1 2014-02-21 Catherine Moore <clm@codesourcery.com>
3 * doc/invoke.texi (mvirt, mno-virt): Document.
4 * config/mips/mips.opt (mvirt): New option.
5 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
7 2014-02-21 Richard Biener <rguenther@suse.de>
9 PR tree-optimization/60276
10 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
11 (STMT_VINFO_MIN_NEG_DIST): New macro.
12 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
13 STMT_VINFO_MIN_NEG_DIST.
14 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
15 made for negative dependence distances still hold.
17 2014-02-21 Richard Biener <rguenther@suse.de>
20 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
21 DECL_INITIAL for globals not in the current function context.
23 2014-02-21 Jakub Jelinek <jakub@redhat.com>
25 PR tree-optimization/56490
26 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
27 * tree-ssa-uninit.c: Include params.h.
28 (compute_control_dep_chain): Add num_calls argument, return false
29 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
30 num_calls to recursive call.
31 (find_predicates): Change dep_chain into normal array,
32 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
33 variable and adjust compute_control_dep_chain caller.
34 (find_def_preds): Likewise.
36 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
38 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
39 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
41 2014-02-21 Nick Clifton <nickc@redhat.com>
43 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
45 (popqi1): Add mode to pre_dec.
48 2014-02-21 Jakub Jelinek <jakub@redhat.com>
50 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
51 mode for mask of V8SFmode permutation.
53 2014-02-20 Richard Henderson <rth@redhat.com>
56 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
57 a new pseudo for OLDVAL.
59 2014-02-20 Jakub Jelinek <jakub@redhat.com>
62 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
63 gen_reg_rtx if d->testing_p.
64 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
65 if d->testing_p and we will certainly return true.
66 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
69 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
71 * emit-rtl.c (gen_reg_rtx): Assert that
72 crtl->emit.regno_pointer_align_length is non-zero.
74 2014-02-20 Richard Henderson <rth@redhat.com>
77 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
78 on failure the store back into EXPECT.
80 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
81 Sandra Loosemore <sandra@codesourcery.com>
83 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
84 * config/nios2/nios2.c (nios2_function_profiler):
85 Add -fPIC (flag_pic == 2) support.
86 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
87 (nios2_large_offset_p): New function.
88 (nios2_unspec_reloc_p): Move up position, update to use
90 (nios2_unspec_address): Remove function.
91 (nios2_unspec_offset): New function.
92 (nios2_large_got_address): New function.
93 (nios2_got_address): Add large offset support.
94 (nios2_legitimize_tls_address): Update usage of removed and new
96 (nios2_symbol_binds_local_p): New function.
97 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
98 (nios2_legitimize_address): Update to use nios2_large_offset_p.
99 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
100 (nios2_print_operand): Merge H/L processing, add hiadj/lo
101 processing for (const (unspec ...)).
102 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
104 2014-02-20 Richard Biener <rguenther@suse.de>
106 * tree-cfg.c (replace_uses_by): Mark altered BBs before
107 doing the substitution.
108 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
110 2014-02-20 Martin Jambor <mjambor@suse.cz>
113 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
114 info when checking whether lattices are bottom.
116 2014-02-20 Richard Biener <rguenther@suse.de>
119 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
122 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
125 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
126 parameter specifying the scaling.
127 (inline_call): Update.
128 (want_inline_recursively): Guard division by zero.
129 (recursive_inlining): Update.
130 * ipa-inline.h (clone_inlined_nodes): Update.
132 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
135 * config/i386/i386.c (classify_argument): Pass structures of size
136 64 bytes or less in register.
138 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
139 Kirill Yukhin <kirill.yukhin@intel.com>
141 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
142 (_mm_rcp28_round_ss): Ditto.
143 (_mm_rsqrt28_round_sd): Ditto.
144 (_mm_rsqrt28_round_ss): Ditto.
145 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
146 (_mm_rcp14_round_ss): Ditto.
147 (_mm_rsqrt14_round_sd): Ditto.
148 (_mm_rsqrt14_round_ss): Ditto.
149 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
150 the first input operand, get rid of match_dup.
151 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
153 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
155 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
156 operand as the first input operand, set type attribute.
157 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
159 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
160 operand as the first input operand, set type attribute.
162 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
164 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
167 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
170 * config/i386/i386.c (construct_container): Remove TFmode check
171 for X86_64_INTEGER_CLASS.
173 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
176 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
177 only when -Wpsabi is enabled.
179 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
182 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
183 passing arrays in registers are the same as for structs, so remove the
184 special case for them.
186 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
188 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
189 destination type, extract only the valid bits if the source type is not
190 integral and has a different mode.
192 2014-02-19 Richard Biener <rguenther@suse.de>
195 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
198 2014-02-19 Richard Biener <rguenther@suse.de>
201 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
202 (ipa_modify_call_arguments): Emit an argument load explicitely and
203 preserve virtual SSA form there and for the replacement call.
204 Do not update SSA form nor free dominance info.
206 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
208 * ipa.c (function_and_variable_visibility): Also clear WEAK
209 flag when disolving COMDAT_GROUP.
211 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
213 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
214 * ipa-prop.c (ipa_set_jf_known_type): Return early when
216 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
217 do more sanity checks.
218 (detect_type_change): Return true when giving up early.
219 (compute_complex_assign_jump_func): Fix type parameter of
221 (compute_complex_ancestor_jump_func): Likewise.
222 (update_jump_functions_after_inlining): Fix updating of
224 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
226 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
228 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
229 inline clones when edge disappears.
231 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
234 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
235 Split 64-bit moves into 2 patterns. Do not allow the use of
236 direct move for TDmode in little endian, since the decimal value
237 has little endian bytes within a word, but the 64-bit pieces are
238 ordered in a big endian fashion, and normal subreg's of TDmode are
240 (mov<mode>_64bit_dm): Likewise.
241 (movtd_64bit_nodm): Likewise.
243 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
245 PR tree-optimization/60174
246 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
247 statement of an SSA_NAME that occurs in an abnormal PHI node.
249 2014-02-18 Jakub Jelinek <jakub@redhat.com>
252 * final.c (SEEN_BB): Remove.
253 (SEEN_NOTE, SEEN_EMITTED): Renumber.
254 (final_scan_insn): Don't force_source_line on second
255 NOTE_INSN_BASIC_BLOCK.
257 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
260 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
261 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
262 (type_natural_mode): Warn ABI change when %zmm register is not
263 available for AVX512F vector value passing.
265 2014-02-18 Kai Tietz <ktietz@redhat.com>
268 * config/i386/i386.c (ix86_expand_prologue): Use value in
269 rax register as displacement when restoring %r10 or %rax.
270 Fix wrong offset when restoring both registers.
272 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
274 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
275 assertion with conditional return.
277 2014-02-18 Jakub Jelinek <jakub@redhat.com>
278 Uros Bizjak <ubizjak@gmail.com>
281 * config/i386/driver-i386.c (host_detect_local_cpu): If
282 YMM state is not saved by the OS, also clear has_f16c. Move
283 CPUID 0x80000001 handling before YMM state saving checking.
285 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
287 PR rtl-optimization/58960
288 * haifa-sched.c (alloc_global_sched_pressure_data): New,
289 factored out from ...
290 (sched_init): ... here.
291 (free_global_sched_pressure_data): New, factored out from ...
292 (sched_finish): ... here.
293 * sched-int.h (free_global_sched_pressure_data): Declare.
294 * sched-rgn.c (nr_regions_initial): New static global.
295 (haifa_find_rgns): Initialize it.
296 (schedule_region): Disable sched-pressure for the newly
299 2014-02-17 Richard Biener <rguenther@suse.de>
301 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
302 release SSA defs of pattern stmts.
304 2014-02-17 Richard Biener <rguenther@suse.de>
306 * tree-inline.c (expand_call_inline): Release the virtual
307 operand defined by the call we are about to inline.
309 2014-02-17 Richard Biener <rguenther@suse.de>
311 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
313 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
314 Ilya Tocar <ilya.tocar@intel.com>
316 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
317 arguments order in builtin.
318 (_mm512_permutexvar_epi64): Ditto.
319 (_mm512_mask_permutexvar_epi64): Ditto
320 (_mm512_maskz_permutexvar_epi32): Ditto
321 (_mm512_permutexvar_epi32): Ditto
322 (_mm512_mask_permutexvar_epi32): Ditto
324 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
326 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
327 (p8_vmrgow): Likewise.
329 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
331 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
334 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
337 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
338 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
339 into 64-bit and 32-bit moves. On 64-bit moves, add support for
340 using direct move instructions on ISA 2.07. Also adjust
341 instruction length for 64-bit.
342 (mov<mode>_64bit, TFmode/TDmode): Likewise.
343 (mov<mode>_32bit, TFmode/TDmode): Likewise.
345 2014-02-15 Alan Modra <amodra@gmail.com>
349 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
350 find_replacement on parts of insn rtl that might be reloaded.
352 2014-02-15 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/60183
355 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
356 (tree_ssa_phiprop): Calculate and free post-dominators.
358 2014-02-14 Jeff Law <law@redhat.com>
360 PR rtl-optimization/60131
361 * ree.c (get_extended_src_reg): New function.
362 (combine_reaching_defs): Use it rather than assuming location of REG.
363 (find_and_remove_re): Verify first operand of extension is
364 a REG before adding the insns to the copy list.
366 2014-02-14 Roland McGrath <mcgrathr@google.com>
368 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
369 * configure: Regenerated.
370 * config.in: Regenerated.
371 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
372 instead of ASM_SHORT.
374 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
375 Richard Earnshaw <rearnsha@arm.com>
377 PR rtl-optimization/59535
378 * lra-constraints.c (process_alt_operands): Encourage alternative
379 when unassigned pseudo class is superset of the alternative class.
380 (inherit_reload_reg): Don't inherit when optimizing for code size.
381 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
382 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
383 modes not less than 4 for Thumb1.
385 2014-02-14 Kyle McMartin <kyle@redhat.com>
388 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
390 2014-02-14 Richard Biener <rguenther@suse.de>
392 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
393 (get_frame_arg): Drop the assert with langhook types_compatible_p.
394 Do not strip INDIRECT_REFs.
396 2014-02-14 Richard Biener <rguenther@suse.de>
399 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
400 DECL_FUNCTION_SPECIFIC_TARGET.
401 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
402 * tree-streamer-out.c (pack_ts_target_option): Remove.
403 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
404 (write_ts_function_decl_tree_pointers): Do not stream
405 DECL_FUNCTION_SPECIFIC_TARGET.
406 * tree-streamer-in.c (unpack_ts_target_option): Remove.
407 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
408 (lto_input_ts_function_decl_tree_pointers): Do not stream
409 DECL_FUNCTION_SPECIFIC_TARGET.
411 2014-02-14 Jakub Jelinek <jakub@redhat.com>
413 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
414 (get_initial_def_for_induction, vectorizable_induction): Ignore
415 debug stmts when looking for exit_phi.
416 (vectorizable_live_operation): Fix up condition.
418 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
420 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
421 nreverse() because it changes the content of original tree list.
423 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
425 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
426 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
428 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
430 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
431 GNU coding standards.
433 2014-02-13 Jakub Jelinek <jakub@redhat.com>
436 * dwarf2out.c (gen_subprogram_die): Don't call
437 add_calling_convention_attribute if subr_die is old_die.
439 2014-02-13 Sharad Singhai <singhai@google.com>
441 * doc/optinfo.texi: Fix order of nodes.
443 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
445 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
446 operands[2], not operands[3].
448 2014-02-13 Richard Biener <rguenther@suse.de>
451 * doc/install.texi (ISL): Update recommended version to 0.12.2,
452 mention the possibility of an in-tree build.
453 (CLooG): Update recommended version to 0.18.1, mention the
454 possibility of an in-tree build and clarify that the ISL
455 bundled with CLooG does not work.
457 2014-02-13 Jakub Jelinek <jakub@redhat.com>
460 * expr.c (compress_float_constant): If x is a hard register,
461 extend into a pseudo and then move to x.
463 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
465 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
466 caused by bad second argument to warning_at() with -mhotpatch and
467 nested functions (e.g. with gfortran).
469 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
471 * opts.c (option_name): Remove "enabled by default" rider.
473 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
475 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
477 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
478 Uros Bizjak <ubizjak@gmail.com>
481 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
482 * configure: Regenerated.
484 2014-02-12 Richard Biener <rguenther@suse.de>
486 * vec.c (vec_prefix::calculate_allocation): Move as
487 inline variant to vec.h.
488 (vec_prefix::calculate_allocation_1): New out-of-line version.
489 * vec.h (vec_prefix::calculate_allocation_1): Declare.
490 (vec_prefix::m_has_auto_buf): Rename to ...
491 (vec_prefix::m_using_auto_storage): ... this.
492 (vec_prefix::calculate_allocation): Inline the easy cases
493 and dispatch to calculate_allocation_1 which doesn't need the
495 (va_heap::reserve): Use gcc_checking_assert.
496 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
497 m_using_auto_storage.
498 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
500 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
501 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
502 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
504 2014-02-12 Richard Biener <rguenther@suse.de>
506 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
507 when we found a dependence.
509 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
511 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
513 (maybe_fold_stmt): ... into this new function.
514 * omp-low.c (lower_omp): Update comment.
516 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
519 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
522 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
524 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
525 identifiers in comments.
526 (cortexa53_extra_costs): Likewise.
527 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
528 (cortexa7_extra_costs): Likewise.
529 (cortexa12_extra_costs): Likewise.
530 (cortexa15_extra_costs): Likewise.
531 (v7m_extra_costs): Likewise.
533 2014-02-12 Richard Biener <rguenther@suse.de>
536 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
537 of posix_memalign being successful.
538 (lower_stmt): Restrict lowering of posix_memalign to when
539 -ftree-bit-ccp is enabled.
541 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
543 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
545 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
547 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
549 PR rtl-optimization/60116
550 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
551 other_insn once the combination has been validated.
553 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
556 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
558 * ipa-devirt.c: Include demangle.h
559 (odr_violation_reported): New static variable.
560 (add_type_duplicate): Update odr_violations.
561 (maybe_record_node): Add completep parameter; update it.
562 (record_target_from_binfo): Add COMPLETEP parameter;
564 (possible_polymorphic_call_targets_1): Likewise.
565 (struct polymorphic_call_target_d): Add nonconstruction_targets;
566 rename FINAL to COMPLETE.
567 (record_targets_from_bases): Sanity check we found the binfo;
568 fix COMPLETEP updating.
569 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
570 parameter, fix computing of COMPLETEP.
571 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
572 at LTO time do demangling.
573 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
574 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
576 (gimple_get_virt_method_for_binfo): Likewise.
577 * gimple-fold.h (gimple_get_virt_method_for_binfo,
578 gimple_get_virt_method_for_vtable): Update prototypes.
580 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
583 * genautomata.c (add_presence_absence): Fix typo with
584 {final_}presence_list.
586 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
589 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
590 for VSX/Altivec vectors that land in GPR registers.
592 2014-02-11 Richard Henderson <rth@redhat.com>
593 Jakub Jelinek <jakub@redhat.com>
596 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
597 around drhs if type conversion to lacc->type is not useless.
599 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
601 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
603 (cortex-a57.cortex-a53): Likewise.
604 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
606 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
608 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
611 2014-02-11 Renlin Li <Renlin.Li@arm.com>
613 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
614 add_options_for_arm_vfp3.
616 2014-02-11 Jeff Law <law@redhat.com>
619 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
620 object with an undesirable mode.
622 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
625 * config/i386/sol2-9.h: New file.
626 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
627 *-*-solaris2.9*): Use it.
629 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
631 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
632 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
634 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
636 * config/microblaze/microblaze.c: Extend mcpu version format
638 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
640 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
642 2014-02-10 Richard Henderson <rth@redhat.com>
645 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
646 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
647 ms-abi vs -mno-accumulate-outgoing-args.
648 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
649 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
652 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
655 * cfgexpand.c (expand_asm_operands): Attach source location to
656 ASM_INPUT rtx objects.
657 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
659 2014-02-10 Nick Clifton <nickc@redhat.com>
661 * config/mn10300/mn10300.c (popcount): New function.
662 (mn10300_expand_prologue): Include saved registers in stack usage
665 2014-02-10 Jeff Law <law@redhat.com>
668 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
669 when changing the SET_DEST of a prior insn to avoid an input reload.
671 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
673 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
674 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
675 -mcall-openbsd, or -mcall-linux.
676 (CC1_ENDIAN_BIG_SPEC): Remove.
677 (CC1_ENDIAN_LITTLE_SPEC): Remove.
678 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
679 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
680 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
681 and %cc1_endian_default.
682 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
684 2014-02-10 Richard Biener <rguenther@suse.de>
686 PR tree-optimization/60115
687 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
688 MEM_REF handling. Properly verify that the accesses are not
689 out of the objects bound.
691 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
693 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
696 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
698 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
699 proper constants and fix formatting.
700 (possible_polymorphic_call_targets): Fix formatting.
702 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
703 Ilya Tocar <ilya.tocar@intel.com>
705 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
706 (_mm512_loadu_epi32): Renamed into...
707 (_mm512_loadu_si512): This.
708 (_mm512_storeu_epi32): Renamed into...
709 (_mm512_storeu_si512): This.
710 (_mm512_maskz_ceil_ps): Removed.
711 (_mm512_maskz_ceil_pd): Ditto.
712 (_mm512_maskz_floor_ps): Ditto.
713 (_mm512_maskz_floor_pd): Ditto.
714 (_mm512_floor_round_ps): Ditto.
715 (_mm512_floor_round_pd): Ditto.
716 (_mm512_ceil_round_ps): Ditto.
717 (_mm512_ceil_round_pd): Ditto.
718 (_mm512_mask_floor_round_ps): Ditto.
719 (_mm512_mask_floor_round_pd): Ditto.
720 (_mm512_mask_ceil_round_ps): Ditto.
721 (_mm512_mask_ceil_round_pd): Ditto.
722 (_mm512_maskz_floor_round_ps): Ditto.
723 (_mm512_maskz_floor_round_pd): Ditto.
724 (_mm512_maskz_ceil_round_ps): Ditto.
725 (_mm512_maskz_ceil_round_pd): Ditto.
726 (_mm512_expand_pd): Ditto.
727 (_mm512_expand_ps): Ditto.
728 * config/i386/i386.c (ix86_builtins): Remove
729 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
731 * config/i386/predicates.md (const1256_operand): New.
732 (const_1_to_2_operand): Ditto.
733 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
734 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
735 (*avx512pf_gatherpf<mode>sf): Ditto.
736 (avx512pf_gatherpf<mode>df): Ditto.
737 (*avx512pf_gatherpf<mode>df_mask): Ditto.
738 (*avx512pf_gatherpf<mode>df): Ditto.
739 (avx512pf_scatterpf<mode>sf): Ditto.
740 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
741 (*avx512pf_scatterpf<mode>sf): Ditto.
742 (avx512pf_scatterpf<mode>df): Ditto.
743 (*avx512pf_scatterpf<mode>df_mask): Ditto.
744 (*avx512pf_scatterpf<mode>df): Ditto.
745 (avx512f_expand<mode>): Removed.
746 (<shift_insn><mode>3<mask_name>): Change predicate type.
748 2014-02-08 Jakub Jelinek <jakub@redhat.com>
750 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
751 not at the end of datarefs vector use ordered_remove to avoid
752 reordering datarefs vector.
755 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
756 mark local addressable non-static vars as GOVD_PRIVATE
757 instead of GOVD_LOCAL.
758 * omp-low.c (lower_omp_for): Move gimple_bind_vars
759 and BLOCK_VARS of gimple_bind_block to new_stmt rather
763 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
764 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
765 assume_aligned or alloc_align attributes.
766 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
767 arguments. Handle also assume_aligned and alloc_align attributes.
768 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
769 calls to functions with assume_aligned or alloc_align attributes.
770 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
772 2014-02-08 Terry Guo <terry.guo@arm.com>
774 * doc/invoke.texi: Document ARM -march=armv7e-m.
776 2014-02-08 Jakub Jelinek <jakub@redhat.com>
778 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
779 flag on __cilkrts_rethrow builtin.
782 * ipa-cp.c (determine_versionability): Fail at -O0
783 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
784 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
787 2014-02-04 Jakub Jelinek <jakub@redhat.com>
790 * tree-inline.c (copy_forbidden): Fail for
791 __attribute__((optimize (0))) functions.
793 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
795 * varpool.c: Include pointer-set.h.
796 (varpool_remove_unreferenced_decls): Variables in other partitions
797 will not be output; be however careful to not lose information
800 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
802 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
803 lookup in the vtable constructor.
805 2014-02-07 Jeff Law <law@redhat.com>
808 * config/m68k/m68k.md (ashldi_extsi): Turn into a
809 define_insn_and_split.
811 * ipa-inline.c (inline_small_functions): Fix typos.
813 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
815 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
816 (s390_can_use_return_insn): Declare.
817 * config/s390/s390.h (EPILOGUE_USES): Define.
818 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
820 (s390_chunkify_start): Handle return JUMP_LABELs.
821 (s390_early_mach): Emit a main_pool instruction on the entry edge.
822 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
823 (s390_can_use_return_insn): New functions.
824 (s390_fix_long_loop_prediction): Handle conditional returns.
825 (TARGET_SET_UP_BY_PROLOGUE): Define.
826 * config/s390/s390.md (ANY_RETURN): New code iterator.
827 (*creturn, *csimple_return, return, simple_return): New patterns.
829 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
831 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
832 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
833 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
834 REG_CFA_RESTORE list when deciding not to restore a register.
836 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
838 * config/s390/s390.c: Include tree-pass.h and context.h.
839 (s390_early_mach): New function, split out from...
840 (s390_emit_prologue): ...here.
841 (pass_data_s390_early_mach): New pass structure.
842 (pass_s390_early_mach): New class.
843 (s390_option_override): Create and register early_mach pass.
846 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
848 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
849 to match for the exit block.
851 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
853 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
854 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
855 Reject misaligned operands.
857 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
859 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
861 2014-02-07 Richard Biener <rguenther@suse.de>
864 * gimple-low.c (lower_builtin_posix_memalign): New function.
865 (lower_stmt): Call it to lower posix_memalign in a way
866 to make alignment info accessible.
868 2014-02-07 Jakub Jelinek <jakub@redhat.com>
871 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
872 __builtin_setjmp_receiver.
874 2014-02-07 Richard Biener <rguenther@suse.de>
877 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
878 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
879 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
880 Handle BUILT_IN_POSIX_MEMALIGN.
881 (find_func_clobbers): Likewise.
882 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
883 (call_may_clobber_ref_p_1): Likewise.
885 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
888 * ipa-devirt.c (record_target_from_binfo): Remove overactive
891 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
894 * lto-cgraph.c (lto_output_node): Use
895 symtab_get_symbol_partitioning_class.
896 (lto_output_varpool_node): likewise.
897 (symtab_get_symbol_partitioning_class): Move here from
899 * cgraph.h (symbol_partitioning_class): Likewise.
900 (symtab_get_symbol_partitioning_class): Declare.
902 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
904 * ggc.h (ggc_internal_cleared_alloc): New macro.
905 * vec.h (vec_safe_copy): Handle memory stats.
906 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
907 * target-globals.c (save_target_globals): Likewise.
909 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
912 * expr.c (emit_move_resolve_push): Export; be bit more selective
913 on when to clear alias set.
914 * expr.h (emit_move_resolve_push): Declare.
915 * function.h (struct function): Add tail_call_marked.
916 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
917 * config/i386/i386-protos.h (ix86_expand_push): Remove.
918 * config/i386/i386.md (TImode move expander): De not call
920 (FP push expanders): Preserve memory attributes.
921 * config/i386/sse.md (push<mode>1): Remove.
922 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
923 (ix86_expand_push): Remove.
924 * config/i386/mmx.md (push<mode>1): Remove.
926 2014-02-06 Jakub Jelinek <jakub@redhat.com>
928 PR rtl-optimization/60030
929 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
930 lopart with paradoxical subreg before shifting it up by hprec.
932 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
934 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
935 Remove extra newline at end of file.
936 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
937 (arm_issue_rate): Handle cortexa57.
938 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
939 (cortex-a57.cortex-a53): Likewise.
941 2014-02-06 Jakub Jelinek <jakub@redhat.com>
944 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
945 don't record in REG_FRAME_RELATED_EXPR registers not set in that
947 (arm_expand_prologue): Adjust all callers.
948 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
949 info, registers also at the lowest numbered registers side. Use
950 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
954 * var-tracking.c (adjust_mems): Before adding a SET to
955 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
957 2014-02-06 Alan Modra <amodra@gmail.com>
960 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
961 change SDmode to DDmode when lra_in_progress.
963 2014-02-06 Jakub Jelinek <jakub@redhat.com>
966 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
967 free_data_ref on the dr first, and before goto again also set dr
968 to the next dr. For simd_lane_access, free old datarefs[i] before
969 overwriting it. For get_vectype_for_scalar_type failure, don't
970 free_data_ref if simd_lane_access.
972 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
975 * tree.h (opts_for_fn): New inline function.
976 (opt_for_fn): Define.
977 * config/i386/i386.c (ix86_function_regparm): Use
978 opt_for_fn (decl, optimize) instead of optimize.
980 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
982 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
983 for SYMBOL_REF in large memory model.
985 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
987 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
989 (cortex-a57): Likewise.
990 (cortex-a57.cortex-a53): Likewise.
992 2014-02-06 Yury Gribov <y.gribov@samsung.com>
993 Kugan Vivekanandarajah <kuganv@linaro.org>
995 * config/arm/arm.c (arm_vector_alignment_reachable): Check
997 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
999 2014-02-06 Richard Biener <rguenther@suse.de>
1001 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1002 set_loop_copy and initialize_original_copy_tables.
1004 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1006 * config/aarch64/aarch64-simd.md
1007 (aarch64_ashr_simddi): Change QI to SI.
1009 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1010 Jakub Jelinek <jakub@redhat.com>
1013 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1016 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1018 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1019 CODE_FOR_altivec_vpku[hw]um to
1020 CODE_FOR_altivec_vpku[hw]um_direct.
1021 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1022 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1023 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1024 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1026 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1028 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1029 generation for -maltivec=be.
1030 (altivec_vsumsws): Simplify redundant test.
1032 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1034 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1035 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1036 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1037 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1038 gen_altivec_vpkuwum.
1039 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1041 (altivec_vpks<VI_char>ss): Likewise.
1042 (altivec_vpks<VI_char>us): Likewise.
1043 (altivec_vpku<VI_char>us): Likewise.
1044 (altivec_vpku<VI_char>um): Likewise.
1045 (altivec_vpku<VI_char>um_direct): New (copy of
1046 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1048 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1049 target is little endian and -maltivec=be is not specified.
1050 (*altivec_vupkhs<VU_char>_direct): New (copy of
1051 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1052 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1053 target is little endian and -maltivec=be is not specified.
1054 (*altivec_vupkls<VU_char>_direct): New (copy of
1055 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1056 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1057 little endian and -maltivec=be is not specified.
1058 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1059 little endian and -maltivec=be is not specified.
1061 2014-02-05 Richard Henderson <rth@redhat.com>
1064 * combine-stack-adj.c: Revert r206943.
1065 * sched-int.h (struct deps_desc): Add last_args_size.
1066 * sched-deps.c (init_deps): Initialize it.
1067 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1068 contain REG_ARGS_SIZE notes.
1070 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1072 * lto-cgraph.c (asm_nodes_output): Make global.
1073 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1074 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1075 (driver_handle_option): Handle OPT_fwpa.
1077 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1080 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1081 a comment typo and formatting issue. If odr_hash hasn't been
1082 created, return vNULL and set *completep to false.
1085 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1086 bb with no successors.
1088 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1091 * doc/invoke.texi (-march): Clarify documentation for ARM.
1095 2014-02-05 Richard Biener <rguenther@suse.de>
1097 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1098 when not vectorizing because of too many alias checks.
1099 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1100 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1102 2014-02-05 Nick Clifton <nickc@redhat.com>
1104 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1105 accept extended registers in any mode when compiling for the MN10300.
1107 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1109 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1110 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1111 sanitization attributes.
1112 (can_inline_edge_p): Likewise.
1113 (sanitize_attrs_match_for_inline_p): New function.
1115 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1117 * ipa-prop.c (detect_type_change): Shor circuit testing of
1118 type changes on THIS pointer.
1120 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1123 * config/pa/pa.c (legitimize_tls_address): Return original address
1124 if not passed a SYMBOL_REF rtx.
1125 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1127 (pa_emit_move_sequence): Simplify TLS source operands.
1128 (pa_legitimate_constant_p): Reject all TLS constants.
1129 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1130 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1132 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1134 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1135 groups when we know they are controlled by LTO.
1136 * varasm.c (default_binds_local_p_1): If object is in other partition,
1137 it will be resolved locally.
1139 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1141 * config/host-linux.c (linux_gt_pch_use_address): Don't
1142 use SSIZE_MAX because it is not always defined.
1144 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1147 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1148 threshold for pseudo splitting.
1149 (update_ebb_live_info): Process call argument hard registers and
1150 hard registers from insn definition too.
1151 (max_small_class_regs_num): New constant.
1152 (inherit_in_ebb): Update live hard regs through EBBs. Update
1153 reloads_num only for small register classes. Don't split for
1156 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1159 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1162 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1164 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1167 2014-02-04 Marek Polacek <polacek@redhat.com>
1169 * gdbinit.in (pel): Define.
1171 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1173 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1176 2014-02-04 Richard Biener <rguenther@suse.de>
1179 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1180 in function context local.
1181 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1182 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1183 similar to LTO_imported_decl_ref.
1185 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1187 PR tree-optimization/60002
1188 * cgraphclones.c (build_function_decl_skip_args): Clear
1191 PR tree-optimization/60023
1192 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1193 false to gsi_replace.
1194 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1195 has been in some EH region and vec_stmt could throw, add
1196 vec_stmt into the same EH region.
1197 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1198 has no lhs, ignore it.
1199 * internal-fn.c (expand_MASK_LOAD): Likewise.
1202 * tree-inline.c (copy_forbidden): Fail for
1203 __attribute__((optimize (0))) functions.
1206 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1207 is set, copy one less argument.
1208 (expand_simd_clones): Don't subtract clone_info->inbranch
1209 from simd_clone_struct_alloc argument.
1211 PR rtl-optimization/57915
1212 * recog.c (simplify_while_replacing): If all unary/binary/relational
1213 operation arguments are constant, attempt to simplify those.
1216 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1217 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1219 2014-02-04 Richard Biener <rguenther@suse.de>
1221 PR tree-optimization/60012
1222 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1223 TBAA disambiguation to all DDRs.
1225 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1228 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1229 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1231 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1234 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1236 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1238 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1239 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1241 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1244 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1245 to figure out targets of polymorphic calls with known decl.
1246 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1247 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1248 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1249 (get_polymorphic_call_info): ... here.
1250 (get_polymorphic_call_info_from_invariant): New function.
1252 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1254 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1255 lookup via vtable pointer; check for type consistency
1256 and turn inconsitent facts into UNREACHABLE.
1257 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1258 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1259 type inconsistent querries; return UNREACHABLE instead.
1261 2014-02-03 Richard Henderson <rth@twiddle.net>
1264 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1265 already processed this node.
1266 (normalize_one_pred_1): Pass along mark_set.
1267 (normalize_one_pred): Create and destroy a pointer_set_t.
1268 (normalize_one_pred_chain): Likewise.
1270 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1272 PR gcov-profile/58602
1273 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1275 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1278 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1279 -fno-devirtualize; try to devirtualize by the knowledge of
1280 virtual table pointer given by aggregate propagation.
1281 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1282 (ipa_print_node_jump_functions): Dump also offset that
1283 is relevant for polymorphic calls.
1284 (determine_known_aggregate_parts): Add arg_type parameter; use it
1285 instead of determining the type from pointer type.
1286 (ipa_compute_jump_functions_for_edge): Update call of
1287 determine_known_aggregate_parts.
1288 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1289 (gimple_get_virt_method_for_binfo): ... here; simplify using
1290 vtable_pointer_value_to_vtable.
1291 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1292 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1293 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1294 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1296 (vtable_pointer_value_to_binfo): ... here.
1297 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1299 2014-02-03 Teresa Johnson <tejohnson@google.com>
1301 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1302 redef of outer loop index variable.
1304 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1308 * doc/extend.texi (Function Attributes): Typo.
1310 2014-02-03 Cong Hou <congh@google.com>
1312 PR tree-optimization/60000
1313 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1314 if the vectorized statement is a store. A store statement can only
1315 appear at the end of pattern statements.
1317 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1319 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1320 (ix86_option_override_internal): Default long double to 64-bit for
1321 32-bit Bionic and to 128-bit for 64-bit Bionic.
1323 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1324 TARGET_LONG_DOUBLE_128 is true.
1325 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1327 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1328 (mlong-double-64): Negate -mlong-double-128.
1329 (mlong-double-128): New option.
1331 * config/i386/i386-c.c (ix86_target_macros): Define
1332 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1334 * doc/invoke.texi: Document -mlong-double-128.
1336 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1338 PR rtl-optimization/60024
1339 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1341 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1343 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1345 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1347 PR rtl-optimization/57662
1348 * sel-sched.c (code_motion_path_driver): Do not mark already not
1349 existing blocks in the visiting bitmap.
1351 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1353 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1354 on the insn being emitted.
1356 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1357 Will Deacon <will.deacon@arm.com>
1359 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1361 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1363 * config/arm/arm-tables.opt: Regenerate.
1365 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1367 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1368 for vector types other than V16QImode.
1369 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1370 define_expand, and call altivec_expand_vec_perm_le when producing
1371 code with little endian element order.
1372 (*altivec_vperm_<mode>_internal): New insn having previous
1373 behavior of altivec_vperm_<mode>.
1374 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1375 altivec_expand_vec_perm_le when producing code with little endian
1377 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1378 behavior of altivec_vperm_<mode>_uns.
1380 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1382 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1383 (altivec_vsumsws): Add handling for -maltivec=be with a little
1385 (altivec_vsumsws_direct): New.
1386 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1387 gen_altivec_vsumsws.
1389 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1391 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1392 vtable_pointer_value_to_binfo): New functions.
1393 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1394 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1396 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1398 * config/nios2/nios2.md (load_got_register): Initialize GOT
1399 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1400 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1402 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1404 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1405 preserverd by passthrough, do not propagate the type.
1407 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1409 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1410 (mips_atomic_assign_expand_fenv): New function.
1411 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1413 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1415 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1416 (__builtin_mips_set_fcsr): Likewise.
1417 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1418 MIPS_USI_FTYPE_VOID.
1419 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1420 (mips16_expand_set_fcsr): Likewise.
1421 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1422 (mips16_set_fcsr_stub): Likewise.
1423 (mips16_get_fcsr_one_only_stub): New class.
1424 (mips16_set_fcsr_one_only_stub): Likewise.
1425 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1426 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1427 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1428 (hard_float): New availability predicate.
1429 (mips_builtins): Add get_fcsr and set_fcsr.
1430 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1431 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1432 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1433 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1434 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1437 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1439 * config/mips/mips.c (mips_one_only_stub): New class.
1440 (mips_need_mips16_rdhwr_p): Replace with...
1441 (mips16_rdhwr_stub): ...this new variable.
1442 (mips16_stub_call_address): New function.
1443 (mips16_rdhwr_one_only_stub): New class.
1444 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1445 (mips_output_mips16_rdhwr): Delete.
1446 (mips_finish_stub): New function.
1447 (mips_code_end): Use it to handle rdhwr stubs.
1449 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1452 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1453 when calculating size of integer atomic types.
1455 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1457 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1459 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1461 PR tree-optimization/60003
1462 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1463 * profile.c (branch_prob): Use gimple_call_builtin_p
1464 to check for BUILT_IN_SETJMP_RECEIVER.
1465 * tree-inline.c (copy_bb): Call notice_special_calls.
1467 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1470 * lra-constraints.c (process_alt_operands): Update reload_sum only
1473 2014-01-31 Richard Henderson <rth@redhat.com>
1476 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1477 until after else_eh is processed.
1479 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1481 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1482 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1483 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1484 in smmintrin.h, remove them.
1485 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1486 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1487 * config/i386/i386.md (ROUND_SAE): Fix value.
1488 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1489 (const48_operand): New.
1490 * config/i386/subst.md (round), (round_expand): Use
1491 const_4_or_8_to_11_operand.
1492 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1494 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1496 * config/i386/constraints.md (Yk): Swap meaning with k.
1497 * config/i386/i386.md (movhi_internal): Change Yk to k.
1498 (movqi_internal): Ditto.
1499 (*k<logic><mode>): Ditto.
1502 (kandn<mode>): Ditto.
1503 (*<code>hi_1): Ditto.
1504 (*<code>qi_1): Ditto.
1505 (kxnor<mode>): Ditto.
1506 (kortestzhi): Ditto.
1507 (kortestchi): Ditto.
1509 (*one_cmplhi2_1): Ditto.
1510 (*one_cmplqi2_1): Ditto.
1511 * config/i386/sse.md (): Change k to Yk.
1512 (avx512f_load<mode>_mask): Ditto.
1513 (avx512f_blendm<mode>): Ditto.
1514 (avx512f_store<mode>_mask): Ditto.
1515 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1516 (avx512f_storedqu<mode>_mask): Ditto.
1517 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1519 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1520 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1521 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1522 (avx512f_maskcmp<mode>3): Ditto.
1523 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1524 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1525 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1526 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1527 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1528 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1529 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1530 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1531 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1532 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1533 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1534 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1535 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1536 (vec_extract_lo_<mode>_maskm): Ditto.
1537 (vec_extract_hi_<mode>_maskm): Ditto.
1538 (avx512f_vternlog<mode>_mask): Ditto.
1539 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1540 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1541 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1542 (avx512f_<code>v8div16qi2_mask): Ditto.
1543 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1544 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1545 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1546 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1547 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1548 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1549 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1550 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1551 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1552 (avx512cd_maskb_vec_dupv8di): Ditto.
1553 (avx512cd_maskw_vec_dupv16si): Ditto.
1554 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1555 (avx512f_vpermi2var<mode>3_mask): Ditto.
1556 (avx512f_vpermi2var<mode>3_mask): Ditto.
1557 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1558 (*avx512f_gathersi<mode>): Ditto.
1559 (*avx512f_gathersi<mode>_2): Ditto.
1560 (*avx512f_gatherdi<mode>): Ditto.
1561 (*avx512f_gatherdi<mode>_2): Ditto.
1562 (*avx512f_scattersi<mode>): Ditto.
1563 (*avx512f_scatterdi<mode>): Ditto.
1564 (avx512f_compress<mode>_mask): Ditto.
1565 (avx512f_compressstore<mode>_mask): Ditto.
1566 (avx512f_expand<mode>_mask): Ditto.
1567 * config/i386/subst.md (mask): Change k to Yk.
1568 (mask_scalar_merge): Ditto.
1571 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1573 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1575 2014-01-31 Richard Biener <rguenther@suse.de>
1578 * builtins.c (fold_builtin_memory_op): Make sure to not
1579 use a floating-point mode or a boolean or enumeral type for
1582 2014-01-30 DJ Delorie <dj@redhat.com>
1584 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1585 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1586 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1587 whenever main() has an epilogue.
1589 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1591 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1592 unused variable "field".
1593 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1594 (vsx_mergeh_<mode>): Likewise.
1595 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1596 (altivec_vmrghh): Likewise.
1597 (altivec_vmrghw): Likewise.
1598 (altivec_vmrglb): Likewise.
1599 (altivec_vmrglh): Likewise.
1600 (altivec_vmrglw): Likewise.
1601 (altivec_vspltb): Add missing uses.
1602 (altivec_vsplth): Likewise.
1603 (altivec_vspltw): Likewise.
1604 (altivec_vspltsf): Likewise.
1606 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1609 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1610 frame related instructions.
1612 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1614 PR rtl-optimization/59959
1615 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1616 any reload of register whose subreg is invalid.
1618 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1620 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1621 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1622 Add missing return type - void.
1624 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1626 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1627 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1628 remove element index adjustment for endian (now handled in vsx.md
1630 (altivec_expand_vec_perm_const): Use
1631 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1632 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1633 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1634 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1635 define_expand and a new define_insn *altivec_vspltb_internal;
1636 adjust for -maltivec=be on a little endian target.
1637 (altivec_vspltb_direct): New.
1638 (altivec_vsplth): Divide into a define_expand and a new
1639 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1640 little endian target.
1641 (altivec_vsplth_direct): New.
1642 (altivec_vspltw): Divide into a define_expand and a new
1643 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1644 little endian target.
1645 (altivec_vspltw_direct): New.
1646 (altivec_vspltsf): Divide into a define_expand and a new
1647 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1648 a little endian target.
1650 2014-01-30 Richard Biener <rguenther@suse.de>
1652 PR tree-optimization/59993
1653 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1654 can propagate form the earlier stmt and avoid the transform
1655 when the intermediate result is needed.
1657 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1659 * README.Portability: Fix typo.
1661 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1663 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1664 comparison_operator with ordered_comparison_operator.
1666 2014-01-30 Nick Clifton <nickc@redhat.com>
1668 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1669 Rename to mn10300_store_multiple_regs.
1670 * config/mn10300/mn10300.c: Likewise.
1671 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1672 store_multiple_regs.
1673 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1674 Call mn10300_store_multiple_regs.
1676 2014-01-30 Nick Clifton <nickc@redhat.com>
1677 DJ Delorie <dj@redhat.com>
1679 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1680 %fp 2 to keep registers after it properly word-aligned.
1681 (rl78_alloc_physical_registers_umul): Handle the case where both
1682 input operands are the same.
1684 2014-01-30 Richard Biener <rguenther@suse.de>
1686 PR tree-optimization/59903
1687 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1690 2014-01-30 Jason Merrill <jason@redhat.com>
1693 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1696 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1698 2014-01-30 Richard Biener <rguenther@suse.de>
1700 PR tree-optimization/59951
1701 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1703 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1706 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1707 SFmode to DFmode case.
1709 2014-01-29 DJ Delorie <dj@redhat.com>
1711 * config/msp430/msp430.opt (-minrt): New.
1712 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1714 (ENDFILE_SPEC): Likewise.
1716 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1718 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1719 (estimate_function_body_sizes): Use it.
1721 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1724 * dwarf2out.c (is_cxx_auto): New.
1725 (is_base_type): Use it.
1726 (gen_type_die_with_usage): Likewise.
1728 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1730 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1731 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1732 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1733 -maltivec=be with LE targets.
1734 (vsx_mergeh_<mode>): Likewise.
1735 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1736 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1737 (altivec_vmrghb): Replace with define_expand and new
1738 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1739 (altivec_vmrghb_direct): New define_insn.
1740 (altivec_vmrghh): Replace with define_expand and new
1741 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1742 (altivec_vmrghh_direct): New define_insn.
1743 (altivec_vmrghw): Replace with define_expand and new
1744 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1745 (altivec_vmrghw_direct): New define_insn.
1746 (*altivec_vmrghsf): Adjust for endianness.
1747 (altivec_vmrglb): Replace with define_expand and new
1748 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1749 (altivec_vmrglb_direct): New define_insn.
1750 (altivec_vmrglh): Replace with define_expand and new
1751 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1752 (altivec_vmrglh_direct): New define_insn.
1753 (altivec_vmrglw): Replace with define_expand and new
1754 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1755 (altivec_vmrglw_direct): New define_insn.
1756 (*altivec_vmrglsf): Adjust for endianness.
1757 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1758 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1759 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1760 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1761 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1762 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1763 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1764 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1766 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1768 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1769 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1772 2014-01-29 Richard Biener <rguenther@suse.de>
1774 PR tree-optimization/58742
1775 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1776 associate_pointerplus_align.
1777 (associate_pointerplus_diff): New function.
1778 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1779 and associate_pointerplus_diff.
1781 2014-01-29 Richard Biener <rguenther@suse.de>
1783 * lto-streamer.h (LTO_major_version): Bump to 3.
1784 (LTO_minor_version): Reset to 0.
1786 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1788 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1789 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1790 (arm_file_start): Generate correct asm header for armv7ve.
1791 * config/arm/bpabi.h: Add multilib support for armv7ve.
1792 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1793 and cortex-a15 to armv7ve.
1794 * config/arm/t-aprofile: Add multilib support for armv7ve.
1795 * doc/invoke.texi: Document -march=armv7ve.
1797 2014-01-29 Richard Biener <rguenther@suse.de>
1799 PR tree-optimization/58742
1800 * tree-ssa-forwprop.c (associate_plusminus): Return true
1801 if we changed sth, defer EH cleanup to ...
1802 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1803 (simplify_mult): New function.
1805 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1808 PR tree-optimization/59920
1809 * tree.c (build_common_builtin_nodes): Remove
1810 __builtin_setjmp_dispatcher initialization.
1811 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1812 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1813 instead of gsi_after_labels + manually skipping debug stmts.
1814 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1815 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1816 * tree-inline.c (copy_edges_for_bb): Remove
1817 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1818 argument. Ignore computed_goto_p stmts. Don't call
1819 make_abnormal_goto_edges. If a call might need abnormal edges
1820 for non-local gotos, see if it already has an edge to
1821 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1822 with true argument, don't do anything then, otherwise add
1823 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1824 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1826 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1827 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1828 (lower_stmt): Don't set data->calls_builtin_setjmp.
1829 (lower_builtin_setjmp): Adjust comment.
1830 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1831 * tree-cfg.c (found_computed_goto): Remove.
1832 (factor_computed_gotos): Remove.
1833 (make_goto_expr_edges): Return bool, true for computed gotos.
1834 Don't call make_abnormal_goto_edges.
1835 (build_gimple_cfg): Don't set found_computed_goto, don't call
1836 factor_computed_gotos.
1837 (computed_goto_p): No longer static.
1838 (make_blocks): Don't set found_computed_goto.
1839 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1840 (make_edges): If make_goto_expr_edges returns true, push bb
1841 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1842 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1843 vector. Record mapping between bbs and OpenMP regions if there
1844 are any, adjust make_gimple_omp_edges caller. Call
1845 handle_abnormal_edges.
1846 (make_abnormal_goto_edges): Remove.
1847 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1848 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1849 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1850 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1851 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1852 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1853 filling *region also set *region_idx to (*region)->entry->index.
1856 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1857 For REGs set ORIGINAL_REGNO.
1859 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1861 * doc/md.texi: Mention that a target shouldn't implement
1862 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1865 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1867 PR tree-optimization/59594
1868 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1869 a copy of the datarefs vector rather than the vector itself.
1871 2014-01-28 Jason Merrill <jason@redhat.com>
1874 * dwarf2out.c (auto_die): New static.
1875 (gen_type_die_with_usage): Handle C++1y 'auto'.
1876 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1879 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1882 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1883 (SPEC_X32): Likewise.
1884 (SPEC_64): Likewise.
1885 * config/i386/i386.c (ix86_option_override_internal): Turn off
1886 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1888 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1889 * config/i386/i386.h (TARGET_16BIT): New macro.
1890 (TARGET_16BIT_P): Likewise.
1891 * config/i386/i386.opt: Add m16.
1892 * doc/invoke.texi: Document -m16.
1894 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1896 PR preprocessor/59935
1897 * input.c (location_get_source_line): Bail out on when line number
1898 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1900 2014-01-28 Richard Biener <rguenther@suse.de>
1902 PR tree-optimization/58742
1903 * tree-ssa-forwprop.c (associate_plusminus): Handle
1904 pointer subtraction of the form (T)(P + A) - (T)P.
1906 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1908 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1911 2014-01-28 Richard Biener <rguenther@suse.de>
1914 2014-01-28 Richard Biener <rguenther@suse.de>
1916 PR rtl-optimization/45364
1917 PR rtl-optimization/59890
1918 * var-tracking.c (local_get_addr_clear_given_value): Handle
1919 already cleared slot.
1920 (val_reset): Handle not allocated local_get_addr_cache.
1921 (vt_find_locations): Use post-order on the inverted CFG.
1923 2014-01-28 Richard Biener <rguenther@suse.de>
1925 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1927 2014-01-28 Richard Biener <rguenther@suse.de>
1929 PR rtl-optimization/45364
1930 PR rtl-optimization/59890
1931 * var-tracking.c (local_get_addr_clear_given_value): Handle
1932 already cleared slot.
1933 (val_reset): Handle not allocated local_get_addr_cache.
1934 (vt_find_locations): Use post-order on the inverted CFG.
1936 2014-01-28 Alan Modra <amodra@gmail.com>
1938 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1939 * configure.ac <recursive call for build != host>: Define
1940 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1941 and LD_FOR_BUILD too.
1942 * configure: Regenerate.
1944 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1946 * config/i386/i386.c (get_builtin_code_for_version): Separate
1947 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1948 Broadwell from Haswell.
1950 2014-01-27 Steve Ellcey <sellcey@mips.com>
1952 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1953 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1954 * config/mips/mips.c (mips_option_override): Change setting
1956 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1957 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1958 Change from Mask to Var.
1960 2014-01-27 Jeff Law <law@redhat.com>
1962 * ipa-inline.c (inline_small_functions): Fix typo.
1964 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1966 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1967 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1968 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1969 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1970 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1971 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1972 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1973 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1974 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1975 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1976 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1977 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1978 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1979 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1980 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1981 (_mm512_storeu_epi64): Ditto.
1982 (_mm512_cmpge_epi32_mask): Ditto.
1983 (_mm512_cmpge_epu32_mask): Ditto.
1984 (_mm512_cmpge_epi64_mask): Ditto.
1985 (_mm512_cmpge_epu64_mask): Ditto.
1986 (_mm512_cmple_epi32_mask): Ditto.
1987 (_mm512_cmple_epu32_mask): Ditto.
1988 (_mm512_cmple_epi64_mask): Ditto.
1989 (_mm512_cmple_epu64_mask): Ditto.
1990 (_mm512_cmplt_epi32_mask): Ditto.
1991 (_mm512_cmplt_epu32_mask): Ditto.
1992 (_mm512_cmplt_epi64_mask): Ditto.
1993 (_mm512_cmplt_epu64_mask): Ditto.
1994 (_mm512_cmpneq_epi32_mask): Ditto.
1995 (_mm512_cmpneq_epu32_mask): Ditto.
1996 (_mm512_cmpneq_epi64_mask): Ditto.
1997 (_mm512_cmpneq_epu64_mask): Ditto.
1998 (_mm512_expand_pd): Ditto.
1999 (_mm512_expand_ps): Ditto.
2000 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2001 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2002 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2003 * config/i386/i386.c (ix86_builtins): Add
2004 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2005 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2006 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2007 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2008 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2009 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2010 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2011 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2012 IX86_BUILTIN_PMOVUSQW512_MEM.
2013 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2014 __builtin_ia32_pmovsqd512mem_mask,
2015 __builtin_ia32_pmovqd512mem_mask,
2016 __builtin_ia32_pmovusqw512mem_mask,
2017 __builtin_ia32_pmovsqw512mem_mask,
2018 __builtin_ia32_pmovqw512mem_mask,
2019 __builtin_ia32_pmovusdw512mem_mask,
2020 __builtin_ia32_pmovsdw512mem_mask,
2021 __builtin_ia32_pmovdw512mem_mask,
2022 __builtin_ia32_pmovqb512mem_mask,
2023 __builtin_ia32_pmovusqb512mem_mask,
2024 __builtin_ia32_pmovsqb512mem_mask,
2025 __builtin_ia32_pmovusdb512mem_mask,
2026 __builtin_ia32_pmovsdb512mem_mask,
2027 __builtin_ia32_pmovdb512mem_mask.
2028 (bdesc_args): Add __builtin_ia32_expanddf512,
2029 __builtin_ia32_expandsf512.
2030 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2031 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2032 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2033 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2034 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2035 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2036 (avx512f_<code>v8div16qi2_mask_store): This.
2037 (avx512f_expand<mode>): New.
2039 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2041 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2043 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2044 (_mm512_prefetch_i32scatter_pd): Ditto.
2045 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2046 (_mm512_prefetch_i64scatter_pd): Ditto.
2047 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2048 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2049 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2050 (_mm512_prefetch_i32scatter_ps): Ditto.
2051 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2052 (_mm512_prefetch_i64scatter_ps): Ditto.
2053 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2054 * config/i386/i386-builtin-types.def: Define
2055 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2056 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2057 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2058 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2059 IX86_BUILTIN_SCATTERPFQPD.
2060 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2061 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2062 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2063 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2064 __builtin_ia32_scatterpfqps.
2065 (ix86_expand_builtin): Expand new built-ins.
2066 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2067 fix memory access data type.
2068 (*avx512pf_gatherpf<mode>_mask): Ditto.
2069 (*avx512pf_gatherpf<mode>): Ditto.
2070 (avx512pf_scatterpf<mode>): Ditto.
2071 (*avx512pf_scatterpf<mode>_mask): Ditto.
2072 (*avx512pf_scatterpf<mode>): Ditto.
2073 (GATHER_SCATTER_SF_MEM_MODE): New.
2074 (avx512pf_gatherpf<mode>df): Ditto.
2075 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2076 (*avx512pf_scatterpf<mode>df): Ditto.
2078 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2081 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2082 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2085 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2087 * common/config/arm/arm-common.c
2088 (arm_rewrite_mcpu): Handle multiple names.
2090 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2092 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2094 * gimple-builder.h (create_gimple_tmp): Delete.
2096 2014-01-27 Christian Bruel <christian.bruel@st.com>
2098 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2101 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2103 * config/pa/pa.md (call): Generate indirect long calls to non-local
2104 functions when outputing 32-bit code.
2105 (call_value): Likewise except for special call to buggy powf function.
2107 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2108 portable runtime and PIC indirect calls.
2109 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2110 and PIC call sequences. Use ldo instead of blr to set return register
2111 in PIC call sequence.
2113 2014-01-25 Walter Lee <walt@tilera.com>
2115 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2116 avoid clobbering a live register.
2118 2014-01-25 Walter Lee <walt@tilera.com>
2120 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2121 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2122 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2123 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2125 2014-01-25 Walter Lee <walt@tilera.com>
2127 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2128 arguments on even registers.
2129 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2131 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2132 (BIGGEST_ALIGNMENT): Ditto.
2133 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2135 2014-01-25 Walter Lee <walt@tilera.com>
2137 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2138 insns before bundling.
2139 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2141 2014-01-25 Walter Lee <walt@tilera.com>
2143 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2144 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2145 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2147 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2149 * config/mips/constraints.md (kl): Delete.
2150 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2151 define expands, using...
2152 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2153 instructions for MIPS16.
2154 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2155 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2157 2014-01-25 Walter Lee <walt@tilera.com>
2159 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2163 2014-01-25 Walter Lee <walt@tilera.com>
2165 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2166 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2168 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2170 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2173 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2175 * print-rtl.c (in_call_function_usage): New var.
2176 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2177 EXPR_LIST mode as mode and not as reg note name.
2180 * cfgloopmanip.c (copy_loop_info): If
2181 loop->warned_aggressive_loop_optimizations, make sure
2182 the flag is set in target loop too.
2184 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2186 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2188 * builtins.def: Likewise.
2189 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2190 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2191 * ira.c (ira_setup_eliminable_regset): Likewise.
2192 * omp-low.c (gate_expand_omp): Likewise.
2193 (execute_lower_omp): Likewise.
2194 (diagnose_sb_0): Likewise.
2195 (gate_diagnose_omp_blocks): Likewise.
2196 (simd_clone_clauses_extract): Likewise.
2199 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2201 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2202 correction for little endian...
2203 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2206 2014-01-24 Jeff Law <law@redhat.com>
2208 PR tree-optimization/59919
2209 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2210 for non-returning calls.
2212 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2214 * common/config/aarch64/aarch64-common.c
2215 (aarch64_rewrite_mcpu): Handle multiple names.
2216 * config/aarch64/aarch64.h
2217 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2219 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2221 * input.c (add_file_to_cache_tab): Handle the case where fopen
2224 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2227 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2228 from push operand if code of push isn't PRE_DEC.
2230 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2233 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2234 -mquad-memory-atomic. Update -mquad-memory documentation to say
2235 it is only used for non-atomic loads/stores.
2237 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2238 -mquad-memory or -mquad-memory-atomic switches.
2240 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2241 -mquad-memory-atomic to ISA 2.07 support.
2243 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2244 to separate support of normal quad word memory operations (ldq, stq)
2245 from the atomic quad word memory operations.
2247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2248 support to separate non-atomic quad word operations from atomic
2249 quad word operations. Disable non-atomic quad word operations in
2250 little endian mode so that we don't have to swap words after the
2251 load and before the store.
2252 (quad_load_store_p): Add comment about atomic quad word support.
2253 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2254 options printed with -mdebug=reg.
2256 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2257 -mquad-memory-atomic as the test for whether we have quad word
2258 atomic instructions.
2259 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2260 or -mp8-vector are used, allow byte/half-word atomic operations.
2262 * config/rs6000/sync.md (load_lockedti): Insure that the address
2263 is a proper indexed or indirect address for the lqarx instruction.
2264 On little endian systems, swap the hi/lo registers after the lqarx
2266 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2267 insure the address is valid for the lqarx instruction.
2268 (store_conditionalti): Insure that the address is a proper indexed
2269 or indirect address for the stqcrx. instruction. On little endian
2270 systems, swap the hi/lo registers before doing the stqcrx.
2272 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2273 insure the address is valid for the stqcrx. instruction.
2275 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2276 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2277 type of quad memory support is available.
2279 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2282 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2283 there is a danger of looping.
2285 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2288 force flag_ira_loop_pressure if set via command line.
2290 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2292 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2293 (ashr_simd): New builtin handling DI mode.
2294 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2295 (aarch64_sshr_simddi): New match pattern.
2296 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2297 (vshrd_n_s64): Likewise.
2298 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2300 2014-01-23 Nick Clifton <nickc@redhat.com>
2302 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2303 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2304 favour of mcu specific scripts.
2305 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2308 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2309 Alex Velenko <Alex.Velenko@arm.com>
2311 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2312 (vaddv_s16): Likewise.
2313 (vaddv_s32): Likewise.
2314 (vaddv_u8): Likewise.
2315 (vaddv_u16): Likewise.
2316 (vaddv_u32): Likewise.
2317 (vaddvq_s8): Likewise.
2318 (vaddvq_s16): Likewise.
2319 (vaddvq_s32): Likewise.
2320 (vaddvq_s64): Likewise.
2321 (vaddvq_u8): Likewise.
2322 (vaddvq_u16): Likewise.
2323 (vaddvq_u32): Likewise.
2324 (vaddvq_u64): Likewise.
2325 (vaddv_f32): Likewise.
2326 (vaddvq_f32): Likewise.
2327 (vaddvq_f64): Likewise.
2328 (vmaxv_f32): Likewise.
2329 (vmaxv_s8): Likewise.
2330 (vmaxv_s16): Likewise.
2331 (vmaxv_s32): Likewise.
2332 (vmaxv_u8): Likewise.
2333 (vmaxv_u16): Likewise.
2334 (vmaxv_u32): Likewise.
2335 (vmaxvq_f32): Likewise.
2336 (vmaxvq_f64): Likewise.
2337 (vmaxvq_s8): Likewise.
2338 (vmaxvq_s16): Likewise.
2339 (vmaxvq_s32): Likewise.
2340 (vmaxvq_u8): Likewise.
2341 (vmaxvq_u16): Likewise.
2342 (vmaxvq_u32): Likewise.
2343 (vmaxnmv_f32): Likewise.
2344 (vmaxnmvq_f32): Likewise.
2345 (vmaxnmvq_f64): Likewise.
2346 (vminv_f32): Likewise.
2347 (vminv_s8): Likewise.
2348 (vminv_s16): Likewise.
2349 (vminv_s32): Likewise.
2350 (vminv_u8): Likewise.
2351 (vminv_u16): Likewise.
2352 (vminv_u32): Likewise.
2353 (vminvq_f32): Likewise.
2354 (vminvq_f64): Likewise.
2355 (vminvq_s8): Likewise.
2356 (vminvq_s16): Likewise.
2357 (vminvq_s32): Likewise.
2358 (vminvq_u8): Likewise.
2359 (vminvq_u16): Likewise.
2360 (vminvq_u32): Likewise.
2361 (vminnmv_f32): Likewise.
2362 (vminnmvq_f32): Likewise.
2363 (vminnmvq_f64): Likewise.
2365 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2367 * config/aarch64/aarch64-simd.md
2368 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2369 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2370 (*aarch64_mul3_elt<mode>): Likewise.
2371 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2372 (*aarch64_mul3_elt_to_64v2df): Likewise.
2373 (*aarch64_mla_elt<mode>): Likewise.
2374 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2375 (*aarch64_mls_elt<mode>): Likewise.
2376 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2377 (*aarch64_fma4_elt<mode>): Likewise.
2378 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2379 (*aarch64_fma4_elt_to_64v2df): Likewise.
2380 (*aarch64_fnma4_elt<mode>): Likewise.
2381 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2382 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2383 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2384 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2385 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2386 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2387 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2388 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2389 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2391 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2393 * config/aarch64/aarch64-simd.md
2394 (aarch64_be_checked_get_lane<mode>): New define_expand.
2395 * config/aarch64/aarch64-simd-builtins.def
2396 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2397 New builtin definition.
2398 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2399 Use new safe be builtin.
2401 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2403 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2405 (aarch64_be_st1<mode>): Likewise.
2406 (aarch_ld1<VALL:mode>): Define_expand modified.
2407 (aarch_st1<VALL:mode>): Likewise.
2408 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2409 (UNSPEC_ST1): Likewise.
2411 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2413 * config/microblaze/microblaze.md: Add trap insn and attribute
2415 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2417 PR preprocessor/58580
2418 * input.h (location_get_source_line): Take an additional line_size
2420 (void diagnostics_file_cache_fini): Declare new function.
2421 * input.c (struct fcache): New type.
2422 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2423 New static constants.
2424 (diagnostic_file_cache_init, total_lines_num)
2425 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2426 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2427 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2428 (get_next_line, read_next_line, goto_next_line, read_line_num):
2429 New static function definitions.
2430 (diagnostic_file_cache_fini): New function.
2431 (location_get_source_line): Take an additional output line_len
2432 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2434 * diagnostic.c (diagnostic_finish): Call
2435 diagnostic_file_cache_fini.
2436 (adjust_line): Take an additional input parameter for the length
2437 of the line, rather than calculating it with strlen.
2438 (diagnostic_show_locus): Adjust the use of
2439 location_get_source_line and adjust_line with respect to their new
2440 signature. While displaying a line now, do not stop at the first
2441 null byte. Rather, display the zero byte as a space and keep
2442 going until we reach the size of the line.
2443 * Makefile.in: Add vec.o to OBJS-libcommon
2445 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2446 Ilya Tocar <ilya.tocar@intel.com>
2448 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2449 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2450 (__builtin_ia32_kmov16): Ditto.
2451 * config/i386/i386.md (UNSPEC_KMOV): New.
2454 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2456 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2457 (_mm512_storeu_si512): Ditto.
2459 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2462 * rtl.h (get_referenced_operands): Declare.
2463 * recog.c (get_referenced_operands): New function.
2464 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2465 operands have been referenced when recording LO_SUM references.
2467 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2469 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2471 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2473 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2474 Enable for generic and recent AMD targets.
2476 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2478 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2479 ARG_SIZE note when adjustment was eliminated.
2481 2014-01-22 Jeff Law <law@redhat.com>
2483 PR tree-optimization/59597
2484 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2485 in file. Accept new argument REGISTERING and use it to modify
2486 dump output appropriately.
2487 (register_jump_thread): Corresponding changes.
2488 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2489 thread paths involving joiner blocks. Add code to dump cancelled
2490 jump threading paths.
2492 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2494 PR rtl-optimization/59477
2495 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2496 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2498 2014-01-22 Tom Tromey <tromey@redhat.com>
2500 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2502 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2504 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2506 PR rtl-optimization/59896
2507 * lra-constraints.c (process_alt_operands): Check unused note for
2508 matched operands of insn with no output reloads.
2510 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2512 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2513 (mips_move_from_gpr_cost): Likewise.
2515 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2517 PR rtl-optimization/59858
2518 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2519 ira_class_hard_regs_num.
2520 (process_alt_operands): Increase reject for dying matched operand.
2522 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2525 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2526 smaller than size, perform several stores or loads and stores
2527 at dst + count - size to store or copy all of size bytes, rather
2528 than just last modesize bytes.
2530 2014-01-20 DJ Delorie <dj@redhat.com>
2532 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2533 that CLOBBERs are REGs before propogating their values.
2535 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2538 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2539 (cgraph_inline_failed_type): New function.
2540 * cgraph.h (DEFCIFCODE): Add type.
2541 (cgraph_inline_failed_type_t): New enum.
2542 (cgraph_inline_failed_type): New prototype.
2543 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2544 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2545 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2546 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2547 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2548 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2549 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2550 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2551 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2552 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2553 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2554 OPTIMIZATION_MISMATCH.
2555 * tree-inline.c (expand_call_inline): Emit errors during
2556 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2558 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2561 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2562 mode attribute in insn output.
2564 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2566 * output.h (output_constant): Delete.
2567 * varasm.c (output_constant): Make private.
2569 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2571 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2573 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2576 * tree.h (fold_builtin_strcat): New prototype.
2577 * builtins.c (fold_builtin_strcat): No longer static. Add len
2578 argument, if non-NULL, don't call c_strlen. Optimize
2579 directly into __builtin_memcpy instead of __builtin_strcpy.
2580 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2581 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2583 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2585 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2586 for SImode_address_operand operands, having only a REG argument.
2588 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2590 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2591 loader name using mbig-endian.
2592 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2594 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2596 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2600 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2602 * config/aarch64/aarch64-protos.h
2603 (aarch64_cannot_change_mode_class_ptr): Declare.
2604 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2605 aarch64_cannot_change_mode_class_ptr): New.
2606 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2607 backend hook aarch64_cannot_change_mode_class.
2609 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2611 * common/config/aarch64/aarch64-common.c
2612 (aarch64_handle_option): Don't handle any option order logic here.
2613 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2614 selected_cpu, warn on architecture version mismatch.
2615 (aarch64_override_options): Fix parsing order for option strings.
2617 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2618 Iain Sandoe <iain@codesourcery.com>
2621 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2622 warning. Amend comment to reflect current functionality.
2624 2014-01-20 Richard Biener <rguenther@suse.de>
2627 * builtins.c (fold_builtin_strcat): Remove case better handled
2628 by tree-ssa-strlen.c.
2630 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2632 * config/aarch64/aarch64.opt
2633 (mcpu, march, mtune): Make case-insensitive.
2635 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2638 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2639 if operands[1] is a REG or ZERO_EXTEND of a REG.
2641 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2643 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2645 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2647 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2648 long non-pic millicode calls.
2650 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2652 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2654 2014-01-19 Kito Cheng <kito@0xlab.org>
2656 * builtins.c (expand_movstr): Check movstr expand done or fail.
2658 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2659 H.J. Lu <hongjiu.lu@intel.com>
2662 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2663 to DImode for zero-extended addresses.
2665 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2667 PR rtl-optimization/57763
2668 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2669 on the new indirect jump_insn and increment LABEL_NUSES (label).
2671 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2675 * config.gcc (x86_archs): New variable.
2676 (x86_64_archs): Likewise.
2677 (x86_cpus): Likewise.
2678 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2679 --with-arch/--with-cpu= options.
2680 Support --with-arch=/--with-cpu={nehalem,westmere,
2681 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2683 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2685 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2686 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2688 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2690 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2692 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2695 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2696 clear cpp_get_options (parse_in)->warn_unused_macros for
2697 ix86_target_macros_internal with cpp_define.
2699 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2701 * jump.c (delete_related_insns): Keep (use (insn))s.
2702 * reorg.c (redundant_insn): Check for barriers too.
2704 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2706 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2708 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2710 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2711 call to $$dyncall when TARGET_LONG_CALLS is true.
2713 2014-01-17 Jeff Law <law@redhat.com>
2715 * ree.c (combine_set_extension): Temporarily disable test for
2716 changing number of hard registers.
2718 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2721 * ipa-inline-analysis.c (inline_free_summary):
2722 Do not free summary of aliases.
2724 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2727 * gimplify.c (gimplify_expr): Use create_tmp_var
2728 instead of create_tmp_var_raw. If cond doesn't have
2729 integral type, don't add the IFN_ANNOTATE builtin at all.
2731 2014-01-17 Martin Jambor <mjambor@suse.cz>
2734 * ipa-cp.c (prev_edge_clone): New variable.
2735 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2736 Also resize prev_edge_clone vector.
2737 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2738 (ipcp_edge_removal_hook): New function.
2739 (ipcp_driver): Register ipcp_edge_removal_hook.
2741 2014-01-17 Andrew Pinski <apinski@cavium.com>
2742 Steve Ellcey <sellcey@mips.com>
2745 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2748 2014-01-17 Jeff Law <law@redhat.com>
2751 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2752 so that pass_ccp runs first.
2754 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2756 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2757 (ix86_adjust_cost): Use !TARGET_XXX.
2758 (do_reorder_for_imul): Likewise.
2759 (swap_top_of_ready_list): Likewise.
2760 (ix86_sched_reorder): Likewise.
2762 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2764 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2765 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2766 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2767 (intel_memset): New. Duplicate slm_memset.
2768 (intel_cost): New. Duplicate slm_cost.
2769 (m_INTEL): New macro.
2770 (processor_target_table): Add "intel".
2771 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2772 with PROCESSOR_INTEL for "intel".
2773 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2774 PROCESSOR_SILVERMONT.
2775 (ix86_issue_rate): Likewise.
2776 (ix86_adjust_cost): Likewise.
2777 (ia32_multipass_dfa_lookahead): Likewise.
2778 (swap_top_of_ready_list): Likewise.
2779 (ix86_sched_reorder): Likewise.
2780 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2781 instead of TARGET_OPT_AGU.
2782 * config/i386/i386.h (TARGET_INTEL): New.
2783 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2784 (processor_type): Add PROCESSOR_INTEL.
2785 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2786 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2788 2014-01-17 Marek Polacek <polacek@redhat.com>
2791 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2794 2014-01-17 Richard Biener <rguenther@suse.de>
2796 PR tree-optimization/46590
2797 * opts.c (default_options_table): Add entries for
2798 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2799 all enabled at -O1 but not for -Og.
2800 * common.opt (fbranch-count-reg): Remove Init(1).
2801 (fmove-loop-invariants): Likewise.
2802 (ftree-pta): Likewise.
2804 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2806 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2807 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2808 decls to at least the GCC 4.8 used alignments.
2811 * tree-nested.c (convert_nonlocal_reference_stmt,
2812 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2813 of GIMPLE_BIND stmts, adjust associated decls.
2815 2014-01-17 Richard Biener <rguenther@suse.de>
2817 PR tree-optimization/46590
2818 * vec.h (vec<>::bseach): New member function implementing
2819 binary search according to C89 bsearch.
2820 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2821 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2822 bitmap pointer again. Make accesses_in_loop a flat array.
2823 (mem_ref_obstack): New global.
2824 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2825 (mark_ref_stored): Likewise.
2826 (ref_indep_loop_p_2): Likewise.
2827 (set_ref_stored_in_loop): New helper function.
2828 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2829 (memref_free): Adjust.
2830 (record_mem_ref_loc): Simplify.
2831 (gather_mem_refs_stmt): Adjust.
2832 (sort_locs_in_loop_postorder_cmp): New function.
2833 (analyze_memory_references): Sort accesses_in_loop after
2834 loop postorder number.
2835 (find_ref_loc_in_loop_cmp): New function.
2836 (for_all_locs_in_loop): Find relevant cluster of locs in
2837 accesses_in_loop and iterate without recursion.
2838 (execute_sm): Avoid uninit warning.
2839 (struct ref_always_accessed): Simplify.
2840 (ref_always_accessed::operator ()): Likewise.
2841 (ref_always_accessed_p): Likewise.
2842 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2843 loop postorder numbers here.
2844 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2847 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2850 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2851 on decls for which assemble_alias has been called.
2853 2014-01-17 Nick Clifton <nickc@redhat.com>
2855 * config/msp430/msp430.opt: (mcpu): New option.
2856 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2857 (msp430_option_override): Parse target_cpu. If the MCU name
2858 matches a generic string, clear target_mcu.
2859 (msp430_attr): Allow numeric interrupt values up to 63.
2860 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2861 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2863 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2865 * config/msp430/msp430.md (popm): Use %J rather than %I.
2866 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2867 (addhi_cy_i): Use immediate_operand for operand 2.
2868 * doc/invoke.texi: Document -mcpu option.
2870 2014-01-17 Richard Biener <rguenther@suse.de>
2872 PR rtl-optimization/38518
2873 * df.h (df_analyze_loop): Declare.
2874 * df-core.c: Include cfgloop.h.
2875 (df_analyze_1): Split out main part of df_analyze.
2876 (df_analyze): Adjust.
2877 (loop_inverted_post_order_compute): New function.
2878 (loop_post_order_compute): Likewise.
2879 (df_analyze_loop): New function avoiding whole-function
2881 * loop-invariant.c (find_defs): Use df_analyze_loop.
2882 (find_invariants): Adjust.
2883 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2885 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2887 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2888 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2890 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2892 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2893 traversal when removing references.
2895 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2898 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2900 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2903 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2904 pushing a reload for an autoinc when we had previously reloaded an
2905 inner part of the address.
2907 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2909 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2911 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2912 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2913 when not giving up or versioning for alias only because of
2915 (vect_analyze_data_ref_dependences): Set to true.
2916 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2918 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2919 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2923 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2926 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2927 operand 0 predicate for gathers, use a new pseudo as subtarget.
2929 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2932 * lra-constraints.c (process_alt_operands): Add printing debug info.
2933 Check absence of input/output reloads for matched operands too.
2935 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2937 PR rtl-optimization/59835
2938 * ira.c (ira_init_register_move_cost): Increase cost for
2941 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2943 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2945 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2948 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2949 non-register objects. Use gen_(high/low)part more consistently.
2952 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2955 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2956 endian support, remove tests for WORDS_BIG_ENDIAN.
2957 (p8_mfvsrd_3_<mode>): Likewise.
2958 (reload_gpr_from_vsx<mode>): Likewise.
2959 (reload_gpr_from_vsxsf): Likewise.
2960 (p8_mfvsrd_4_disf): Likewise.
2962 2014-01-16 Richard Biener <rguenther@suse.de>
2964 PR rtl-optimization/46590
2965 * lcm.c (compute_antinout_edge): Use postorder iteration.
2966 (compute_laterin): Use inverted postorder iteration.
2968 2014-01-16 Nick Clifton <nickc@redhat.com>
2971 * varasm.c (output_constant): Return the number of bytes actually
2973 (output_constructor_array_range): Update the field size with the
2974 number of bytes emitted by output_constant.
2975 (output_constructor_regular_field): Likewise. Also do not
2976 complain if the total number of bytes emitted is now greater
2977 than the expected fieldpos.
2978 * output.h (output_constant): Update prototype and descriptive comment.
2980 2014-01-16 Marek Polacek <polacek@redhat.com>
2983 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2984 it is error_mark_node.
2986 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2988 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2989 VALID_AVX256_REG_OR_OI_MODE.
2991 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2993 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2994 current procedure should be profiled.
2996 2014-01-15 Andrew Pinski <apinski@cavium.com>
2998 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2999 of moving from/to the STACK_REG register class.
3001 2014-01-15 Richard Henderson <rth@redhat.com>
3004 * reginfo.c (global_regs_decl): Globalize.
3005 * rtl.h (global_regs_decl): Declare.
3006 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3007 reserved via global_regs.
3009 2014-01-15 Teresa Johnson <tejohnson@google.com>
3011 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3013 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3015 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3016 and vmulosh rather than call gen_vec_widen_smult_*.
3017 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3018 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3019 (vec_widen_smult_even_v16qi): Likewise.
3020 (vec_widen_umult_even_v8hi): Likewise.
3021 (vec_widen_smult_even_v8hi): Likewise.
3022 (vec_widen_umult_odd_v16qi): Likewise.
3023 (vec_widen_smult_odd_v16qi): Likewise.
3024 (vec_widen_umult_odd_v8hi): Likewise.
3025 (vec_widen_smult_odd_v8hi): Likewise.
3026 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3027 vmuloub rather than call gen_vec_widen_umult_*.
3028 (vec_widen_umult_lo_v16qi): Likewise.
3029 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3030 vmulosb rather than call gen_vec_widen_smult_*.
3031 (vec_widen_smult_lo_v16qi): Likewise.
3032 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3033 rather than call gen_vec_widen_umult_*.
3034 (vec_widen_umult_lo_v8hi): Likewise.
3035 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3036 rather than call gen_vec_widen_smult_*.
3037 (vec_widen_smult_lo_v8hi): Likewise.
3039 2014-01-15 Jeff Law <law@redhat.com>
3041 PR tree-optimization/59747
3042 * ree.c (find_and_remove_re): Properly handle case where a second
3043 eliminated extension requires widening a copy created for elimination
3044 of a prior extension.
3045 (combine_set_extension): Ensure that the number of hard regs needed
3046 for a destination register does not change when we widen it.
3048 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3050 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3051 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3052 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3053 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3054 (avr-*-rtems*): Likewise.
3055 (bfin*-rtems*): Likewise.
3056 (moxie-*-rtems*): Likewise.
3057 (h8300-*-rtems*): Likewise.
3058 (i[34567]86-*-rtems*): Likewise.
3059 (lm32-*-rtems*): Likewise.
3060 (m32r-*-rtems*): Likewise.
3061 (m68k-*-rtems*): Likewise.
3062 (microblaze*-*-rtems*): Likewise.
3063 (mips*-*-rtems*): Likewise.
3064 (powerpc-*-rtems*): Likewise.
3065 (sh-*-rtems*): Likewise.
3066 (sparc-*-rtems*): Likewise.
3067 (sparc64-*-rtems*): Likewise.
3068 (v850-*-rtems*): Likewise.
3069 (m32c-*-rtems*): Likewise.
3071 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3073 PR rtl-optimization/59511
3074 * ira.c (ira_init_register_move_cost): Use memory costs for some
3075 cases of register move cost calculations.
3076 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3077 instead of BB frequency.
3078 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3079 * lra-assigns.c (find_hard_regno_for): Ditto.
3081 2014-01-15 Richard Biener <rguenther@suse.de>
3083 PR tree-optimization/59822
3084 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3085 (vectorizable_load): Use it to hoist defs of uses of invariant
3086 loads out of the loop.
3088 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3089 Kugan Vivekanandarajah <kuganv@linaro.org>
3092 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3095 2014-01-15 Richard Biener <rguenther@suse.de>
3097 PR rtl-optimization/59802
3098 * lcm.c (compute_available): Use inverted postorder to seed
3099 the initial worklist.
3101 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3104 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3105 ADDR_REGS for invalid symrefs in non-PIC code.
3107 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3110 * builtins.c (determine_block_size): Initialize *probable_max_size
3111 even if len_rtx is CONST_INT.
3113 2014-01-14 Andrew Pinski <apinski@cavium.com>
3115 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3116 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3117 (cortexa53_tunings): Likewise.
3118 (aarch64_sched_issue_rate): New function.
3119 (TARGET_SCHED_ISSUE_RATE): Define.
3121 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3123 * ira-costs.c (find_costs_and_classes): Add missed
3124 ira_init_register_move_cost_if_necessary.
3126 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3129 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3131 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3134 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3135 to indicate if type is used for function return value. Warn ABI
3136 change if the vector mode isn't available for function return value.
3137 (ix86_function_arg_advance): Pass false to type_natural_mode.
3138 (ix86_function_arg): Likewise.
3139 (ix86_gimplify_va_arg): Likewise.
3140 (function_arg_32): Don't warn ABI change.
3141 (ix86_function_value): Pass true to type_natural_mode.
3142 (ix86_return_in_memory): Likewise.
3143 (ix86_struct_value_rtx): Removed.
3144 (TARGET_STRUCT_VALUE_RTX): Likewise.
3146 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3148 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3149 converting a conditional jump into a conditional return.
3151 2014-01-14 Richard Biener <rguenther@suse.de>
3153 PR tree-optimization/58921
3154 PR tree-optimization/59006
3155 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3156 hoisting invariant stmts.
3157 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3158 invariant loads on the preheader edge if possible.
3160 2014-01-14 Joey Ye <joey.ye@arm.com>
3162 * doc/plugin.texi (Building GCC plugins): Update to C++.
3164 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3166 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3167 (_mm_rcp28_round_ss): Ditto.
3168 (_mm_rsqrt28_round_sd): Ditto.
3169 (_mm_rsqrt28_round_ss): Ditto.
3170 (_mm_rcp28_sd): Ditto.
3171 (_mm_rcp28_ss): Ditto.
3172 (_mm_rsqrt28_sd): Ditto.
3173 (_mm_rsqrt28_ss): Ditto.
3174 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3175 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3176 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3177 (IX86_BUILTIN_RCP28SD): Ditto.
3178 (IX86_BUILTIN_RCP28SS): Ditto.
3179 (IX86_BUILTIN_RSQRT28SD): Ditto.
3180 (IX86_BUILTIN_RSQRT28SS): Ditto.
3181 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3182 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3183 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3184 (ix86_expand_special_args_builtin): Expand new FTYPE.
3185 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3186 (srcp14<mode>): Make insn unary.
3187 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3188 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3189 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3190 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3191 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3192 Fix rounding: make it SAE only.
3193 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3195 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3197 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3198 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3199 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3200 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3201 (round_saeonly_mask_scalar_operand4): Ditto.
3202 (round_saeonly_mask_scalar_op3): Ditto.
3203 (round_saeonly_mask_scalar_op4): Ditto.
3205 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3207 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3208 Implement -maltivec=be for vec_insert and vec_extract.
3210 2014-01-10 DJ Delorie <dj@redhat.com>
3212 * config/msp430/msp430.md (call_internal): Don't allow memory
3213 references with SP as the base register.
3214 (call_value_internal): Likewise.
3215 * config/msp430/constraints.md (Yc): New. For memory references
3216 that don't use SP as a base register.
3218 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3219 "an integer without a # prefix"
3220 * config/msp430/msp430.md (epilogue_helper): Use it.
3222 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3225 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3226 AVX512F gather builtins.
3227 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3228 on gather decls with INTEGER_TYPE masktype.
3229 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3230 directly into the builtin rather than hoisting it before loop.
3232 PR tree-optimization/59387
3233 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3234 (scev_const_prop): If folded_casts and type has undefined overflow,
3235 use force_gimple_operand instead of force_gimple_operand_gsi and
3236 for each added stmt if it is assign with
3237 arith_code_with_undefined_signed_overflow, call
3238 rewrite_to_defined_overflow.
3239 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3240 gimple-fold.h instead.
3241 (arith_code_with_undefined_signed_overflow,
3242 rewrite_to_defined_overflow): Moved to ...
3243 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3244 rewrite_to_defined_overflow): ... here. No longer static.
3245 Include gimplify-me.h.
3246 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3247 rewrite_to_defined_overflow): New prototypes.
3249 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3251 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3253 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3255 * builtins.c (get_object_alignment_2): Minor tweak.
3256 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3258 2014-01-13 Christian Bruel <christian.bruel@st.com>
3260 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3261 optimized non constant lengths.
3263 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3266 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3267 load as __atomic_load_N if possible.
3269 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3271 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3273 (rs6000_expand_builtin): Adjust call.
3275 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3278 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3279 * config/rs6000/rs6000.c: Include target-globals.h.
3280 (rs6000_set_current_function): Instead of doing target_reinit
3281 unconditionally, use save_target_globals_default_opts and
3282 restore_target_globals.
3284 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3286 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3287 (rs6000_expand_builtin): Handle mffs and mtfsf.
3288 (rs6000_init_builtins): Define mffs and mtfsf.
3289 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3290 (rs6000_mffs): New pattern.
3291 (rs6000_mtfsf): New pattern.
3293 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3295 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3296 Start narrowing with START. Apply candidate-use pair
3297 and check overall cost in narrowing.
3298 (iv_ca_prune): Pass new argument.
3300 2014-01-10 Jeff Law <law@redhat.com>
3303 * ree.c (combine_reaching_defs): Ensure the defining statement
3304 occurs before the extension when optimizing extensions with
3305 different source and destination hard registers.
3307 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3310 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3311 vtables into the type inheritance graph.
3313 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3315 PR rtl-optimization/59754
3316 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3317 modes in the REGNO != REGNO case.
3319 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3321 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3323 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3325 PR tree-optimization/59745
3326 * tree-predcom.c (tree_predictive_commoning_loop): Call
3327 free_affine_expand_cache if giving up because components is NULL.
3329 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3330 GC in payload of target_globals struct instead of allocating them on
3331 the heap and the larger structs separately using GC.
3332 * target-globals.h (struct target_globals): Make regs, hard_regs,
3333 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3334 of GTY((skip)) and change type to void *.
3335 (reset_target_globals): Cast loads from those fields to corresponding
3338 2014-01-10 Steve Ellcey <sellcey@mips.com>
3341 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3342 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3343 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3345 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3348 * aarch64-modes.def (CC_Zmode): New flags mode.
3349 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3350 represents an equality.
3351 (aarch64_get_condition_code): Handle CC_Zmode.
3352 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3354 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3356 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3357 extraction in good case.
3359 2014-01-10 Richard Biener <rguenther@suse.de>
3361 PR tree-optimization/59374
3362 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3363 checking after SLP discovery. Mark stmts not participating
3364 in any SLP instance properly.
3366 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3368 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3369 when handling a SET rtx.
3371 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3373 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3374 (cortex-a57): Likewise.
3375 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3377 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3379 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3380 non-iwmmxt builtins.
3382 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3386 * ipa-devirt.c record_target_from_binfo): Take as argument
3387 stack of binfos and lookup matching one for virtual inheritance.
3388 (possible_polymorphic_call_targets_1): Update.
3390 2014-01-10 Huacai Chen <chenhc@lemote.com>
3392 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3393 kernel strings for Loongson-2E/2F/3A.
3395 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3398 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3399 is_gimple_call before calling gimple_call_internal_p.
3401 2014-01-09 Steve Ellcey <sellcey@mips.com>
3403 * Makefile.in (TREE_FLOW_H): Remove.
3404 (TREE_SSA_H): Add file names from tree-flow.h.
3405 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3406 * tree.h: Remove tree-flow.h reference.
3407 * hash-table.h: Remove tree-flow.h reference.
3408 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3409 reference with tree-ssa-loop.h.
3411 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3413 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3414 default element-order behavior for -maltivec.
3415 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3417 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3418 when targeting big endian, at least for now.
3419 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3421 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3424 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3425 var satisfies use_register_for_decl, just take into account type
3426 alignment, rather than decl alignment.
3428 PR tree-optimization/59622
3429 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3430 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3431 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3432 Don't devirtualize for inplace at all. For targets.length () == 1,
3433 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3435 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3437 * config/i386/i386.md (cpu): Remove the unused btver1.
3439 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3441 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3443 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3446 * tree-core.h (struct target_globals): New forward declaration.
3447 (struct tree_target_option): Add globals field.
3448 * tree.h (TREE_TARGET_GLOBALS): Define.
3449 (prepare_target_option_nodes_for_pch): New prototype.
3450 * target-globals.h (struct target_globals): Define even if
3452 * tree.c (prepare_target_option_node_for_pch,
3453 prepare_target_option_nodes_for_pch): New functions.
3454 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3455 * config/i386/i386.c: Include target-globals.h.
3456 (ix86_set_current_function): Instead of doing target_reinit
3457 unconditionally, use save_target_globals_default_opts and
3458 restore_target_globals.
3460 2014-01-09 Richard Biener <rguenther@suse.de>
3462 PR tree-optimization/59715
3463 * tree-cfg.h (split_critical_edges): Declare.
3464 * tree-cfg.c (split_critical_edges): Export.
3465 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3467 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3469 * cfgexpand.c (expand_stack_vars): Optionally disable
3470 asan stack protection.
3471 (expand_used_vars): Likewise.
3472 (partition_stack_vars): Likewise.
3473 * asan.c (asan_emit_stack_protection): Optionally disable
3474 after return stack usage.
3475 (instrument_derefs): Optionally disable memory access instrumentation.
3476 (instrument_builtin_call): Likewise.
3477 (instrument_strlen_call): Likewise.
3478 (asan_protect_global): Optionally disable global variables protection.
3479 * doc/invoke.texi: Added doc for new options.
3480 * params.def: Added new options.
3481 * params.h: Likewise.
3483 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3485 PR rtl-optimization/59724
3486 * ifcvt.c (cond_exec_process_if_block): Don't call
3487 flow_find_head_matching_sequence with 0 longest_match.
3488 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3489 non-active insns if !stop_after.
3490 (try_head_merge_bb): Revert 2014-01-07 changes.
3492 2014-01-08 Jeff Law <law@redhat.com>
3494 * ree.c (get_sub_rtx): New function, extracted from...
3495 (merge_def_and_ext): Here.
3496 (combine_reaching_defs): Use get_sub_rtx.
3498 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3500 * cgraph.h (varpool_variable_node): Do not choke on null node.
3502 2014-01-08 Catherine Moore <clm@codesourcery.com>
3504 * config/mips/mips.md (simple_return): Attempt to use JRC
3506 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3508 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3510 PR rtl-optimization/59137
3511 * reorg.c (steal_delay_list_from_target): Call update_block for
3513 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3515 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3517 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3518 two duplicate entries.
3520 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3523 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3525 * config/mips/mips.c (mips_truncated_op_cost): New function.
3526 (mips_rtx_costs): Adjust test for BADDU.
3527 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3529 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3531 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3532 (*baddu_si): ...this new pattern.
3534 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3537 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3539 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3542 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3544 (expand_expr, expand_normal): Adjust.
3545 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3546 inner_reference_p. Use inner_reference_p to expand inner references.
3547 (store_expr): Adjust.
3548 * cfgexpand.c (expand_call_stmt): Adjust.
3550 2014-01-08 Rong Xu <xur@google.com>
3552 * gcov-io.c (gcov_var): Move from gcov-io.h.
3553 (gcov_position): Ditto.
3554 (gcov_is_error): Ditto.
3555 (gcov_rewrite): Ditto.
3556 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3557 only part to libgcc/libgcov.h.
3559 2014-01-08 Marek Polacek <polacek@redhat.com>
3562 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3564 2014-01-08 Marek Polacek <polacek@redhat.com>
3567 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3569 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3571 PR rtl-optimization/59649
3572 * stor-layout.c (get_mode_bounds): For BImode return
3573 0 and STORE_FLAG_VALUE.
3575 2014-01-08 Richard Biener <rguenther@suse.de>
3578 * gimple.h (is_gimple_builtin_call): Remove.
3579 (gimple_builtin_call_types_compatible_p): New.
3580 (gimple_call_builtin_p): New overload.
3581 * gimple.c (is_gimple_builtin_call): Remove.
3582 (validate_call): Rename to ...
3583 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3585 (validate_type): New static function.
3586 (gimple_call_builtin_p): New overload and adjust.
3587 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3588 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3589 (gimple_fold_stmt_to_constant_1): Likewise.
3590 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3592 2014-01-08 Richard Biener <rguenther@suse.de>
3595 * gimplify.c (gimplify_expr): Gimplify register-register type
3596 VIEW_CONVERT_EXPRs to separate stmts.
3598 2014-01-07 Jeff Law <law@redhat.com>
3601 * ree.c (combine_set_extension): Handle case where source
3602 and destination registers in an extension insn are different.
3603 (combine_reaching_defs): Allow source and destination registers
3604 in extension to be different under limited circumstances.
3605 (add_removable_extension): Remove restriction that the
3606 source and destination registers in the extension are the same.
3607 (find_and_remove_re): Emit a copy from the extension's
3608 destination to its source after the defining insn if
3609 the source and destination registers are different.
3612 * ifcvt.c (merge_if_block): If we are merging a block with more than
3613 one successor with a block with no successors, remove any BARRIER
3614 after the second block.
3616 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3618 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3620 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3623 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3624 for 14-bit register offsets when INT14_OK_STRICT is false.
3626 2014-01-07 Roland Stigge <stigge@antcom.de>
3627 Michael Meissner <meissner@linux.vnet.ibm.com>
3630 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3631 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3633 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3635 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3638 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3640 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3641 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3642 rtx is const0_rtx or not.
3644 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3647 * target-globals.c (save_target_globals): Remove this_fn_optab
3649 * toplev.c: Include optabs.h.
3650 (target_reinit): Temporarily restore the global options if another
3651 set of options are in force.
3653 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3655 PR rtl-optimization/58668
3656 * cfgcleanup.c (flow_find_cross_jump): Don't count
3657 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3658 to determine what is counted.
3659 (flow_find_head_matching_sequence): Use active_insn_p to determine
3661 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3663 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3664 determine what is counted.
3666 PR tree-optimization/59643
3667 * tree-predcom.c (split_data_refs_to_components): If one dr is
3668 read and one write, determine_offset fails and the write isn't
3669 in the bad component, just put the read into the bad component.
3671 2014-01-07 Mike Stump <mikestump@comcast.net>
3672 Jakub Jelinek <jakub@redhat.com>
3675 * tree-core.h (struct tree_optimization_option): Change optabs
3676 type from unsigned char * to void *.
3677 * optabs.c (init_tree_optimization_optabs): Adjust
3678 TREE_OPTIMIZATION_OPTABS initialization.
3680 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3683 * config/i386/i386.h (struct machine_function): Add
3684 no_drap_save_restore field.
3685 * config/i386/i386.c (ix86_save_reg): Use
3686 !cfun->machine->no_drap_save_restore instead of
3687 crtl->stack_realign_needed.
3688 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3689 this function clears frame_pointer_needed. Set
3690 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3691 and DRAP reg is needed.
3693 2014-01-06 Marek Polacek <polacek@redhat.com>
3696 * doc/implement-c.texi: Mention that other integer types are
3697 permitted as bit-field types in strictly conforming mode.
3699 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3701 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3704 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3706 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3708 2014-01-06 Martin Jambor <mjambor@suse.cz>
3711 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3713 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3715 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3719 * var-tracking.c (add_stores): Preserve the value of the source even if
3720 we don't record the store.
3722 2014-01-06 Terry Guo <terry.guo@arm.com>
3724 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3726 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3729 * config/darwin.c (darwin_function_section): Adjust return values to
3730 correspond to optimisation changes made in r206070.
3732 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3734 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3735 from prefetch_block tune setting.
3736 (nocona_cost): Correct size of prefetch block to 64.
3738 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3740 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3741 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3742 used to save the static chain register in the computation of the offset
3743 from which the FP registers need to be restored.
3745 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3747 PR tree-optimization/59519
3748 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3749 ICE if get_current_def (current_new_name) is already non-NULL, as long
3750 as it is a phi result of some other phi in *new_exit_bb that has
3753 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3754 or vmovdqu* for misaligned_operand.
3755 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3756 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3757 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3758 aligned_mem for AVX512F masked aligned load and store builtins and for
3761 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3763 PR tree-optimization/59651
3764 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3765 Address range for negative step should be added by TYPE_SIZE_UNIT.
3767 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3769 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3770 overlapping registers also for registers other than the stack pointer.
3772 2014-01-03 Marek Polacek <polacek@redhat.com>
3775 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3778 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3781 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3784 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3785 (push splitter): Use <P:MODE_SIZE> instead of
3786 GET_MODE_SIZE (<P:MODE>mode).
3787 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3788 (mov -1, reg peephole2): Likewise.
3789 * config/i386/sse.md (*mov<mode>_internal,
3790 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3791 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3792 *<code><mode>3, *andnot<mode>3<mask_name>,
3793 <mask_codefor><code><mode>3<mask_name>): Likewise.
3794 * config/i386/subst.md (mask_mode512bit_condition,
3795 sd_mask_mode512bit_condition): Likewise.
3797 2014-01-02 Xinliang David Li <davidxl@google.com>
3799 PR tree-optimization/59303
3800 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3801 (dump_predicates): Better output format.
3802 (pred_equal_p): New function.
3803 (is_neq_relop_p): Ditto.
3804 (is_neq_zero_form_p): Ditto.
3805 (pred_expr_equal_p): Ditto.
3806 (pred_neg_p): Ditto.
3807 (simplify_pred): Ditto.
3808 (simplify_preds_2): Ditto.
3809 (simplify_preds_3): Ditto.
3810 (simplify_preds_4): Ditto.
3811 (simplify_preds): Ditto.
3813 (push_to_worklist): Ditto.
3814 (get_pred_info_from_cmp): Ditto.
3815 (is_degenerated_phi): Ditto.
3816 (normalize_one_pred_1): Ditto.
3817 (normalize_one_pred): Ditto.
3818 (normalize_one_pred_chain): Ditto.
3819 (normalize_preds): Ditto.
3820 (normalize_cond_1): Remove function.
3821 (normalize_cond): Ditto.
3822 (is_gcond_subset_of): Ditto.
3823 (is_subset_of_any): Ditto.
3824 (is_or_set_subset_of): Ditto.
3825 (is_and_set_subset_of): Ditto.
3826 (is_norm_cond_subset_of): Ditto.
3827 (pred_chain_length_cmp): Ditto.
3828 (convert_control_dep_chain_into_preds): Type change.
3829 (find_predicates): Ditto.
3830 (find_def_preds): Ditto.
3831 (destroy_predicates_vecs): Ditto.
3832 (find_matching_predicates_in_rest_chains): Ditto.
3833 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3834 (is_pred_expr_subset): Ditto.
3835 (is_pred_chain_subset_of): Ditto.
3836 (is_included_in): Ditto.
3837 (is_superset_of): Ditto.
3839 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3841 Update copyright years.
3843 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3845 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3846 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3847 config/arc/arc.md, config/arc/arc.opt,
3848 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3849 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3850 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3851 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3852 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3853 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3854 vtable-verify.c, vtable-verify.h: Use the standard form for the
3857 2014-01-02 Tobias Burnus <burnus@net-b.de>
3859 * gcc.c (process_command): Update copyright notice dates.
3860 * gcov-dump.c: Ditto.
3862 * doc/cpp.texi: Bump @copying's copyright year.
3863 * doc/cppinternals.texi: Ditto.
3864 * doc/gcc.texi: Ditto.
3865 * doc/gccint.texi: Ditto.
3866 * doc/gcov.texi: Ditto.
3867 * doc/install.texi: Ditto.
3868 * doc/invoke.texi: Ditto.
3870 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3872 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3874 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3876 * config/i386/sse.md (*mov<mode>_internal): Guard
3877 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3879 PR rtl-optimization/59647
3880 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3881 new_rtx into UNSIGNED_FLOAT rtxes.
3883 Copyright (C) 2014 Free Software Foundation, Inc.
3885 Copying and distribution of this file, with or without modification,
3886 are permitted in any medium without royalty provided the copyright
3887 notice and this notice are preserved.