1 2014-02-21 Richard Biener <rguenther@suse.de>
3 PR tree-optimization/60276
4 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
5 (STMT_VINFO_MIN_NEG_DIST): New macro.
6 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
7 STMT_VINFO_MIN_NEG_DIST.
8 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
9 made for negative dependence distances still hold.
11 2014-02-21 Richard Biener <rguenther@suse.de>
14 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
15 DECL_INITIAL for globals not in the current function context.
17 2014-02-21 Jakub Jelinek <jakub@redhat.com>
19 PR tree-optimization/56490
20 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
21 * tree-ssa-uninit.c: Include params.h.
22 (compute_control_dep_chain): Add num_calls argument, return false
23 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
24 num_calls to recursive call.
25 (find_predicates): Change dep_chain into normal array,
26 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
27 variable and adjust compute_control_dep_chain caller.
28 (find_def_preds): Likewise.
30 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
32 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
33 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
35 2014-02-21 Nick Clifton <nickc@redhat.com>
37 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
39 (popqi1): Add mode to pre_dec.
42 2014-02-21 Jakub Jelinek <jakub@redhat.com>
44 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
45 mode for mask of V8SFmode permutation.
47 2014-02-20 Richard Henderson <rth@redhat.com>
50 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
51 a new pseudo for OLDVAL.
53 2014-02-20 Jakub Jelinek <jakub@redhat.com>
56 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
57 gen_reg_rtx if d->testing_p.
58 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
59 if d->testing_p and we will certainly return true.
60 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
63 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
65 * emit-rtl.c (gen_reg_rtx): Assert that
66 crtl->emit.regno_pointer_align_length is non-zero.
68 2014-02-20 Richard Henderson <rth@redhat.com>
71 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
72 on failure the store back into EXPECT.
74 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
75 Sandra Loosemore <sandra@codesourcery.com>
77 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
78 * config/nios2/nios2.c (nios2_function_profiler):
79 Add -fPIC (flag_pic == 2) support.
80 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
81 (nios2_large_offset_p): New function.
82 (nios2_unspec_reloc_p): Move up position, update to use
84 (nios2_unspec_address): Remove function.
85 (nios2_unspec_offset): New function.
86 (nios2_large_got_address): New function.
87 (nios2_got_address): Add large offset support.
88 (nios2_legitimize_tls_address): Update usage of removed and new
90 (nios2_symbol_binds_local_p): New function.
91 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
92 (nios2_legitimize_address): Update to use nios2_large_offset_p.
93 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
94 (nios2_print_operand): Merge H/L processing, add hiadj/lo
95 processing for (const (unspec ...)).
96 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
98 2014-02-20 Richard Biener <rguenther@suse.de>
100 * tree-cfg.c (replace_uses_by): Mark altered BBs before
101 doing the substitution.
102 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
104 2014-02-20 Martin Jambor <mjambor@suse.cz>
107 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
108 info when checking whether lattices are bottom.
110 2014-02-20 Richard Biener <rguenther@suse.de>
113 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
116 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
119 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
120 parameter specifying the scaling.
121 (inline_call): Update.
122 (want_inline_recursively): Guard division by zero.
123 (recursive_inlining): Update.
124 * ipa-inline.h (clone_inlined_nodes): Update.
126 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
129 * config/i386/i386.c (classify_argument): Pass structures of size
130 64 bytes or less in register.
132 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
133 Kirill Yukhin <kirill.yukhin@intel.com>
135 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
136 (_mm_rcp28_round_ss): Ditto.
137 (_mm_rsqrt28_round_sd): Ditto.
138 (_mm_rsqrt28_round_ss): Ditto.
139 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
140 (_mm_rcp14_round_ss): Ditto.
141 (_mm_rsqrt14_round_sd): Ditto.
142 (_mm_rsqrt14_round_ss): Ditto.
143 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
144 the first input operand, get rid of match_dup.
145 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
147 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
149 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
150 operand as the first input operand, set type attribute.
151 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
153 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
154 operand as the first input operand, set type attribute.
156 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
158 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
161 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
164 * config/i386/i386.c (construct_container): Remove TFmode check
165 for X86_64_INTEGER_CLASS.
167 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
170 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
171 only when -Wpsabi is enabled.
173 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
176 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
177 passing arrays in registers are the same as for structs, so remove the
178 special case for them.
180 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
182 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
183 destination type, extract only the valid bits if the source type is not
184 integral and has a different mode.
186 2014-02-19 Richard Biener <rguenther@suse.de>
189 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
192 2014-02-19 Richard Biener <rguenther@suse.de>
195 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
196 (ipa_modify_call_arguments): Emit an argument load explicitely and
197 preserve virtual SSA form there and for the replacement call.
198 Do not update SSA form nor free dominance info.
200 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
202 * ipa.c (function_and_variable_visibility): Also clear WEAK
203 flag when disolving COMDAT_GROUP.
205 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
207 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
208 * ipa-prop.c (ipa_set_jf_known_type): Return early when
210 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
211 do more sanity checks.
212 (detect_type_change): Return true when giving up early.
213 (compute_complex_assign_jump_func): Fix type parameter of
215 (compute_complex_ancestor_jump_func): Likewise.
216 (update_jump_functions_after_inlining): Fix updating of
218 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
220 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
222 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
223 inline clones when edge disappears.
225 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
228 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
229 Split 64-bit moves into 2 patterns. Do not allow the use of
230 direct move for TDmode in little endian, since the decimal value
231 has little endian bytes within a word, but the 64-bit pieces are
232 ordered in a big endian fashion, and normal subreg's of TDmode are
234 (mov<mode>_64bit_dm): Likewise.
235 (movtd_64bit_nodm): Likewise.
237 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
239 PR tree-optimization/60174
240 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
241 statement of an SSA_NAME that occurs in an abnormal PHI node.
243 2014-02-18 Jakub Jelinek <jakub@redhat.com>
246 * final.c (SEEN_BB): Remove.
247 (SEEN_NOTE, SEEN_EMITTED): Renumber.
248 (final_scan_insn): Don't force_source_line on second
249 NOTE_INSN_BASIC_BLOCK.
251 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
254 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
255 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
256 (type_natural_mode): Warn ABI change when %zmm register is not
257 available for AVX512F vector value passing.
259 2014-02-18 Kai Tietz <ktietz@redhat.com>
262 * config/i386/i386.c (ix86_expand_prologue): Use value in
263 rax register as displacement when restoring %r10 or %rax.
264 Fix wrong offset when restoring both registers.
266 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
268 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
269 assertion with conditional return.
271 2014-02-18 Jakub Jelinek <jakub@redhat.com>
272 Uros Bizjak <ubizjak@gmail.com>
275 * config/i386/driver-i386.c (host_detect_local_cpu): If
276 YMM state is not saved by the OS, also clear has_f16c. Move
277 CPUID 0x80000001 handling before YMM state saving checking.
279 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
281 PR rtl-optimization/58960
282 * haifa-sched.c (alloc_global_sched_pressure_data): New,
283 factored out from ...
284 (sched_init): ... here.
285 (free_global_sched_pressure_data): New, factored out from ...
286 (sched_finish): ... here.
287 * sched-int.h (free_global_sched_pressure_data): Declare.
288 * sched-rgn.c (nr_regions_initial): New static global.
289 (haifa_find_rgns): Initialize it.
290 (schedule_region): Disable sched-pressure for the newly
293 2014-02-17 Richard Biener <rguenther@suse.de>
295 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
296 release SSA defs of pattern stmts.
298 2014-02-17 Richard Biener <rguenther@suse.de>
300 * tree-inline.c (expand_call_inline): Release the virtual
301 operand defined by the call we are about to inline.
303 2014-02-17 Richard Biener <rguenther@suse.de>
305 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
307 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
308 Ilya Tocar <ilya.tocar@intel.com>
310 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
311 arguments order in builtin.
312 (_mm512_permutexvar_epi64): Ditto.
313 (_mm512_mask_permutexvar_epi64): Ditto
314 (_mm512_maskz_permutexvar_epi32): Ditto
315 (_mm512_permutexvar_epi32): Ditto
316 (_mm512_mask_permutexvar_epi32): Ditto
318 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
320 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
321 (p8_vmrgow): Likewise.
323 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
325 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
328 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
331 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
332 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
333 into 64-bit and 32-bit moves. On 64-bit moves, add support for
334 using direct move instructions on ISA 2.07. Also adjust
335 instruction length for 64-bit.
336 (mov<mode>_64bit, TFmode/TDmode): Likewise.
337 (mov<mode>_32bit, TFmode/TDmode): Likewise.
339 2014-02-15 Alan Modra <amodra@gmail.com>
343 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
344 find_replacement on parts of insn rtl that might be reloaded.
346 2014-02-15 Richard Biener <rguenther@suse.de>
348 PR tree-optimization/60183
349 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
350 (tree_ssa_phiprop): Calculate and free post-dominators.
352 2014-02-14 Jeff Law <law@redhat.com>
354 PR rtl-optimization/60131
355 * ree.c (get_extended_src_reg): New function.
356 (combine_reaching_defs): Use it rather than assuming location of REG.
357 (find_and_remove_re): Verify first operand of extension is
358 a REG before adding the insns to the copy list.
360 2014-02-14 Roland McGrath <mcgrathr@google.com>
362 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
363 * configure: Regenerated.
364 * config.in: Regenerated.
365 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
366 instead of ASM_SHORT.
368 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
369 Richard Earnshaw <rearnsha@arm.com>
371 PR rtl-optimization/59535
372 * lra-constraints.c (process_alt_operands): Encourage alternative
373 when unassigned pseudo class is superset of the alternative class.
374 (inherit_reload_reg): Don't inherit when optimizing for code size.
375 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
376 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
377 modes not less than 4 for Thumb1.
379 2014-02-14 Kyle McMartin <kyle@redhat.com>
382 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
384 2014-02-14 Richard Biener <rguenther@suse.de>
386 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
387 (get_frame_arg): Drop the assert with langhook types_compatible_p.
388 Do not strip INDIRECT_REFs.
390 2014-02-14 Richard Biener <rguenther@suse.de>
393 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
394 DECL_FUNCTION_SPECIFIC_TARGET.
395 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
396 * tree-streamer-out.c (pack_ts_target_option): Remove.
397 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
398 (write_ts_function_decl_tree_pointers): Do not stream
399 DECL_FUNCTION_SPECIFIC_TARGET.
400 * tree-streamer-in.c (unpack_ts_target_option): Remove.
401 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
402 (lto_input_ts_function_decl_tree_pointers): Do not stream
403 DECL_FUNCTION_SPECIFIC_TARGET.
405 2014-02-14 Jakub Jelinek <jakub@redhat.com>
407 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
408 (get_initial_def_for_induction, vectorizable_induction): Ignore
409 debug stmts when looking for exit_phi.
410 (vectorizable_live_operation): Fix up condition.
412 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
414 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
415 nreverse() because it changes the content of original tree list.
417 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
419 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
420 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
422 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
424 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
425 GNU coding standards.
427 2014-02-13 Jakub Jelinek <jakub@redhat.com>
430 * dwarf2out.c (gen_subprogram_die): Don't call
431 add_calling_convention_attribute if subr_die is old_die.
433 2014-02-13 Sharad Singhai <singhai@google.com>
435 * doc/optinfo.texi: Fix order of nodes.
437 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
439 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
440 operands[2], not operands[3].
442 2014-02-13 Richard Biener <rguenther@suse.de>
445 * doc/install.texi (ISL): Update recommended version to 0.12.2,
446 mention the possibility of an in-tree build.
447 (CLooG): Update recommended version to 0.18.1, mention the
448 possibility of an in-tree build and clarify that the ISL
449 bundled with CLooG does not work.
451 2014-02-13 Jakub Jelinek <jakub@redhat.com>
454 * expr.c (compress_float_constant): If x is a hard register,
455 extend into a pseudo and then move to x.
457 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
459 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
460 caused by bad second argument to warning_at() with -mhotpatch and
461 nested functions (e.g. with gfortran).
463 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
465 * opts.c (option_name): Remove "enabled by default" rider.
467 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
469 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
471 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
472 Uros Bizjak <ubizjak@gmail.com>
475 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
476 * configure: Regenerated.
478 2014-02-12 Richard Biener <rguenther@suse.de>
480 * vec.c (vec_prefix::calculate_allocation): Move as
481 inline variant to vec.h.
482 (vec_prefix::calculate_allocation_1): New out-of-line version.
483 * vec.h (vec_prefix::calculate_allocation_1): Declare.
484 (vec_prefix::m_has_auto_buf): Rename to ...
485 (vec_prefix::m_using_auto_storage): ... this.
486 (vec_prefix::calculate_allocation): Inline the easy cases
487 and dispatch to calculate_allocation_1 which doesn't need the
489 (va_heap::reserve): Use gcc_checking_assert.
490 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
491 m_using_auto_storage.
492 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
494 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
495 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
496 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
498 2014-02-12 Richard Biener <rguenther@suse.de>
500 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
501 when we found a dependence.
503 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
505 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
507 (maybe_fold_stmt): ... into this new function.
508 * omp-low.c (lower_omp): Update comment.
510 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
513 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
516 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
518 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
519 identifiers in comments.
520 (cortexa53_extra_costs): Likewise.
521 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
522 (cortexa7_extra_costs): Likewise.
523 (cortexa12_extra_costs): Likewise.
524 (cortexa15_extra_costs): Likewise.
525 (v7m_extra_costs): Likewise.
527 2014-02-12 Richard Biener <rguenther@suse.de>
530 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
531 of posix_memalign being successful.
532 (lower_stmt): Restrict lowering of posix_memalign to when
533 -ftree-bit-ccp is enabled.
535 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
537 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
539 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
541 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
543 PR rtl-optimization/60116
544 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
545 other_insn once the combination has been validated.
547 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
550 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
552 * ipa-devirt.c: Include demangle.h
553 (odr_violation_reported): New static variable.
554 (add_type_duplicate): Update odr_violations.
555 (maybe_record_node): Add completep parameter; update it.
556 (record_target_from_binfo): Add COMPLETEP parameter;
558 (possible_polymorphic_call_targets_1): Likewise.
559 (struct polymorphic_call_target_d): Add nonconstruction_targets;
560 rename FINAL to COMPLETE.
561 (record_targets_from_bases): Sanity check we found the binfo;
562 fix COMPLETEP updating.
563 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
564 parameter, fix computing of COMPLETEP.
565 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
566 at LTO time do demangling.
567 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
568 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
570 (gimple_get_virt_method_for_binfo): Likewise.
571 * gimple-fold.h (gimple_get_virt_method_for_binfo,
572 gimple_get_virt_method_for_vtable): Update prototypes.
574 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
577 * genautomata.c (add_presence_absence): Fix typo with
578 {final_}presence_list.
580 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
583 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
584 for VSX/Altivec vectors that land in GPR registers.
586 2014-02-11 Richard Henderson <rth@redhat.com>
587 Jakub Jelinek <jakub@redhat.com>
590 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
591 around drhs if type conversion to lacc->type is not useless.
593 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
595 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
597 (cortex-a57.cortex-a53): Likewise.
598 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
600 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
602 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
605 2014-02-11 Renlin Li <Renlin.Li@arm.com>
607 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
608 add_options_for_arm_vfp3.
610 2014-02-11 Jeff Law <law@redhat.com>
613 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
614 object with an undesirable mode.
616 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
619 * config/i386/sol2-9.h: New file.
620 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
621 *-*-solaris2.9*): Use it.
623 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
625 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
626 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
628 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
630 * config/microblaze/microblaze.c: Extend mcpu version format
632 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
634 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
636 2014-02-10 Richard Henderson <rth@redhat.com>
639 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
640 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
641 ms-abi vs -mno-accumulate-outgoing-args.
642 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
643 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
646 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
649 * cfgexpand.c (expand_asm_operands): Attach source location to
650 ASM_INPUT rtx objects.
651 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
653 2014-02-10 Nick Clifton <nickc@redhat.com>
655 * config/mn10300/mn10300.c (popcount): New function.
656 (mn10300_expand_prologue): Include saved registers in stack usage
659 2014-02-10 Jeff Law <law@redhat.com>
662 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
663 when changing the SET_DEST of a prior insn to avoid an input reload.
665 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
667 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
668 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
669 -mcall-openbsd, or -mcall-linux.
670 (CC1_ENDIAN_BIG_SPEC): Remove.
671 (CC1_ENDIAN_LITTLE_SPEC): Remove.
672 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
673 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
674 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
675 and %cc1_endian_default.
676 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
678 2014-02-10 Richard Biener <rguenther@suse.de>
680 PR tree-optimization/60115
681 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
682 MEM_REF handling. Properly verify that the accesses are not
683 out of the objects bound.
685 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
687 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
690 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
692 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
693 proper constants and fix formatting.
694 (possible_polymorphic_call_targets): Fix formatting.
696 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
697 Ilya Tocar <ilya.tocar@intel.com>
699 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
700 (_mm512_loadu_epi32): Renamed into...
701 (_mm512_loadu_si512): This.
702 (_mm512_storeu_epi32): Renamed into...
703 (_mm512_storeu_si512): This.
704 (_mm512_maskz_ceil_ps): Removed.
705 (_mm512_maskz_ceil_pd): Ditto.
706 (_mm512_maskz_floor_ps): Ditto.
707 (_mm512_maskz_floor_pd): Ditto.
708 (_mm512_floor_round_ps): Ditto.
709 (_mm512_floor_round_pd): Ditto.
710 (_mm512_ceil_round_ps): Ditto.
711 (_mm512_ceil_round_pd): Ditto.
712 (_mm512_mask_floor_round_ps): Ditto.
713 (_mm512_mask_floor_round_pd): Ditto.
714 (_mm512_mask_ceil_round_ps): Ditto.
715 (_mm512_mask_ceil_round_pd): Ditto.
716 (_mm512_maskz_floor_round_ps): Ditto.
717 (_mm512_maskz_floor_round_pd): Ditto.
718 (_mm512_maskz_ceil_round_ps): Ditto.
719 (_mm512_maskz_ceil_round_pd): Ditto.
720 (_mm512_expand_pd): Ditto.
721 (_mm512_expand_ps): Ditto.
722 * config/i386/i386.c (ix86_builtins): Remove
723 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
725 * config/i386/predicates.md (const1256_operand): New.
726 (const_1_to_2_operand): Ditto.
727 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
728 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
729 (*avx512pf_gatherpf<mode>sf): Ditto.
730 (avx512pf_gatherpf<mode>df): Ditto.
731 (*avx512pf_gatherpf<mode>df_mask): Ditto.
732 (*avx512pf_gatherpf<mode>df): Ditto.
733 (avx512pf_scatterpf<mode>sf): Ditto.
734 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
735 (*avx512pf_scatterpf<mode>sf): Ditto.
736 (avx512pf_scatterpf<mode>df): Ditto.
737 (*avx512pf_scatterpf<mode>df_mask): Ditto.
738 (*avx512pf_scatterpf<mode>df): Ditto.
739 (avx512f_expand<mode>): Removed.
740 (<shift_insn><mode>3<mask_name>): Change predicate type.
742 2014-02-08 Jakub Jelinek <jakub@redhat.com>
744 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
745 not at the end of datarefs vector use ordered_remove to avoid
746 reordering datarefs vector.
749 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
750 mark local addressable non-static vars as GOVD_PRIVATE
751 instead of GOVD_LOCAL.
752 * omp-low.c (lower_omp_for): Move gimple_bind_vars
753 and BLOCK_VARS of gimple_bind_block to new_stmt rather
757 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
758 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
759 assume_aligned or alloc_align attributes.
760 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
761 arguments. Handle also assume_aligned and alloc_align attributes.
762 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
763 calls to functions with assume_aligned or alloc_align attributes.
764 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
766 2014-02-08 Terry Guo <terry.guo@arm.com>
768 * doc/invoke.texi: Document ARM -march=armv7e-m.
770 2014-02-08 Jakub Jelinek <jakub@redhat.com>
772 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
773 flag on __cilkrts_rethrow builtin.
776 * ipa-cp.c (determine_versionability): Fail at -O0
777 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
778 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
781 2014-02-04 Jakub Jelinek <jakub@redhat.com>
784 * tree-inline.c (copy_forbidden): Fail for
785 __attribute__((optimize (0))) functions.
787 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
789 * varpool.c: Include pointer-set.h.
790 (varpool_remove_unreferenced_decls): Variables in other partitions
791 will not be output; be however careful to not lose information
794 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
796 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
797 lookup in the vtable constructor.
799 2014-02-07 Jeff Law <law@redhat.com>
802 * config/m68k/m68k.md (ashldi_extsi): Turn into a
803 define_insn_and_split.
805 * ipa-inline.c (inline_small_functions): Fix typos.
807 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
809 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
810 (s390_can_use_return_insn): Declare.
811 * config/s390/s390.h (EPILOGUE_USES): Define.
812 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
814 (s390_chunkify_start): Handle return JUMP_LABELs.
815 (s390_early_mach): Emit a main_pool instruction on the entry edge.
816 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
817 (s390_can_use_return_insn): New functions.
818 (s390_fix_long_loop_prediction): Handle conditional returns.
819 (TARGET_SET_UP_BY_PROLOGUE): Define.
820 * config/s390/s390.md (ANY_RETURN): New code iterator.
821 (*creturn, *csimple_return, return, simple_return): New patterns.
823 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
825 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
826 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
827 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
828 REG_CFA_RESTORE list when deciding not to restore a register.
830 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
832 * config/s390/s390.c: Include tree-pass.h and context.h.
833 (s390_early_mach): New function, split out from...
834 (s390_emit_prologue): ...here.
835 (pass_data_s390_early_mach): New pass structure.
836 (pass_s390_early_mach): New class.
837 (s390_option_override): Create and register early_mach pass.
840 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
842 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
843 to match for the exit block.
845 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
847 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
848 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
849 Reject misaligned operands.
851 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
853 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
855 2014-02-07 Richard Biener <rguenther@suse.de>
858 * gimple-low.c (lower_builtin_posix_memalign): New function.
859 (lower_stmt): Call it to lower posix_memalign in a way
860 to make alignment info accessible.
862 2014-02-07 Jakub Jelinek <jakub@redhat.com>
865 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
866 __builtin_setjmp_receiver.
868 2014-02-07 Richard Biener <rguenther@suse.de>
871 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
872 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
873 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
874 Handle BUILT_IN_POSIX_MEMALIGN.
875 (find_func_clobbers): Likewise.
876 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
877 (call_may_clobber_ref_p_1): Likewise.
879 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
882 * ipa-devirt.c (record_target_from_binfo): Remove overactive
885 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
888 * lto-cgraph.c (lto_output_node): Use
889 symtab_get_symbol_partitioning_class.
890 (lto_output_varpool_node): likewise.
891 (symtab_get_symbol_partitioning_class): Move here from
893 * cgraph.h (symbol_partitioning_class): Likewise.
894 (symtab_get_symbol_partitioning_class): Declare.
896 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
898 * ggc.h (ggc_internal_cleared_alloc): New macro.
899 * vec.h (vec_safe_copy): Handle memory stats.
900 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
901 * target-globals.c (save_target_globals): Likewise.
903 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
906 * expr.c (emit_move_resolve_push): Export; be bit more selective
907 on when to clear alias set.
908 * expr.h (emit_move_resolve_push): Declare.
909 * function.h (struct function): Add tail_call_marked.
910 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
911 * config/i386/i386-protos.h (ix86_expand_push): Remove.
912 * config/i386/i386.md (TImode move expander): De not call
914 (FP push expanders): Preserve memory attributes.
915 * config/i386/sse.md (push<mode>1): Remove.
916 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
917 (ix86_expand_push): Remove.
918 * config/i386/mmx.md (push<mode>1): Remove.
920 2014-02-06 Jakub Jelinek <jakub@redhat.com>
922 PR rtl-optimization/60030
923 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
924 lopart with paradoxical subreg before shifting it up by hprec.
926 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
928 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
929 Remove extra newline at end of file.
930 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
931 (arm_issue_rate): Handle cortexa57.
932 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
933 (cortex-a57.cortex-a53): Likewise.
935 2014-02-06 Jakub Jelinek <jakub@redhat.com>
938 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
939 don't record in REG_FRAME_RELATED_EXPR registers not set in that
941 (arm_expand_prologue): Adjust all callers.
942 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
943 info, registers also at the lowest numbered registers side. Use
944 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
948 * var-tracking.c (adjust_mems): Before adding a SET to
949 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
951 2014-02-06 Alan Modra <amodra@gmail.com>
954 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
955 change SDmode to DDmode when lra_in_progress.
957 2014-02-06 Jakub Jelinek <jakub@redhat.com>
960 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
961 free_data_ref on the dr first, and before goto again also set dr
962 to the next dr. For simd_lane_access, free old datarefs[i] before
963 overwriting it. For get_vectype_for_scalar_type failure, don't
964 free_data_ref if simd_lane_access.
966 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
969 * tree.h (opts_for_fn): New inline function.
970 (opt_for_fn): Define.
971 * config/i386/i386.c (ix86_function_regparm): Use
972 opt_for_fn (decl, optimize) instead of optimize.
974 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
976 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
977 for SYMBOL_REF in large memory model.
979 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
981 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
983 (cortex-a57): Likewise.
984 (cortex-a57.cortex-a53): Likewise.
986 2014-02-06 Yury Gribov <y.gribov@samsung.com>
987 Kugan Vivekanandarajah <kuganv@linaro.org>
989 * config/arm/arm.c (arm_vector_alignment_reachable): Check
991 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
993 2014-02-06 Richard Biener <rguenther@suse.de>
995 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
996 set_loop_copy and initialize_original_copy_tables.
998 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1000 * config/aarch64/aarch64-simd.md
1001 (aarch64_ashr_simddi): Change QI to SI.
1003 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1004 Jakub Jelinek <jakub@redhat.com>
1007 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1010 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1012 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1013 CODE_FOR_altivec_vpku[hw]um to
1014 CODE_FOR_altivec_vpku[hw]um_direct.
1015 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1016 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1017 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1018 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1020 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1022 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1023 generation for -maltivec=be.
1024 (altivec_vsumsws): Simplify redundant test.
1026 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1028 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1029 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1030 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1031 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1032 gen_altivec_vpkuwum.
1033 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1035 (altivec_vpks<VI_char>ss): Likewise.
1036 (altivec_vpks<VI_char>us): Likewise.
1037 (altivec_vpku<VI_char>us): Likewise.
1038 (altivec_vpku<VI_char>um): Likewise.
1039 (altivec_vpku<VI_char>um_direct): New (copy of
1040 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1042 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1043 target is little endian and -maltivec=be is not specified.
1044 (*altivec_vupkhs<VU_char>_direct): New (copy of
1045 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1046 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1047 target is little endian and -maltivec=be is not specified.
1048 (*altivec_vupkls<VU_char>_direct): New (copy of
1049 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1050 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1051 little endian and -maltivec=be is not specified.
1052 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1053 little endian and -maltivec=be is not specified.
1055 2014-02-05 Richard Henderson <rth@redhat.com>
1058 * combine-stack-adj.c: Revert r206943.
1059 * sched-int.h (struct deps_desc): Add last_args_size.
1060 * sched-deps.c (init_deps): Initialize it.
1061 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1062 contain REG_ARGS_SIZE notes.
1064 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1066 * lto-cgraph.c (asm_nodes_output): Make global.
1067 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1068 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1069 (driver_handle_option): Handle OPT_fwpa.
1071 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1074 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1075 a comment typo and formatting issue. If odr_hash hasn't been
1076 created, return vNULL and set *completep to false.
1079 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1080 bb with no successors.
1082 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1085 * doc/invoke.texi (-march): Clarify documentation for ARM.
1089 2014-02-05 Richard Biener <rguenther@suse.de>
1091 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1092 when not vectorizing because of too many alias checks.
1093 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1094 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1096 2014-02-05 Nick Clifton <nickc@redhat.com>
1098 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1099 accept extended registers in any mode when compiling for the MN10300.
1101 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1103 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1104 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1105 sanitization attributes.
1106 (can_inline_edge_p): Likewise.
1107 (sanitize_attrs_match_for_inline_p): New function.
1109 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1111 * ipa-prop.c (detect_type_change): Shor circuit testing of
1112 type changes on THIS pointer.
1114 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1117 * config/pa/pa.c (legitimize_tls_address): Return original address
1118 if not passed a SYMBOL_REF rtx.
1119 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1121 (pa_emit_move_sequence): Simplify TLS source operands.
1122 (pa_legitimate_constant_p): Reject all TLS constants.
1123 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1124 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1126 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1128 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1129 groups when we know they are controlled by LTO.
1130 * varasm.c (default_binds_local_p_1): If object is in other partition,
1131 it will be resolved locally.
1133 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1135 * config/host-linux.c (linux_gt_pch_use_address): Don't
1136 use SSIZE_MAX because it is not always defined.
1138 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1141 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1142 threshold for pseudo splitting.
1143 (update_ebb_live_info): Process call argument hard registers and
1144 hard registers from insn definition too.
1145 (max_small_class_regs_num): New constant.
1146 (inherit_in_ebb): Update live hard regs through EBBs. Update
1147 reloads_num only for small register classes. Don't split for
1150 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1153 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1156 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1158 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1161 2014-02-04 Marek Polacek <polacek@redhat.com>
1163 * gdbinit.in (pel): Define.
1165 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1167 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1170 2014-02-04 Richard Biener <rguenther@suse.de>
1173 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1174 in function context local.
1175 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1176 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1177 similar to LTO_imported_decl_ref.
1179 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1181 PR tree-optimization/60002
1182 * cgraphclones.c (build_function_decl_skip_args): Clear
1185 PR tree-optimization/60023
1186 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1187 false to gsi_replace.
1188 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1189 has been in some EH region and vec_stmt could throw, add
1190 vec_stmt into the same EH region.
1191 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1192 has no lhs, ignore it.
1193 * internal-fn.c (expand_MASK_LOAD): Likewise.
1196 * tree-inline.c (copy_forbidden): Fail for
1197 __attribute__((optimize (0))) functions.
1200 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1201 is set, copy one less argument.
1202 (expand_simd_clones): Don't subtract clone_info->inbranch
1203 from simd_clone_struct_alloc argument.
1205 PR rtl-optimization/57915
1206 * recog.c (simplify_while_replacing): If all unary/binary/relational
1207 operation arguments are constant, attempt to simplify those.
1210 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1211 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1213 2014-02-04 Richard Biener <rguenther@suse.de>
1215 PR tree-optimization/60012
1216 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1217 TBAA disambiguation to all DDRs.
1219 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1222 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1223 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1225 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1228 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1230 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1232 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1233 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1235 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1238 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1239 to figure out targets of polymorphic calls with known decl.
1240 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1241 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1242 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1243 (get_polymorphic_call_info): ... here.
1244 (get_polymorphic_call_info_from_invariant): New function.
1246 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1248 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1249 lookup via vtable pointer; check for type consistency
1250 and turn inconsitent facts into UNREACHABLE.
1251 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1252 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1253 type inconsistent querries; return UNREACHABLE instead.
1255 2014-02-03 Richard Henderson <rth@twiddle.net>
1258 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1259 already processed this node.
1260 (normalize_one_pred_1): Pass along mark_set.
1261 (normalize_one_pred): Create and destroy a pointer_set_t.
1262 (normalize_one_pred_chain): Likewise.
1264 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1266 PR gcov-profile/58602
1267 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1269 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1272 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1273 -fno-devirtualize; try to devirtualize by the knowledge of
1274 virtual table pointer given by aggregate propagation.
1275 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1276 (ipa_print_node_jump_functions): Dump also offset that
1277 is relevant for polymorphic calls.
1278 (determine_known_aggregate_parts): Add arg_type parameter; use it
1279 instead of determining the type from pointer type.
1280 (ipa_compute_jump_functions_for_edge): Update call of
1281 determine_known_aggregate_parts.
1282 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1283 (gimple_get_virt_method_for_binfo): ... here; simplify using
1284 vtable_pointer_value_to_vtable.
1285 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1286 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1287 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1288 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1290 (vtable_pointer_value_to_binfo): ... here.
1291 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1293 2014-02-03 Teresa Johnson <tejohnson@google.com>
1295 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1296 redef of outer loop index variable.
1298 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1302 * doc/extend.texi (Function Attributes): Typo.
1304 2014-02-03 Cong Hou <congh@google.com>
1306 PR tree-optimization/60000
1307 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1308 if the vectorized statement is a store. A store statement can only
1309 appear at the end of pattern statements.
1311 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1313 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1314 (ix86_option_override_internal): Default long double to 64-bit for
1315 32-bit Bionic and to 128-bit for 64-bit Bionic.
1317 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1318 TARGET_LONG_DOUBLE_128 is true.
1319 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1321 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1322 (mlong-double-64): Negate -mlong-double-128.
1323 (mlong-double-128): New option.
1325 * config/i386/i386-c.c (ix86_target_macros): Define
1326 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1328 * doc/invoke.texi: Document -mlong-double-128.
1330 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1332 PR rtl-optimization/60024
1333 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1335 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1337 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1339 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1341 PR rtl-optimization/57662
1342 * sel-sched.c (code_motion_path_driver): Do not mark already not
1343 existing blocks in the visiting bitmap.
1345 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1347 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1348 on the insn being emitted.
1350 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1351 Will Deacon <will.deacon@arm.com>
1353 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1355 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1357 * config/arm/arm-tables.opt: Regenerate.
1359 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1361 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1362 for vector types other than V16QImode.
1363 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1364 define_expand, and call altivec_expand_vec_perm_le when producing
1365 code with little endian element order.
1366 (*altivec_vperm_<mode>_internal): New insn having previous
1367 behavior of altivec_vperm_<mode>.
1368 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1369 altivec_expand_vec_perm_le when producing code with little endian
1371 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1372 behavior of altivec_vperm_<mode>_uns.
1374 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1376 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1377 (altivec_vsumsws): Add handling for -maltivec=be with a little
1379 (altivec_vsumsws_direct): New.
1380 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1381 gen_altivec_vsumsws.
1383 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1385 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1386 vtable_pointer_value_to_binfo): New functions.
1387 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1388 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1390 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1392 * config/nios2/nios2.md (load_got_register): Initialize GOT
1393 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1394 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1396 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1398 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1399 preserverd by passthrough, do not propagate the type.
1401 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1403 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1404 (mips_atomic_assign_expand_fenv): New function.
1405 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1407 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1409 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1410 (__builtin_mips_set_fcsr): Likewise.
1411 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1412 MIPS_USI_FTYPE_VOID.
1413 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1414 (mips16_expand_set_fcsr): Likewise.
1415 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1416 (mips16_set_fcsr_stub): Likewise.
1417 (mips16_get_fcsr_one_only_stub): New class.
1418 (mips16_set_fcsr_one_only_stub): Likewise.
1419 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1420 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1421 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1422 (hard_float): New availability predicate.
1423 (mips_builtins): Add get_fcsr and set_fcsr.
1424 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1425 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1426 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1427 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1428 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1431 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1433 * config/mips/mips.c (mips_one_only_stub): New class.
1434 (mips_need_mips16_rdhwr_p): Replace with...
1435 (mips16_rdhwr_stub): ...this new variable.
1436 (mips16_stub_call_address): New function.
1437 (mips16_rdhwr_one_only_stub): New class.
1438 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1439 (mips_output_mips16_rdhwr): Delete.
1440 (mips_finish_stub): New function.
1441 (mips_code_end): Use it to handle rdhwr stubs.
1443 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1446 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1447 when calculating size of integer atomic types.
1449 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1451 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1453 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1455 PR tree-optimization/60003
1456 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1457 * profile.c (branch_prob): Use gimple_call_builtin_p
1458 to check for BUILT_IN_SETJMP_RECEIVER.
1459 * tree-inline.c (copy_bb): Call notice_special_calls.
1461 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1464 * lra-constraints.c (process_alt_operands): Update reload_sum only
1467 2014-01-31 Richard Henderson <rth@redhat.com>
1470 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1471 until after else_eh is processed.
1473 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1475 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1476 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1477 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1478 in smmintrin.h, remove them.
1479 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1480 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1481 * config/i386/i386.md (ROUND_SAE): Fix value.
1482 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1483 (const48_operand): New.
1484 * config/i386/subst.md (round), (round_expand): Use
1485 const_4_or_8_to_11_operand.
1486 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1488 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1490 * config/i386/constraints.md (Yk): Swap meaning with k.
1491 * config/i386/i386.md (movhi_internal): Change Yk to k.
1492 (movqi_internal): Ditto.
1493 (*k<logic><mode>): Ditto.
1496 (kandn<mode>): Ditto.
1497 (*<code>hi_1): Ditto.
1498 (*<code>qi_1): Ditto.
1499 (kxnor<mode>): Ditto.
1500 (kortestzhi): Ditto.
1501 (kortestchi): Ditto.
1503 (*one_cmplhi2_1): Ditto.
1504 (*one_cmplqi2_1): Ditto.
1505 * config/i386/sse.md (): Change k to Yk.
1506 (avx512f_load<mode>_mask): Ditto.
1507 (avx512f_blendm<mode>): Ditto.
1508 (avx512f_store<mode>_mask): Ditto.
1509 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1510 (avx512f_storedqu<mode>_mask): Ditto.
1511 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1513 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1514 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1515 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1516 (avx512f_maskcmp<mode>3): Ditto.
1517 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1518 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1519 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1520 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1521 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1522 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1523 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1524 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1525 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1526 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1527 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1528 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1529 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1530 (vec_extract_lo_<mode>_maskm): Ditto.
1531 (vec_extract_hi_<mode>_maskm): Ditto.
1532 (avx512f_vternlog<mode>_mask): Ditto.
1533 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1534 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1535 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1536 (avx512f_<code>v8div16qi2_mask): Ditto.
1537 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1538 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1539 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1540 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1541 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1542 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1543 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1544 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1545 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1546 (avx512cd_maskb_vec_dupv8di): Ditto.
1547 (avx512cd_maskw_vec_dupv16si): Ditto.
1548 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1549 (avx512f_vpermi2var<mode>3_mask): Ditto.
1550 (avx512f_vpermi2var<mode>3_mask): Ditto.
1551 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1552 (*avx512f_gathersi<mode>): Ditto.
1553 (*avx512f_gathersi<mode>_2): Ditto.
1554 (*avx512f_gatherdi<mode>): Ditto.
1555 (*avx512f_gatherdi<mode>_2): Ditto.
1556 (*avx512f_scattersi<mode>): Ditto.
1557 (*avx512f_scatterdi<mode>): Ditto.
1558 (avx512f_compress<mode>_mask): Ditto.
1559 (avx512f_compressstore<mode>_mask): Ditto.
1560 (avx512f_expand<mode>_mask): Ditto.
1561 * config/i386/subst.md (mask): Change k to Yk.
1562 (mask_scalar_merge): Ditto.
1565 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1567 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1569 2014-01-31 Richard Biener <rguenther@suse.de>
1572 * builtins.c (fold_builtin_memory_op): Make sure to not
1573 use a floating-point mode or a boolean or enumeral type for
1576 2014-01-30 DJ Delorie <dj@redhat.com>
1578 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1579 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1580 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1581 whenever main() has an epilogue.
1583 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1585 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1586 unused variable "field".
1587 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1588 (vsx_mergeh_<mode>): Likewise.
1589 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1590 (altivec_vmrghh): Likewise.
1591 (altivec_vmrghw): Likewise.
1592 (altivec_vmrglb): Likewise.
1593 (altivec_vmrglh): Likewise.
1594 (altivec_vmrglw): Likewise.
1595 (altivec_vspltb): Add missing uses.
1596 (altivec_vsplth): Likewise.
1597 (altivec_vspltw): Likewise.
1598 (altivec_vspltsf): Likewise.
1600 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1603 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1604 frame related instructions.
1606 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1608 PR rtl-optimization/59959
1609 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1610 any reload of register whose subreg is invalid.
1612 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1614 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1615 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1616 Add missing return type - void.
1618 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1620 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1621 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1622 remove element index adjustment for endian (now handled in vsx.md
1624 (altivec_expand_vec_perm_const): Use
1625 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1626 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1627 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1628 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1629 define_expand and a new define_insn *altivec_vspltb_internal;
1630 adjust for -maltivec=be on a little endian target.
1631 (altivec_vspltb_direct): New.
1632 (altivec_vsplth): Divide into a define_expand and a new
1633 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1634 little endian target.
1635 (altivec_vsplth_direct): New.
1636 (altivec_vspltw): Divide into a define_expand and a new
1637 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1638 little endian target.
1639 (altivec_vspltw_direct): New.
1640 (altivec_vspltsf): Divide into a define_expand and a new
1641 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1642 a little endian target.
1644 2014-01-30 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/59993
1647 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1648 can propagate form the earlier stmt and avoid the transform
1649 when the intermediate result is needed.
1651 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1653 * README.Portability: Fix typo.
1655 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1657 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1658 comparison_operator with ordered_comparison_operator.
1660 2014-01-30 Nick Clifton <nickc@redhat.com>
1662 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1663 Rename to mn10300_store_multiple_regs.
1664 * config/mn10300/mn10300.c: Likewise.
1665 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1666 store_multiple_regs.
1667 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1668 Call mn10300_store_multiple_regs.
1670 2014-01-30 Nick Clifton <nickc@redhat.com>
1671 DJ Delorie <dj@redhat.com>
1673 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1674 %fp 2 to keep registers after it properly word-aligned.
1675 (rl78_alloc_physical_registers_umul): Handle the case where both
1676 input operands are the same.
1678 2014-01-30 Richard Biener <rguenther@suse.de>
1680 PR tree-optimization/59903
1681 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1684 2014-01-30 Jason Merrill <jason@redhat.com>
1687 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1690 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1692 2014-01-30 Richard Biener <rguenther@suse.de>
1694 PR tree-optimization/59951
1695 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1697 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1700 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1701 SFmode to DFmode case.
1703 2014-01-29 DJ Delorie <dj@redhat.com>
1705 * config/msp430/msp430.opt (-minrt): New.
1706 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1708 (ENDFILE_SPEC): Likewise.
1710 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1712 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1713 (estimate_function_body_sizes): Use it.
1715 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1718 * dwarf2out.c (is_cxx_auto): New.
1719 (is_base_type): Use it.
1720 (gen_type_die_with_usage): Likewise.
1722 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1724 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1725 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1726 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1727 -maltivec=be with LE targets.
1728 (vsx_mergeh_<mode>): Likewise.
1729 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1730 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1731 (altivec_vmrghb): Replace with define_expand and new
1732 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1733 (altivec_vmrghb_direct): New define_insn.
1734 (altivec_vmrghh): Replace with define_expand and new
1735 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1736 (altivec_vmrghh_direct): New define_insn.
1737 (altivec_vmrghw): Replace with define_expand and new
1738 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1739 (altivec_vmrghw_direct): New define_insn.
1740 (*altivec_vmrghsf): Adjust for endianness.
1741 (altivec_vmrglb): Replace with define_expand and new
1742 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1743 (altivec_vmrglb_direct): New define_insn.
1744 (altivec_vmrglh): Replace with define_expand and new
1745 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1746 (altivec_vmrglh_direct): New define_insn.
1747 (altivec_vmrglw): Replace with define_expand and new
1748 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1749 (altivec_vmrglw_direct): New define_insn.
1750 (*altivec_vmrglsf): Adjust for endianness.
1751 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1752 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1753 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1754 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1755 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1756 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1757 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1758 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1760 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1762 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1763 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1766 2014-01-29 Richard Biener <rguenther@suse.de>
1768 PR tree-optimization/58742
1769 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1770 associate_pointerplus_align.
1771 (associate_pointerplus_diff): New function.
1772 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1773 and associate_pointerplus_diff.
1775 2014-01-29 Richard Biener <rguenther@suse.de>
1777 * lto-streamer.h (LTO_major_version): Bump to 3.
1778 (LTO_minor_version): Reset to 0.
1780 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1782 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1783 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1784 (arm_file_start): Generate correct asm header for armv7ve.
1785 * config/arm/bpabi.h: Add multilib support for armv7ve.
1786 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1787 and cortex-a15 to armv7ve.
1788 * config/arm/t-aprofile: Add multilib support for armv7ve.
1789 * doc/invoke.texi: Document -march=armv7ve.
1791 2014-01-29 Richard Biener <rguenther@suse.de>
1793 PR tree-optimization/58742
1794 * tree-ssa-forwprop.c (associate_plusminus): Return true
1795 if we changed sth, defer EH cleanup to ...
1796 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1797 (simplify_mult): New function.
1799 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1802 PR tree-optimization/59920
1803 * tree.c (build_common_builtin_nodes): Remove
1804 __builtin_setjmp_dispatcher initialization.
1805 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1806 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1807 instead of gsi_after_labels + manually skipping debug stmts.
1808 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1809 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1810 * tree-inline.c (copy_edges_for_bb): Remove
1811 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1812 argument. Ignore computed_goto_p stmts. Don't call
1813 make_abnormal_goto_edges. If a call might need abnormal edges
1814 for non-local gotos, see if it already has an edge to
1815 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1816 with true argument, don't do anything then, otherwise add
1817 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1818 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1820 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1821 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1822 (lower_stmt): Don't set data->calls_builtin_setjmp.
1823 (lower_builtin_setjmp): Adjust comment.
1824 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1825 * tree-cfg.c (found_computed_goto): Remove.
1826 (factor_computed_gotos): Remove.
1827 (make_goto_expr_edges): Return bool, true for computed gotos.
1828 Don't call make_abnormal_goto_edges.
1829 (build_gimple_cfg): Don't set found_computed_goto, don't call
1830 factor_computed_gotos.
1831 (computed_goto_p): No longer static.
1832 (make_blocks): Don't set found_computed_goto.
1833 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1834 (make_edges): If make_goto_expr_edges returns true, push bb
1835 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1836 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1837 vector. Record mapping between bbs and OpenMP regions if there
1838 are any, adjust make_gimple_omp_edges caller. Call
1839 handle_abnormal_edges.
1840 (make_abnormal_goto_edges): Remove.
1841 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1842 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1843 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1844 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1845 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1846 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1847 filling *region also set *region_idx to (*region)->entry->index.
1850 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1851 For REGs set ORIGINAL_REGNO.
1853 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1855 * doc/md.texi: Mention that a target shouldn't implement
1856 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1859 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1861 PR tree-optimization/59594
1862 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1863 a copy of the datarefs vector rather than the vector itself.
1865 2014-01-28 Jason Merrill <jason@redhat.com>
1868 * dwarf2out.c (auto_die): New static.
1869 (gen_type_die_with_usage): Handle C++1y 'auto'.
1870 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1873 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1876 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1877 (SPEC_X32): Likewise.
1878 (SPEC_64): Likewise.
1879 * config/i386/i386.c (ix86_option_override_internal): Turn off
1880 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1882 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1883 * config/i386/i386.h (TARGET_16BIT): New macro.
1884 (TARGET_16BIT_P): Likewise.
1885 * config/i386/i386.opt: Add m16.
1886 * doc/invoke.texi: Document -m16.
1888 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1890 PR preprocessor/59935
1891 * input.c (location_get_source_line): Bail out on when line number
1892 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1894 2014-01-28 Richard Biener <rguenther@suse.de>
1896 PR tree-optimization/58742
1897 * tree-ssa-forwprop.c (associate_plusminus): Handle
1898 pointer subtraction of the form (T)(P + A) - (T)P.
1900 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1902 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1905 2014-01-28 Richard Biener <rguenther@suse.de>
1908 2014-01-28 Richard Biener <rguenther@suse.de>
1910 PR rtl-optimization/45364
1911 PR rtl-optimization/59890
1912 * var-tracking.c (local_get_addr_clear_given_value): Handle
1913 already cleared slot.
1914 (val_reset): Handle not allocated local_get_addr_cache.
1915 (vt_find_locations): Use post-order on the inverted CFG.
1917 2014-01-28 Richard Biener <rguenther@suse.de>
1919 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1921 2014-01-28 Richard Biener <rguenther@suse.de>
1923 PR rtl-optimization/45364
1924 PR rtl-optimization/59890
1925 * var-tracking.c (local_get_addr_clear_given_value): Handle
1926 already cleared slot.
1927 (val_reset): Handle not allocated local_get_addr_cache.
1928 (vt_find_locations): Use post-order on the inverted CFG.
1930 2014-01-28 Alan Modra <amodra@gmail.com>
1932 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1933 * configure.ac <recursive call for build != host>: Define
1934 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1935 and LD_FOR_BUILD too.
1936 * configure: Regenerate.
1938 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1940 * config/i386/i386.c (get_builtin_code_for_version): Separate
1941 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1942 Broadwell from Haswell.
1944 2014-01-27 Steve Ellcey <sellcey@mips.com>
1946 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1947 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1948 * config/mips/mips.c (mips_option_override): Change setting
1950 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1951 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1952 Change from Mask to Var.
1954 2014-01-27 Jeff Law <law@redhat.com>
1956 * ipa-inline.c (inline_small_functions): Fix typo.
1958 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1960 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1961 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1962 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1963 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1964 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1965 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1966 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1967 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1968 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1969 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1970 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1971 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1972 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1973 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1974 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1975 (_mm512_storeu_epi64): Ditto.
1976 (_mm512_cmpge_epi32_mask): Ditto.
1977 (_mm512_cmpge_epu32_mask): Ditto.
1978 (_mm512_cmpge_epi64_mask): Ditto.
1979 (_mm512_cmpge_epu64_mask): Ditto.
1980 (_mm512_cmple_epi32_mask): Ditto.
1981 (_mm512_cmple_epu32_mask): Ditto.
1982 (_mm512_cmple_epi64_mask): Ditto.
1983 (_mm512_cmple_epu64_mask): Ditto.
1984 (_mm512_cmplt_epi32_mask): Ditto.
1985 (_mm512_cmplt_epu32_mask): Ditto.
1986 (_mm512_cmplt_epi64_mask): Ditto.
1987 (_mm512_cmplt_epu64_mask): Ditto.
1988 (_mm512_cmpneq_epi32_mask): Ditto.
1989 (_mm512_cmpneq_epu32_mask): Ditto.
1990 (_mm512_cmpneq_epi64_mask): Ditto.
1991 (_mm512_cmpneq_epu64_mask): Ditto.
1992 (_mm512_expand_pd): Ditto.
1993 (_mm512_expand_ps): Ditto.
1994 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1995 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1996 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1997 * config/i386/i386.c (ix86_builtins): Add
1998 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1999 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2000 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2001 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2002 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2003 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2004 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2005 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2006 IX86_BUILTIN_PMOVUSQW512_MEM.
2007 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2008 __builtin_ia32_pmovsqd512mem_mask,
2009 __builtin_ia32_pmovqd512mem_mask,
2010 __builtin_ia32_pmovusqw512mem_mask,
2011 __builtin_ia32_pmovsqw512mem_mask,
2012 __builtin_ia32_pmovqw512mem_mask,
2013 __builtin_ia32_pmovusdw512mem_mask,
2014 __builtin_ia32_pmovsdw512mem_mask,
2015 __builtin_ia32_pmovdw512mem_mask,
2016 __builtin_ia32_pmovqb512mem_mask,
2017 __builtin_ia32_pmovusqb512mem_mask,
2018 __builtin_ia32_pmovsqb512mem_mask,
2019 __builtin_ia32_pmovusdb512mem_mask,
2020 __builtin_ia32_pmovsdb512mem_mask,
2021 __builtin_ia32_pmovdb512mem_mask.
2022 (bdesc_args): Add __builtin_ia32_expanddf512,
2023 __builtin_ia32_expandsf512.
2024 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2025 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2026 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2027 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2028 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2029 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2030 (avx512f_<code>v8div16qi2_mask_store): This.
2031 (avx512f_expand<mode>): New.
2033 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2035 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2037 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2038 (_mm512_prefetch_i32scatter_pd): Ditto.
2039 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2040 (_mm512_prefetch_i64scatter_pd): Ditto.
2041 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2042 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2043 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2044 (_mm512_prefetch_i32scatter_ps): Ditto.
2045 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2046 (_mm512_prefetch_i64scatter_ps): Ditto.
2047 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2048 * config/i386/i386-builtin-types.def: Define
2049 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2050 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2051 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2052 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2053 IX86_BUILTIN_SCATTERPFQPD.
2054 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2055 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2056 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2057 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2058 __builtin_ia32_scatterpfqps.
2059 (ix86_expand_builtin): Expand new built-ins.
2060 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2061 fix memory access data type.
2062 (*avx512pf_gatherpf<mode>_mask): Ditto.
2063 (*avx512pf_gatherpf<mode>): Ditto.
2064 (avx512pf_scatterpf<mode>): Ditto.
2065 (*avx512pf_scatterpf<mode>_mask): Ditto.
2066 (*avx512pf_scatterpf<mode>): Ditto.
2067 (GATHER_SCATTER_SF_MEM_MODE): New.
2068 (avx512pf_gatherpf<mode>df): Ditto.
2069 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2070 (*avx512pf_scatterpf<mode>df): Ditto.
2072 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2075 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2076 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2079 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2081 * common/config/arm/arm-common.c
2082 (arm_rewrite_mcpu): Handle multiple names.
2084 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2086 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2088 * gimple-builder.h (create_gimple_tmp): Delete.
2090 2014-01-27 Christian Bruel <christian.bruel@st.com>
2092 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2095 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2097 * config/pa/pa.md (call): Generate indirect long calls to non-local
2098 functions when outputing 32-bit code.
2099 (call_value): Likewise except for special call to buggy powf function.
2101 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2102 portable runtime and PIC indirect calls.
2103 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2104 and PIC call sequences. Use ldo instead of blr to set return register
2105 in PIC call sequence.
2107 2014-01-25 Walter Lee <walt@tilera.com>
2109 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2110 avoid clobbering a live register.
2112 2014-01-25 Walter Lee <walt@tilera.com>
2114 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2115 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2116 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2117 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2119 2014-01-25 Walter Lee <walt@tilera.com>
2121 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2122 arguments on even registers.
2123 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2125 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2126 (BIGGEST_ALIGNMENT): Ditto.
2127 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2129 2014-01-25 Walter Lee <walt@tilera.com>
2131 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2132 insns before bundling.
2133 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2135 2014-01-25 Walter Lee <walt@tilera.com>
2137 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2138 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2139 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2141 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2143 * config/mips/constraints.md (kl): Delete.
2144 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2145 define expands, using...
2146 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2147 instructions for MIPS16.
2148 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2149 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2151 2014-01-25 Walter Lee <walt@tilera.com>
2153 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2157 2014-01-25 Walter Lee <walt@tilera.com>
2159 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2160 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2162 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2164 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2167 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2169 * print-rtl.c (in_call_function_usage): New var.
2170 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2171 EXPR_LIST mode as mode and not as reg note name.
2174 * cfgloopmanip.c (copy_loop_info): If
2175 loop->warned_aggressive_loop_optimizations, make sure
2176 the flag is set in target loop too.
2178 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2180 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2182 * builtins.def: Likewise.
2183 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2184 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2185 * ira.c (ira_setup_eliminable_regset): Likewise.
2186 * omp-low.c (gate_expand_omp): Likewise.
2187 (execute_lower_omp): Likewise.
2188 (diagnose_sb_0): Likewise.
2189 (gate_diagnose_omp_blocks): Likewise.
2190 (simd_clone_clauses_extract): Likewise.
2193 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2195 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2196 correction for little endian...
2197 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2200 2014-01-24 Jeff Law <law@redhat.com>
2202 PR tree-optimization/59919
2203 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2204 for non-returning calls.
2206 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2208 * common/config/aarch64/aarch64-common.c
2209 (aarch64_rewrite_mcpu): Handle multiple names.
2210 * config/aarch64/aarch64.h
2211 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2213 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2215 * input.c (add_file_to_cache_tab): Handle the case where fopen
2218 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2221 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2222 from push operand if code of push isn't PRE_DEC.
2224 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2227 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2228 -mquad-memory-atomic. Update -mquad-memory documentation to say
2229 it is only used for non-atomic loads/stores.
2231 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2232 -mquad-memory or -mquad-memory-atomic switches.
2234 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2235 -mquad-memory-atomic to ISA 2.07 support.
2237 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2238 to separate support of normal quad word memory operations (ldq, stq)
2239 from the atomic quad word memory operations.
2241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2242 support to separate non-atomic quad word operations from atomic
2243 quad word operations. Disable non-atomic quad word operations in
2244 little endian mode so that we don't have to swap words after the
2245 load and before the store.
2246 (quad_load_store_p): Add comment about atomic quad word support.
2247 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2248 options printed with -mdebug=reg.
2250 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2251 -mquad-memory-atomic as the test for whether we have quad word
2252 atomic instructions.
2253 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2254 or -mp8-vector are used, allow byte/half-word atomic operations.
2256 * config/rs6000/sync.md (load_lockedti): Insure that the address
2257 is a proper indexed or indirect address for the lqarx instruction.
2258 On little endian systems, swap the hi/lo registers after the lqarx
2260 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2261 insure the address is valid for the lqarx instruction.
2262 (store_conditionalti): Insure that the address is a proper indexed
2263 or indirect address for the stqcrx. instruction. On little endian
2264 systems, swap the hi/lo registers before doing the stqcrx.
2266 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2267 insure the address is valid for the stqcrx. instruction.
2269 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2270 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2271 type of quad memory support is available.
2273 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2276 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2277 there is a danger of looping.
2279 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2282 force flag_ira_loop_pressure if set via command line.
2284 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2286 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2287 (ashr_simd): New builtin handling DI mode.
2288 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2289 (aarch64_sshr_simddi): New match pattern.
2290 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2291 (vshrd_n_s64): Likewise.
2292 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2294 2014-01-23 Nick Clifton <nickc@redhat.com>
2296 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2297 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2298 favour of mcu specific scripts.
2299 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2302 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2303 Alex Velenko <Alex.Velenko@arm.com>
2305 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2306 (vaddv_s16): Likewise.
2307 (vaddv_s32): Likewise.
2308 (vaddv_u8): Likewise.
2309 (vaddv_u16): Likewise.
2310 (vaddv_u32): Likewise.
2311 (vaddvq_s8): Likewise.
2312 (vaddvq_s16): Likewise.
2313 (vaddvq_s32): Likewise.
2314 (vaddvq_s64): Likewise.
2315 (vaddvq_u8): Likewise.
2316 (vaddvq_u16): Likewise.
2317 (vaddvq_u32): Likewise.
2318 (vaddvq_u64): Likewise.
2319 (vaddv_f32): Likewise.
2320 (vaddvq_f32): Likewise.
2321 (vaddvq_f64): Likewise.
2322 (vmaxv_f32): Likewise.
2323 (vmaxv_s8): Likewise.
2324 (vmaxv_s16): Likewise.
2325 (vmaxv_s32): Likewise.
2326 (vmaxv_u8): Likewise.
2327 (vmaxv_u16): Likewise.
2328 (vmaxv_u32): Likewise.
2329 (vmaxvq_f32): Likewise.
2330 (vmaxvq_f64): Likewise.
2331 (vmaxvq_s8): Likewise.
2332 (vmaxvq_s16): Likewise.
2333 (vmaxvq_s32): Likewise.
2334 (vmaxvq_u8): Likewise.
2335 (vmaxvq_u16): Likewise.
2336 (vmaxvq_u32): Likewise.
2337 (vmaxnmv_f32): Likewise.
2338 (vmaxnmvq_f32): Likewise.
2339 (vmaxnmvq_f64): Likewise.
2340 (vminv_f32): Likewise.
2341 (vminv_s8): Likewise.
2342 (vminv_s16): Likewise.
2343 (vminv_s32): Likewise.
2344 (vminv_u8): Likewise.
2345 (vminv_u16): Likewise.
2346 (vminv_u32): Likewise.
2347 (vminvq_f32): Likewise.
2348 (vminvq_f64): Likewise.
2349 (vminvq_s8): Likewise.
2350 (vminvq_s16): Likewise.
2351 (vminvq_s32): Likewise.
2352 (vminvq_u8): Likewise.
2353 (vminvq_u16): Likewise.
2354 (vminvq_u32): Likewise.
2355 (vminnmv_f32): Likewise.
2356 (vminnmvq_f32): Likewise.
2357 (vminnmvq_f64): Likewise.
2359 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2361 * config/aarch64/aarch64-simd.md
2362 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2363 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2364 (*aarch64_mul3_elt<mode>): Likewise.
2365 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2366 (*aarch64_mul3_elt_to_64v2df): Likewise.
2367 (*aarch64_mla_elt<mode>): Likewise.
2368 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2369 (*aarch64_mls_elt<mode>): Likewise.
2370 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2371 (*aarch64_fma4_elt<mode>): Likewise.
2372 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2373 (*aarch64_fma4_elt_to_64v2df): Likewise.
2374 (*aarch64_fnma4_elt<mode>): Likewise.
2375 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2376 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2377 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2378 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2379 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2380 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2381 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2382 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2383 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2385 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2387 * config/aarch64/aarch64-simd.md
2388 (aarch64_be_checked_get_lane<mode>): New define_expand.
2389 * config/aarch64/aarch64-simd-builtins.def
2390 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2391 New builtin definition.
2392 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2393 Use new safe be builtin.
2395 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2397 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2399 (aarch64_be_st1<mode>): Likewise.
2400 (aarch_ld1<VALL:mode>): Define_expand modified.
2401 (aarch_st1<VALL:mode>): Likewise.
2402 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2403 (UNSPEC_ST1): Likewise.
2405 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2407 * config/microblaze/microblaze.md: Add trap insn and attribute
2409 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2411 PR preprocessor/58580
2412 * input.h (location_get_source_line): Take an additional line_size
2414 (void diagnostics_file_cache_fini): Declare new function.
2415 * input.c (struct fcache): New type.
2416 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2417 New static constants.
2418 (diagnostic_file_cache_init, total_lines_num)
2419 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2420 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2421 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2422 (get_next_line, read_next_line, goto_next_line, read_line_num):
2423 New static function definitions.
2424 (diagnostic_file_cache_fini): New function.
2425 (location_get_source_line): Take an additional output line_len
2426 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2428 * diagnostic.c (diagnostic_finish): Call
2429 diagnostic_file_cache_fini.
2430 (adjust_line): Take an additional input parameter for the length
2431 of the line, rather than calculating it with strlen.
2432 (diagnostic_show_locus): Adjust the use of
2433 location_get_source_line and adjust_line with respect to their new
2434 signature. While displaying a line now, do not stop at the first
2435 null byte. Rather, display the zero byte as a space and keep
2436 going until we reach the size of the line.
2437 * Makefile.in: Add vec.o to OBJS-libcommon
2439 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2440 Ilya Tocar <ilya.tocar@intel.com>
2442 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2443 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2444 (__builtin_ia32_kmov16): Ditto.
2445 * config/i386/i386.md (UNSPEC_KMOV): New.
2448 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2450 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2451 (_mm512_storeu_si512): Ditto.
2453 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2456 * rtl.h (get_referenced_operands): Declare.
2457 * recog.c (get_referenced_operands): New function.
2458 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2459 operands have been referenced when recording LO_SUM references.
2461 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2463 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2465 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2467 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2468 Enable for generic and recent AMD targets.
2470 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2472 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2473 ARG_SIZE note when adjustment was eliminated.
2475 2014-01-22 Jeff Law <law@redhat.com>
2477 PR tree-optimization/59597
2478 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2479 in file. Accept new argument REGISTERING and use it to modify
2480 dump output appropriately.
2481 (register_jump_thread): Corresponding changes.
2482 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2483 thread paths involving joiner blocks. Add code to dump cancelled
2484 jump threading paths.
2486 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2488 PR rtl-optimization/59477
2489 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2490 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2492 2014-01-22 Tom Tromey <tromey@redhat.com>
2494 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2496 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2498 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2500 PR rtl-optimization/59896
2501 * lra-constraints.c (process_alt_operands): Check unused note for
2502 matched operands of insn with no output reloads.
2504 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2506 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2507 (mips_move_from_gpr_cost): Likewise.
2509 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2511 PR rtl-optimization/59858
2512 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2513 ira_class_hard_regs_num.
2514 (process_alt_operands): Increase reject for dying matched operand.
2516 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2519 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2520 smaller than size, perform several stores or loads and stores
2521 at dst + count - size to store or copy all of size bytes, rather
2522 than just last modesize bytes.
2524 2014-01-20 DJ Delorie <dj@redhat.com>
2526 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2527 that CLOBBERs are REGs before propogating their values.
2529 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2532 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2533 (cgraph_inline_failed_type): New function.
2534 * cgraph.h (DEFCIFCODE): Add type.
2535 (cgraph_inline_failed_type_t): New enum.
2536 (cgraph_inline_failed_type): New prototype.
2537 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2538 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2539 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2540 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2541 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2542 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2543 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2544 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2545 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2546 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2547 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2548 OPTIMIZATION_MISMATCH.
2549 * tree-inline.c (expand_call_inline): Emit errors during
2550 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2552 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2555 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2556 mode attribute in insn output.
2558 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2560 * output.h (output_constant): Delete.
2561 * varasm.c (output_constant): Make private.
2563 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2565 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2567 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2570 * tree.h (fold_builtin_strcat): New prototype.
2571 * builtins.c (fold_builtin_strcat): No longer static. Add len
2572 argument, if non-NULL, don't call c_strlen. Optimize
2573 directly into __builtin_memcpy instead of __builtin_strcpy.
2574 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2575 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2577 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2579 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2580 for SImode_address_operand operands, having only a REG argument.
2582 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2584 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2585 loader name using mbig-endian.
2586 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2588 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2590 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2594 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2596 * config/aarch64/aarch64-protos.h
2597 (aarch64_cannot_change_mode_class_ptr): Declare.
2598 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2599 aarch64_cannot_change_mode_class_ptr): New.
2600 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2601 backend hook aarch64_cannot_change_mode_class.
2603 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2605 * common/config/aarch64/aarch64-common.c
2606 (aarch64_handle_option): Don't handle any option order logic here.
2607 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2608 selected_cpu, warn on architecture version mismatch.
2609 (aarch64_override_options): Fix parsing order for option strings.
2611 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2612 Iain Sandoe <iain@codesourcery.com>
2615 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2616 warning. Amend comment to reflect current functionality.
2618 2014-01-20 Richard Biener <rguenther@suse.de>
2621 * builtins.c (fold_builtin_strcat): Remove case better handled
2622 by tree-ssa-strlen.c.
2624 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2626 * config/aarch64/aarch64.opt
2627 (mcpu, march, mtune): Make case-insensitive.
2629 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2632 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2633 if operands[1] is a REG or ZERO_EXTEND of a REG.
2635 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2637 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2639 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2641 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2642 long non-pic millicode calls.
2644 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2646 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2648 2014-01-19 Kito Cheng <kito@0xlab.org>
2650 * builtins.c (expand_movstr): Check movstr expand done or fail.
2652 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2653 H.J. Lu <hongjiu.lu@intel.com>
2656 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2657 to DImode for zero-extended addresses.
2659 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2661 PR rtl-optimization/57763
2662 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2663 on the new indirect jump_insn and increment LABEL_NUSES (label).
2665 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2669 * config.gcc (x86_archs): New variable.
2670 (x86_64_archs): Likewise.
2671 (x86_cpus): Likewise.
2672 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2673 --with-arch/--with-cpu= options.
2674 Support --with-arch=/--with-cpu={nehalem,westmere,
2675 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2677 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2679 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2680 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2682 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2684 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2686 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2689 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2690 clear cpp_get_options (parse_in)->warn_unused_macros for
2691 ix86_target_macros_internal with cpp_define.
2693 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2695 * jump.c (delete_related_insns): Keep (use (insn))s.
2696 * reorg.c (redundant_insn): Check for barriers too.
2698 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2700 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2702 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2704 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2705 call to $$dyncall when TARGET_LONG_CALLS is true.
2707 2014-01-17 Jeff Law <law@redhat.com>
2709 * ree.c (combine_set_extension): Temporarily disable test for
2710 changing number of hard registers.
2712 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2715 * ipa-inline-analysis.c (inline_free_summary):
2716 Do not free summary of aliases.
2718 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2721 * gimplify.c (gimplify_expr): Use create_tmp_var
2722 instead of create_tmp_var_raw. If cond doesn't have
2723 integral type, don't add the IFN_ANNOTATE builtin at all.
2725 2014-01-17 Martin Jambor <mjambor@suse.cz>
2728 * ipa-cp.c (prev_edge_clone): New variable.
2729 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2730 Also resize prev_edge_clone vector.
2731 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2732 (ipcp_edge_removal_hook): New function.
2733 (ipcp_driver): Register ipcp_edge_removal_hook.
2735 2014-01-17 Andrew Pinski <apinski@cavium.com>
2736 Steve Ellcey <sellcey@mips.com>
2739 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2742 2014-01-17 Jeff Law <law@redhat.com>
2745 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2746 so that pass_ccp runs first.
2748 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2750 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2751 (ix86_adjust_cost): Use !TARGET_XXX.
2752 (do_reorder_for_imul): Likewise.
2753 (swap_top_of_ready_list): Likewise.
2754 (ix86_sched_reorder): Likewise.
2756 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2758 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2759 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2760 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2761 (intel_memset): New. Duplicate slm_memset.
2762 (intel_cost): New. Duplicate slm_cost.
2763 (m_INTEL): New macro.
2764 (processor_target_table): Add "intel".
2765 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2766 with PROCESSOR_INTEL for "intel".
2767 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2768 PROCESSOR_SILVERMONT.
2769 (ix86_issue_rate): Likewise.
2770 (ix86_adjust_cost): Likewise.
2771 (ia32_multipass_dfa_lookahead): Likewise.
2772 (swap_top_of_ready_list): Likewise.
2773 (ix86_sched_reorder): Likewise.
2774 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2775 instead of TARGET_OPT_AGU.
2776 * config/i386/i386.h (TARGET_INTEL): New.
2777 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2778 (processor_type): Add PROCESSOR_INTEL.
2779 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2780 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2782 2014-01-17 Marek Polacek <polacek@redhat.com>
2785 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2788 2014-01-17 Richard Biener <rguenther@suse.de>
2790 PR tree-optimization/46590
2791 * opts.c (default_options_table): Add entries for
2792 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2793 all enabled at -O1 but not for -Og.
2794 * common.opt (fbranch-count-reg): Remove Init(1).
2795 (fmove-loop-invariants): Likewise.
2796 (ftree-pta): Likewise.
2798 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2800 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2801 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2802 decls to at least the GCC 4.8 used alignments.
2805 * tree-nested.c (convert_nonlocal_reference_stmt,
2806 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2807 of GIMPLE_BIND stmts, adjust associated decls.
2809 2014-01-17 Richard Biener <rguenther@suse.de>
2811 PR tree-optimization/46590
2812 * vec.h (vec<>::bseach): New member function implementing
2813 binary search according to C89 bsearch.
2814 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2815 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2816 bitmap pointer again. Make accesses_in_loop a flat array.
2817 (mem_ref_obstack): New global.
2818 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2819 (mark_ref_stored): Likewise.
2820 (ref_indep_loop_p_2): Likewise.
2821 (set_ref_stored_in_loop): New helper function.
2822 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2823 (memref_free): Adjust.
2824 (record_mem_ref_loc): Simplify.
2825 (gather_mem_refs_stmt): Adjust.
2826 (sort_locs_in_loop_postorder_cmp): New function.
2827 (analyze_memory_references): Sort accesses_in_loop after
2828 loop postorder number.
2829 (find_ref_loc_in_loop_cmp): New function.
2830 (for_all_locs_in_loop): Find relevant cluster of locs in
2831 accesses_in_loop and iterate without recursion.
2832 (execute_sm): Avoid uninit warning.
2833 (struct ref_always_accessed): Simplify.
2834 (ref_always_accessed::operator ()): Likewise.
2835 (ref_always_accessed_p): Likewise.
2836 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2837 loop postorder numbers here.
2838 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2841 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2844 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2845 on decls for which assemble_alias has been called.
2847 2014-01-17 Nick Clifton <nickc@redhat.com>
2849 * config/msp430/msp430.opt: (mcpu): New option.
2850 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2851 (msp430_option_override): Parse target_cpu. If the MCU name
2852 matches a generic string, clear target_mcu.
2853 (msp430_attr): Allow numeric interrupt values up to 63.
2854 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2855 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2857 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2859 * config/msp430/msp430.md (popm): Use %J rather than %I.
2860 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2861 (addhi_cy_i): Use immediate_operand for operand 2.
2862 * doc/invoke.texi: Document -mcpu option.
2864 2014-01-17 Richard Biener <rguenther@suse.de>
2866 PR rtl-optimization/38518
2867 * df.h (df_analyze_loop): Declare.
2868 * df-core.c: Include cfgloop.h.
2869 (df_analyze_1): Split out main part of df_analyze.
2870 (df_analyze): Adjust.
2871 (loop_inverted_post_order_compute): New function.
2872 (loop_post_order_compute): Likewise.
2873 (df_analyze_loop): New function avoiding whole-function
2875 * loop-invariant.c (find_defs): Use df_analyze_loop.
2876 (find_invariants): Adjust.
2877 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2879 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2881 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2882 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2884 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2886 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2887 traversal when removing references.
2889 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2892 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2894 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2897 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2898 pushing a reload for an autoinc when we had previously reloaded an
2899 inner part of the address.
2901 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2903 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2905 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2906 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2907 when not giving up or versioning for alias only because of
2909 (vect_analyze_data_ref_dependences): Set to true.
2910 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2912 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2913 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2917 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2920 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2921 operand 0 predicate for gathers, use a new pseudo as subtarget.
2923 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2926 * lra-constraints.c (process_alt_operands): Add printing debug info.
2927 Check absence of input/output reloads for matched operands too.
2929 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2931 PR rtl-optimization/59835
2932 * ira.c (ira_init_register_move_cost): Increase cost for
2935 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2937 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2939 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2942 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2943 non-register objects. Use gen_(high/low)part more consistently.
2946 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2949 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2950 endian support, remove tests for WORDS_BIG_ENDIAN.
2951 (p8_mfvsrd_3_<mode>): Likewise.
2952 (reload_gpr_from_vsx<mode>): Likewise.
2953 (reload_gpr_from_vsxsf): Likewise.
2954 (p8_mfvsrd_4_disf): Likewise.
2956 2014-01-16 Richard Biener <rguenther@suse.de>
2958 PR rtl-optimization/46590
2959 * lcm.c (compute_antinout_edge): Use postorder iteration.
2960 (compute_laterin): Use inverted postorder iteration.
2962 2014-01-16 Nick Clifton <nickc@redhat.com>
2965 * varasm.c (output_constant): Return the number of bytes actually
2967 (output_constructor_array_range): Update the field size with the
2968 number of bytes emitted by output_constant.
2969 (output_constructor_regular_field): Likewise. Also do not
2970 complain if the total number of bytes emitted is now greater
2971 than the expected fieldpos.
2972 * output.h (output_constant): Update prototype and descriptive comment.
2974 2014-01-16 Marek Polacek <polacek@redhat.com>
2977 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2978 it is error_mark_node.
2980 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2982 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2983 VALID_AVX256_REG_OR_OI_MODE.
2985 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2987 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2988 current procedure should be profiled.
2990 2014-01-15 Andrew Pinski <apinski@cavium.com>
2992 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2993 of moving from/to the STACK_REG register class.
2995 2014-01-15 Richard Henderson <rth@redhat.com>
2998 * reginfo.c (global_regs_decl): Globalize.
2999 * rtl.h (global_regs_decl): Declare.
3000 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3001 reserved via global_regs.
3003 2014-01-15 Teresa Johnson <tejohnson@google.com>
3005 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3007 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3009 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3010 and vmulosh rather than call gen_vec_widen_smult_*.
3011 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3012 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3013 (vec_widen_smult_even_v16qi): Likewise.
3014 (vec_widen_umult_even_v8hi): Likewise.
3015 (vec_widen_smult_even_v8hi): Likewise.
3016 (vec_widen_umult_odd_v16qi): Likewise.
3017 (vec_widen_smult_odd_v16qi): Likewise.
3018 (vec_widen_umult_odd_v8hi): Likewise.
3019 (vec_widen_smult_odd_v8hi): Likewise.
3020 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3021 vmuloub rather than call gen_vec_widen_umult_*.
3022 (vec_widen_umult_lo_v16qi): Likewise.
3023 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3024 vmulosb rather than call gen_vec_widen_smult_*.
3025 (vec_widen_smult_lo_v16qi): Likewise.
3026 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3027 rather than call gen_vec_widen_umult_*.
3028 (vec_widen_umult_lo_v8hi): Likewise.
3029 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3030 rather than call gen_vec_widen_smult_*.
3031 (vec_widen_smult_lo_v8hi): Likewise.
3033 2014-01-15 Jeff Law <law@redhat.com>
3035 PR tree-optimization/59747
3036 * ree.c (find_and_remove_re): Properly handle case where a second
3037 eliminated extension requires widening a copy created for elimination
3038 of a prior extension.
3039 (combine_set_extension): Ensure that the number of hard regs needed
3040 for a destination register does not change when we widen it.
3042 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3044 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3045 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3046 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3047 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3048 (avr-*-rtems*): Likewise.
3049 (bfin*-rtems*): Likewise.
3050 (moxie-*-rtems*): Likewise.
3051 (h8300-*-rtems*): Likewise.
3052 (i[34567]86-*-rtems*): Likewise.
3053 (lm32-*-rtems*): Likewise.
3054 (m32r-*-rtems*): Likewise.
3055 (m68k-*-rtems*): Likewise.
3056 (microblaze*-*-rtems*): Likewise.
3057 (mips*-*-rtems*): Likewise.
3058 (powerpc-*-rtems*): Likewise.
3059 (sh-*-rtems*): Likewise.
3060 (sparc-*-rtems*): Likewise.
3061 (sparc64-*-rtems*): Likewise.
3062 (v850-*-rtems*): Likewise.
3063 (m32c-*-rtems*): Likewise.
3065 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3067 PR rtl-optimization/59511
3068 * ira.c (ira_init_register_move_cost): Use memory costs for some
3069 cases of register move cost calculations.
3070 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3071 instead of BB frequency.
3072 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3073 * lra-assigns.c (find_hard_regno_for): Ditto.
3075 2014-01-15 Richard Biener <rguenther@suse.de>
3077 PR tree-optimization/59822
3078 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3079 (vectorizable_load): Use it to hoist defs of uses of invariant
3080 loads out of the loop.
3082 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3083 Kugan Vivekanandarajah <kuganv@linaro.org>
3086 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3089 2014-01-15 Richard Biener <rguenther@suse.de>
3091 PR rtl-optimization/59802
3092 * lcm.c (compute_available): Use inverted postorder to seed
3093 the initial worklist.
3095 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3098 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3099 ADDR_REGS for invalid symrefs in non-PIC code.
3101 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3104 * builtins.c (determine_block_size): Initialize *probable_max_size
3105 even if len_rtx is CONST_INT.
3107 2014-01-14 Andrew Pinski <apinski@cavium.com>
3109 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3110 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3111 (cortexa53_tunings): Likewise.
3112 (aarch64_sched_issue_rate): New function.
3113 (TARGET_SCHED_ISSUE_RATE): Define.
3115 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3117 * ira-costs.c (find_costs_and_classes): Add missed
3118 ira_init_register_move_cost_if_necessary.
3120 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3123 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3125 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3128 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3129 to indicate if type is used for function return value. Warn ABI
3130 change if the vector mode isn't available for function return value.
3131 (ix86_function_arg_advance): Pass false to type_natural_mode.
3132 (ix86_function_arg): Likewise.
3133 (ix86_gimplify_va_arg): Likewise.
3134 (function_arg_32): Don't warn ABI change.
3135 (ix86_function_value): Pass true to type_natural_mode.
3136 (ix86_return_in_memory): Likewise.
3137 (ix86_struct_value_rtx): Removed.
3138 (TARGET_STRUCT_VALUE_RTX): Likewise.
3140 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3142 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3143 converting a conditional jump into a conditional return.
3145 2014-01-14 Richard Biener <rguenther@suse.de>
3147 PR tree-optimization/58921
3148 PR tree-optimization/59006
3149 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3150 hoisting invariant stmts.
3151 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3152 invariant loads on the preheader edge if possible.
3154 2014-01-14 Joey Ye <joey.ye@arm.com>
3156 * doc/plugin.texi (Building GCC plugins): Update to C++.
3158 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3160 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3161 (_mm_rcp28_round_ss): Ditto.
3162 (_mm_rsqrt28_round_sd): Ditto.
3163 (_mm_rsqrt28_round_ss): Ditto.
3164 (_mm_rcp28_sd): Ditto.
3165 (_mm_rcp28_ss): Ditto.
3166 (_mm_rsqrt28_sd): Ditto.
3167 (_mm_rsqrt28_ss): Ditto.
3168 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3169 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3170 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3171 (IX86_BUILTIN_RCP28SD): Ditto.
3172 (IX86_BUILTIN_RCP28SS): Ditto.
3173 (IX86_BUILTIN_RSQRT28SD): Ditto.
3174 (IX86_BUILTIN_RSQRT28SS): Ditto.
3175 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3176 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3177 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3178 (ix86_expand_special_args_builtin): Expand new FTYPE.
3179 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3180 (srcp14<mode>): Make insn unary.
3181 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3182 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3183 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3184 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3185 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3186 Fix rounding: make it SAE only.
3187 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3189 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3191 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3192 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3193 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3194 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3195 (round_saeonly_mask_scalar_operand4): Ditto.
3196 (round_saeonly_mask_scalar_op3): Ditto.
3197 (round_saeonly_mask_scalar_op4): Ditto.
3199 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3201 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3202 Implement -maltivec=be for vec_insert and vec_extract.
3204 2014-01-10 DJ Delorie <dj@redhat.com>
3206 * config/msp430/msp430.md (call_internal): Don't allow memory
3207 references with SP as the base register.
3208 (call_value_internal): Likewise.
3209 * config/msp430/constraints.md (Yc): New. For memory references
3210 that don't use SP as a base register.
3212 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3213 "an integer without a # prefix"
3214 * config/msp430/msp430.md (epilogue_helper): Use it.
3216 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3219 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3220 AVX512F gather builtins.
3221 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3222 on gather decls with INTEGER_TYPE masktype.
3223 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3224 directly into the builtin rather than hoisting it before loop.
3226 PR tree-optimization/59387
3227 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3228 (scev_const_prop): If folded_casts and type has undefined overflow,
3229 use force_gimple_operand instead of force_gimple_operand_gsi and
3230 for each added stmt if it is assign with
3231 arith_code_with_undefined_signed_overflow, call
3232 rewrite_to_defined_overflow.
3233 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3234 gimple-fold.h instead.
3235 (arith_code_with_undefined_signed_overflow,
3236 rewrite_to_defined_overflow): Moved to ...
3237 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3238 rewrite_to_defined_overflow): ... here. No longer static.
3239 Include gimplify-me.h.
3240 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3241 rewrite_to_defined_overflow): New prototypes.
3243 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3245 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3247 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3249 * builtins.c (get_object_alignment_2): Minor tweak.
3250 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3252 2014-01-13 Christian Bruel <christian.bruel@st.com>
3254 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3255 optimized non constant lengths.
3257 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3260 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3261 load as __atomic_load_N if possible.
3263 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3265 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3267 (rs6000_expand_builtin): Adjust call.
3269 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3272 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3273 * config/rs6000/rs6000.c: Include target-globals.h.
3274 (rs6000_set_current_function): Instead of doing target_reinit
3275 unconditionally, use save_target_globals_default_opts and
3276 restore_target_globals.
3278 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3280 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3281 (rs6000_expand_builtin): Handle mffs and mtfsf.
3282 (rs6000_init_builtins): Define mffs and mtfsf.
3283 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3284 (rs6000_mffs): New pattern.
3285 (rs6000_mtfsf): New pattern.
3287 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3289 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3290 Start narrowing with START. Apply candidate-use pair
3291 and check overall cost in narrowing.
3292 (iv_ca_prune): Pass new argument.
3294 2014-01-10 Jeff Law <law@redhat.com>
3297 * ree.c (combine_reaching_defs): Ensure the defining statement
3298 occurs before the extension when optimizing extensions with
3299 different source and destination hard registers.
3301 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3304 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3305 vtables into the type inheritance graph.
3307 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3309 PR rtl-optimization/59754
3310 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3311 modes in the REGNO != REGNO case.
3313 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3315 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3317 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3319 PR tree-optimization/59745
3320 * tree-predcom.c (tree_predictive_commoning_loop): Call
3321 free_affine_expand_cache if giving up because components is NULL.
3323 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3324 GC in payload of target_globals struct instead of allocating them on
3325 the heap and the larger structs separately using GC.
3326 * target-globals.h (struct target_globals): Make regs, hard_regs,
3327 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3328 of GTY((skip)) and change type to void *.
3329 (reset_target_globals): Cast loads from those fields to corresponding
3332 2014-01-10 Steve Ellcey <sellcey@mips.com>
3335 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3336 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3337 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3339 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3342 * aarch64-modes.def (CC_Zmode): New flags mode.
3343 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3344 represents an equality.
3345 (aarch64_get_condition_code): Handle CC_Zmode.
3346 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3348 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3350 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3351 extraction in good case.
3353 2014-01-10 Richard Biener <rguenther@suse.de>
3355 PR tree-optimization/59374
3356 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3357 checking after SLP discovery. Mark stmts not participating
3358 in any SLP instance properly.
3360 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3362 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3363 when handling a SET rtx.
3365 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3367 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3368 (cortex-a57): Likewise.
3369 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3371 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3373 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3374 non-iwmmxt builtins.
3376 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3380 * ipa-devirt.c record_target_from_binfo): Take as argument
3381 stack of binfos and lookup matching one for virtual inheritance.
3382 (possible_polymorphic_call_targets_1): Update.
3384 2014-01-10 Huacai Chen <chenhc@lemote.com>
3386 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3387 kernel strings for Loongson-2E/2F/3A.
3389 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3392 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3393 is_gimple_call before calling gimple_call_internal_p.
3395 2014-01-09 Steve Ellcey <sellcey@mips.com>
3397 * Makefile.in (TREE_FLOW_H): Remove.
3398 (TREE_SSA_H): Add file names from tree-flow.h.
3399 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3400 * tree.h: Remove tree-flow.h reference.
3401 * hash-table.h: Remove tree-flow.h reference.
3402 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3403 reference with tree-ssa-loop.h.
3405 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3407 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3408 default element-order behavior for -maltivec.
3409 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3410 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3411 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3412 when targeting big endian, at least for now.
3413 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3415 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3418 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3419 var satisfies use_register_for_decl, just take into account type
3420 alignment, rather than decl alignment.
3422 PR tree-optimization/59622
3423 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3424 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3425 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3426 Don't devirtualize for inplace at all. For targets.length () == 1,
3427 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3429 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3431 * config/i386/i386.md (cpu): Remove the unused btver1.
3433 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3435 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3437 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3440 * tree-core.h (struct target_globals): New forward declaration.
3441 (struct tree_target_option): Add globals field.
3442 * tree.h (TREE_TARGET_GLOBALS): Define.
3443 (prepare_target_option_nodes_for_pch): New prototype.
3444 * target-globals.h (struct target_globals): Define even if
3446 * tree.c (prepare_target_option_node_for_pch,
3447 prepare_target_option_nodes_for_pch): New functions.
3448 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3449 * config/i386/i386.c: Include target-globals.h.
3450 (ix86_set_current_function): Instead of doing target_reinit
3451 unconditionally, use save_target_globals_default_opts and
3452 restore_target_globals.
3454 2014-01-09 Richard Biener <rguenther@suse.de>
3456 PR tree-optimization/59715
3457 * tree-cfg.h (split_critical_edges): Declare.
3458 * tree-cfg.c (split_critical_edges): Export.
3459 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3461 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3463 * cfgexpand.c (expand_stack_vars): Optionally disable
3464 asan stack protection.
3465 (expand_used_vars): Likewise.
3466 (partition_stack_vars): Likewise.
3467 * asan.c (asan_emit_stack_protection): Optionally disable
3468 after return stack usage.
3469 (instrument_derefs): Optionally disable memory access instrumentation.
3470 (instrument_builtin_call): Likewise.
3471 (instrument_strlen_call): Likewise.
3472 (asan_protect_global): Optionally disable global variables protection.
3473 * doc/invoke.texi: Added doc for new options.
3474 * params.def: Added new options.
3475 * params.h: Likewise.
3477 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3479 PR rtl-optimization/59724
3480 * ifcvt.c (cond_exec_process_if_block): Don't call
3481 flow_find_head_matching_sequence with 0 longest_match.
3482 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3483 non-active insns if !stop_after.
3484 (try_head_merge_bb): Revert 2014-01-07 changes.
3486 2014-01-08 Jeff Law <law@redhat.com>
3488 * ree.c (get_sub_rtx): New function, extracted from...
3489 (merge_def_and_ext): Here.
3490 (combine_reaching_defs): Use get_sub_rtx.
3492 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3494 * cgraph.h (varpool_variable_node): Do not choke on null node.
3496 2014-01-08 Catherine Moore <clm@codesourcery.com>
3498 * config/mips/mips.md (simple_return): Attempt to use JRC
3500 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3502 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3504 PR rtl-optimization/59137
3505 * reorg.c (steal_delay_list_from_target): Call update_block for
3507 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3509 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3511 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3512 two duplicate entries.
3514 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3517 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3519 * config/mips/mips.c (mips_truncated_op_cost): New function.
3520 (mips_rtx_costs): Adjust test for BADDU.
3521 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3523 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3525 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3526 (*baddu_si): ...this new pattern.
3528 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3531 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3533 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3536 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3538 (expand_expr, expand_normal): Adjust.
3539 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3540 inner_reference_p. Use inner_reference_p to expand inner references.
3541 (store_expr): Adjust.
3542 * cfgexpand.c (expand_call_stmt): Adjust.
3544 2014-01-08 Rong Xu <xur@google.com>
3546 * gcov-io.c (gcov_var): Move from gcov-io.h.
3547 (gcov_position): Ditto.
3548 (gcov_is_error): Ditto.
3549 (gcov_rewrite): Ditto.
3550 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3551 only part to libgcc/libgcov.h.
3553 2014-01-08 Marek Polacek <polacek@redhat.com>
3556 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3558 2014-01-08 Marek Polacek <polacek@redhat.com>
3561 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3563 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3565 PR rtl-optimization/59649
3566 * stor-layout.c (get_mode_bounds): For BImode return
3567 0 and STORE_FLAG_VALUE.
3569 2014-01-08 Richard Biener <rguenther@suse.de>
3572 * gimple.h (is_gimple_builtin_call): Remove.
3573 (gimple_builtin_call_types_compatible_p): New.
3574 (gimple_call_builtin_p): New overload.
3575 * gimple.c (is_gimple_builtin_call): Remove.
3576 (validate_call): Rename to ...
3577 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3579 (validate_type): New static function.
3580 (gimple_call_builtin_p): New overload and adjust.
3581 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3582 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3583 (gimple_fold_stmt_to_constant_1): Likewise.
3584 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3586 2014-01-08 Richard Biener <rguenther@suse.de>
3589 * gimplify.c (gimplify_expr): Gimplify register-register type
3590 VIEW_CONVERT_EXPRs to separate stmts.
3592 2014-01-07 Jeff Law <law@redhat.com>
3595 * ree.c (combine_set_extension): Handle case where source
3596 and destination registers in an extension insn are different.
3597 (combine_reaching_defs): Allow source and destination registers
3598 in extension to be different under limited circumstances.
3599 (add_removable_extension): Remove restriction that the
3600 source and destination registers in the extension are the same.
3601 (find_and_remove_re): Emit a copy from the extension's
3602 destination to its source after the defining insn if
3603 the source and destination registers are different.
3606 * ifcvt.c (merge_if_block): If we are merging a block with more than
3607 one successor with a block with no successors, remove any BARRIER
3608 after the second block.
3610 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3612 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3614 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3617 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3618 for 14-bit register offsets when INT14_OK_STRICT is false.
3620 2014-01-07 Roland Stigge <stigge@antcom.de>
3621 Michael Meissner <meissner@linux.vnet.ibm.com>
3624 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3625 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3627 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3629 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3632 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3634 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3635 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3636 rtx is const0_rtx or not.
3638 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3641 * target-globals.c (save_target_globals): Remove this_fn_optab
3643 * toplev.c: Include optabs.h.
3644 (target_reinit): Temporarily restore the global options if another
3645 set of options are in force.
3647 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3649 PR rtl-optimization/58668
3650 * cfgcleanup.c (flow_find_cross_jump): Don't count
3651 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3652 to determine what is counted.
3653 (flow_find_head_matching_sequence): Use active_insn_p to determine
3655 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3657 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3658 determine what is counted.
3660 PR tree-optimization/59643
3661 * tree-predcom.c (split_data_refs_to_components): If one dr is
3662 read and one write, determine_offset fails and the write isn't
3663 in the bad component, just put the read into the bad component.
3665 2014-01-07 Mike Stump <mikestump@comcast.net>
3666 Jakub Jelinek <jakub@redhat.com>
3669 * tree-core.h (struct tree_optimization_option): Change optabs
3670 type from unsigned char * to void *.
3671 * optabs.c (init_tree_optimization_optabs): Adjust
3672 TREE_OPTIMIZATION_OPTABS initialization.
3674 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3677 * config/i386/i386.h (struct machine_function): Add
3678 no_drap_save_restore field.
3679 * config/i386/i386.c (ix86_save_reg): Use
3680 !cfun->machine->no_drap_save_restore instead of
3681 crtl->stack_realign_needed.
3682 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3683 this function clears frame_pointer_needed. Set
3684 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3685 and DRAP reg is needed.
3687 2014-01-06 Marek Polacek <polacek@redhat.com>
3690 * doc/implement-c.texi: Mention that other integer types are
3691 permitted as bit-field types in strictly conforming mode.
3693 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3695 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3698 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3700 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3702 2014-01-06 Martin Jambor <mjambor@suse.cz>
3705 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3707 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3709 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3713 * var-tracking.c (add_stores): Preserve the value of the source even if
3714 we don't record the store.
3716 2014-01-06 Terry Guo <terry.guo@arm.com>
3718 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3720 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3723 * config/darwin.c (darwin_function_section): Adjust return values to
3724 correspond to optimisation changes made in r206070.
3726 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3728 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3729 from prefetch_block tune setting.
3730 (nocona_cost): Correct size of prefetch block to 64.
3732 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3734 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3735 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3736 used to save the static chain register in the computation of the offset
3737 from which the FP registers need to be restored.
3739 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3741 PR tree-optimization/59519
3742 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3743 ICE if get_current_def (current_new_name) is already non-NULL, as long
3744 as it is a phi result of some other phi in *new_exit_bb that has
3747 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3748 or vmovdqu* for misaligned_operand.
3749 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3750 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3751 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3752 aligned_mem for AVX512F masked aligned load and store builtins and for
3755 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3757 PR tree-optimization/59651
3758 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3759 Address range for negative step should be added by TYPE_SIZE_UNIT.
3761 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3763 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3764 overlapping registers also for registers other than the stack pointer.
3766 2014-01-03 Marek Polacek <polacek@redhat.com>
3769 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3772 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3775 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3778 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3779 (push splitter): Use <P:MODE_SIZE> instead of
3780 GET_MODE_SIZE (<P:MODE>mode).
3781 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3782 (mov -1, reg peephole2): Likewise.
3783 * config/i386/sse.md (*mov<mode>_internal,
3784 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3785 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3786 *<code><mode>3, *andnot<mode>3<mask_name>,
3787 <mask_codefor><code><mode>3<mask_name>): Likewise.
3788 * config/i386/subst.md (mask_mode512bit_condition,
3789 sd_mask_mode512bit_condition): Likewise.
3791 2014-01-02 Xinliang David Li <davidxl@google.com>
3793 PR tree-optimization/59303
3794 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3795 (dump_predicates): Better output format.
3796 (pred_equal_p): New function.
3797 (is_neq_relop_p): Ditto.
3798 (is_neq_zero_form_p): Ditto.
3799 (pred_expr_equal_p): Ditto.
3800 (pred_neg_p): Ditto.
3801 (simplify_pred): Ditto.
3802 (simplify_preds_2): Ditto.
3803 (simplify_preds_3): Ditto.
3804 (simplify_preds_4): Ditto.
3805 (simplify_preds): Ditto.
3807 (push_to_worklist): Ditto.
3808 (get_pred_info_from_cmp): Ditto.
3809 (is_degenerated_phi): Ditto.
3810 (normalize_one_pred_1): Ditto.
3811 (normalize_one_pred): Ditto.
3812 (normalize_one_pred_chain): Ditto.
3813 (normalize_preds): Ditto.
3814 (normalize_cond_1): Remove function.
3815 (normalize_cond): Ditto.
3816 (is_gcond_subset_of): Ditto.
3817 (is_subset_of_any): Ditto.
3818 (is_or_set_subset_of): Ditto.
3819 (is_and_set_subset_of): Ditto.
3820 (is_norm_cond_subset_of): Ditto.
3821 (pred_chain_length_cmp): Ditto.
3822 (convert_control_dep_chain_into_preds): Type change.
3823 (find_predicates): Ditto.
3824 (find_def_preds): Ditto.
3825 (destroy_predicates_vecs): Ditto.
3826 (find_matching_predicates_in_rest_chains): Ditto.
3827 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3828 (is_pred_expr_subset): Ditto.
3829 (is_pred_chain_subset_of): Ditto.
3830 (is_included_in): Ditto.
3831 (is_superset_of): Ditto.
3833 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3835 Update copyright years.
3837 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3839 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3840 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3841 config/arc/arc.md, config/arc/arc.opt,
3842 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3843 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3844 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3845 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3846 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3847 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3848 vtable-verify.c, vtable-verify.h: Use the standard form for the
3851 2014-01-02 Tobias Burnus <burnus@net-b.de>
3853 * gcc.c (process_command): Update copyright notice dates.
3854 * gcov-dump.c: Ditto.
3856 * doc/cpp.texi: Bump @copying's copyright year.
3857 * doc/cppinternals.texi: Ditto.
3858 * doc/gcc.texi: Ditto.
3859 * doc/gccint.texi: Ditto.
3860 * doc/gcov.texi: Ditto.
3861 * doc/install.texi: Ditto.
3862 * doc/invoke.texi: Ditto.
3864 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3866 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3868 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3870 * config/i386/sse.md (*mov<mode>_internal): Guard
3871 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3873 PR rtl-optimization/59647
3874 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3875 new_rtx into UNSIGNED_FLOAT rtxes.
3877 Copyright (C) 2014 Free Software Foundation, Inc.
3879 Copying and distribution of this file, with or without modification,
3880 are permitted in any medium without royalty provided the copyright
3881 notice and this notice are preserved.