1 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
6 2014-02-21 Nick Clifton <nickc@redhat.com>
8 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
10 (popqi1): Add mode to pre_dec.
13 2014-02-21 Jakub Jelinek <jakub@redhat.com>
15 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
16 mode for mask of V8SFmode permutation.
18 2014-02-20 Richard Henderson <rth@redhat.com>
21 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
22 a new pseudo for OLDVAL.
24 2014-02-20 Jakub Jelinek <jakub@redhat.com>
27 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
28 gen_reg_rtx if d->testing_p.
29 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
30 if d->testing_p and we will certainly return true.
31 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
34 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
36 * emit-rtl.c (gen_reg_rtx): Assert that
37 crtl->emit.regno_pointer_align_length is non-zero.
39 2014-02-20 Richard Henderson <rth@redhat.com>
42 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
43 on failure the store back into EXPECT.
45 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
46 Sandra Loosemore <sandra@codesourcery.com>
48 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
49 * config/nios2/nios2.c (nios2_function_profiler):
50 Add -fPIC (flag_pic == 2) support.
51 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
52 (nios2_large_offset_p): New function.
53 (nios2_unspec_reloc_p): Move up position, update to use
55 (nios2_unspec_address): Remove function.
56 (nios2_unspec_offset): New function.
57 (nios2_large_got_address): New function.
58 (nios2_got_address): Add large offset support.
59 (nios2_legitimize_tls_address): Update usage of removed and new
61 (nios2_symbol_binds_local_p): New function.
62 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
63 (nios2_legitimize_address): Update to use nios2_large_offset_p.
64 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
65 (nios2_print_operand): Merge H/L processing, add hiadj/lo
66 processing for (const (unspec ...)).
67 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
69 2014-02-20 Richard Biener <rguenther@suse.de>
71 * tree-cfg.c (replace_uses_by): Mark altered BBs before
72 doing the substitution.
73 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
75 2014-02-20 Martin Jambor <mjambor@suse.cz>
78 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
79 info when checking whether lattices are bottom.
81 2014-02-20 Richard Biener <rguenther@suse.de>
84 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
87 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
90 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
91 parameter specifying the scaling.
92 (inline_call): Update.
93 (want_inline_recursively): Guard division by zero.
94 (recursive_inlining): Update.
95 * ipa-inline.h (clone_inlined_nodes): Update.
97 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
100 * config/i386/i386.c (classify_argument): Pass structures of size
101 64 bytes or less in register.
103 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
104 Kirill Yukhin <kirill.yukhin@intel.com>
106 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
107 (_mm_rcp28_round_ss): Ditto.
108 (_mm_rsqrt28_round_sd): Ditto.
109 (_mm_rsqrt28_round_ss): Ditto.
110 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
111 (_mm_rcp14_round_ss): Ditto.
112 (_mm_rsqrt14_round_sd): Ditto.
113 (_mm_rsqrt14_round_ss): Ditto.
114 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
115 the first input operand, get rid of match_dup.
116 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
118 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
120 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
121 operand as the first input operand, set type attribute.
122 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
124 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
125 operand as the first input operand, set type attribute.
127 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
129 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
132 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
135 * config/i386/i386.c (construct_container): Remove TFmode check
136 for X86_64_INTEGER_CLASS.
138 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
141 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
142 only when -Wpsabi is enabled.
144 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
147 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
148 passing arrays in registers are the same as for structs, so remove the
149 special case for them.
151 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
153 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
154 destination type, extract only the valid bits if the source type is not
155 integral and has a different mode.
157 2014-02-19 Richard Biener <rguenther@suse.de>
160 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
163 2014-02-19 Richard Biener <rguenther@suse.de>
166 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
167 (ipa_modify_call_arguments): Emit an argument load explicitely and
168 preserve virtual SSA form there and for the replacement call.
169 Do not update SSA form nor free dominance info.
171 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
173 * ipa.c (function_and_variable_visibility): Also clear WEAK
174 flag when disolving COMDAT_GROUP.
176 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
178 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
179 * ipa-prop.c (ipa_set_jf_known_type): Return early when
181 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
182 do more sanity checks.
183 (detect_type_change): Return true when giving up early.
184 (compute_complex_assign_jump_func): Fix type parameter of
186 (compute_complex_ancestor_jump_func): Likewise.
187 (update_jump_functions_after_inlining): Fix updating of
189 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
191 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
193 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
194 inline clones when edge disappears.
196 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
199 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
200 Split 64-bit moves into 2 patterns. Do not allow the use of
201 direct move for TDmode in little endian, since the decimal value
202 has little endian bytes within a word, but the 64-bit pieces are
203 ordered in a big endian fashion, and normal subreg's of TDmode are
205 (mov<mode>_64bit_dm): Likewise.
206 (movtd_64bit_nodm): Likewise.
208 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
210 PR tree-optimization/60174
211 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
212 statement of an SSA_NAME that occurs in an abnormal PHI node.
214 2014-02-18 Jakub Jelinek <jakub@redhat.com>
217 * final.c (SEEN_BB): Remove.
218 (SEEN_NOTE, SEEN_EMITTED): Renumber.
219 (final_scan_insn): Don't force_source_line on second
220 NOTE_INSN_BASIC_BLOCK.
222 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
225 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
226 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
227 (type_natural_mode): Warn ABI change when %zmm register is not
228 available for AVX512F vector value passing.
230 2014-02-18 Kai Tietz <ktietz@redhat.com>
233 * config/i386/i386.c (ix86_expand_prologue): Use value in
234 rax register as displacement when restoring %r10 or %rax.
235 Fix wrong offset when restoring both registers.
237 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
239 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
240 assertion with conditional return.
242 2014-02-18 Jakub Jelinek <jakub@redhat.com>
243 Uros Bizjak <ubizjak@gmail.com>
246 * config/i386/driver-i386.c (host_detect_local_cpu): If
247 YMM state is not saved by the OS, also clear has_f16c. Move
248 CPUID 0x80000001 handling before YMM state saving checking.
250 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
252 PR rtl-optimization/58960
253 * haifa-sched.c (alloc_global_sched_pressure_data): New,
254 factored out from ...
255 (sched_init): ... here.
256 (free_global_sched_pressure_data): New, factored out from ...
257 (sched_finish): ... here.
258 * sched-int.h (free_global_sched_pressure_data): Declare.
259 * sched-rgn.c (nr_regions_initial): New static global.
260 (haifa_find_rgns): Initialize it.
261 (schedule_region): Disable sched-pressure for the newly
264 2014-02-17 Richard Biener <rguenther@suse.de>
266 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
267 release SSA defs of pattern stmts.
269 2014-02-17 Richard Biener <rguenther@suse.de>
271 * tree-inline.c (expand_call_inline): Release the virtual
272 operand defined by the call we are about to inline.
274 2014-02-17 Richard Biener <rguenther@suse.de>
276 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
278 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
279 Ilya Tocar <ilya.tocar@intel.com>
281 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
282 arguments order in builtin.
283 (_mm512_permutexvar_epi64): Ditto.
284 (_mm512_mask_permutexvar_epi64): Ditto
285 (_mm512_maskz_permutexvar_epi32): Ditto
286 (_mm512_permutexvar_epi32): Ditto
287 (_mm512_mask_permutexvar_epi32): Ditto
289 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
291 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
292 (p8_vmrgow): Likewise.
294 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
296 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
299 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
302 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
303 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
304 into 64-bit and 32-bit moves. On 64-bit moves, add support for
305 using direct move instructions on ISA 2.07. Also adjust
306 instruction length for 64-bit.
307 (mov<mode>_64bit, TFmode/TDmode): Likewise.
308 (mov<mode>_32bit, TFmode/TDmode): Likewise.
310 2014-02-15 Alan Modra <amodra@gmail.com>
314 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
315 find_replacement on parts of insn rtl that might be reloaded.
317 2014-02-15 Richard Biener <rguenther@suse.de>
319 PR tree-optimization/60183
320 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
321 (tree_ssa_phiprop): Calculate and free post-dominators.
323 2014-02-14 Jeff Law <law@redhat.com>
325 PR rtl-optimization/60131
326 * ree.c (get_extended_src_reg): New function.
327 (combine_reaching_defs): Use it rather than assuming location of REG.
328 (find_and_remove_re): Verify first operand of extension is
329 a REG before adding the insns to the copy list.
331 2014-02-14 Roland McGrath <mcgrathr@google.com>
333 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
334 * configure: Regenerated.
335 * config.in: Regenerated.
336 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
337 instead of ASM_SHORT.
339 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
340 Richard Earnshaw <rearnsha@arm.com>
342 PR rtl-optimization/59535
343 * lra-constraints.c (process_alt_operands): Encourage alternative
344 when unassigned pseudo class is superset of the alternative class.
345 (inherit_reload_reg): Don't inherit when optimizing for code size.
346 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
347 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
348 modes not less than 4 for Thumb1.
350 2014-02-14 Kyle McMartin <kyle@redhat.com>
353 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
355 2014-02-14 Richard Biener <rguenther@suse.de>
357 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
358 (get_frame_arg): Drop the assert with langhook types_compatible_p.
359 Do not strip INDIRECT_REFs.
361 2014-02-14 Richard Biener <rguenther@suse.de>
364 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
365 DECL_FUNCTION_SPECIFIC_TARGET.
366 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
367 * tree-streamer-out.c (pack_ts_target_option): Remove.
368 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
369 (write_ts_function_decl_tree_pointers): Do not stream
370 DECL_FUNCTION_SPECIFIC_TARGET.
371 * tree-streamer-in.c (unpack_ts_target_option): Remove.
372 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
373 (lto_input_ts_function_decl_tree_pointers): Do not stream
374 DECL_FUNCTION_SPECIFIC_TARGET.
376 2014-02-14 Jakub Jelinek <jakub@redhat.com>
378 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
379 (get_initial_def_for_induction, vectorizable_induction): Ignore
380 debug stmts when looking for exit_phi.
381 (vectorizable_live_operation): Fix up condition.
383 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
385 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
386 nreverse() because it changes the content of original tree list.
388 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
390 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
391 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
393 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
395 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
396 GNU coding standards.
398 2014-02-13 Jakub Jelinek <jakub@redhat.com>
401 * dwarf2out.c (gen_subprogram_die): Don't call
402 add_calling_convention_attribute if subr_die is old_die.
404 2014-02-13 Sharad Singhai <singhai@google.com>
406 * doc/optinfo.texi: Fix order of nodes.
408 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
410 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
411 operands[2], not operands[3].
413 2014-02-13 Richard Biener <rguenther@suse.de>
416 * doc/install.texi (ISL): Update recommended version to 0.12.2,
417 mention the possibility of an in-tree build.
418 (CLooG): Update recommended version to 0.18.1, mention the
419 possibility of an in-tree build and clarify that the ISL
420 bundled with CLooG does not work.
422 2014-02-13 Jakub Jelinek <jakub@redhat.com>
425 * expr.c (compress_float_constant): If x is a hard register,
426 extend into a pseudo and then move to x.
428 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
430 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
431 caused by bad second argument to warning_at() with -mhotpatch and
432 nested functions (e.g. with gfortran).
434 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
436 * opts.c (option_name): Remove "enabled by default" rider.
438 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
440 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
442 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
443 Uros Bizjak <ubizjak@gmail.com>
446 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
447 * configure: Regenerated.
449 2014-02-12 Richard Biener <rguenther@suse.de>
451 * vec.c (vec_prefix::calculate_allocation): Move as
452 inline variant to vec.h.
453 (vec_prefix::calculate_allocation_1): New out-of-line version.
454 * vec.h (vec_prefix::calculate_allocation_1): Declare.
455 (vec_prefix::m_has_auto_buf): Rename to ...
456 (vec_prefix::m_using_auto_storage): ... this.
457 (vec_prefix::calculate_allocation): Inline the easy cases
458 and dispatch to calculate_allocation_1 which doesn't need the
460 (va_heap::reserve): Use gcc_checking_assert.
461 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
462 m_using_auto_storage.
463 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
465 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
466 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
467 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
469 2014-02-12 Richard Biener <rguenther@suse.de>
471 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
472 when we found a dependence.
474 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
476 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
478 (maybe_fold_stmt): ... into this new function.
479 * omp-low.c (lower_omp): Update comment.
481 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
484 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
487 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
489 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
490 identifiers in comments.
491 (cortexa53_extra_costs): Likewise.
492 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
493 (cortexa7_extra_costs): Likewise.
494 (cortexa12_extra_costs): Likewise.
495 (cortexa15_extra_costs): Likewise.
496 (v7m_extra_costs): Likewise.
498 2014-02-12 Richard Biener <rguenther@suse.de>
501 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
502 of posix_memalign being successful.
503 (lower_stmt): Restrict lowering of posix_memalign to when
504 -ftree-bit-ccp is enabled.
506 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
508 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
510 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
512 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
514 PR rtl-optimization/60116
515 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
516 other_insn once the combination has been validated.
518 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
521 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
523 * ipa-devirt.c: Include demangle.h
524 (odr_violation_reported): New static variable.
525 (add_type_duplicate): Update odr_violations.
526 (maybe_record_node): Add completep parameter; update it.
527 (record_target_from_binfo): Add COMPLETEP parameter;
529 (possible_polymorphic_call_targets_1): Likewise.
530 (struct polymorphic_call_target_d): Add nonconstruction_targets;
531 rename FINAL to COMPLETE.
532 (record_targets_from_bases): Sanity check we found the binfo;
533 fix COMPLETEP updating.
534 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
535 parameter, fix computing of COMPLETEP.
536 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
537 at LTO time do demangling.
538 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
539 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
541 (gimple_get_virt_method_for_binfo): Likewise.
542 * gimple-fold.h (gimple_get_virt_method_for_binfo,
543 gimple_get_virt_method_for_vtable): Update prototypes.
545 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
548 * genautomata.c (add_presence_absence): Fix typo with
549 {final_}presence_list.
551 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
554 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
555 for VSX/Altivec vectors that land in GPR registers.
557 2014-02-11 Richard Henderson <rth@redhat.com>
558 Jakub Jelinek <jakub@redhat.com>
561 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
562 around drhs if type conversion to lacc->type is not useless.
564 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
566 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
568 (cortex-a57.cortex-a53): Likewise.
569 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
571 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
573 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
576 2014-02-11 Renlin Li <Renlin.Li@arm.com>
578 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
579 add_options_for_arm_vfp3.
581 2014-02-11 Jeff Law <law@redhat.com>
584 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
585 object with an undesirable mode.
587 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
590 * config/i386/sol2-9.h: New file.
591 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
592 *-*-solaris2.9*): Use it.
594 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
596 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
597 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
599 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
601 * config/microblaze/microblaze.c: Extend mcpu version format
603 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
605 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
607 2014-02-10 Richard Henderson <rth@redhat.com>
610 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
611 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
612 ms-abi vs -mno-accumulate-outgoing-args.
613 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
614 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
617 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
620 * cfgexpand.c (expand_asm_operands): Attach source location to
621 ASM_INPUT rtx objects.
622 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
624 2014-02-10 Nick Clifton <nickc@redhat.com>
626 * config/mn10300/mn10300.c (popcount): New function.
627 (mn10300_expand_prologue): Include saved registers in stack usage
630 2014-02-10 Jeff Law <law@redhat.com>
633 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
634 when changing the SET_DEST of a prior insn to avoid an input reload.
636 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
638 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
639 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
640 -mcall-openbsd, or -mcall-linux.
641 (CC1_ENDIAN_BIG_SPEC): Remove.
642 (CC1_ENDIAN_LITTLE_SPEC): Remove.
643 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
644 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
645 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
646 and %cc1_endian_default.
647 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
649 2014-02-10 Richard Biener <rguenther@suse.de>
651 PR tree-optimization/60115
652 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
653 MEM_REF handling. Properly verify that the accesses are not
654 out of the objects bound.
656 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
658 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
661 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
663 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
664 proper constants and fix formatting.
665 (possible_polymorphic_call_targets): Fix formatting.
667 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
668 Ilya Tocar <ilya.tocar@intel.com>
670 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
671 (_mm512_loadu_epi32): Renamed into...
672 (_mm512_loadu_si512): This.
673 (_mm512_storeu_epi32): Renamed into...
674 (_mm512_storeu_si512): This.
675 (_mm512_maskz_ceil_ps): Removed.
676 (_mm512_maskz_ceil_pd): Ditto.
677 (_mm512_maskz_floor_ps): Ditto.
678 (_mm512_maskz_floor_pd): Ditto.
679 (_mm512_floor_round_ps): Ditto.
680 (_mm512_floor_round_pd): Ditto.
681 (_mm512_ceil_round_ps): Ditto.
682 (_mm512_ceil_round_pd): Ditto.
683 (_mm512_mask_floor_round_ps): Ditto.
684 (_mm512_mask_floor_round_pd): Ditto.
685 (_mm512_mask_ceil_round_ps): Ditto.
686 (_mm512_mask_ceil_round_pd): Ditto.
687 (_mm512_maskz_floor_round_ps): Ditto.
688 (_mm512_maskz_floor_round_pd): Ditto.
689 (_mm512_maskz_ceil_round_ps): Ditto.
690 (_mm512_maskz_ceil_round_pd): Ditto.
691 (_mm512_expand_pd): Ditto.
692 (_mm512_expand_ps): Ditto.
693 * config/i386/i386.c (ix86_builtins): Remove
694 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
696 * config/i386/predicates.md (const1256_operand): New.
697 (const_1_to_2_operand): Ditto.
698 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
699 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
700 (*avx512pf_gatherpf<mode>sf): Ditto.
701 (avx512pf_gatherpf<mode>df): Ditto.
702 (*avx512pf_gatherpf<mode>df_mask): Ditto.
703 (*avx512pf_gatherpf<mode>df): Ditto.
704 (avx512pf_scatterpf<mode>sf): Ditto.
705 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
706 (*avx512pf_scatterpf<mode>sf): Ditto.
707 (avx512pf_scatterpf<mode>df): Ditto.
708 (*avx512pf_scatterpf<mode>df_mask): Ditto.
709 (*avx512pf_scatterpf<mode>df): Ditto.
710 (avx512f_expand<mode>): Removed.
711 (<shift_insn><mode>3<mask_name>): Change predicate type.
713 2014-02-08 Jakub Jelinek <jakub@redhat.com>
715 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
716 not at the end of datarefs vector use ordered_remove to avoid
717 reordering datarefs vector.
720 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
721 mark local addressable non-static vars as GOVD_PRIVATE
722 instead of GOVD_LOCAL.
723 * omp-low.c (lower_omp_for): Move gimple_bind_vars
724 and BLOCK_VARS of gimple_bind_block to new_stmt rather
728 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
729 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
730 assume_aligned or alloc_align attributes.
731 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
732 arguments. Handle also assume_aligned and alloc_align attributes.
733 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
734 calls to functions with assume_aligned or alloc_align attributes.
735 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
737 2014-02-08 Terry Guo <terry.guo@arm.com>
739 * doc/invoke.texi: Document ARM -march=armv7e-m.
741 2014-02-08 Jakub Jelinek <jakub@redhat.com>
743 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
744 flag on __cilkrts_rethrow builtin.
747 * ipa-cp.c (determine_versionability): Fail at -O0
748 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
749 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
752 2014-02-04 Jakub Jelinek <jakub@redhat.com>
755 * tree-inline.c (copy_forbidden): Fail for
756 __attribute__((optimize (0))) functions.
758 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
760 * varpool.c: Include pointer-set.h.
761 (varpool_remove_unreferenced_decls): Variables in other partitions
762 will not be output; be however careful to not lose information
765 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
767 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
768 lookup in the vtable constructor.
770 2014-02-07 Jeff Law <law@redhat.com>
773 * config/m68k/m68k.md (ashldi_extsi): Turn into a
774 define_insn_and_split.
776 * ipa-inline.c (inline_small_functions): Fix typos.
778 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
780 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
781 (s390_can_use_return_insn): Declare.
782 * config/s390/s390.h (EPILOGUE_USES): Define.
783 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
785 (s390_chunkify_start): Handle return JUMP_LABELs.
786 (s390_early_mach): Emit a main_pool instruction on the entry edge.
787 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
788 (s390_can_use_return_insn): New functions.
789 (s390_fix_long_loop_prediction): Handle conditional returns.
790 (TARGET_SET_UP_BY_PROLOGUE): Define.
791 * config/s390/s390.md (ANY_RETURN): New code iterator.
792 (*creturn, *csimple_return, return, simple_return): New patterns.
794 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
796 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
797 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
798 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
799 REG_CFA_RESTORE list when deciding not to restore a register.
801 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
803 * config/s390/s390.c: Include tree-pass.h and context.h.
804 (s390_early_mach): New function, split out from...
805 (s390_emit_prologue): ...here.
806 (pass_data_s390_early_mach): New pass structure.
807 (pass_s390_early_mach): New class.
808 (s390_option_override): Create and register early_mach pass.
811 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
813 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
814 to match for the exit block.
816 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
818 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
819 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
820 Reject misaligned operands.
822 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
824 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
826 2014-02-07 Richard Biener <rguenther@suse.de>
829 * gimple-low.c (lower_builtin_posix_memalign): New function.
830 (lower_stmt): Call it to lower posix_memalign in a way
831 to make alignment info accessible.
833 2014-02-07 Jakub Jelinek <jakub@redhat.com>
836 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
837 __builtin_setjmp_receiver.
839 2014-02-07 Richard Biener <rguenther@suse.de>
842 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
843 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
844 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
845 Handle BUILT_IN_POSIX_MEMALIGN.
846 (find_func_clobbers): Likewise.
847 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
848 (call_may_clobber_ref_p_1): Likewise.
850 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
853 * ipa-devirt.c (record_target_from_binfo): Remove overactive
856 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
859 * lto-cgraph.c (lto_output_node): Use
860 symtab_get_symbol_partitioning_class.
861 (lto_output_varpool_node): likewise.
862 (symtab_get_symbol_partitioning_class): Move here from
864 * cgraph.h (symbol_partitioning_class): Likewise.
865 (symtab_get_symbol_partitioning_class): Declare.
867 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
869 * ggc.h (ggc_internal_cleared_alloc): New macro.
870 * vec.h (vec_safe_copy): Handle memory stats.
871 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
872 * target-globals.c (save_target_globals): Likewise.
874 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
877 * expr.c (emit_move_resolve_push): Export; be bit more selective
878 on when to clear alias set.
879 * expr.h (emit_move_resolve_push): Declare.
880 * function.h (struct function): Add tail_call_marked.
881 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
882 * config/i386/i386-protos.h (ix86_expand_push): Remove.
883 * config/i386/i386.md (TImode move expander): De not call
885 (FP push expanders): Preserve memory attributes.
886 * config/i386/sse.md (push<mode>1): Remove.
887 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
888 (ix86_expand_push): Remove.
889 * config/i386/mmx.md (push<mode>1): Remove.
891 2014-02-06 Jakub Jelinek <jakub@redhat.com>
893 PR rtl-optimization/60030
894 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
895 lopart with paradoxical subreg before shifting it up by hprec.
897 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
899 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
900 Remove extra newline at end of file.
901 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
902 (arm_issue_rate): Handle cortexa57.
903 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
904 (cortex-a57.cortex-a53): Likewise.
906 2014-02-06 Jakub Jelinek <jakub@redhat.com>
909 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
910 don't record in REG_FRAME_RELATED_EXPR registers not set in that
912 (arm_expand_prologue): Adjust all callers.
913 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
914 info, registers also at the lowest numbered registers side. Use
915 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
919 * var-tracking.c (adjust_mems): Before adding a SET to
920 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
922 2014-02-06 Alan Modra <amodra@gmail.com>
925 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
926 change SDmode to DDmode when lra_in_progress.
928 2014-02-06 Jakub Jelinek <jakub@redhat.com>
931 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
932 free_data_ref on the dr first, and before goto again also set dr
933 to the next dr. For simd_lane_access, free old datarefs[i] before
934 overwriting it. For get_vectype_for_scalar_type failure, don't
935 free_data_ref if simd_lane_access.
937 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
940 * tree.h (opts_for_fn): New inline function.
941 (opt_for_fn): Define.
942 * config/i386/i386.c (ix86_function_regparm): Use
943 opt_for_fn (decl, optimize) instead of optimize.
945 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
947 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
948 for SYMBOL_REF in large memory model.
950 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
952 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
954 (cortex-a57): Likewise.
955 (cortex-a57.cortex-a53): Likewise.
957 2014-02-06 Yury Gribov <y.gribov@samsung.com>
958 Kugan Vivekanandarajah <kuganv@linaro.org>
960 * config/arm/arm.c (arm_vector_alignment_reachable): Check
962 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
964 2014-02-06 Richard Biener <rguenther@suse.de>
966 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
967 set_loop_copy and initialize_original_copy_tables.
969 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
971 * config/aarch64/aarch64-simd.md
972 (aarch64_ashr_simddi): Change QI to SI.
974 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
975 Jakub Jelinek <jakub@redhat.com>
978 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
981 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
983 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
984 CODE_FOR_altivec_vpku[hw]um to
985 CODE_FOR_altivec_vpku[hw]um_direct.
986 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
987 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
988 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
989 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
991 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
993 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
994 generation for -maltivec=be.
995 (altivec_vsumsws): Simplify redundant test.
997 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
999 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1000 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1001 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1002 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1003 gen_altivec_vpkuwum.
1004 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1006 (altivec_vpks<VI_char>ss): Likewise.
1007 (altivec_vpks<VI_char>us): Likewise.
1008 (altivec_vpku<VI_char>us): Likewise.
1009 (altivec_vpku<VI_char>um): Likewise.
1010 (altivec_vpku<VI_char>um_direct): New (copy of
1011 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1013 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1014 target is little endian and -maltivec=be is not specified.
1015 (*altivec_vupkhs<VU_char>_direct): New (copy of
1016 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1017 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1018 target is little endian and -maltivec=be is not specified.
1019 (*altivec_vupkls<VU_char>_direct): New (copy of
1020 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1021 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1022 little endian and -maltivec=be is not specified.
1023 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1024 little endian and -maltivec=be is not specified.
1026 2014-02-05 Richard Henderson <rth@redhat.com>
1029 * combine-stack-adj.c: Revert r206943.
1030 * sched-int.h (struct deps_desc): Add last_args_size.
1031 * sched-deps.c (init_deps): Initialize it.
1032 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1033 contain REG_ARGS_SIZE notes.
1035 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1037 * lto-cgraph.c (asm_nodes_output): Make global.
1038 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1039 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1040 (driver_handle_option): Handle OPT_fwpa.
1042 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1045 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1046 a comment typo and formatting issue. If odr_hash hasn't been
1047 created, return vNULL and set *completep to false.
1050 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1051 bb with no successors.
1053 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1056 * doc/invoke.texi (-march): Clarify documentation for ARM.
1060 2014-02-05 Richard Biener <rguenther@suse.de>
1062 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1063 when not vectorizing because of too many alias checks.
1064 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1065 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1067 2014-02-05 Nick Clifton <nickc@redhat.com>
1069 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1070 accept extended registers in any mode when compiling for the MN10300.
1072 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1074 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1075 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1076 sanitization attributes.
1077 (can_inline_edge_p): Likewise.
1078 (sanitize_attrs_match_for_inline_p): New function.
1080 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1082 * ipa-prop.c (detect_type_change): Shor circuit testing of
1083 type changes on THIS pointer.
1085 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1088 * config/pa/pa.c (legitimize_tls_address): Return original address
1089 if not passed a SYMBOL_REF rtx.
1090 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1092 (pa_emit_move_sequence): Simplify TLS source operands.
1093 (pa_legitimate_constant_p): Reject all TLS constants.
1094 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1095 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1097 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1099 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1100 groups when we know they are controlled by LTO.
1101 * varasm.c (default_binds_local_p_1): If object is in other partition,
1102 it will be resolved locally.
1104 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1106 * config/host-linux.c (linux_gt_pch_use_address): Don't
1107 use SSIZE_MAX because it is not always defined.
1109 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1112 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1113 threshold for pseudo splitting.
1114 (update_ebb_live_info): Process call argument hard registers and
1115 hard registers from insn definition too.
1116 (max_small_class_regs_num): New constant.
1117 (inherit_in_ebb): Update live hard regs through EBBs. Update
1118 reloads_num only for small register classes. Don't split for
1121 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1124 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1127 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1129 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1132 2014-02-04 Marek Polacek <polacek@redhat.com>
1134 * gdbinit.in (pel): Define.
1136 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1138 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1141 2014-02-04 Richard Biener <rguenther@suse.de>
1144 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1145 in function context local.
1146 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1147 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1148 similar to LTO_imported_decl_ref.
1150 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1152 PR tree-optimization/60002
1153 * cgraphclones.c (build_function_decl_skip_args): Clear
1156 PR tree-optimization/60023
1157 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1158 false to gsi_replace.
1159 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1160 has been in some EH region and vec_stmt could throw, add
1161 vec_stmt into the same EH region.
1162 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1163 has no lhs, ignore it.
1164 * internal-fn.c (expand_MASK_LOAD): Likewise.
1167 * tree-inline.c (copy_forbidden): Fail for
1168 __attribute__((optimize (0))) functions.
1171 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1172 is set, copy one less argument.
1173 (expand_simd_clones): Don't subtract clone_info->inbranch
1174 from simd_clone_struct_alloc argument.
1176 PR rtl-optimization/57915
1177 * recog.c (simplify_while_replacing): If all unary/binary/relational
1178 operation arguments are constant, attempt to simplify those.
1181 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1182 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1184 2014-02-04 Richard Biener <rguenther@suse.de>
1186 PR tree-optimization/60012
1187 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1188 TBAA disambiguation to all DDRs.
1190 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1193 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1194 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1196 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1199 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1201 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1203 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1204 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1206 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1209 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1210 to figure out targets of polymorphic calls with known decl.
1211 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1212 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1213 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1214 (get_polymorphic_call_info): ... here.
1215 (get_polymorphic_call_info_from_invariant): New function.
1217 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1219 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1220 lookup via vtable pointer; check for type consistency
1221 and turn inconsitent facts into UNREACHABLE.
1222 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1223 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1224 type inconsistent querries; return UNREACHABLE instead.
1226 2014-02-03 Richard Henderson <rth@twiddle.net>
1229 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1230 already processed this node.
1231 (normalize_one_pred_1): Pass along mark_set.
1232 (normalize_one_pred): Create and destroy a pointer_set_t.
1233 (normalize_one_pred_chain): Likewise.
1235 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1237 PR gcov-profile/58602
1238 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1240 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1243 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1244 -fno-devirtualize; try to devirtualize by the knowledge of
1245 virtual table pointer given by aggregate propagation.
1246 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1247 (ipa_print_node_jump_functions): Dump also offset that
1248 is relevant for polymorphic calls.
1249 (determine_known_aggregate_parts): Add arg_type parameter; use it
1250 instead of determining the type from pointer type.
1251 (ipa_compute_jump_functions_for_edge): Update call of
1252 determine_known_aggregate_parts.
1253 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1254 (gimple_get_virt_method_for_binfo): ... here; simplify using
1255 vtable_pointer_value_to_vtable.
1256 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1257 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1258 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1259 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1261 (vtable_pointer_value_to_binfo): ... here.
1262 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1264 2014-02-03 Teresa Johnson <tejohnson@google.com>
1266 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1267 redef of outer loop index variable.
1269 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1273 * doc/extend.texi (Function Attributes): Typo.
1275 2014-02-03 Cong Hou <congh@google.com>
1277 PR tree-optimization/60000
1278 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1279 if the vectorized statement is a store. A store statement can only
1280 appear at the end of pattern statements.
1282 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1284 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1285 (ix86_option_override_internal): Default long double to 64-bit for
1286 32-bit Bionic and to 128-bit for 64-bit Bionic.
1288 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1289 TARGET_LONG_DOUBLE_128 is true.
1290 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1292 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1293 (mlong-double-64): Negate -mlong-double-128.
1294 (mlong-double-128): New option.
1296 * config/i386/i386-c.c (ix86_target_macros): Define
1297 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1299 * doc/invoke.texi: Document -mlong-double-128.
1301 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1303 PR rtl-optimization/60024
1304 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1306 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1308 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1310 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1312 PR rtl-optimization/57662
1313 * sel-sched.c (code_motion_path_driver): Do not mark already not
1314 existing blocks in the visiting bitmap.
1316 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1318 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1319 on the insn being emitted.
1321 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1322 Will Deacon <will.deacon@arm.com>
1324 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1326 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1328 * config/arm/arm-tables.opt: Regenerate.
1330 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1332 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1333 for vector types other than V16QImode.
1334 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1335 define_expand, and call altivec_expand_vec_perm_le when producing
1336 code with little endian element order.
1337 (*altivec_vperm_<mode>_internal): New insn having previous
1338 behavior of altivec_vperm_<mode>.
1339 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1340 altivec_expand_vec_perm_le when producing code with little endian
1342 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1343 behavior of altivec_vperm_<mode>_uns.
1345 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1347 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1348 (altivec_vsumsws): Add handling for -maltivec=be with a little
1350 (altivec_vsumsws_direct): New.
1351 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1352 gen_altivec_vsumsws.
1354 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1356 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1357 vtable_pointer_value_to_binfo): New functions.
1358 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1359 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1361 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1363 * config/nios2/nios2.md (load_got_register): Initialize GOT
1364 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1365 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1367 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1369 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1370 preserverd by passthrough, do not propagate the type.
1372 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1374 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1375 (mips_atomic_assign_expand_fenv): New function.
1376 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1378 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1380 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1381 (__builtin_mips_set_fcsr): Likewise.
1382 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1383 MIPS_USI_FTYPE_VOID.
1384 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1385 (mips16_expand_set_fcsr): Likewise.
1386 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1387 (mips16_set_fcsr_stub): Likewise.
1388 (mips16_get_fcsr_one_only_stub): New class.
1389 (mips16_set_fcsr_one_only_stub): Likewise.
1390 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1391 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1392 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1393 (hard_float): New availability predicate.
1394 (mips_builtins): Add get_fcsr and set_fcsr.
1395 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1396 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1397 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1398 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1399 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1402 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1404 * config/mips/mips.c (mips_one_only_stub): New class.
1405 (mips_need_mips16_rdhwr_p): Replace with...
1406 (mips16_rdhwr_stub): ...this new variable.
1407 (mips16_stub_call_address): New function.
1408 (mips16_rdhwr_one_only_stub): New class.
1409 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1410 (mips_output_mips16_rdhwr): Delete.
1411 (mips_finish_stub): New function.
1412 (mips_code_end): Use it to handle rdhwr stubs.
1414 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1417 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1418 when calculating size of integer atomic types.
1420 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1422 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1424 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1426 PR tree-optimization/60003
1427 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1428 * profile.c (branch_prob): Use gimple_call_builtin_p
1429 to check for BUILT_IN_SETJMP_RECEIVER.
1430 * tree-inline.c (copy_bb): Call notice_special_calls.
1432 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1435 * lra-constraints.c (process_alt_operands): Update reload_sum only
1438 2014-01-31 Richard Henderson <rth@redhat.com>
1441 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1442 until after else_eh is processed.
1444 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1446 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1447 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1448 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1449 in smmintrin.h, remove them.
1450 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1451 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1452 * config/i386/i386.md (ROUND_SAE): Fix value.
1453 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1454 (const48_operand): New.
1455 * config/i386/subst.md (round), (round_expand): Use
1456 const_4_or_8_to_11_operand.
1457 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1459 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1461 * config/i386/constraints.md (Yk): Swap meaning with k.
1462 * config/i386/i386.md (movhi_internal): Change Yk to k.
1463 (movqi_internal): Ditto.
1464 (*k<logic><mode>): Ditto.
1467 (kandn<mode>): Ditto.
1468 (*<code>hi_1): Ditto.
1469 (*<code>qi_1): Ditto.
1470 (kxnor<mode>): Ditto.
1471 (kortestzhi): Ditto.
1472 (kortestchi): Ditto.
1474 (*one_cmplhi2_1): Ditto.
1475 (*one_cmplqi2_1): Ditto.
1476 * config/i386/sse.md (): Change k to Yk.
1477 (avx512f_load<mode>_mask): Ditto.
1478 (avx512f_blendm<mode>): Ditto.
1479 (avx512f_store<mode>_mask): Ditto.
1480 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1481 (avx512f_storedqu<mode>_mask): Ditto.
1482 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1484 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1485 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1486 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1487 (avx512f_maskcmp<mode>3): Ditto.
1488 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1489 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1490 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1491 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1492 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1493 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1494 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1495 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1496 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1497 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1498 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1499 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1500 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1501 (vec_extract_lo_<mode>_maskm): Ditto.
1502 (vec_extract_hi_<mode>_maskm): Ditto.
1503 (avx512f_vternlog<mode>_mask): Ditto.
1504 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1505 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1506 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1507 (avx512f_<code>v8div16qi2_mask): Ditto.
1508 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1509 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1510 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1511 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1512 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1513 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1514 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1515 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1516 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1517 (avx512cd_maskb_vec_dupv8di): Ditto.
1518 (avx512cd_maskw_vec_dupv16si): Ditto.
1519 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1520 (avx512f_vpermi2var<mode>3_mask): Ditto.
1521 (avx512f_vpermi2var<mode>3_mask): Ditto.
1522 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1523 (*avx512f_gathersi<mode>): Ditto.
1524 (*avx512f_gathersi<mode>_2): Ditto.
1525 (*avx512f_gatherdi<mode>): Ditto.
1526 (*avx512f_gatherdi<mode>_2): Ditto.
1527 (*avx512f_scattersi<mode>): Ditto.
1528 (*avx512f_scatterdi<mode>): Ditto.
1529 (avx512f_compress<mode>_mask): Ditto.
1530 (avx512f_compressstore<mode>_mask): Ditto.
1531 (avx512f_expand<mode>_mask): Ditto.
1532 * config/i386/subst.md (mask): Change k to Yk.
1533 (mask_scalar_merge): Ditto.
1536 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1538 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1540 2014-01-31 Richard Biener <rguenther@suse.de>
1543 * builtins.c (fold_builtin_memory_op): Make sure to not
1544 use a floating-point mode or a boolean or enumeral type for
1547 2014-01-30 DJ Delorie <dj@redhat.com>
1549 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1550 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1551 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1552 whenever main() has an epilogue.
1554 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1556 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1557 unused variable "field".
1558 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1559 (vsx_mergeh_<mode>): Likewise.
1560 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1561 (altivec_vmrghh): Likewise.
1562 (altivec_vmrghw): Likewise.
1563 (altivec_vmrglb): Likewise.
1564 (altivec_vmrglh): Likewise.
1565 (altivec_vmrglw): Likewise.
1566 (altivec_vspltb): Add missing uses.
1567 (altivec_vsplth): Likewise.
1568 (altivec_vspltw): Likewise.
1569 (altivec_vspltsf): Likewise.
1571 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1574 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1575 frame related instructions.
1577 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1579 PR rtl-optimization/59959
1580 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1581 any reload of register whose subreg is invalid.
1583 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1585 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1586 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1587 Add missing return type - void.
1589 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1591 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1592 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1593 remove element index adjustment for endian (now handled in vsx.md
1595 (altivec_expand_vec_perm_const): Use
1596 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1597 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1598 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1599 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1600 define_expand and a new define_insn *altivec_vspltb_internal;
1601 adjust for -maltivec=be on a little endian target.
1602 (altivec_vspltb_direct): New.
1603 (altivec_vsplth): Divide into a define_expand and a new
1604 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1605 little endian target.
1606 (altivec_vsplth_direct): New.
1607 (altivec_vspltw): Divide into a define_expand and a new
1608 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1609 little endian target.
1610 (altivec_vspltw_direct): New.
1611 (altivec_vspltsf): Divide into a define_expand and a new
1612 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1613 a little endian target.
1615 2014-01-30 Richard Biener <rguenther@suse.de>
1617 PR tree-optimization/59993
1618 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1619 can propagate form the earlier stmt and avoid the transform
1620 when the intermediate result is needed.
1622 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1624 * README.Portability: Fix typo.
1626 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1628 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1629 comparison_operator with ordered_comparison_operator.
1631 2014-01-30 Nick Clifton <nickc@redhat.com>
1633 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1634 Rename to mn10300_store_multiple_regs.
1635 * config/mn10300/mn10300.c: Likewise.
1636 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1637 store_multiple_regs.
1638 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1639 Call mn10300_store_multiple_regs.
1641 2014-01-30 Nick Clifton <nickc@redhat.com>
1642 DJ Delorie <dj@redhat.com>
1644 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1645 %fp 2 to keep registers after it properly word-aligned.
1646 (rl78_alloc_physical_registers_umul): Handle the case where both
1647 input operands are the same.
1649 2014-01-30 Richard Biener <rguenther@suse.de>
1651 PR tree-optimization/59903
1652 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1655 2014-01-30 Jason Merrill <jason@redhat.com>
1658 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1661 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1663 2014-01-30 Richard Biener <rguenther@suse.de>
1665 PR tree-optimization/59951
1666 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1668 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1671 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1672 SFmode to DFmode case.
1674 2014-01-29 DJ Delorie <dj@redhat.com>
1676 * config/msp430/msp430.opt (-minrt): New.
1677 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1679 (ENDFILE_SPEC): Likewise.
1681 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1683 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1684 (estimate_function_body_sizes): Use it.
1686 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1689 * dwarf2out.c (is_cxx_auto): New.
1690 (is_base_type): Use it.
1691 (gen_type_die_with_usage): Likewise.
1693 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1695 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1696 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1697 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1698 -maltivec=be with LE targets.
1699 (vsx_mergeh_<mode>): Likewise.
1700 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1701 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1702 (altivec_vmrghb): Replace with define_expand and new
1703 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1704 (altivec_vmrghb_direct): New define_insn.
1705 (altivec_vmrghh): Replace with define_expand and new
1706 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1707 (altivec_vmrghh_direct): New define_insn.
1708 (altivec_vmrghw): Replace with define_expand and new
1709 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1710 (altivec_vmrghw_direct): New define_insn.
1711 (*altivec_vmrghsf): Adjust for endianness.
1712 (altivec_vmrglb): Replace with define_expand and new
1713 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1714 (altivec_vmrglb_direct): New define_insn.
1715 (altivec_vmrglh): Replace with define_expand and new
1716 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1717 (altivec_vmrglh_direct): New define_insn.
1718 (altivec_vmrglw): Replace with define_expand and new
1719 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1720 (altivec_vmrglw_direct): New define_insn.
1721 (*altivec_vmrglsf): Adjust for endianness.
1722 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1723 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1724 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1725 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1726 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1727 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1728 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1729 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1731 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1733 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1734 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1737 2014-01-29 Richard Biener <rguenther@suse.de>
1739 PR tree-optimization/58742
1740 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1741 associate_pointerplus_align.
1742 (associate_pointerplus_diff): New function.
1743 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1744 and associate_pointerplus_diff.
1746 2014-01-29 Richard Biener <rguenther@suse.de>
1748 * lto-streamer.h (LTO_major_version): Bump to 3.
1749 (LTO_minor_version): Reset to 0.
1751 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1753 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1754 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1755 (arm_file_start): Generate correct asm header for armv7ve.
1756 * config/arm/bpabi.h: Add multilib support for armv7ve.
1757 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1758 and cortex-a15 to armv7ve.
1759 * config/arm/t-aprofile: Add multilib support for armv7ve.
1760 * doc/invoke.texi: Document -march=armv7ve.
1762 2014-01-29 Richard Biener <rguenther@suse.de>
1764 PR tree-optimization/58742
1765 * tree-ssa-forwprop.c (associate_plusminus): Return true
1766 if we changed sth, defer EH cleanup to ...
1767 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1768 (simplify_mult): New function.
1770 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1773 PR tree-optimization/59920
1774 * tree.c (build_common_builtin_nodes): Remove
1775 __builtin_setjmp_dispatcher initialization.
1776 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1777 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1778 instead of gsi_after_labels + manually skipping debug stmts.
1779 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1780 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1781 * tree-inline.c (copy_edges_for_bb): Remove
1782 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1783 argument. Ignore computed_goto_p stmts. Don't call
1784 make_abnormal_goto_edges. If a call might need abnormal edges
1785 for non-local gotos, see if it already has an edge to
1786 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1787 with true argument, don't do anything then, otherwise add
1788 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1789 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1791 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1792 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1793 (lower_stmt): Don't set data->calls_builtin_setjmp.
1794 (lower_builtin_setjmp): Adjust comment.
1795 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1796 * tree-cfg.c (found_computed_goto): Remove.
1797 (factor_computed_gotos): Remove.
1798 (make_goto_expr_edges): Return bool, true for computed gotos.
1799 Don't call make_abnormal_goto_edges.
1800 (build_gimple_cfg): Don't set found_computed_goto, don't call
1801 factor_computed_gotos.
1802 (computed_goto_p): No longer static.
1803 (make_blocks): Don't set found_computed_goto.
1804 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1805 (make_edges): If make_goto_expr_edges returns true, push bb
1806 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1807 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1808 vector. Record mapping between bbs and OpenMP regions if there
1809 are any, adjust make_gimple_omp_edges caller. Call
1810 handle_abnormal_edges.
1811 (make_abnormal_goto_edges): Remove.
1812 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1813 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1814 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1815 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1816 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1817 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1818 filling *region also set *region_idx to (*region)->entry->index.
1821 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1822 For REGs set ORIGINAL_REGNO.
1824 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1826 * doc/md.texi: Mention that a target shouldn't implement
1827 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1830 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1832 PR tree-optimization/59594
1833 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1834 a copy of the datarefs vector rather than the vector itself.
1836 2014-01-28 Jason Merrill <jason@redhat.com>
1839 * dwarf2out.c (auto_die): New static.
1840 (gen_type_die_with_usage): Handle C++1y 'auto'.
1841 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1844 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1847 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1848 (SPEC_X32): Likewise.
1849 (SPEC_64): Likewise.
1850 * config/i386/i386.c (ix86_option_override_internal): Turn off
1851 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1853 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1854 * config/i386/i386.h (TARGET_16BIT): New macro.
1855 (TARGET_16BIT_P): Likewise.
1856 * config/i386/i386.opt: Add m16.
1857 * doc/invoke.texi: Document -m16.
1859 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1861 PR preprocessor/59935
1862 * input.c (location_get_source_line): Bail out on when line number
1863 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1865 2014-01-28 Richard Biener <rguenther@suse.de>
1867 PR tree-optimization/58742
1868 * tree-ssa-forwprop.c (associate_plusminus): Handle
1869 pointer subtraction of the form (T)(P + A) - (T)P.
1871 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1873 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1876 2014-01-28 Richard Biener <rguenther@suse.de>
1879 2014-01-28 Richard Biener <rguenther@suse.de>
1881 PR rtl-optimization/45364
1882 PR rtl-optimization/59890
1883 * var-tracking.c (local_get_addr_clear_given_value): Handle
1884 already cleared slot.
1885 (val_reset): Handle not allocated local_get_addr_cache.
1886 (vt_find_locations): Use post-order on the inverted CFG.
1888 2014-01-28 Richard Biener <rguenther@suse.de>
1890 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1892 2014-01-28 Richard Biener <rguenther@suse.de>
1894 PR rtl-optimization/45364
1895 PR rtl-optimization/59890
1896 * var-tracking.c (local_get_addr_clear_given_value): Handle
1897 already cleared slot.
1898 (val_reset): Handle not allocated local_get_addr_cache.
1899 (vt_find_locations): Use post-order on the inverted CFG.
1901 2014-01-28 Alan Modra <amodra@gmail.com>
1903 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1904 * configure.ac <recursive call for build != host>: Define
1905 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1906 and LD_FOR_BUILD too.
1907 * configure: Regenerate.
1909 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1911 * config/i386/i386.c (get_builtin_code_for_version): Separate
1912 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1913 Broadwell from Haswell.
1915 2014-01-27 Steve Ellcey <sellcey@mips.com>
1917 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1918 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1919 * config/mips/mips.c (mips_option_override): Change setting
1921 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1922 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1923 Change from Mask to Var.
1925 2014-01-27 Jeff Law <law@redhat.com>
1927 * ipa-inline.c (inline_small_functions): Fix typo.
1929 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1931 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1932 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1933 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1934 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1935 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1936 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1937 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1938 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1939 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1940 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1941 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1942 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1943 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1944 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1945 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1946 (_mm512_storeu_epi64): Ditto.
1947 (_mm512_cmpge_epi32_mask): Ditto.
1948 (_mm512_cmpge_epu32_mask): Ditto.
1949 (_mm512_cmpge_epi64_mask): Ditto.
1950 (_mm512_cmpge_epu64_mask): Ditto.
1951 (_mm512_cmple_epi32_mask): Ditto.
1952 (_mm512_cmple_epu32_mask): Ditto.
1953 (_mm512_cmple_epi64_mask): Ditto.
1954 (_mm512_cmple_epu64_mask): Ditto.
1955 (_mm512_cmplt_epi32_mask): Ditto.
1956 (_mm512_cmplt_epu32_mask): Ditto.
1957 (_mm512_cmplt_epi64_mask): Ditto.
1958 (_mm512_cmplt_epu64_mask): Ditto.
1959 (_mm512_cmpneq_epi32_mask): Ditto.
1960 (_mm512_cmpneq_epu32_mask): Ditto.
1961 (_mm512_cmpneq_epi64_mask): Ditto.
1962 (_mm512_cmpneq_epu64_mask): Ditto.
1963 (_mm512_expand_pd): Ditto.
1964 (_mm512_expand_ps): Ditto.
1965 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1966 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1967 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1968 * config/i386/i386.c (ix86_builtins): Add
1969 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1970 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1971 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1972 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1973 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1974 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1975 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1976 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1977 IX86_BUILTIN_PMOVUSQW512_MEM.
1978 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1979 __builtin_ia32_pmovsqd512mem_mask,
1980 __builtin_ia32_pmovqd512mem_mask,
1981 __builtin_ia32_pmovusqw512mem_mask,
1982 __builtin_ia32_pmovsqw512mem_mask,
1983 __builtin_ia32_pmovqw512mem_mask,
1984 __builtin_ia32_pmovusdw512mem_mask,
1985 __builtin_ia32_pmovsdw512mem_mask,
1986 __builtin_ia32_pmovdw512mem_mask,
1987 __builtin_ia32_pmovqb512mem_mask,
1988 __builtin_ia32_pmovusqb512mem_mask,
1989 __builtin_ia32_pmovsqb512mem_mask,
1990 __builtin_ia32_pmovusdb512mem_mask,
1991 __builtin_ia32_pmovsdb512mem_mask,
1992 __builtin_ia32_pmovdb512mem_mask.
1993 (bdesc_args): Add __builtin_ia32_expanddf512,
1994 __builtin_ia32_expandsf512.
1995 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1996 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1997 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1998 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1999 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2000 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2001 (avx512f_<code>v8div16qi2_mask_store): This.
2002 (avx512f_expand<mode>): New.
2004 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2006 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2008 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2009 (_mm512_prefetch_i32scatter_pd): Ditto.
2010 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2011 (_mm512_prefetch_i64scatter_pd): Ditto.
2012 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2013 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2014 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2015 (_mm512_prefetch_i32scatter_ps): Ditto.
2016 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2017 (_mm512_prefetch_i64scatter_ps): Ditto.
2018 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2019 * config/i386/i386-builtin-types.def: Define
2020 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2021 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2022 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2023 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2024 IX86_BUILTIN_SCATTERPFQPD.
2025 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2026 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2027 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2028 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2029 __builtin_ia32_scatterpfqps.
2030 (ix86_expand_builtin): Expand new built-ins.
2031 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2032 fix memory access data type.
2033 (*avx512pf_gatherpf<mode>_mask): Ditto.
2034 (*avx512pf_gatherpf<mode>): Ditto.
2035 (avx512pf_scatterpf<mode>): Ditto.
2036 (*avx512pf_scatterpf<mode>_mask): Ditto.
2037 (*avx512pf_scatterpf<mode>): Ditto.
2038 (GATHER_SCATTER_SF_MEM_MODE): New.
2039 (avx512pf_gatherpf<mode>df): Ditto.
2040 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2041 (*avx512pf_scatterpf<mode>df): Ditto.
2043 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2046 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2047 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2050 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2052 * common/config/arm/arm-common.c
2053 (arm_rewrite_mcpu): Handle multiple names.
2055 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2057 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2059 * gimple-builder.h (create_gimple_tmp): Delete.
2061 2014-01-27 Christian Bruel <christian.bruel@st.com>
2063 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2066 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2068 * config/pa/pa.md (call): Generate indirect long calls to non-local
2069 functions when outputing 32-bit code.
2070 (call_value): Likewise except for special call to buggy powf function.
2072 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2073 portable runtime and PIC indirect calls.
2074 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2075 and PIC call sequences. Use ldo instead of blr to set return register
2076 in PIC call sequence.
2078 2014-01-25 Walter Lee <walt@tilera.com>
2080 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2081 avoid clobbering a live register.
2083 2014-01-25 Walter Lee <walt@tilera.com>
2085 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2086 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2087 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2088 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2090 2014-01-25 Walter Lee <walt@tilera.com>
2092 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2093 arguments on even registers.
2094 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2096 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2097 (BIGGEST_ALIGNMENT): Ditto.
2098 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2100 2014-01-25 Walter Lee <walt@tilera.com>
2102 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2103 insns before bundling.
2104 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2106 2014-01-25 Walter Lee <walt@tilera.com>
2108 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2109 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2110 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2112 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2114 * config/mips/constraints.md (kl): Delete.
2115 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2116 define expands, using...
2117 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2118 instructions for MIPS16.
2119 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2120 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2122 2014-01-25 Walter Lee <walt@tilera.com>
2124 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2128 2014-01-25 Walter Lee <walt@tilera.com>
2130 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2131 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2133 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2135 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2138 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2140 * print-rtl.c (in_call_function_usage): New var.
2141 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2142 EXPR_LIST mode as mode and not as reg note name.
2145 * cfgloopmanip.c (copy_loop_info): If
2146 loop->warned_aggressive_loop_optimizations, make sure
2147 the flag is set in target loop too.
2149 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2151 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2153 * builtins.def: Likewise.
2154 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2155 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2156 * ira.c (ira_setup_eliminable_regset): Likewise.
2157 * omp-low.c (gate_expand_omp): Likewise.
2158 (execute_lower_omp): Likewise.
2159 (diagnose_sb_0): Likewise.
2160 (gate_diagnose_omp_blocks): Likewise.
2161 (simd_clone_clauses_extract): Likewise.
2164 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2166 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2167 correction for little endian...
2168 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2171 2014-01-24 Jeff Law <law@redhat.com>
2173 PR tree-optimization/59919
2174 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2175 for non-returning calls.
2177 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2179 * common/config/aarch64/aarch64-common.c
2180 (aarch64_rewrite_mcpu): Handle multiple names.
2181 * config/aarch64/aarch64.h
2182 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2184 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2186 * input.c (add_file_to_cache_tab): Handle the case where fopen
2189 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2192 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2193 from push operand if code of push isn't PRE_DEC.
2195 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2198 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2199 -mquad-memory-atomic. Update -mquad-memory documentation to say
2200 it is only used for non-atomic loads/stores.
2202 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2203 -mquad-memory or -mquad-memory-atomic switches.
2205 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2206 -mquad-memory-atomic to ISA 2.07 support.
2208 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2209 to separate support of normal quad word memory operations (ldq, stq)
2210 from the atomic quad word memory operations.
2212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2213 support to separate non-atomic quad word operations from atomic
2214 quad word operations. Disable non-atomic quad word operations in
2215 little endian mode so that we don't have to swap words after the
2216 load and before the store.
2217 (quad_load_store_p): Add comment about atomic quad word support.
2218 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2219 options printed with -mdebug=reg.
2221 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2222 -mquad-memory-atomic as the test for whether we have quad word
2223 atomic instructions.
2224 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2225 or -mp8-vector are used, allow byte/half-word atomic operations.
2227 * config/rs6000/sync.md (load_lockedti): Insure that the address
2228 is a proper indexed or indirect address for the lqarx instruction.
2229 On little endian systems, swap the hi/lo registers after the lqarx
2231 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2232 insure the address is valid for the lqarx instruction.
2233 (store_conditionalti): Insure that the address is a proper indexed
2234 or indirect address for the stqcrx. instruction. On little endian
2235 systems, swap the hi/lo registers before doing the stqcrx.
2237 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2238 insure the address is valid for the stqcrx. instruction.
2240 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2241 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2242 type of quad memory support is available.
2244 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2247 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2248 there is a danger of looping.
2250 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2253 force flag_ira_loop_pressure if set via command line.
2255 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2257 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2258 (ashr_simd): New builtin handling DI mode.
2259 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2260 (aarch64_sshr_simddi): New match pattern.
2261 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2262 (vshrd_n_s64): Likewise.
2263 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2265 2014-01-23 Nick Clifton <nickc@redhat.com>
2267 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2268 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2269 favour of mcu specific scripts.
2270 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2273 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2274 Alex Velenko <Alex.Velenko@arm.com>
2276 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2277 (vaddv_s16): Likewise.
2278 (vaddv_s32): Likewise.
2279 (vaddv_u8): Likewise.
2280 (vaddv_u16): Likewise.
2281 (vaddv_u32): Likewise.
2282 (vaddvq_s8): Likewise.
2283 (vaddvq_s16): Likewise.
2284 (vaddvq_s32): Likewise.
2285 (vaddvq_s64): Likewise.
2286 (vaddvq_u8): Likewise.
2287 (vaddvq_u16): Likewise.
2288 (vaddvq_u32): Likewise.
2289 (vaddvq_u64): Likewise.
2290 (vaddv_f32): Likewise.
2291 (vaddvq_f32): Likewise.
2292 (vaddvq_f64): Likewise.
2293 (vmaxv_f32): Likewise.
2294 (vmaxv_s8): Likewise.
2295 (vmaxv_s16): Likewise.
2296 (vmaxv_s32): Likewise.
2297 (vmaxv_u8): Likewise.
2298 (vmaxv_u16): Likewise.
2299 (vmaxv_u32): Likewise.
2300 (vmaxvq_f32): Likewise.
2301 (vmaxvq_f64): Likewise.
2302 (vmaxvq_s8): Likewise.
2303 (vmaxvq_s16): Likewise.
2304 (vmaxvq_s32): Likewise.
2305 (vmaxvq_u8): Likewise.
2306 (vmaxvq_u16): Likewise.
2307 (vmaxvq_u32): Likewise.
2308 (vmaxnmv_f32): Likewise.
2309 (vmaxnmvq_f32): Likewise.
2310 (vmaxnmvq_f64): Likewise.
2311 (vminv_f32): Likewise.
2312 (vminv_s8): Likewise.
2313 (vminv_s16): Likewise.
2314 (vminv_s32): Likewise.
2315 (vminv_u8): Likewise.
2316 (vminv_u16): Likewise.
2317 (vminv_u32): Likewise.
2318 (vminvq_f32): Likewise.
2319 (vminvq_f64): Likewise.
2320 (vminvq_s8): Likewise.
2321 (vminvq_s16): Likewise.
2322 (vminvq_s32): Likewise.
2323 (vminvq_u8): Likewise.
2324 (vminvq_u16): Likewise.
2325 (vminvq_u32): Likewise.
2326 (vminnmv_f32): Likewise.
2327 (vminnmvq_f32): Likewise.
2328 (vminnmvq_f64): Likewise.
2330 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2332 * config/aarch64/aarch64-simd.md
2333 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2334 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2335 (*aarch64_mul3_elt<mode>): Likewise.
2336 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2337 (*aarch64_mul3_elt_to_64v2df): Likewise.
2338 (*aarch64_mla_elt<mode>): Likewise.
2339 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2340 (*aarch64_mls_elt<mode>): Likewise.
2341 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2342 (*aarch64_fma4_elt<mode>): Likewise.
2343 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2344 (*aarch64_fma4_elt_to_64v2df): Likewise.
2345 (*aarch64_fnma4_elt<mode>): Likewise.
2346 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2347 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2348 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2349 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2350 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2351 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2352 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2353 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2354 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2356 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2358 * config/aarch64/aarch64-simd.md
2359 (aarch64_be_checked_get_lane<mode>): New define_expand.
2360 * config/aarch64/aarch64-simd-builtins.def
2361 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2362 New builtin definition.
2363 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2364 Use new safe be builtin.
2366 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2368 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2370 (aarch64_be_st1<mode>): Likewise.
2371 (aarch_ld1<VALL:mode>): Define_expand modified.
2372 (aarch_st1<VALL:mode>): Likewise.
2373 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2374 (UNSPEC_ST1): Likewise.
2376 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2378 * config/microblaze/microblaze.md: Add trap insn and attribute
2380 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2382 PR preprocessor/58580
2383 * input.h (location_get_source_line): Take an additional line_size
2385 (void diagnostics_file_cache_fini): Declare new function.
2386 * input.c (struct fcache): New type.
2387 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2388 New static constants.
2389 (diagnostic_file_cache_init, total_lines_num)
2390 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2391 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2392 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2393 (get_next_line, read_next_line, goto_next_line, read_line_num):
2394 New static function definitions.
2395 (diagnostic_file_cache_fini): New function.
2396 (location_get_source_line): Take an additional output line_len
2397 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2399 * diagnostic.c (diagnostic_finish): Call
2400 diagnostic_file_cache_fini.
2401 (adjust_line): Take an additional input parameter for the length
2402 of the line, rather than calculating it with strlen.
2403 (diagnostic_show_locus): Adjust the use of
2404 location_get_source_line and adjust_line with respect to their new
2405 signature. While displaying a line now, do not stop at the first
2406 null byte. Rather, display the zero byte as a space and keep
2407 going until we reach the size of the line.
2408 * Makefile.in: Add vec.o to OBJS-libcommon
2410 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2411 Ilya Tocar <ilya.tocar@intel.com>
2413 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2414 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2415 (__builtin_ia32_kmov16): Ditto.
2416 * config/i386/i386.md (UNSPEC_KMOV): New.
2419 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2421 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2422 (_mm512_storeu_si512): Ditto.
2424 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2427 * rtl.h (get_referenced_operands): Declare.
2428 * recog.c (get_referenced_operands): New function.
2429 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2430 operands have been referenced when recording LO_SUM references.
2432 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2434 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2436 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2438 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2439 Enable for generic and recent AMD targets.
2441 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2443 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2444 ARG_SIZE note when adjustment was eliminated.
2446 2014-01-22 Jeff Law <law@redhat.com>
2448 PR tree-optimization/59597
2449 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2450 in file. Accept new argument REGISTERING and use it to modify
2451 dump output appropriately.
2452 (register_jump_thread): Corresponding changes.
2453 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2454 thread paths involving joiner blocks. Add code to dump cancelled
2455 jump threading paths.
2457 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2459 PR rtl-optimization/59477
2460 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2461 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2463 2014-01-22 Tom Tromey <tromey@redhat.com>
2465 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2467 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2469 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2471 PR rtl-optimization/59896
2472 * lra-constraints.c (process_alt_operands): Check unused note for
2473 matched operands of insn with no output reloads.
2475 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2477 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2478 (mips_move_from_gpr_cost): Likewise.
2480 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2482 PR rtl-optimization/59858
2483 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2484 ira_class_hard_regs_num.
2485 (process_alt_operands): Increase reject for dying matched operand.
2487 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2490 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2491 smaller than size, perform several stores or loads and stores
2492 at dst + count - size to store or copy all of size bytes, rather
2493 than just last modesize bytes.
2495 2014-01-20 DJ Delorie <dj@redhat.com>
2497 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2498 that CLOBBERs are REGs before propogating their values.
2500 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2503 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2504 (cgraph_inline_failed_type): New function.
2505 * cgraph.h (DEFCIFCODE): Add type.
2506 (cgraph_inline_failed_type_t): New enum.
2507 (cgraph_inline_failed_type): New prototype.
2508 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2509 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2510 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2511 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2512 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2513 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2514 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2515 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2516 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2517 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2518 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2519 OPTIMIZATION_MISMATCH.
2520 * tree-inline.c (expand_call_inline): Emit errors during
2521 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2523 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2526 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2527 mode attribute in insn output.
2529 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2531 * output.h (output_constant): Delete.
2532 * varasm.c (output_constant): Make private.
2534 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2536 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2538 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2541 * tree.h (fold_builtin_strcat): New prototype.
2542 * builtins.c (fold_builtin_strcat): No longer static. Add len
2543 argument, if non-NULL, don't call c_strlen. Optimize
2544 directly into __builtin_memcpy instead of __builtin_strcpy.
2545 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2546 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2548 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2550 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2551 for SImode_address_operand operands, having only a REG argument.
2553 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2555 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2556 loader name using mbig-endian.
2557 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2559 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2561 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2565 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2567 * config/aarch64/aarch64-protos.h
2568 (aarch64_cannot_change_mode_class_ptr): Declare.
2569 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2570 aarch64_cannot_change_mode_class_ptr): New.
2571 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2572 backend hook aarch64_cannot_change_mode_class.
2574 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2576 * common/config/aarch64/aarch64-common.c
2577 (aarch64_handle_option): Don't handle any option order logic here.
2578 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2579 selected_cpu, warn on architecture version mismatch.
2580 (aarch64_override_options): Fix parsing order for option strings.
2582 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2583 Iain Sandoe <iain@codesourcery.com>
2586 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2587 warning. Amend comment to reflect current functionality.
2589 2014-01-20 Richard Biener <rguenther@suse.de>
2592 * builtins.c (fold_builtin_strcat): Remove case better handled
2593 by tree-ssa-strlen.c.
2595 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2597 * config/aarch64/aarch64.opt
2598 (mcpu, march, mtune): Make case-insensitive.
2600 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2603 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2604 if operands[1] is a REG or ZERO_EXTEND of a REG.
2606 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2608 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2610 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2612 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2613 long non-pic millicode calls.
2615 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2617 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2619 2014-01-19 Kito Cheng <kito@0xlab.org>
2621 * builtins.c (expand_movstr): Check movstr expand done or fail.
2623 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2624 H.J. Lu <hongjiu.lu@intel.com>
2627 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2628 to DImode for zero-extended addresses.
2630 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2632 PR rtl-optimization/57763
2633 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2634 on the new indirect jump_insn and increment LABEL_NUSES (label).
2636 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2640 * config.gcc (x86_archs): New variable.
2641 (x86_64_archs): Likewise.
2642 (x86_cpus): Likewise.
2643 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2644 --with-arch/--with-cpu= options.
2645 Support --with-arch=/--with-cpu={nehalem,westmere,
2646 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2648 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2650 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2651 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2653 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2655 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2657 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2660 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2661 clear cpp_get_options (parse_in)->warn_unused_macros for
2662 ix86_target_macros_internal with cpp_define.
2664 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2666 * jump.c (delete_related_insns): Keep (use (insn))s.
2667 * reorg.c (redundant_insn): Check for barriers too.
2669 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2671 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2673 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2675 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2676 call to $$dyncall when TARGET_LONG_CALLS is true.
2678 2014-01-17 Jeff Law <law@redhat.com>
2680 * ree.c (combine_set_extension): Temporarily disable test for
2681 changing number of hard registers.
2683 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2686 * ipa-inline-analysis.c (inline_free_summary):
2687 Do not free summary of aliases.
2689 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2692 * gimplify.c (gimplify_expr): Use create_tmp_var
2693 instead of create_tmp_var_raw. If cond doesn't have
2694 integral type, don't add the IFN_ANNOTATE builtin at all.
2696 2014-01-17 Martin Jambor <mjambor@suse.cz>
2699 * ipa-cp.c (prev_edge_clone): New variable.
2700 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2701 Also resize prev_edge_clone vector.
2702 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2703 (ipcp_edge_removal_hook): New function.
2704 (ipcp_driver): Register ipcp_edge_removal_hook.
2706 2014-01-17 Andrew Pinski <apinski@cavium.com>
2707 Steve Ellcey <sellcey@mips.com>
2710 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2713 2014-01-17 Jeff Law <law@redhat.com>
2716 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2717 so that pass_ccp runs first.
2719 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2721 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2722 (ix86_adjust_cost): Use !TARGET_XXX.
2723 (do_reorder_for_imul): Likewise.
2724 (swap_top_of_ready_list): Likewise.
2725 (ix86_sched_reorder): Likewise.
2727 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2729 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2730 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2731 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2732 (intel_memset): New. Duplicate slm_memset.
2733 (intel_cost): New. Duplicate slm_cost.
2734 (m_INTEL): New macro.
2735 (processor_target_table): Add "intel".
2736 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2737 with PROCESSOR_INTEL for "intel".
2738 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2739 PROCESSOR_SILVERMONT.
2740 (ix86_issue_rate): Likewise.
2741 (ix86_adjust_cost): Likewise.
2742 (ia32_multipass_dfa_lookahead): Likewise.
2743 (swap_top_of_ready_list): Likewise.
2744 (ix86_sched_reorder): Likewise.
2745 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2746 instead of TARGET_OPT_AGU.
2747 * config/i386/i386.h (TARGET_INTEL): New.
2748 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2749 (processor_type): Add PROCESSOR_INTEL.
2750 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2751 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2753 2014-01-17 Marek Polacek <polacek@redhat.com>
2756 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2759 2014-01-17 Richard Biener <rguenther@suse.de>
2761 PR tree-optimization/46590
2762 * opts.c (default_options_table): Add entries for
2763 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2764 all enabled at -O1 but not for -Og.
2765 * common.opt (fbranch-count-reg): Remove Init(1).
2766 (fmove-loop-invariants): Likewise.
2767 (ftree-pta): Likewise.
2769 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2771 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2772 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2773 decls to at least the GCC 4.8 used alignments.
2776 * tree-nested.c (convert_nonlocal_reference_stmt,
2777 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2778 of GIMPLE_BIND stmts, adjust associated decls.
2780 2014-01-17 Richard Biener <rguenther@suse.de>
2782 PR tree-optimization/46590
2783 * vec.h (vec<>::bseach): New member function implementing
2784 binary search according to C89 bsearch.
2785 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2786 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2787 bitmap pointer again. Make accesses_in_loop a flat array.
2788 (mem_ref_obstack): New global.
2789 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2790 (mark_ref_stored): Likewise.
2791 (ref_indep_loop_p_2): Likewise.
2792 (set_ref_stored_in_loop): New helper function.
2793 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2794 (memref_free): Adjust.
2795 (record_mem_ref_loc): Simplify.
2796 (gather_mem_refs_stmt): Adjust.
2797 (sort_locs_in_loop_postorder_cmp): New function.
2798 (analyze_memory_references): Sort accesses_in_loop after
2799 loop postorder number.
2800 (find_ref_loc_in_loop_cmp): New function.
2801 (for_all_locs_in_loop): Find relevant cluster of locs in
2802 accesses_in_loop and iterate without recursion.
2803 (execute_sm): Avoid uninit warning.
2804 (struct ref_always_accessed): Simplify.
2805 (ref_always_accessed::operator ()): Likewise.
2806 (ref_always_accessed_p): Likewise.
2807 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2808 loop postorder numbers here.
2809 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2812 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2815 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2816 on decls for which assemble_alias has been called.
2818 2014-01-17 Nick Clifton <nickc@redhat.com>
2820 * config/msp430/msp430.opt: (mcpu): New option.
2821 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2822 (msp430_option_override): Parse target_cpu. If the MCU name
2823 matches a generic string, clear target_mcu.
2824 (msp430_attr): Allow numeric interrupt values up to 63.
2825 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2826 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2828 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2830 * config/msp430/msp430.md (popm): Use %J rather than %I.
2831 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2832 (addhi_cy_i): Use immediate_operand for operand 2.
2833 * doc/invoke.texi: Document -mcpu option.
2835 2014-01-17 Richard Biener <rguenther@suse.de>
2837 PR rtl-optimization/38518
2838 * df.h (df_analyze_loop): Declare.
2839 * df-core.c: Include cfgloop.h.
2840 (df_analyze_1): Split out main part of df_analyze.
2841 (df_analyze): Adjust.
2842 (loop_inverted_post_order_compute): New function.
2843 (loop_post_order_compute): Likewise.
2844 (df_analyze_loop): New function avoiding whole-function
2846 * loop-invariant.c (find_defs): Use df_analyze_loop.
2847 (find_invariants): Adjust.
2848 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2850 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2852 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2853 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2855 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2857 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2858 traversal when removing references.
2860 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2863 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2865 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2868 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2869 pushing a reload for an autoinc when we had previously reloaded an
2870 inner part of the address.
2872 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2874 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2876 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2877 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2878 when not giving up or versioning for alias only because of
2880 (vect_analyze_data_ref_dependences): Set to true.
2881 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2883 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2884 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2888 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2891 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2892 operand 0 predicate for gathers, use a new pseudo as subtarget.
2894 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2897 * lra-constraints.c (process_alt_operands): Add printing debug info.
2898 Check absence of input/output reloads for matched operands too.
2900 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2902 PR rtl-optimization/59835
2903 * ira.c (ira_init_register_move_cost): Increase cost for
2906 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2908 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2910 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2913 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2914 non-register objects. Use gen_(high/low)part more consistently.
2917 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2920 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2921 endian support, remove tests for WORDS_BIG_ENDIAN.
2922 (p8_mfvsrd_3_<mode>): Likewise.
2923 (reload_gpr_from_vsx<mode>): Likewise.
2924 (reload_gpr_from_vsxsf): Likewise.
2925 (p8_mfvsrd_4_disf): Likewise.
2927 2014-01-16 Richard Biener <rguenther@suse.de>
2929 PR rtl-optimization/46590
2930 * lcm.c (compute_antinout_edge): Use postorder iteration.
2931 (compute_laterin): Use inverted postorder iteration.
2933 2014-01-16 Nick Clifton <nickc@redhat.com>
2936 * varasm.c (output_constant): Return the number of bytes actually
2938 (output_constructor_array_range): Update the field size with the
2939 number of bytes emitted by output_constant.
2940 (output_constructor_regular_field): Likewise. Also do not
2941 complain if the total number of bytes emitted is now greater
2942 than the expected fieldpos.
2943 * output.h (output_constant): Update prototype and descriptive comment.
2945 2014-01-16 Marek Polacek <polacek@redhat.com>
2948 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2949 it is error_mark_node.
2951 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2953 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2954 VALID_AVX256_REG_OR_OI_MODE.
2956 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2958 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2959 current procedure should be profiled.
2961 2014-01-15 Andrew Pinski <apinski@cavium.com>
2963 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2964 of moving from/to the STACK_REG register class.
2966 2014-01-15 Richard Henderson <rth@redhat.com>
2969 * reginfo.c (global_regs_decl): Globalize.
2970 * rtl.h (global_regs_decl): Declare.
2971 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2972 reserved via global_regs.
2974 2014-01-15 Teresa Johnson <tejohnson@google.com>
2976 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2978 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2980 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2981 and vmulosh rather than call gen_vec_widen_smult_*.
2982 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2983 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2984 (vec_widen_smult_even_v16qi): Likewise.
2985 (vec_widen_umult_even_v8hi): Likewise.
2986 (vec_widen_smult_even_v8hi): Likewise.
2987 (vec_widen_umult_odd_v16qi): Likewise.
2988 (vec_widen_smult_odd_v16qi): Likewise.
2989 (vec_widen_umult_odd_v8hi): Likewise.
2990 (vec_widen_smult_odd_v8hi): Likewise.
2991 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2992 vmuloub rather than call gen_vec_widen_umult_*.
2993 (vec_widen_umult_lo_v16qi): Likewise.
2994 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2995 vmulosb rather than call gen_vec_widen_smult_*.
2996 (vec_widen_smult_lo_v16qi): Likewise.
2997 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2998 rather than call gen_vec_widen_umult_*.
2999 (vec_widen_umult_lo_v8hi): Likewise.
3000 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3001 rather than call gen_vec_widen_smult_*.
3002 (vec_widen_smult_lo_v8hi): Likewise.
3004 2014-01-15 Jeff Law <law@redhat.com>
3006 PR tree-optimization/59747
3007 * ree.c (find_and_remove_re): Properly handle case where a second
3008 eliminated extension requires widening a copy created for elimination
3009 of a prior extension.
3010 (combine_set_extension): Ensure that the number of hard regs needed
3011 for a destination register does not change when we widen it.
3013 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3015 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3016 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3017 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3018 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3019 (avr-*-rtems*): Likewise.
3020 (bfin*-rtems*): Likewise.
3021 (moxie-*-rtems*): Likewise.
3022 (h8300-*-rtems*): Likewise.
3023 (i[34567]86-*-rtems*): Likewise.
3024 (lm32-*-rtems*): Likewise.
3025 (m32r-*-rtems*): Likewise.
3026 (m68k-*-rtems*): Likewise.
3027 (microblaze*-*-rtems*): Likewise.
3028 (mips*-*-rtems*): Likewise.
3029 (powerpc-*-rtems*): Likewise.
3030 (sh-*-rtems*): Likewise.
3031 (sparc-*-rtems*): Likewise.
3032 (sparc64-*-rtems*): Likewise.
3033 (v850-*-rtems*): Likewise.
3034 (m32c-*-rtems*): Likewise.
3036 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3038 PR rtl-optimization/59511
3039 * ira.c (ira_init_register_move_cost): Use memory costs for some
3040 cases of register move cost calculations.
3041 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3042 instead of BB frequency.
3043 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3044 * lra-assigns.c (find_hard_regno_for): Ditto.
3046 2014-01-15 Richard Biener <rguenther@suse.de>
3048 PR tree-optimization/59822
3049 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3050 (vectorizable_load): Use it to hoist defs of uses of invariant
3051 loads out of the loop.
3053 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3054 Kugan Vivekanandarajah <kuganv@linaro.org>
3057 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3060 2014-01-15 Richard Biener <rguenther@suse.de>
3062 PR rtl-optimization/59802
3063 * lcm.c (compute_available): Use inverted postorder to seed
3064 the initial worklist.
3066 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3069 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3070 ADDR_REGS for invalid symrefs in non-PIC code.
3072 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3075 * builtins.c (determine_block_size): Initialize *probable_max_size
3076 even if len_rtx is CONST_INT.
3078 2014-01-14 Andrew Pinski <apinski@cavium.com>
3080 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3081 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3082 (cortexa53_tunings): Likewise.
3083 (aarch64_sched_issue_rate): New function.
3084 (TARGET_SCHED_ISSUE_RATE): Define.
3086 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3088 * ira-costs.c (find_costs_and_classes): Add missed
3089 ira_init_register_move_cost_if_necessary.
3091 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3094 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3096 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3099 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3100 to indicate if type is used for function return value. Warn ABI
3101 change if the vector mode isn't available for function return value.
3102 (ix86_function_arg_advance): Pass false to type_natural_mode.
3103 (ix86_function_arg): Likewise.
3104 (ix86_gimplify_va_arg): Likewise.
3105 (function_arg_32): Don't warn ABI change.
3106 (ix86_function_value): Pass true to type_natural_mode.
3107 (ix86_return_in_memory): Likewise.
3108 (ix86_struct_value_rtx): Removed.
3109 (TARGET_STRUCT_VALUE_RTX): Likewise.
3111 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3113 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3114 converting a conditional jump into a conditional return.
3116 2014-01-14 Richard Biener <rguenther@suse.de>
3118 PR tree-optimization/58921
3119 PR tree-optimization/59006
3120 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3121 hoisting invariant stmts.
3122 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3123 invariant loads on the preheader edge if possible.
3125 2014-01-14 Joey Ye <joey.ye@arm.com>
3127 * doc/plugin.texi (Building GCC plugins): Update to C++.
3129 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3131 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3132 (_mm_rcp28_round_ss): Ditto.
3133 (_mm_rsqrt28_round_sd): Ditto.
3134 (_mm_rsqrt28_round_ss): Ditto.
3135 (_mm_rcp28_sd): Ditto.
3136 (_mm_rcp28_ss): Ditto.
3137 (_mm_rsqrt28_sd): Ditto.
3138 (_mm_rsqrt28_ss): Ditto.
3139 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3140 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3141 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3142 (IX86_BUILTIN_RCP28SD): Ditto.
3143 (IX86_BUILTIN_RCP28SS): Ditto.
3144 (IX86_BUILTIN_RSQRT28SD): Ditto.
3145 (IX86_BUILTIN_RSQRT28SS): Ditto.
3146 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3147 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3148 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3149 (ix86_expand_special_args_builtin): Expand new FTYPE.
3150 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3151 (srcp14<mode>): Make insn unary.
3152 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3153 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3154 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3155 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3156 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3157 Fix rounding: make it SAE only.
3158 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3160 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3162 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3163 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3164 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3165 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3166 (round_saeonly_mask_scalar_operand4): Ditto.
3167 (round_saeonly_mask_scalar_op3): Ditto.
3168 (round_saeonly_mask_scalar_op4): Ditto.
3170 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3172 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3173 Implement -maltivec=be for vec_insert and vec_extract.
3175 2014-01-10 DJ Delorie <dj@redhat.com>
3177 * config/msp430/msp430.md (call_internal): Don't allow memory
3178 references with SP as the base register.
3179 (call_value_internal): Likewise.
3180 * config/msp430/constraints.md (Yc): New. For memory references
3181 that don't use SP as a base register.
3183 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3184 "an integer without a # prefix"
3185 * config/msp430/msp430.md (epilogue_helper): Use it.
3187 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3190 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3191 AVX512F gather builtins.
3192 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3193 on gather decls with INTEGER_TYPE masktype.
3194 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3195 directly into the builtin rather than hoisting it before loop.
3197 PR tree-optimization/59387
3198 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3199 (scev_const_prop): If folded_casts and type has undefined overflow,
3200 use force_gimple_operand instead of force_gimple_operand_gsi and
3201 for each added stmt if it is assign with
3202 arith_code_with_undefined_signed_overflow, call
3203 rewrite_to_defined_overflow.
3204 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3205 gimple-fold.h instead.
3206 (arith_code_with_undefined_signed_overflow,
3207 rewrite_to_defined_overflow): Moved to ...
3208 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3209 rewrite_to_defined_overflow): ... here. No longer static.
3210 Include gimplify-me.h.
3211 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3212 rewrite_to_defined_overflow): New prototypes.
3214 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3216 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3218 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3220 * builtins.c (get_object_alignment_2): Minor tweak.
3221 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3223 2014-01-13 Christian Bruel <christian.bruel@st.com>
3225 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3226 optimized non constant lengths.
3228 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3231 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3232 load as __atomic_load_N if possible.
3234 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3236 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3238 (rs6000_expand_builtin): Adjust call.
3240 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3243 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3244 * config/rs6000/rs6000.c: Include target-globals.h.
3245 (rs6000_set_current_function): Instead of doing target_reinit
3246 unconditionally, use save_target_globals_default_opts and
3247 restore_target_globals.
3249 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3251 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3252 (rs6000_expand_builtin): Handle mffs and mtfsf.
3253 (rs6000_init_builtins): Define mffs and mtfsf.
3254 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3255 (rs6000_mffs): New pattern.
3256 (rs6000_mtfsf): New pattern.
3258 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3260 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3261 Start narrowing with START. Apply candidate-use pair
3262 and check overall cost in narrowing.
3263 (iv_ca_prune): Pass new argument.
3265 2014-01-10 Jeff Law <law@redhat.com>
3268 * ree.c (combine_reaching_defs): Ensure the defining statement
3269 occurs before the extension when optimizing extensions with
3270 different source and destination hard registers.
3272 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3275 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3276 vtables into the type inheritance graph.
3278 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3280 PR rtl-optimization/59754
3281 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3282 modes in the REGNO != REGNO case.
3284 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3286 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3288 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3290 PR tree-optimization/59745
3291 * tree-predcom.c (tree_predictive_commoning_loop): Call
3292 free_affine_expand_cache if giving up because components is NULL.
3294 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3295 GC in payload of target_globals struct instead of allocating them on
3296 the heap and the larger structs separately using GC.
3297 * target-globals.h (struct target_globals): Make regs, hard_regs,
3298 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3299 of GTY((skip)) and change type to void *.
3300 (reset_target_globals): Cast loads from those fields to corresponding
3303 2014-01-10 Steve Ellcey <sellcey@mips.com>
3306 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3307 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3308 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3310 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3313 * aarch64-modes.def (CC_Zmode): New flags mode.
3314 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3315 represents an equality.
3316 (aarch64_get_condition_code): Handle CC_Zmode.
3317 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3319 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3321 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3322 extraction in good case.
3324 2014-01-10 Richard Biener <rguenther@suse.de>
3326 PR tree-optimization/59374
3327 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3328 checking after SLP discovery. Mark stmts not participating
3329 in any SLP instance properly.
3331 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3333 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3334 when handling a SET rtx.
3336 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3338 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3339 (cortex-a57): Likewise.
3340 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3342 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3344 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3345 non-iwmmxt builtins.
3347 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3351 * ipa-devirt.c record_target_from_binfo): Take as argument
3352 stack of binfos and lookup matching one for virtual inheritance.
3353 (possible_polymorphic_call_targets_1): Update.
3355 2014-01-10 Huacai Chen <chenhc@lemote.com>
3357 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3358 kernel strings for Loongson-2E/2F/3A.
3360 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3363 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3364 is_gimple_call before calling gimple_call_internal_p.
3366 2014-01-09 Steve Ellcey <sellcey@mips.com>
3368 * Makefile.in (TREE_FLOW_H): Remove.
3369 (TREE_SSA_H): Add file names from tree-flow.h.
3370 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3371 * tree.h: Remove tree-flow.h reference.
3372 * hash-table.h: Remove tree-flow.h reference.
3373 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3374 reference with tree-ssa-loop.h.
3376 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3378 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3379 default element-order behavior for -maltivec.
3380 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3382 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3383 when targeting big endian, at least for now.
3384 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3386 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3389 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3390 var satisfies use_register_for_decl, just take into account type
3391 alignment, rather than decl alignment.
3393 PR tree-optimization/59622
3394 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3395 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3396 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3397 Don't devirtualize for inplace at all. For targets.length () == 1,
3398 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3400 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3402 * config/i386/i386.md (cpu): Remove the unused btver1.
3404 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3406 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3408 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3411 * tree-core.h (struct target_globals): New forward declaration.
3412 (struct tree_target_option): Add globals field.
3413 * tree.h (TREE_TARGET_GLOBALS): Define.
3414 (prepare_target_option_nodes_for_pch): New prototype.
3415 * target-globals.h (struct target_globals): Define even if
3417 * tree.c (prepare_target_option_node_for_pch,
3418 prepare_target_option_nodes_for_pch): New functions.
3419 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3420 * config/i386/i386.c: Include target-globals.h.
3421 (ix86_set_current_function): Instead of doing target_reinit
3422 unconditionally, use save_target_globals_default_opts and
3423 restore_target_globals.
3425 2014-01-09 Richard Biener <rguenther@suse.de>
3427 PR tree-optimization/59715
3428 * tree-cfg.h (split_critical_edges): Declare.
3429 * tree-cfg.c (split_critical_edges): Export.
3430 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3432 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3434 * cfgexpand.c (expand_stack_vars): Optionally disable
3435 asan stack protection.
3436 (expand_used_vars): Likewise.
3437 (partition_stack_vars): Likewise.
3438 * asan.c (asan_emit_stack_protection): Optionally disable
3439 after return stack usage.
3440 (instrument_derefs): Optionally disable memory access instrumentation.
3441 (instrument_builtin_call): Likewise.
3442 (instrument_strlen_call): Likewise.
3443 (asan_protect_global): Optionally disable global variables protection.
3444 * doc/invoke.texi: Added doc for new options.
3445 * params.def: Added new options.
3446 * params.h: Likewise.
3448 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3450 PR rtl-optimization/59724
3451 * ifcvt.c (cond_exec_process_if_block): Don't call
3452 flow_find_head_matching_sequence with 0 longest_match.
3453 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3454 non-active insns if !stop_after.
3455 (try_head_merge_bb): Revert 2014-01-07 changes.
3457 2014-01-08 Jeff Law <law@redhat.com>
3459 * ree.c (get_sub_rtx): New function, extracted from...
3460 (merge_def_and_ext): Here.
3461 (combine_reaching_defs): Use get_sub_rtx.
3463 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3465 * cgraph.h (varpool_variable_node): Do not choke on null node.
3467 2014-01-08 Catherine Moore <clm@codesourcery.com>
3469 * config/mips/mips.md (simple_return): Attempt to use JRC
3471 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3473 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3475 PR rtl-optimization/59137
3476 * reorg.c (steal_delay_list_from_target): Call update_block for
3478 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3480 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3482 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3483 two duplicate entries.
3485 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3488 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3490 * config/mips/mips.c (mips_truncated_op_cost): New function.
3491 (mips_rtx_costs): Adjust test for BADDU.
3492 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3494 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3496 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3497 (*baddu_si): ...this new pattern.
3499 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3502 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3504 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3507 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3509 (expand_expr, expand_normal): Adjust.
3510 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3511 inner_reference_p. Use inner_reference_p to expand inner references.
3512 (store_expr): Adjust.
3513 * cfgexpand.c (expand_call_stmt): Adjust.
3515 2014-01-08 Rong Xu <xur@google.com>
3517 * gcov-io.c (gcov_var): Move from gcov-io.h.
3518 (gcov_position): Ditto.
3519 (gcov_is_error): Ditto.
3520 (gcov_rewrite): Ditto.
3521 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3522 only part to libgcc/libgcov.h.
3524 2014-01-08 Marek Polacek <polacek@redhat.com>
3527 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3529 2014-01-08 Marek Polacek <polacek@redhat.com>
3532 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3534 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3536 PR rtl-optimization/59649
3537 * stor-layout.c (get_mode_bounds): For BImode return
3538 0 and STORE_FLAG_VALUE.
3540 2014-01-08 Richard Biener <rguenther@suse.de>
3543 * gimple.h (is_gimple_builtin_call): Remove.
3544 (gimple_builtin_call_types_compatible_p): New.
3545 (gimple_call_builtin_p): New overload.
3546 * gimple.c (is_gimple_builtin_call): Remove.
3547 (validate_call): Rename to ...
3548 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3550 (validate_type): New static function.
3551 (gimple_call_builtin_p): New overload and adjust.
3552 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3553 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3554 (gimple_fold_stmt_to_constant_1): Likewise.
3555 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3557 2014-01-08 Richard Biener <rguenther@suse.de>
3560 * gimplify.c (gimplify_expr): Gimplify register-register type
3561 VIEW_CONVERT_EXPRs to separate stmts.
3563 2014-01-07 Jeff Law <law@redhat.com>
3566 * ree.c (combine_set_extension): Handle case where source
3567 and destination registers in an extension insn are different.
3568 (combine_reaching_defs): Allow source and destination registers
3569 in extension to be different under limited circumstances.
3570 (add_removable_extension): Remove restriction that the
3571 source and destination registers in the extension are the same.
3572 (find_and_remove_re): Emit a copy from the extension's
3573 destination to its source after the defining insn if
3574 the source and destination registers are different.
3577 * ifcvt.c (merge_if_block): If we are merging a block with more than
3578 one successor with a block with no successors, remove any BARRIER
3579 after the second block.
3581 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3583 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3585 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3588 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3589 for 14-bit register offsets when INT14_OK_STRICT is false.
3591 2014-01-07 Roland Stigge <stigge@antcom.de>
3592 Michael Meissner <meissner@linux.vnet.ibm.com>
3595 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3596 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3598 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3600 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3603 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3605 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3606 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3607 rtx is const0_rtx or not.
3609 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3612 * target-globals.c (save_target_globals): Remove this_fn_optab
3614 * toplev.c: Include optabs.h.
3615 (target_reinit): Temporarily restore the global options if another
3616 set of options are in force.
3618 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3620 PR rtl-optimization/58668
3621 * cfgcleanup.c (flow_find_cross_jump): Don't count
3622 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3623 to determine what is counted.
3624 (flow_find_head_matching_sequence): Use active_insn_p to determine
3626 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3628 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3629 determine what is counted.
3631 PR tree-optimization/59643
3632 * tree-predcom.c (split_data_refs_to_components): If one dr is
3633 read and one write, determine_offset fails and the write isn't
3634 in the bad component, just put the read into the bad component.
3636 2014-01-07 Mike Stump <mikestump@comcast.net>
3637 Jakub Jelinek <jakub@redhat.com>
3640 * tree-core.h (struct tree_optimization_option): Change optabs
3641 type from unsigned char * to void *.
3642 * optabs.c (init_tree_optimization_optabs): Adjust
3643 TREE_OPTIMIZATION_OPTABS initialization.
3645 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3648 * config/i386/i386.h (struct machine_function): Add
3649 no_drap_save_restore field.
3650 * config/i386/i386.c (ix86_save_reg): Use
3651 !cfun->machine->no_drap_save_restore instead of
3652 crtl->stack_realign_needed.
3653 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3654 this function clears frame_pointer_needed. Set
3655 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3656 and DRAP reg is needed.
3658 2014-01-06 Marek Polacek <polacek@redhat.com>
3661 * doc/implement-c.texi: Mention that other integer types are
3662 permitted as bit-field types in strictly conforming mode.
3664 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3666 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3669 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3671 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3673 2014-01-06 Martin Jambor <mjambor@suse.cz>
3676 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3678 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3680 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3684 * var-tracking.c (add_stores): Preserve the value of the source even if
3685 we don't record the store.
3687 2014-01-06 Terry Guo <terry.guo@arm.com>
3689 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3691 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3694 * config/darwin.c (darwin_function_section): Adjust return values to
3695 correspond to optimisation changes made in r206070.
3697 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3699 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3700 from prefetch_block tune setting.
3701 (nocona_cost): Correct size of prefetch block to 64.
3703 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3705 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3706 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3707 used to save the static chain register in the computation of the offset
3708 from which the FP registers need to be restored.
3710 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3712 PR tree-optimization/59519
3713 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3714 ICE if get_current_def (current_new_name) is already non-NULL, as long
3715 as it is a phi result of some other phi in *new_exit_bb that has
3718 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3719 or vmovdqu* for misaligned_operand.
3720 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3721 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3722 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3723 aligned_mem for AVX512F masked aligned load and store builtins and for
3726 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3728 PR tree-optimization/59651
3729 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3730 Address range for negative step should be added by TYPE_SIZE_UNIT.
3732 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3734 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3735 overlapping registers also for registers other than the stack pointer.
3737 2014-01-03 Marek Polacek <polacek@redhat.com>
3740 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3743 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3746 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3749 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3750 (push splitter): Use <P:MODE_SIZE> instead of
3751 GET_MODE_SIZE (<P:MODE>mode).
3752 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3753 (mov -1, reg peephole2): Likewise.
3754 * config/i386/sse.md (*mov<mode>_internal,
3755 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3756 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3757 *<code><mode>3, *andnot<mode>3<mask_name>,
3758 <mask_codefor><code><mode>3<mask_name>): Likewise.
3759 * config/i386/subst.md (mask_mode512bit_condition,
3760 sd_mask_mode512bit_condition): Likewise.
3762 2014-01-02 Xinliang David Li <davidxl@google.com>
3764 PR tree-optimization/59303
3765 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3766 (dump_predicates): Better output format.
3767 (pred_equal_p): New function.
3768 (is_neq_relop_p): Ditto.
3769 (is_neq_zero_form_p): Ditto.
3770 (pred_expr_equal_p): Ditto.
3771 (pred_neg_p): Ditto.
3772 (simplify_pred): Ditto.
3773 (simplify_preds_2): Ditto.
3774 (simplify_preds_3): Ditto.
3775 (simplify_preds_4): Ditto.
3776 (simplify_preds): Ditto.
3778 (push_to_worklist): Ditto.
3779 (get_pred_info_from_cmp): Ditto.
3780 (is_degenerated_phi): Ditto.
3781 (normalize_one_pred_1): Ditto.
3782 (normalize_one_pred): Ditto.
3783 (normalize_one_pred_chain): Ditto.
3784 (normalize_preds): Ditto.
3785 (normalize_cond_1): Remove function.
3786 (normalize_cond): Ditto.
3787 (is_gcond_subset_of): Ditto.
3788 (is_subset_of_any): Ditto.
3789 (is_or_set_subset_of): Ditto.
3790 (is_and_set_subset_of): Ditto.
3791 (is_norm_cond_subset_of): Ditto.
3792 (pred_chain_length_cmp): Ditto.
3793 (convert_control_dep_chain_into_preds): Type change.
3794 (find_predicates): Ditto.
3795 (find_def_preds): Ditto.
3796 (destroy_predicates_vecs): Ditto.
3797 (find_matching_predicates_in_rest_chains): Ditto.
3798 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3799 (is_pred_expr_subset): Ditto.
3800 (is_pred_chain_subset_of): Ditto.
3801 (is_included_in): Ditto.
3802 (is_superset_of): Ditto.
3804 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3806 Update copyright years.
3808 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3810 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3811 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3812 config/arc/arc.md, config/arc/arc.opt,
3813 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3814 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3815 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3816 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3817 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3818 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3819 vtable-verify.c, vtable-verify.h: Use the standard form for the
3822 2014-01-02 Tobias Burnus <burnus@net-b.de>
3824 * gcc.c (process_command): Update copyright notice dates.
3825 * gcov-dump.c: Ditto.
3827 * doc/cpp.texi: Bump @copying's copyright year.
3828 * doc/cppinternals.texi: Ditto.
3829 * doc/gcc.texi: Ditto.
3830 * doc/gccint.texi: Ditto.
3831 * doc/gcov.texi: Ditto.
3832 * doc/install.texi: Ditto.
3833 * doc/invoke.texi: Ditto.
3835 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3837 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3839 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3841 * config/i386/sse.md (*mov<mode>_internal): Guard
3842 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3844 PR rtl-optimization/59647
3845 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3846 new_rtx into UNSIGNED_FLOAT rtxes.
3848 Copyright (C) 2014 Free Software Foundation, Inc.
3850 Copying and distribution of this file, with or without modification,
3851 are permitted in any medium without royalty provided the copyright
3852 notice and this notice are preserved.