR rtl-optimization/59477
2549 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2550 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2552 2014-01-22 Tom Tromey <tromey@redhat.com>
2554 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2556 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2558 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2560 PR rtl-optimization/59896
2561 * lra-constraints.c (process_alt_operands): Check unused note for
2562 matched operands of insn with no output reloads.
2564 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2566 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2567 (mips_move_from_gpr_cost): Likewise.
2569 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2571 PR rtl-optimization/59858
2572 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2573 ira_class_hard_regs_num.
2574 (process_alt_operands): Increase reject for dying matched operand.
2576 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2579 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2580 smaller than size, perform several stores or loads and stores
2581 at dst + count - size to store or copy all of size bytes, rather
2582 than just last modesize bytes.
2584 2014-01-20 DJ Delorie <dj@redhat.com>
2586 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2587 that CLOBBERs are REGs before propogating their values.
2589 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2592 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2593 (cgraph_inline_failed_type): New function.
2594 * cgraph.h (DEFCIFCODE): Add type.
2595 (cgraph_inline_failed_type_t): New enum.
2596 (cgraph_inline_failed_type): New prototype.
2597 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2598 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2599 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2600 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2601 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2602 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2603 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2604 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2605 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2606 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2607 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2608 OPTIMIZATION_MISMATCH.
2609 * tree-inline.c (expand_call_inline): Emit errors during
2610 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2612 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2615 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2616 mode attribute in insn output.
2618 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2620 * output.h (output_constant): Delete.
2621 * varasm.c (output_constant): Make private.
2623 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2625 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2627 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2630 * tree.h (fold_builtin_strcat): New prototype.
2631 * builtins.c (fold_builtin_strcat): No longer static. Add len
2632 argument, if non-NULL, don't call c_strlen. Optimize
2633 directly into __builtin_memcpy instead of __builtin_strcpy.
2634 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2635 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2637 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2639 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2640 for SImode_address_operand operands, having only a REG argument.
2642 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2644 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2645 loader name using mbig-endian.
2646 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2648 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2650 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2654 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2656 * config/aarch64/aarch64-protos.h
2657 (aarch64_cannot_change_mode_class_ptr): Declare.
2658 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2659 aarch64_cannot_change_mode_class_ptr): New.
2660 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2661 backend hook aarch64_cannot_change_mode_class.
2663 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2665 * common/config/aarch64/aarch64-common.c
2666 (aarch64_handle_option): Don't handle any option order logic here.
2667 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2668 selected_cpu, warn on architecture version mismatch.
2669 (aarch64_override_options): Fix parsing order for option strings.
2671 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2672 Iain Sandoe <iain@codesourcery.com>
2675 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2676 warning. Amend comment to reflect current functionality.
2678 2014-01-20 Richard Biener <rguenther@suse.de>
2681 * builtins.c (fold_builtin_strcat): Remove case better handled
2682 by tree-ssa-strlen.c.
2684 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2686 * config/aarch64/aarch64.opt
2687 (mcpu, march, mtune): Make case-insensitive.
2689 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2692 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2693 if operands[1] is a REG or ZERO_EXTEND of a REG.
2695 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2697 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2699 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2701 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2702 long non-pic millicode calls.
2704 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2706 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2708 2014-01-19 Kito Cheng <kito@0xlab.org>
2710 * builtins.c (expand_movstr): Check movstr expand done or fail.
2712 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2713 H.J. Lu <hongjiu.lu@intel.com>
2716 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2717 to DImode for zero-extended addresses.
2719 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2721 PR rtl-optimization/57763
2722 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2723 on the new indirect jump_insn and increment LABEL_NUSES (label).
2725 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2729 * config.gcc (x86_archs): New variable.
2730 (x86_64_archs): Likewise.
2731 (x86_cpus): Likewise.
2732 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2733 --with-arch/--with-cpu= options.
2734 Support --with-arch=/--with-cpu={nehalem,westmere,
2735 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2737 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2739 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2740 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2742 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2744 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2746 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2749 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2750 clear cpp_get_options (parse_in)->warn_unused_macros for
2751 ix86_target_macros_internal with cpp_define.
2753 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2755 * jump.c (delete_related_insns): Keep (use (insn))s.
2756 * reorg.c (redundant_insn): Check for barriers too.
2758 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2760 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2762 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2764 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2765 call to $$dyncall when TARGET_LONG_CALLS is true.
2767 2014-01-17 Jeff Law <law@redhat.com>
2769 * ree.c (combine_set_extension): Temporarily disable test for
2770 changing number of hard registers.
2772 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2775 * ipa-inline-analysis.c (inline_free_summary):
2776 Do not free summary of aliases.
2778 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2781 * gimplify.c (gimplify_expr): Use create_tmp_var
2782 instead of create_tmp_var_raw. If cond doesn't have
2783 integral type, don't add the IFN_ANNOTATE builtin at all.
2785 2014-01-17 Martin Jambor <mjambor@suse.cz>
2788 * ipa-cp.c (prev_edge_clone): New variable.
2789 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2790 Also resize prev_edge_clone vector.
2791 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2792 (ipcp_edge_removal_hook): New function.
2793 (ipcp_driver): Register ipcp_edge_removal_hook.
2795 2014-01-17 Andrew Pinski <apinski@cavium.com>
2796 Steve Ellcey <sellcey@mips.com>
2799 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2802 2014-01-17 Jeff Law <law@redhat.com>
2805 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2806 so that pass_ccp runs first.
2808 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2810 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2811 (ix86_adjust_cost): Use !TARGET_XXX.
2812 (do_reorder_for_imul): Likewise.
2813 (swap_top_of_ready_list): Likewise.
2814 (ix86_sched_reorder): Likewise.
2816 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2818 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2819 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2820 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2821 (intel_memset): New. Duplicate slm_memset.
2822 (intel_cost): New. Duplicate slm_cost.
2823 (m_INTEL): New macro.
2824 (processor_target_table): Add "intel".
2825 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2826 with PROCESSOR_INTEL for "intel".
2827 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2828 PROCESSOR_SILVERMONT.
2829 (ix86_issue_rate): Likewise.
2830 (ix86_adjust_cost): Likewise.
2831 (ia32_multipass_dfa_lookahead): Likewise.
2832 (swap_top_of_ready_list): Likewise.
2833 (ix86_sched_reorder): Likewise.
2834 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2835 instead of TARGET_OPT_AGU.
2836 * config/i386/i386.h (TARGET_INTEL): New.
2837 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2838 (processor_type): Add PROCESSOR_INTEL.
2839 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2840 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2842 2014-01-17 Marek Polacek <polacek@redhat.com>
2845 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2848 2014-01-17 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/46590
2851 * opts.c (default_options_table): Add entries for
2852 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2853 all enabled at -O1 but not for -Og.
2854 * common.opt (fbranch-count-reg): Remove Init(1).
2855 (fmove-loop-invariants): Likewise.
2856 (ftree-pta): Likewise.
2858 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2860 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2861 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2862 decls to at least the GCC 4.8 used alignments.
2865 * tree-nested.c (convert_nonlocal_reference_stmt,
2866 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2867 of GIMPLE_BIND stmts, adjust associated decls.
2869 2014-01-17 Richard Biener <rguenther@suse.de>
2871 PR tree-optimization/46590
2872 * vec.h (vec<>::bseach): New member function implementing
2873 binary search according to C89 bsearch.
2874 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2875 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2876 bitmap pointer again. Make accesses_in_loop a flat array.
2877 (mem_ref_obstack): New global.
2878 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2879 (mark_ref_stored): Likewise.
2880 (ref_indep_loop_p_2): Likewise.
2881 (set_ref_stored_in_loop): New helper function.
2882 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2883 (memref_free): Adjust.
2884 (record_mem_ref_loc): Simplify.
2885 (gather_mem_refs_stmt): Adjust.
2886 (sort_locs_in_loop_postorder_cmp): New function.
2887 (analyze_memory_references): Sort accesses_in_loop after
2888 loop postorder number.
2889 (find_ref_loc_in_loop_cmp): New function.
2890 (for_all_locs_in_loop): Find relevant cluster of locs in
2891 accesses_in_loop and iterate without recursion.
2892 (execute_sm): Avoid uninit warning.
2893 (struct ref_always_accessed): Simplify.
2894 (ref_always_accessed::operator ()): Likewise.
2895 (ref_always_accessed_p): Likewise.
2896 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2897 loop postorder numbers here.
2898 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2901 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2904 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2905 on decls for which assemble_alias has been called.
2907 2014-01-17 Nick Clifton <nickc@redhat.com>
2909 * config/msp430/msp430.opt: (mcpu): New option.
2910 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2911 (msp430_option_override): Parse target_cpu. If the MCU name
2912 matches a generic string, clear target_mcu.
2913 (msp430_attr): Allow numeric interrupt values up to 63.
2914 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2915 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2917 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2919 * config/msp430/msp430.md (popm): Use %J rather than %I.
2920 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2921 (addhi_cy_i): Use immediate_operand for operand 2.
2922 * doc/invoke.texi: Document -mcpu option.
2924 2014-01-17 Richard Biener <rguenther@suse.de>
2926 PR rtl-optimization/38518
2927 * df.h (df_analyze_loop): Declare.
2928 * df-core.c: Include cfgloop.h.
2929 (df_analyze_1): Split out main part of df_analyze.
2930 (df_analyze): Adjust.
2931 (loop_inverted_post_order_compute): New function.
2932 (loop_post_order_compute): Likewise.
2933 (df_analyze_loop): New function avoiding whole-function
2935 * loop-invariant.c (find_defs): Use df_analyze_loop.
2936 (find_invariants): Adjust.
2937 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2939 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2941 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2942 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2944 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2946 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2947 traversal when removing references.
2949 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2952 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2954 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2957 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2958 pushing a reload for an autoinc when we had previously reloaded an
2959 inner part of the address.
2961 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2963 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2965 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2966 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2967 when not giving up or versioning for alias only because of
2969 (vect_analyze_data_ref_dependences): Set to true.
2970 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2972 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2973 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2977 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2980 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2981 operand 0 predicate for gathers, use a new pseudo as subtarget.
2983 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2986 * lra-constraints.c (process_alt_operands): Add printing debug info.
2987 Check absence of input/output reloads for matched operands too.
2989 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2991 PR rtl-optimization/59835
2992 * ira.c (ira_init_register_move_cost): Increase cost for
2995 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2997 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2999 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3002 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3003 non-register objects. Use gen_(high/low)part more consistently.
3006 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3009 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3010 endian support, remove tests for WORDS_BIG_ENDIAN.
3011 (p8_mfvsrd_3_<mode>): Likewise.
3012 (reload_gpr_from_vsx<mode>): Likewise.
3013 (reload_gpr_from_vsxsf): Likewise.
3014 (p8_mfvsrd_4_disf): Likewise.
3016 2014-01-16 Richard Biener <rguenther@suse.de>
3018 PR rtl-optimization/46590
3019 * lcm.c (compute_antinout_edge): Use postorder iteration.
3020 (compute_laterin): Use inverted postorder iteration.
3022 2014-01-16 Nick Clifton <nickc@redhat.com>
3025 * varasm.c (output_constant): Return the number of bytes actually
3027 (output_constructor_array_range): Update the field size with the
3028 number of bytes emitted by output_constant.
3029 (output_constructor_regular_field): Likewise. Also do not
3030 complain if the total number of bytes emitted is now greater
3031 than the expected fieldpos.
3032 * output.h (output_constant): Update prototype and descriptive comment.
3034 2014-01-16 Marek Polacek <polacek@redhat.com>
3037 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3038 it is error_mark_node.
3040 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3042 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3043 VALID_AVX256_REG_OR_OI_MODE.
3045 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3047 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3048 current procedure should be profiled.
3050 2014-01-15 Andrew Pinski <apinski@cavium.com>
3052 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3053 of moving from/to the STACK_REG register class.
3055 2014-01-15 Richard Henderson <rth@redhat.com>
3058 * reginfo.c (global_regs_decl): Globalize.
3059 * rtl.h (global_regs_decl): Declare.
3060 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3061 reserved via global_regs.
3063 2014-01-15 Teresa Johnson <tejohnson@google.com>
3065 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3067 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3069 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3070 and vmulosh rather than call gen_vec_widen_smult_*.
3071 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3072 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3073 (vec_widen_smult_even_v16qi): Likewise.
3074 (vec_widen_umult_even_v8hi): Likewise.
3075 (vec_widen_smult_even_v8hi): Likewise.
3076 (vec_widen_umult_odd_v16qi): Likewise.
3077 (vec_widen_smult_odd_v16qi): Likewise.
3078 (vec_widen_umult_odd_v8hi): Likewise.
3079 (vec_widen_smult_odd_v8hi): Likewise.
3080 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3081 vmuloub rather than call gen_vec_widen_umult_*.
3082 (vec_widen_umult_lo_v16qi): Likewise.
3083 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3084 vmulosb rather than call gen_vec_widen_smult_*.
3085 (vec_widen_smult_lo_v16qi): Likewise.
3086 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3087 rather than call gen_vec_widen_umult_*.
3088 (vec_widen_umult_lo_v8hi): Likewise.
3089 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3090 rather than call gen_vec_widen_smult_*.
3091 (vec_widen_smult_lo_v8hi): Likewise.
3093 2014-01-15 Jeff Law <law@redhat.com>
3095 PR tree-optimization/59747
3096 * ree.c (find_and_remove_re): Properly handle case where a second
3097 eliminated extension requires widening a copy created for elimination
3098 of a prior extension.
3099 (combine_set_extension): Ensure that the number of hard regs needed
3100 for a destination register does not change when we widen it.
3102 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3104 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3105 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3106 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3107 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3108 (avr-*-rtems*): Likewise.
3109 (bfin*-rtems*): Likewise.
3110 (moxie-*-rtems*): Likewise.
3111 (h8300-*-rtems*): Likewise.
3112 (i[34567]86-*-rtems*): Likewise.
3113 (lm32-*-rtems*): Likewise.
3114 (m32r-*-rtems*): Likewise.
3115 (m68k-*-rtems*): Likewise.
3116 (microblaze*-*-rtems*): Likewise.
3117 (mips*-*-rtems*): Likewise.
3118 (powerpc-*-rtems*): Likewise.
3119 (sh-*-rtems*): Likewise.
3120 (sparc-*-rtems*): Likewise.
3121 (sparc64-*-rtems*): Likewise.
3122 (v850-*-rtems*): Likewise.
3123 (m32c-*-rtems*): Likewise.
3125 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3127 PR rtl-optimization/59511
3128 * ira.c (ira_init_register_move_cost): Use memory costs for some
3129 cases of register move cost calculations.
3130 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3131 instead of BB frequency.
3132 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3133 * lra-assigns.c (find_hard_regno_for): Ditto.
3135 2014-01-15 Richard Biener <rguenther@suse.de>
3137 PR tree-optimization/59822
3138 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3139 (vectorizable_load): Use it to hoist defs of uses of invariant
3140 loads out of the loop.
3142 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3143 Kugan Vivekanandarajah <kuganv@linaro.org>
3146 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3149 2014-01-15 Richard Biener <rguenther@suse.de>
3151 PR rtl-optimization/59802
3152 * lcm.c (compute_available): Use inverted postorder to seed
3153 the initial worklist.
3155 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3158 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3159 ADDR_REGS for invalid symrefs in non-PIC code.
3161 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3164 * builtins.c (determine_block_size): Initialize *probable_max_size
3165 even if len_rtx is CONST_INT.
3167 2014-01-14 Andrew Pinski <apinski@cavium.com>
3169 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3170 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3171 (cortexa53_tunings): Likewise.
3172 (aarch64_sched_issue_rate): New function.
3173 (TARGET_SCHED_ISSUE_RATE): Define.
3175 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3177 * ira-costs.c (find_costs_and_classes): Add missed
3178 ira_init_register_move_cost_if_necessary.
3180 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3183 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3185 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3188 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3189 to indicate if type is used for function return value. Warn ABI
3190 change if the vector mode isn't available for function return value.
3191 (ix86_function_arg_advance): Pass false to type_natural_mode.
3192 (ix86_function_arg): Likewise.
3193 (ix86_gimplify_va_arg): Likewise.
3194 (function_arg_32): Don't warn ABI change.
3195 (ix86_function_value): Pass true to type_natural_mode.
3196 (ix86_return_in_memory): Likewise.
3197 (ix86_struct_value_rtx): Removed.
3198 (TARGET_STRUCT_VALUE_RTX): Likewise.
3200 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3202 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3203 converting a conditional jump into a conditional return.
3205 2014-01-14 Richard Biener <rguenther@suse.de>
3207 PR tree-optimization/58921
3208 PR tree-optimization/59006
3209 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3210 hoisting invariant stmts.
3211 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3212 invariant loads on the preheader edge if possible.
3214 2014-01-14 Joey Ye <joey.ye@arm.com>
3216 * doc/plugin.texi (Building GCC plugins): Update to C++.
3218 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3220 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3221 (_mm_rcp28_round_ss): Ditto.
3222 (_mm_rsqrt28_round_sd): Ditto.
3223 (_mm_rsqrt28_round_ss): Ditto.
3224 (_mm_rcp28_sd): Ditto.
3225 (_mm_rcp28_ss): Ditto.
3226 (_mm_rsqrt28_sd): Ditto.
3227 (_mm_rsqrt28_ss): Ditto.
3228 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3229 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3230 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3231 (IX86_BUILTIN_RCP28SD): Ditto.
3232 (IX86_BUILTIN_RCP28SS): Ditto.
3233 (IX86_BUILTIN_RSQRT28SD): Ditto.
3234 (IX86_BUILTIN_RSQRT28SS): Ditto.
3235 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3236 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3237 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3238 (ix86_expand_special_args_builtin): Expand new FTYPE.
3239 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3240 (srcp14<mode>): Make insn unary.
3241 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3242 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3243 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3244 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3245 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3246 Fix rounding: make it SAE only.
3247 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3249 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3251 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3252 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3253 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3254 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3255 (round_saeonly_mask_scalar_operand4): Ditto.
3256 (round_saeonly_mask_scalar_op3): Ditto.
3257 (round_saeonly_mask_scalar_op4): Ditto.
3259 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3261 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3262 Implement -maltivec=be for vec_insert and vec_extract.
3264 2014-01-10 DJ Delorie <dj@redhat.com>
3266 * config/msp430/msp430.md (call_internal): Don't allow memory
3267 references with SP as the base register.
3268 (call_value_internal): Likewise.
3269 * config/msp430/constraints.md (Yc): New. For memory references
3270 that don't use SP as a base register.
3272 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3273 "an integer without a # prefix"
3274 * config/msp430/msp430.md (epilogue_helper): Use it.
3276 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3279 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3280 AVX512F gather builtins.
3281 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3282 on gather decls with INTEGER_TYPE masktype.
3283 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3284 directly into the builtin rather than hoisting it before loop.
3286 PR tree-optimization/59387
3287 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3288 (scev_const_prop): If folded_casts and type has undefined overflow,
3289 use force_gimple_operand instead of force_gimple_operand_gsi and
3290 for each added stmt if it is assign with
3291 arith_code_with_undefined_signed_overflow, call
3292 rewrite_to_defined_overflow.
3293 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3294 gimple-fold.h instead.
3295 (arith_code_with_undefined_signed_overflow,
3296 rewrite_to_defined_overflow): Moved to ...
3297 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3298 rewrite_to_defined_overflow): ... here. No longer static.
3299 Include gimplify-me.h.
3300 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3301 rewrite_to_defined_overflow): New prototypes.
3303 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3305 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3307 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3309 * builtins.c (get_object_alignment_2): Minor tweak.
3310 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3312 2014-01-13 Christian Bruel <christian.bruel@st.com>
3314 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3315 optimized non constant lengths.
3317 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3320 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3321 load as __atomic_load_N if possible.
3323 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3325 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3327 (rs6000_expand_builtin): Adjust call.
3329 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3332 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3333 * config/rs6000/rs6000.c: Include target-globals.h.
3334 (rs6000_set_current_function): Instead of doing target_reinit
3335 unconditionally, use save_target_globals_default_opts and
3336 restore_target_globals.
3338 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3340 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3341 (rs6000_expand_builtin): Handle mffs and mtfsf.
3342 (rs6000_init_builtins): Define mffs and mtfsf.
3343 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3344 (rs6000_mffs): New pattern.
3345 (rs6000_mtfsf): New pattern.
3347 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3349 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3350 Start narrowing with START. Apply candidate-use pair
3351 and check overall cost in narrowing.
3352 (iv_ca_prune): Pass new argument.
3354 2014-01-10 Jeff Law <law@redhat.com>
3357 * ree.c (combine_reaching_defs): Ensure the defining statement
3358 occurs before the extension when optimizing extensions with
3359 different source and destination hard registers.
3361 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3364 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3365 vtables into the type inheritance graph.
3367 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3369 PR rtl-optimization/59754
3370 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3371 modes in the REGNO != REGNO case.
3373 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3375 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3377 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3379 PR tree-optimization/59745
3380 * tree-predcom.c (tree_predictive_commoning_loop): Call
3381 free_affine_expand_cache if giving up because components is NULL.
3383 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3384 GC in payload of target_globals struct instead of allocating them on
3385 the heap and the larger structs separately using GC.
3386 * target-globals.h (struct target_globals): Make regs, hard_regs,
3387 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3388 of GTY((skip)) and change type to void *.
3389 (reset_target_globals): Cast loads from those fields to corresponding
3392 2014-01-10 Steve Ellcey <sellcey@mips.com>
3395 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3396 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3397 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3399 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3402 * aarch64-modes.def (CC_Zmode): New flags mode.
3403 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3404 represents an equality.
3405 (aarch64_get_condition_code): Handle CC_Zmode.
3406 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3408 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3410 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3411 extraction in good case.
3413 2014-01-10 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/59374
3416 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3417 checking after SLP discovery. Mark stmts not participating
3418 in any SLP instance properly.
3420 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3422 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3423 when handling a SET rtx.
3425 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3427 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3428 (cortex-a57): Likewise.
3429 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3431 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3433 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3434 non-iwmmxt builtins.
3436 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3440 * ipa-devirt.c record_target_from_binfo): Take as argument
3441 stack of binfos and lookup matching one for virtual inheritance.
3442 (possible_polymorphic_call_targets_1): Update.
3444 2014-01-10 Huacai Chen <chenhc@lemote.com>
3446 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3447 kernel strings for Loongson-2E/2F/3A.
3449 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3452 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3453 is_gimple_call before calling gimple_call_internal_p.
3455 2014-01-09 Steve Ellcey <sellcey@mips.com>
3457 * Makefile.in (TREE_FLOW_H): Remove.
3458 (TREE_SSA_H): Add file names from tree-flow.h.
3459 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3460 * tree.h: Remove tree-flow.h reference.
3461 * hash-table.h: Remove tree-flow.h reference.
3462 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3463 reference with tree-ssa-loop.h.
3465 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3467 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3468 default element-order behavior for -maltivec.
3469 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3470 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3471 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3472 when targeting big endian, at least for now.
3473 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3475 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3478 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3479 var satisfies use_register_for_decl, just take into account type
3480 alignment, rather than decl alignment.
3482 PR tree-optimization/59622
3483 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3484 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3485 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3486 Don't devirtualize for inplace at all. For targets.length () == 1,
3487 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3489 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3491 * config/i386/i386.md (cpu): Remove the unused btver1.
3493 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3495 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3497 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3500 * tree-core.h (struct target_globals): New forward declaration.
3501 (struct tree_target_option): Add globals field.
3502 * tree.h (TREE_TARGET_GLOBALS): Define.
3503 (prepare_target_option_nodes_for_pch): New prototype.
3504 * target-globals.h (struct target_globals): Define even if
3506 * tree.c (prepare_target_option_node_for_pch,
3507 prepare_target_option_nodes_for_pch): New functions.
3508 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3509 * config/i386/i386.c: Include target-globals.h.
3510 (ix86_set_current_function): Instead of doing target_reinit
3511 unconditionally, use save_target_globals_default_opts and
3512 restore_target_globals.
3514 2014-01-09 Richard Biener <rguenther@suse.de>
3516 PR tree-optimization/59715
3517 * tree-cfg.h (split_critical_edges): Declare.
3518 * tree-cfg.c (split_critical_edges): Export.
3519 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3521 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3523 * cfgexpand.c (expand_stack_vars): Optionally disable
3524 asan stack protection.
3525 (expand_used_vars): Likewise.
3526 (partition_stack_vars): Likewise.
3527 * asan.c (asan_emit_stack_protection): Optionally disable
3528 after return stack usage.
3529 (instrument_derefs): Optionally disable memory access instrumentation.
3530 (instrument_builtin_call): Likewise.
3531 (instrument_strlen_call): Likewise.
3532 (asan_protect_global): Optionally disable global variables protection.
3533 * doc/invoke.texi: Added doc for new options.
3534 * params.def: Added new options.
3535 * params.h: Likewise.
3537 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3539 PR rtl-optimization/59724
3540 * ifcvt.c (cond_exec_process_if_block): Don't call
3541 flow_find_head_matching_sequence with 0 longest_match.
3542 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3543 non-active insns if !stop_after.
3544 (try_head_merge_bb): Revert 2014-01-07 changes.
3546 2014-01-08 Jeff Law <law@redhat.com>
3548 * ree.c (get_sub_rtx): New function, extracted from...
3549 (merge_def_and_ext): Here.
3550 (combine_reaching_defs): Use get_sub_rtx.
3552 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3554 * cgraph.h (varpool_variable_node): Do not choke on null node.
3556 2014-01-08 Catherine Moore <clm@codesourcery.com>
3558 * config/mips/mips.md (simple_return): Attempt to use JRC
3560 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3562 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3564 PR rtl-optimization/59137
3565 * reorg.c (steal_delay_list_from_target): Call update_block for
3567 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3569 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3571 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3572 two duplicate entries.
3574 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3577 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3579 * config/mips/mips.c (mips_truncated_op_cost): New function.
3580 (mips_rtx_costs): Adjust test for BADDU.
3581 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3583 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3585 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3586 (*baddu_si): ...this new pattern.
3588 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3591 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3593 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3596 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3598 (expand_expr, expand_normal): Adjust.
3599 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3600 inner_reference_p. Use inner_reference_p to expand inner references.
3601 (store_expr): Adjust.
3602 * cfgexpand.c (expand_call_stmt): Adjust.
3604 2014-01-08 Rong Xu <xur@google.com>
3606 * gcov-io.c (gcov_var): Move from gcov-io.h.
3607 (gcov_position): Ditto.
3608 (gcov_is_error): Ditto.
3609 (gcov_rewrite): Ditto.
3610 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3611 only part to libgcc/libgcov.h.
3613 2014-01-08 Marek Polacek <polacek@redhat.com>
3616 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3618 2014-01-08 Marek Polacek <polacek@redhat.com>
3621 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3623 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3625 PR rtl-optimization/59649
3626 * stor-layout.c (get_mode_bounds): For BImode return
3627 0 and STORE_FLAG_VALUE.
3629 2014-01-08 Richard Biener <rguenther@suse.de>
3632 * gimple.h (is_gimple_builtin_call): Remove.
3633 (gimple_builtin_call_types_compatible_p): New.
3634 (gimple_call_builtin_p): New overload.
3635 * gimple.c (is_gimple_builtin_call): Remove.
3636 (validate_call): Rename to ...
3637 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3639 (validate_type): New static function.
3640 (gimple_call_builtin_p): New overload and adjust.
3641 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3642 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3643 (gimple_fold_stmt_to_constant_1): Likewise.
3644 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3646 2014-01-08 Richard Biener <rguenther@suse.de>
3649 * gimplify.c (gimplify_expr): Gimplify register-register type
3650 VIEW_CONVERT_EXPRs to separate stmts.
3652 2014-01-07 Jeff Law <law@redhat.com>
3655 * ree.c (combine_set_extension): Handle case where source
3656 and destination registers in an extension insn are different.
3657 (combine_reaching_defs): Allow source and destination registers
3658 in extension to be different under limited circumstances.
3659 (add_removable_extension): Remove restriction that the
3660 source and destination registers in the extension are the same.
3661 (find_and_remove_re): Emit a copy from the extension's
3662 destination to its source after the defining insn if
3663 the source and destination registers are different.
3666 * ifcvt.c (merge_if_block): If we are merging a block with more than
3667 one successor with a block with no successors, remove any BARRIER
3668 after the second block.
3670 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3672 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3674 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3677 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3678 for 14-bit register offsets when INT14_OK_STRICT is false.
3680 2014-01-07 Roland Stigge <stigge@antcom.de>
3681 Michael Meissner <meissner@linux.vnet.ibm.com>
3684 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3685 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3687 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3689 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3692 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3694 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3695 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3696 rtx is const0_rtx or not.
3698 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3701 * target-globals.c (save_target_globals): Remove this_fn_optab
3703 * toplev.c: Include optabs.h.
3704 (target_reinit): Temporarily restore the global options if another
3705 set of options are in force.
3707 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3709 PR rtl-optimization/58668
3710 * cfgcleanup.c (flow_find_cross_jump): Don't count
3711 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3712 to determine what is counted.
3713 (flow_find_head_matching_sequence): Use active_insn_p to determine
3715 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3717 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3718 determine what is counted.
3720 PR tree-optimization/59643
3721 * tree-predcom.c (split_data_refs_to_components): If one dr is
3722 read and one write, determine_offset fails and the write isn't
3723 in the bad component, just put the read into the bad component.
3725 2014-01-07 Mike Stump <mikestump@comcast.net>
3726 Jakub Jelinek <jakub@redhat.com>
3729 * tree-core.h (struct tree_optimization_option): Change optabs
3730 type from unsigned char * to void *.
3731 * optabs.c (init_tree_optimization_optabs): Adjust
3732 TREE_OPTIMIZATION_OPTABS initialization.
3734 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3737 * config/i386/i386.h (struct machine_function): Add
3738 no_drap_save_restore field.
3739 * config/i386/i386.c (ix86_save_reg): Use
3740 !cfun->machine->no_drap_save_restore instead of
3741 crtl->stack_realign_needed.
3742 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3743 this function clears frame_pointer_needed. Set
3744 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3745 and DRAP reg is needed.
3747 2014-01-06 Marek Polacek <polacek@redhat.com>
3750 * doc/implement-c.texi: Mention that other integer types are
3751 permitted as bit-field types in strictly conforming mode.
3753 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3755 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3758 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3760 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3762 2014-01-06 Martin Jambor <mjambor@suse.cz>
3765 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3767 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3769 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3773 * var-tracking.c (add_stores): Preserve the value of the source even if
3774 we don't record the store.
3776 2014-01-06 Terry Guo <terry.guo@arm.com>
3778 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3780 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3783 * config/darwin.c (darwin_function_section): Adjust return values to
3784 correspond to optimisation changes made in r206070.
3786 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3788 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3789 from prefetch_block tune setting.
3790 (nocona_cost): Correct size of prefetch block to 64.
3792 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3794 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3795 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3796 used to save the static chain register in the computation of the offset
3797 from which the FP registers need to be restored.
3799 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3801 PR tree-optimization/59519
3802 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3803 ICE if get_current_def (current_new_name) is already non-NULL, as long
3804 as it is a phi result of some other phi in *new_exit_bb that has
3807 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3808 or vmovdqu* for misaligned_operand.
3809 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3810 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3811 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3812 aligned_mem for AVX512F masked aligned load and store builtins and for
3815 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3817 PR tree-optimization/59651
3818 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3819 Address range for negative step should be added by TYPE_SIZE_UNIT.
3821 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3823 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3824 overlapping registers also for registers other than the stack pointer.
3826 2014-01-03 Marek Polacek <polacek@redhat.com>
3829 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3832 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3835 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3838 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3839 (push splitter): Use <P:MODE_SIZE> instead of
3840 GET_MODE_SIZE (<P:MODE>mode).
3841 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3842 (mov -1, reg peephole2): Likewise.
3843 * config/i386/sse.md (*mov<mode>_internal,
3844 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3845 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3846 *<code><mode>3, *andnot<mode>3<mask_name>,
3847 <mask_codefor><code><mode>3<mask_name>): Likewise.
3848 * config/i386/subst.md (mask_mode512bit_condition,
3849 sd_mask_mode512bit_condition): Likewise.
3851 2014-01-02 Xinliang David Li <davidxl@google.com>
3853 PR tree-optimization/59303
3854 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3855 (dump_predicates): Better output format.
3856 (pred_equal_p): New function.
3857 (is_neq_relop_p): Ditto.
3858 (is_neq_zero_form_p): Ditto.
3859 (pred_expr_equal_p): Ditto.
3860 (pred_neg_p): Ditto.
3861 (simplify_pred): Ditto.
3862 (simplify_preds_2): Ditto.
3863 (simplify_preds_3): Ditto.
3864 (simplify_preds_4): Ditto.
3865 (simplify_preds): Ditto.
3867 (push_to_worklist): Ditto.
3868 (get_pred_info_from_cmp): Ditto.
3869 (is_degenerated_phi): Ditto.
3870 (normalize_one_pred_1): Ditto.
3871 (normalize_one_pred): Ditto.
3872 (normalize_one_pred_chain): Ditto.
3873 (normalize_preds): Ditto.
3874 (normalize_cond_1): Remove function.
3875 (normalize_cond): Ditto.
3876 (is_gcond_subset_of): Ditto.
3877 (is_subset_of_any): Ditto.
3878 (is_or_set_subset_of): Ditto.
3879 (is_and_set_subset_of): Ditto.
3880 (is_norm_cond_subset_of): Ditto.
3881 (pred_chain_length_cmp): Ditto.
3882 (convert_control_dep_chain_into_preds): Type change.
3883 (find_predicates): Ditto.
3884 (find_def_preds): Ditto.
3885 (destroy_predicates_vecs): Ditto.
3886 (find_matching_predicates_in_rest_chains): Ditto.
3887 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3888 (is_pred_expr_subset): Ditto.
3889 (is_pred_chain_subset_of): Ditto.
3890 (is_included_in): Ditto.
3891 (is_superset_of): Ditto.
3893 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3895 Update copyright years.
3897 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3899 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3900 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3901 config/arc/arc.md, config/arc/arc.opt,
3902 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3903 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3904 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3905 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3906 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3907 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3908 vtable-verify.c, vtable-verify.h: Use the standard form for the
3911 2014-01-02 Tobias Burnus <burnus@net-b.de>
3913 * gcc.c (process_command): Update copyright notice dates.
3914 * gcov-dump.c: Ditto.
3916 * doc/cpp.texi: Bump @copying's copyright year.
3917 * doc/cppinternals.texi: Ditto.
3918 * doc/gcc.texi: Ditto.
3919 * doc/gccint.texi: Ditto.
3920 * doc/gcov.texi: Ditto.
3921 * doc/install.texi: Ditto.
3922 * doc/invoke.texi: Ditto.
3924 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3926 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3928 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3930 * config/i386/sse.md (*mov<mode>_internal): Guard
3931 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3933 PR rtl-optimization/59647
3934 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3935 new_rtx into UNSIGNED_FLOAT rtxes.
3937 Copyright (C) 2014 Free Software Foundation, Inc.
3939 Copying and distribution of this file, with or without modification,
3940 are permitted in any medium without royalty provided the copyright
3941 notice and this notice are preserved.