1 2014-02-21 Nick Clifton <nickc@redhat.com>
3 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
5 (popqi1): Add mode to pre_dec.
8 2014-02-21 Jakub Jelinek <jakub@redhat.com>
10 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
11 mode for mask of V8SFmode permutation.
13 2014-02-20 Richard Henderson <rth@redhat.com>
16 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
17 a new pseudo for OLDVAL.
19 2014-02-20 Jakub Jelinek <jakub@redhat.com>
22 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
23 gen_reg_rtx if d->testing_p.
24 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
25 if d->testing_p and we will certainly return true.
26 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
29 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
31 * emit-rtl.c (gen_reg_rtx): Assert that
32 crtl->emit.regno_pointer_align_length is non-zero.
34 2014-02-20 Richard Henderson <rth@redhat.com>
37 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
38 on failure the store back into EXPECT.
40 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
41 Sandra Loosemore <sandra@codesourcery.com>
43 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
44 * config/nios2/nios2.c (nios2_function_profiler):
45 Add -fPIC (flag_pic == 2) support.
46 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
47 (nios2_large_offset_p): New function.
48 (nios2_unspec_reloc_p): Move up position, update to use
50 (nios2_unspec_address): Remove function.
51 (nios2_unspec_offset): New function.
52 (nios2_large_got_address): New function.
53 (nios2_got_address): Add large offset support.
54 (nios2_legitimize_tls_address): Update usage of removed and new
56 (nios2_symbol_binds_local_p): New function.
57 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
58 (nios2_legitimize_address): Update to use nios2_large_offset_p.
59 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
60 (nios2_print_operand): Merge H/L processing, add hiadj/lo
61 processing for (const (unspec ...)).
62 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
64 2014-02-20 Richard Biener <rguenther@suse.de>
66 * tree-cfg.c (replace_uses_by): Mark altered BBs before
67 doing the substitution.
68 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
70 2014-02-20 Martin Jambor <mjambor@suse.cz>
73 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
74 info when checking whether lattices are bottom.
76 2014-02-20 Richard Biener <rguenther@suse.de>
79 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
82 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
85 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
86 parameter specifying the scaling.
87 (inline_call): Update.
88 (want_inline_recursively): Guard division by zero.
89 (recursive_inlining): Update.
90 * ipa-inline.h (clone_inlined_nodes): Update.
92 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
95 * config/i386/i386.c (classify_argument): Pass structures of size
96 64 bytes or less in register.
98 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
99 Kirill Yukhin <kirill.yukhin@intel.com>
101 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
102 (_mm_rcp28_round_ss): Ditto.
103 (_mm_rsqrt28_round_sd): Ditto.
104 (_mm_rsqrt28_round_ss): Ditto.
105 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
106 (_mm_rcp14_round_ss): Ditto.
107 (_mm_rsqrt14_round_sd): Ditto.
108 (_mm_rsqrt14_round_ss): Ditto.
109 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
110 the first input operand, get rid of match_dup.
111 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
113 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
115 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
116 operand as the first input operand, set type attribute.
117 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
119 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
120 operand as the first input operand, set type attribute.
122 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
124 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
127 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
130 * config/i386/i386.c (construct_container): Remove TFmode check
131 for X86_64_INTEGER_CLASS.
133 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
136 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
137 only when -Wpsabi is enabled.
139 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
142 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
143 passing arrays in registers are the same as for structs, so remove the
144 special case for them.
146 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
148 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
149 destination type, extract only the valid bits if the source type is not
150 integral and has a different mode.
152 2014-02-19 Richard Biener <rguenther@suse.de>
155 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
158 2014-02-19 Richard Biener <rguenther@suse.de>
161 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
162 (ipa_modify_call_arguments): Emit an argument load explicitely and
163 preserve virtual SSA form there and for the replacement call.
164 Do not update SSA form nor free dominance info.
166 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
168 * ipa.c (function_and_variable_visibility): Also clear WEAK
169 flag when disolving COMDAT_GROUP.
171 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
173 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
174 * ipa-prop.c (ipa_set_jf_known_type): Return early when
176 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
177 do more sanity checks.
178 (detect_type_change): Return true when giving up early.
179 (compute_complex_assign_jump_func): Fix type parameter of
181 (compute_complex_ancestor_jump_func): Likewise.
182 (update_jump_functions_after_inlining): Fix updating of
184 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
186 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
188 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
189 inline clones when edge disappears.
191 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
194 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
195 Split 64-bit moves into 2 patterns. Do not allow the use of
196 direct move for TDmode in little endian, since the decimal value
197 has little endian bytes within a word, but the 64-bit pieces are
198 ordered in a big endian fashion, and normal subreg's of TDmode are
200 (mov<mode>_64bit_dm): Likewise.
201 (movtd_64bit_nodm): Likewise.
203 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
205 PR tree-optimization/60174
206 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
207 statement of an SSA_NAME that occurs in an abnormal PHI node.
209 2014-02-18 Jakub Jelinek <jakub@redhat.com>
212 * final.c (SEEN_BB): Remove.
213 (SEEN_NOTE, SEEN_EMITTED): Renumber.
214 (final_scan_insn): Don't force_source_line on second
215 NOTE_INSN_BASIC_BLOCK.
217 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
220 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
221 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
222 (type_natural_mode): Warn ABI change when %zmm register is not
223 available for AVX512F vector value passing.
225 2014-02-18 Kai Tietz <ktietz@redhat.com>
228 * config/i386/i386.c (ix86_expand_prologue): Use value in
229 rax register as displacement when restoring %r10 or %rax.
230 Fix wrong offset when restoring both registers.
232 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
234 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
235 assertion with conditional return.
237 2014-02-18 Jakub Jelinek <jakub@redhat.com>
238 Uros Bizjak <ubizjak@gmail.com>
241 * config/i386/driver-i386.c (host_detect_local_cpu): If
242 YMM state is not saved by the OS, also clear has_f16c. Move
243 CPUID 0x80000001 handling before YMM state saving checking.
245 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
247 PR rtl-optimization/58960
248 * haifa-sched.c (alloc_global_sched_pressure_data): New,
249 factored out from ...
250 (sched_init): ... here.
251 (free_global_sched_pressure_data): New, factored out from ...
252 (sched_finish): ... here.
253 * sched-int.h (free_global_sched_pressure_data): Declare.
254 * sched-rgn.c (nr_regions_initial): New static global.
255 (haifa_find_rgns): Initialize it.
256 (schedule_region): Disable sched-pressure for the newly
259 2014-02-17 Richard Biener <rguenther@suse.de>
261 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
262 release SSA defs of pattern stmts.
264 2014-02-17 Richard Biener <rguenther@suse.de>
266 * tree-inline.c (expand_call_inline): Release the virtual
267 operand defined by the call we are about to inline.
269 2014-02-17 Richard Biener <rguenther@suse.de>
271 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
273 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
274 Ilya Tocar <ilya.tocar@intel.com>
276 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
277 arguments order in builtin.
278 (_mm512_permutexvar_epi64): Ditto.
279 (_mm512_mask_permutexvar_epi64): Ditto
280 (_mm512_maskz_permutexvar_epi32): Ditto
281 (_mm512_permutexvar_epi32): Ditto
282 (_mm512_mask_permutexvar_epi32): Ditto
284 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
286 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
287 (p8_vmrgow): Likewise.
289 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
291 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
294 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
297 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
298 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
299 into 64-bit and 32-bit moves. On 64-bit moves, add support for
300 using direct move instructions on ISA 2.07. Also adjust
301 instruction length for 64-bit.
302 (mov<mode>_64bit, TFmode/TDmode): Likewise.
303 (mov<mode>_32bit, TFmode/TDmode): Likewise.
305 2014-02-15 Alan Modra <amodra@gmail.com>
309 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
310 find_replacement on parts of insn rtl that might be reloaded.
312 2014-02-15 Richard Biener <rguenther@suse.de>
314 PR tree-optimization/60183
315 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
316 (tree_ssa_phiprop): Calculate and free post-dominators.
318 2014-02-14 Jeff Law <law@redhat.com>
320 PR rtl-optimization/60131
321 * ree.c (get_extended_src_reg): New function.
322 (combine_reaching_defs): Use it rather than assuming location of REG.
323 (find_and_remove_re): Verify first operand of extension is
324 a REG before adding the insns to the copy list.
326 2014-02-14 Roland McGrath <mcgrathr@google.com>
328 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
329 * configure: Regenerated.
330 * config.in: Regenerated.
331 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
332 instead of ASM_SHORT.
334 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
335 Richard Earnshaw <rearnsha@arm.com>
337 PR rtl-optimization/59535
338 * lra-constraints.c (process_alt_operands): Encourage alternative
339 when unassigned pseudo class is superset of the alternative class.
340 (inherit_reload_reg): Don't inherit when optimizing for code size.
341 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
342 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
343 modes not less than 4 for Thumb1.
345 2014-02-14 Kyle McMartin <kyle@redhat.com>
348 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
350 2014-02-14 Richard Biener <rguenther@suse.de>
352 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
353 (get_frame_arg): Drop the assert with langhook types_compatible_p.
354 Do not strip INDIRECT_REFs.
356 2014-02-14 Richard Biener <rguenther@suse.de>
359 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
360 DECL_FUNCTION_SPECIFIC_TARGET.
361 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
362 * tree-streamer-out.c (pack_ts_target_option): Remove.
363 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
364 (write_ts_function_decl_tree_pointers): Do not stream
365 DECL_FUNCTION_SPECIFIC_TARGET.
366 * tree-streamer-in.c (unpack_ts_target_option): Remove.
367 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
368 (lto_input_ts_function_decl_tree_pointers): Do not stream
369 DECL_FUNCTION_SPECIFIC_TARGET.
371 2014-02-14 Jakub Jelinek <jakub@redhat.com>
373 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
374 (get_initial_def_for_induction, vectorizable_induction): Ignore
375 debug stmts when looking for exit_phi.
376 (vectorizable_live_operation): Fix up condition.
378 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
380 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
381 nreverse() because it changes the content of original tree list.
383 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
385 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
386 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
388 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
390 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
391 GNU coding standards.
393 2014-02-13 Jakub Jelinek <jakub@redhat.com>
396 * dwarf2out.c (gen_subprogram_die): Don't call
397 add_calling_convention_attribute if subr_die is old_die.
399 2014-02-13 Sharad Singhai <singhai@google.com>
401 * doc/optinfo.texi: Fix order of nodes.
403 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
405 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
406 operands[2], not operands[3].
408 2014-02-13 Richard Biener <rguenther@suse.de>
411 * doc/install.texi (ISL): Update recommended version to 0.12.2,
412 mention the possibility of an in-tree build.
413 (CLooG): Update recommended version to 0.18.1, mention the
414 possibility of an in-tree build and clarify that the ISL
415 bundled with CLooG does not work.
417 2014-02-13 Jakub Jelinek <jakub@redhat.com>
420 * expr.c (compress_float_constant): If x is a hard register,
421 extend into a pseudo and then move to x.
423 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
425 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
426 caused by bad second argument to warning_at() with -mhotpatch and
427 nested functions (e.g. with gfortran).
429 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
431 * opts.c (option_name): Remove "enabled by default" rider.
433 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
435 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
437 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
438 Uros Bizjak <ubizjak@gmail.com>
441 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
442 * configure: Regenerated.
444 2014-02-12 Richard Biener <rguenther@suse.de>
446 * vec.c (vec_prefix::calculate_allocation): Move as
447 inline variant to vec.h.
448 (vec_prefix::calculate_allocation_1): New out-of-line version.
449 * vec.h (vec_prefix::calculate_allocation_1): Declare.
450 (vec_prefix::m_has_auto_buf): Rename to ...
451 (vec_prefix::m_using_auto_storage): ... this.
452 (vec_prefix::calculate_allocation): Inline the easy cases
453 and dispatch to calculate_allocation_1 which doesn't need the
455 (va_heap::reserve): Use gcc_checking_assert.
456 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
457 m_using_auto_storage.
458 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
460 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
461 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
462 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
464 2014-02-12 Richard Biener <rguenther@suse.de>
466 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
467 when we found a dependence.
469 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
471 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
473 (maybe_fold_stmt): ... into this new function.
474 * omp-low.c (lower_omp): Update comment.
476 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
479 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
482 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
484 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
485 identifiers in comments.
486 (cortexa53_extra_costs): Likewise.
487 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
488 (cortexa7_extra_costs): Likewise.
489 (cortexa12_extra_costs): Likewise.
490 (cortexa15_extra_costs): Likewise.
491 (v7m_extra_costs): Likewise.
493 2014-02-12 Richard Biener <rguenther@suse.de>
496 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
497 of posix_memalign being successful.
498 (lower_stmt): Restrict lowering of posix_memalign to when
499 -ftree-bit-ccp is enabled.
501 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
503 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
505 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
507 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
509 PR rtl-optimization/60116
510 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
511 other_insn once the combination has been validated.
513 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
516 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
518 * ipa-devirt.c: Include demangle.h
519 (odr_violation_reported): New static variable.
520 (add_type_duplicate): Update odr_violations.
521 (maybe_record_node): Add completep parameter; update it.
522 (record_target_from_binfo): Add COMPLETEP parameter;
524 (possible_polymorphic_call_targets_1): Likewise.
525 (struct polymorphic_call_target_d): Add nonconstruction_targets;
526 rename FINAL to COMPLETE.
527 (record_targets_from_bases): Sanity check we found the binfo;
528 fix COMPLETEP updating.
529 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
530 parameter, fix computing of COMPLETEP.
531 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
532 at LTO time do demangling.
533 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
534 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
536 (gimple_get_virt_method_for_binfo): Likewise.
537 * gimple-fold.h (gimple_get_virt_method_for_binfo,
538 gimple_get_virt_method_for_vtable): Update prototypes.
540 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
543 * genautomata.c (add_presence_absence): Fix typo with
544 {final_}presence_list.
546 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
549 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
550 for VSX/Altivec vectors that land in GPR registers.
552 2014-02-11 Richard Henderson <rth@redhat.com>
553 Jakub Jelinek <jakub@redhat.com>
556 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
557 around drhs if type conversion to lacc->type is not useless.
559 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
561 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
563 (cortex-a57.cortex-a53): Likewise.
564 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
566 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
568 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
571 2014-02-11 Renlin Li <Renlin.Li@arm.com>
573 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
574 add_options_for_arm_vfp3.
576 2014-02-11 Jeff Law <law@redhat.com>
579 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
580 object with an undesirable mode.
582 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
585 * config/i386/sol2-9.h: New file.
586 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
587 *-*-solaris2.9*): Use it.
589 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
591 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
592 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
594 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
596 * config/microblaze/microblaze.c: Extend mcpu version format
598 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
600 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
602 2014-02-10 Richard Henderson <rth@redhat.com>
605 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
606 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
607 ms-abi vs -mno-accumulate-outgoing-args.
608 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
609 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
612 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
615 * cfgexpand.c (expand_asm_operands): Attach source location to
616 ASM_INPUT rtx objects.
617 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
619 2014-02-10 Nick Clifton <nickc@redhat.com>
621 * config/mn10300/mn10300.c (popcount): New function.
622 (mn10300_expand_prologue): Include saved registers in stack usage
625 2014-02-10 Jeff Law <law@redhat.com>
628 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
629 when changing the SET_DEST of a prior insn to avoid an input reload.
631 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
633 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
634 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
635 -mcall-openbsd, or -mcall-linux.
636 (CC1_ENDIAN_BIG_SPEC): Remove.
637 (CC1_ENDIAN_LITTLE_SPEC): Remove.
638 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
639 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
640 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
641 and %cc1_endian_default.
642 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
644 2014-02-10 Richard Biener <rguenther@suse.de>
646 PR tree-optimization/60115
647 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
648 MEM_REF handling. Properly verify that the accesses are not
649 out of the objects bound.
651 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
653 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
656 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
658 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
659 proper constants and fix formatting.
660 (possible_polymorphic_call_targets): Fix formatting.
662 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
663 Ilya Tocar <ilya.tocar@intel.com>
665 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
666 (_mm512_loadu_epi32): Renamed into...
667 (_mm512_loadu_si512): This.
668 (_mm512_storeu_epi32): Renamed into...
669 (_mm512_storeu_si512): This.
670 (_mm512_maskz_ceil_ps): Removed.
671 (_mm512_maskz_ceil_pd): Ditto.
672 (_mm512_maskz_floor_ps): Ditto.
673 (_mm512_maskz_floor_pd): Ditto.
674 (_mm512_floor_round_ps): Ditto.
675 (_mm512_floor_round_pd): Ditto.
676 (_mm512_ceil_round_ps): Ditto.
677 (_mm512_ceil_round_pd): Ditto.
678 (_mm512_mask_floor_round_ps): Ditto.
679 (_mm512_mask_floor_round_pd): Ditto.
680 (_mm512_mask_ceil_round_ps): Ditto.
681 (_mm512_mask_ceil_round_pd): Ditto.
682 (_mm512_maskz_floor_round_ps): Ditto.
683 (_mm512_maskz_floor_round_pd): Ditto.
684 (_mm512_maskz_ceil_round_ps): Ditto.
685 (_mm512_maskz_ceil_round_pd): Ditto.
686 (_mm512_expand_pd): Ditto.
687 (_mm512_expand_ps): Ditto.
688 * config/i386/i386.c (ix86_builtins): Remove
689 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
691 * config/i386/predicates.md (const1256_operand): New.
692 (const_1_to_2_operand): Ditto.
693 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
694 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
695 (*avx512pf_gatherpf<mode>sf): Ditto.
696 (avx512pf_gatherpf<mode>df): Ditto.
697 (*avx512pf_gatherpf<mode>df_mask): Ditto.
698 (*avx512pf_gatherpf<mode>df): Ditto.
699 (avx512pf_scatterpf<mode>sf): Ditto.
700 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
701 (*avx512pf_scatterpf<mode>sf): Ditto.
702 (avx512pf_scatterpf<mode>df): Ditto.
703 (*avx512pf_scatterpf<mode>df_mask): Ditto.
704 (*avx512pf_scatterpf<mode>df): Ditto.
705 (avx512f_expand<mode>): Removed.
706 (<shift_insn><mode>3<mask_name>): Change predicate type.
708 2014-02-08 Jakub Jelinek <jakub@redhat.com>
710 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
711 not at the end of datarefs vector use ordered_remove to avoid
712 reordering datarefs vector.
715 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
716 mark local addressable non-static vars as GOVD_PRIVATE
717 instead of GOVD_LOCAL.
718 * omp-low.c (lower_omp_for): Move gimple_bind_vars
719 and BLOCK_VARS of gimple_bind_block to new_stmt rather
723 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
724 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
725 assume_aligned or alloc_align attributes.
726 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
727 arguments. Handle also assume_aligned and alloc_align attributes.
728 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
729 calls to functions with assume_aligned or alloc_align attributes.
730 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
732 2014-02-08 Terry Guo <terry.guo@arm.com>
734 * doc/invoke.texi: Document ARM -march=armv7e-m.
736 2014-02-08 Jakub Jelinek <jakub@redhat.com>
738 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
739 flag on __cilkrts_rethrow builtin.
742 * ipa-cp.c (determine_versionability): Fail at -O0
743 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
744 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
747 2014-02-04 Jakub Jelinek <jakub@redhat.com>
750 * tree-inline.c (copy_forbidden): Fail for
751 __attribute__((optimize (0))) functions.
753 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
755 * varpool.c: Include pointer-set.h.
756 (varpool_remove_unreferenced_decls): Variables in other partitions
757 will not be output; be however careful to not lose information
760 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
762 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
763 lookup in the vtable constructor.
765 2014-02-07 Jeff Law <law@redhat.com>
768 * config/m68k/m68k.md (ashldi_extsi): Turn into a
769 define_insn_and_split.
771 * ipa-inline.c (inline_small_functions): Fix typos.
773 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
775 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
776 (s390_can_use_return_insn): Declare.
777 * config/s390/s390.h (EPILOGUE_USES): Define.
778 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
780 (s390_chunkify_start): Handle return JUMP_LABELs.
781 (s390_early_mach): Emit a main_pool instruction on the entry edge.
782 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
783 (s390_can_use_return_insn): New functions.
784 (s390_fix_long_loop_prediction): Handle conditional returns.
785 (TARGET_SET_UP_BY_PROLOGUE): Define.
786 * config/s390/s390.md (ANY_RETURN): New code iterator.
787 (*creturn, *csimple_return, return, simple_return): New patterns.
789 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
791 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
792 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
793 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
794 REG_CFA_RESTORE list when deciding not to restore a register.
796 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
798 * config/s390/s390.c: Include tree-pass.h and context.h.
799 (s390_early_mach): New function, split out from...
800 (s390_emit_prologue): ...here.
801 (pass_data_s390_early_mach): New pass structure.
802 (pass_s390_early_mach): New class.
803 (s390_option_override): Create and register early_mach pass.
806 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
808 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
809 to match for the exit block.
811 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
813 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
814 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
815 Reject misaligned operands.
817 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
819 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
821 2014-02-07 Richard Biener <rguenther@suse.de>
824 * gimple-low.c (lower_builtin_posix_memalign): New function.
825 (lower_stmt): Call it to lower posix_memalign in a way
826 to make alignment info accessible.
828 2014-02-07 Jakub Jelinek <jakub@redhat.com>
831 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
832 __builtin_setjmp_receiver.
834 2014-02-07 Richard Biener <rguenther@suse.de>
837 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
838 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
839 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
840 Handle BUILT_IN_POSIX_MEMALIGN.
841 (find_func_clobbers): Likewise.
842 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
843 (call_may_clobber_ref_p_1): Likewise.
845 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
848 * ipa-devirt.c (record_target_from_binfo): Remove overactive
851 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
854 * lto-cgraph.c (lto_output_node): Use
855 symtab_get_symbol_partitioning_class.
856 (lto_output_varpool_node): likewise.
857 (symtab_get_symbol_partitioning_class): Move here from
859 * cgraph.h (symbol_partitioning_class): Likewise.
860 (symtab_get_symbol_partitioning_class): Declare.
862 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
864 * ggc.h (ggc_internal_cleared_alloc): New macro.
865 * vec.h (vec_safe_copy): Handle memory stats.
866 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
867 * target-globals.c (save_target_globals): Likewise.
869 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
872 * expr.c (emit_move_resolve_push): Export; be bit more selective
873 on when to clear alias set.
874 * expr.h (emit_move_resolve_push): Declare.
875 * function.h (struct function): Add tail_call_marked.
876 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
877 * config/i386/i386-protos.h (ix86_expand_push): Remove.
878 * config/i386/i386.md (TImode move expander): De not call
880 (FP push expanders): Preserve memory attributes.
881 * config/i386/sse.md (push<mode>1): Remove.
882 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
883 (ix86_expand_push): Remove.
884 * config/i386/mmx.md (push<mode>1): Remove.
886 2014-02-06 Jakub Jelinek <jakub@redhat.com>
888 PR rtl-optimization/60030
889 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
890 lopart with paradoxical subreg before shifting it up by hprec.
892 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
894 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
895 Remove extra newline at end of file.
896 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
897 (arm_issue_rate): Handle cortexa57.
898 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
899 (cortex-a57.cortex-a53): Likewise.
901 2014-02-06 Jakub Jelinek <jakub@redhat.com>
904 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
905 don't record in REG_FRAME_RELATED_EXPR registers not set in that
907 (arm_expand_prologue): Adjust all callers.
908 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
909 info, registers also at the lowest numbered registers side. Use
910 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
914 * var-tracking.c (adjust_mems): Before adding a SET to
915 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
917 2014-02-06 Alan Modra <amodra@gmail.com>
920 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
921 change SDmode to DDmode when lra_in_progress.
923 2014-02-06 Jakub Jelinek <jakub@redhat.com>
926 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
927 free_data_ref on the dr first, and before goto again also set dr
928 to the next dr. For simd_lane_access, free old datarefs[i] before
929 overwriting it. For get_vectype_for_scalar_type failure, don't
930 free_data_ref if simd_lane_access.
932 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
935 * tree.h (opts_for_fn): New inline function.
936 (opt_for_fn): Define.
937 * config/i386/i386.c (ix86_function_regparm): Use
938 opt_for_fn (decl, optimize) instead of optimize.
940 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
942 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
943 for SYMBOL_REF in large memory model.
945 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
947 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
949 (cortex-a57): Likewise.
950 (cortex-a57.cortex-a53): Likewise.
952 2014-02-06 Yury Gribov <y.gribov@samsung.com>
953 Kugan Vivekanandarajah <kuganv@linaro.org>
955 * config/arm/arm.c (arm_vector_alignment_reachable): Check
957 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
959 2014-02-06 Richard Biener <rguenther@suse.de>
961 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
962 set_loop_copy and initialize_original_copy_tables.
964 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
966 * config/aarch64/aarch64-simd.md
967 (aarch64_ashr_simddi): Change QI to SI.
969 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
970 Jakub Jelinek <jakub@redhat.com>
973 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
976 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
978 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
979 CODE_FOR_altivec_vpku[hw]um to
980 CODE_FOR_altivec_vpku[hw]um_direct.
981 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
982 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
983 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
984 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
986 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
988 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
989 generation for -maltivec=be.
990 (altivec_vsumsws): Simplify redundant test.
992 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
994 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
995 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
996 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
997 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
999 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1001 (altivec_vpks<VI_char>ss): Likewise.
1002 (altivec_vpks<VI_char>us): Likewise.
1003 (altivec_vpku<VI_char>us): Likewise.
1004 (altivec_vpku<VI_char>um): Likewise.
1005 (altivec_vpku<VI_char>um_direct): New (copy of
1006 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1008 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1009 target is little endian and -maltivec=be is not specified.
1010 (*altivec_vupkhs<VU_char>_direct): New (copy of
1011 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1012 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1013 target is little endian and -maltivec=be is not specified.
1014 (*altivec_vupkls<VU_char>_direct): New (copy of
1015 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1016 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1017 little endian and -maltivec=be is not specified.
1018 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1019 little endian and -maltivec=be is not specified.
1021 2014-02-05 Richard Henderson <rth@redhat.com>
1024 * combine-stack-adj.c: Revert r206943.
1025 * sched-int.h (struct deps_desc): Add last_args_size.
1026 * sched-deps.c (init_deps): Initialize it.
1027 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1028 contain REG_ARGS_SIZE notes.
1030 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1032 * lto-cgraph.c (asm_nodes_output): Make global.
1033 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1034 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1035 (driver_handle_option): Handle OPT_fwpa.
1037 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1040 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1041 a comment typo and formatting issue. If odr_hash hasn't been
1042 created, return vNULL and set *completep to false.
1045 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1046 bb with no successors.
1048 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1051 * doc/invoke.texi (-march): Clarify documentation for ARM.
1055 2014-02-05 Richard Biener <rguenther@suse.de>
1057 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1058 when not vectorizing because of too many alias checks.
1059 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1060 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1062 2014-02-05 Nick Clifton <nickc@redhat.com>
1064 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1065 accept extended registers in any mode when compiling for the MN10300.
1067 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1069 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1070 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1071 sanitization attributes.
1072 (can_inline_edge_p): Likewise.
1073 (sanitize_attrs_match_for_inline_p): New function.
1075 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1077 * ipa-prop.c (detect_type_change): Shor circuit testing of
1078 type changes on THIS pointer.
1080 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1083 * config/pa/pa.c (legitimize_tls_address): Return original address
1084 if not passed a SYMBOL_REF rtx.
1085 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1087 (pa_emit_move_sequence): Simplify TLS source operands.
1088 (pa_legitimate_constant_p): Reject all TLS constants.
1089 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1090 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1092 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1094 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1095 groups when we know they are controlled by LTO.
1096 * varasm.c (default_binds_local_p_1): If object is in other partition,
1097 it will be resolved locally.
1099 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1101 * config/host-linux.c (linux_gt_pch_use_address): Don't
1102 use SSIZE_MAX because it is not always defined.
1104 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1107 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1108 threshold for pseudo splitting.
1109 (update_ebb_live_info): Process call argument hard registers and
1110 hard registers from insn definition too.
1111 (max_small_class_regs_num): New constant.
1112 (inherit_in_ebb): Update live hard regs through EBBs. Update
1113 reloads_num only for small register classes. Don't split for
1116 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1119 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1122 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1124 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1127 2014-02-04 Marek Polacek <polacek@redhat.com>
1129 * gdbinit.in (pel): Define.
1131 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1133 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1136 2014-02-04 Richard Biener <rguenther@suse.de>
1139 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1140 in function context local.
1141 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1142 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1143 similar to LTO_imported_decl_ref.
1145 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1147 PR tree-optimization/60002
1148 * cgraphclones.c (build_function_decl_skip_args): Clear
1151 PR tree-optimization/60023
1152 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1153 false to gsi_replace.
1154 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1155 has been in some EH region and vec_stmt could throw, add
1156 vec_stmt into the same EH region.
1157 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1158 has no lhs, ignore it.
1159 * internal-fn.c (expand_MASK_LOAD): Likewise.
1162 * tree-inline.c (copy_forbidden): Fail for
1163 __attribute__((optimize (0))) functions.
1166 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1167 is set, copy one less argument.
1168 (expand_simd_clones): Don't subtract clone_info->inbranch
1169 from simd_clone_struct_alloc argument.
1171 PR rtl-optimization/57915
1172 * recog.c (simplify_while_replacing): If all unary/binary/relational
1173 operation arguments are constant, attempt to simplify those.
1176 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1177 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1179 2014-02-04 Richard Biener <rguenther@suse.de>
1181 PR tree-optimization/60012
1182 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1183 TBAA disambiguation to all DDRs.
1185 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1188 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1189 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1191 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1194 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1196 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1198 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1199 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1201 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1204 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1205 to figure out targets of polymorphic calls with known decl.
1206 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1207 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1208 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1209 (get_polymorphic_call_info): ... here.
1210 (get_polymorphic_call_info_from_invariant): New function.
1212 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1214 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1215 lookup via vtable pointer; check for type consistency
1216 and turn inconsitent facts into UNREACHABLE.
1217 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1218 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1219 type inconsistent querries; return UNREACHABLE instead.
1221 2014-02-03 Richard Henderson <rth@twiddle.net>
1224 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1225 already processed this node.
1226 (normalize_one_pred_1): Pass along mark_set.
1227 (normalize_one_pred): Create and destroy a pointer_set_t.
1228 (normalize_one_pred_chain): Likewise.
1230 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1232 PR gcov-profile/58602
1233 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1235 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1238 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1239 -fno-devirtualize; try to devirtualize by the knowledge of
1240 virtual table pointer given by aggregate propagation.
1241 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1242 (ipa_print_node_jump_functions): Dump also offset that
1243 is relevant for polymorphic calls.
1244 (determine_known_aggregate_parts): Add arg_type parameter; use it
1245 instead of determining the type from pointer type.
1246 (ipa_compute_jump_functions_for_edge): Update call of
1247 determine_known_aggregate_parts.
1248 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1249 (gimple_get_virt_method_for_binfo): ... here; simplify using
1250 vtable_pointer_value_to_vtable.
1251 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1252 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1253 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1254 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1256 (vtable_pointer_value_to_binfo): ... here.
1257 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1259 2014-02-03 Teresa Johnson <tejohnson@google.com>
1261 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1262 redef of outer loop index variable.
1264 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1268 * doc/extend.texi (Function Attributes): Typo.
1270 2014-02-03 Cong Hou <congh@google.com>
1272 PR tree-optimization/60000
1273 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1274 if the vectorized statement is a store. A store statement can only
1275 appear at the end of pattern statements.
1277 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1279 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1280 (ix86_option_override_internal): Default long double to 64-bit for
1281 32-bit Bionic and to 128-bit for 64-bit Bionic.
1283 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1284 TARGET_LONG_DOUBLE_128 is true.
1285 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1287 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1288 (mlong-double-64): Negate -mlong-double-128.
1289 (mlong-double-128): New option.
1291 * config/i386/i386-c.c (ix86_target_macros): Define
1292 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1294 * doc/invoke.texi: Document -mlong-double-128.
1296 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1298 PR rtl-optimization/60024
1299 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1301 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1303 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1305 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1307 PR rtl-optimization/57662
1308 * sel-sched.c (code_motion_path_driver): Do not mark already not
1309 existing blocks in the visiting bitmap.
1311 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1313 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1314 on the insn being emitted.
1316 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1317 Will Deacon <will.deacon@arm.com>
1319 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1321 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1323 * config/arm/arm-tables.opt: Regenerate.
1325 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1327 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1328 for vector types other than V16QImode.
1329 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1330 define_expand, and call altivec_expand_vec_perm_le when producing
1331 code with little endian element order.
1332 (*altivec_vperm_<mode>_internal): New insn having previous
1333 behavior of altivec_vperm_<mode>.
1334 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1335 altivec_expand_vec_perm_le when producing code with little endian
1337 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1338 behavior of altivec_vperm_<mode>_uns.
1340 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1342 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1343 (altivec_vsumsws): Add handling for -maltivec=be with a little
1345 (altivec_vsumsws_direct): New.
1346 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1347 gen_altivec_vsumsws.
1349 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1351 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1352 vtable_pointer_value_to_binfo): New functions.
1353 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1354 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1356 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1358 * config/nios2/nios2.md (load_got_register): Initialize GOT
1359 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1360 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1362 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1364 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1365 preserverd by passthrough, do not propagate the type.
1367 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1369 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1370 (mips_atomic_assign_expand_fenv): New function.
1371 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1373 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1375 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1376 (__builtin_mips_set_fcsr): Likewise.
1377 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1378 MIPS_USI_FTYPE_VOID.
1379 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1380 (mips16_expand_set_fcsr): Likewise.
1381 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1382 (mips16_set_fcsr_stub): Likewise.
1383 (mips16_get_fcsr_one_only_stub): New class.
1384 (mips16_set_fcsr_one_only_stub): Likewise.
1385 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1386 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1387 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1388 (hard_float): New availability predicate.
1389 (mips_builtins): Add get_fcsr and set_fcsr.
1390 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1391 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1392 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1393 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1394 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1397 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1399 * config/mips/mips.c (mips_one_only_stub): New class.
1400 (mips_need_mips16_rdhwr_p): Replace with...
1401 (mips16_rdhwr_stub): ...this new variable.
1402 (mips16_stub_call_address): New function.
1403 (mips16_rdhwr_one_only_stub): New class.
1404 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1405 (mips_output_mips16_rdhwr): Delete.
1406 (mips_finish_stub): New function.
1407 (mips_code_end): Use it to handle rdhwr stubs.
1409 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1412 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1413 when calculating size of integer atomic types.
1415 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1417 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1419 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1421 PR tree-optimization/60003
1422 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1423 * profile.c (branch_prob): Use gimple_call_builtin_p
1424 to check for BUILT_IN_SETJMP_RECEIVER.
1425 * tree-inline.c (copy_bb): Call notice_special_calls.
1427 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1430 * lra-constraints.c (process_alt_operands): Update reload_sum only
1433 2014-01-31 Richard Henderson <rth@redhat.com>
1436 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1437 until after else_eh is processed.
1439 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1441 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1442 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1443 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1444 in smmintrin.h, remove them.
1445 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1446 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1447 * config/i386/i386.md (ROUND_SAE): Fix value.
1448 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1449 (const48_operand): New.
1450 * config/i386/subst.md (round), (round_expand): Use
1451 const_4_or_8_to_11_operand.
1452 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1454 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1456 * config/i386/constraints.md (Yk): Swap meaning with k.
1457 * config/i386/i386.md (movhi_internal): Change Yk to k.
1458 (movqi_internal): Ditto.
1459 (*k<logic><mode>): Ditto.
1462 (kandn<mode>): Ditto.
1463 (*<code>hi_1): Ditto.
1464 (*<code>qi_1): Ditto.
1465 (kxnor<mode>): Ditto.
1466 (kortestzhi): Ditto.
1467 (kortestchi): Ditto.
1469 (*one_cmplhi2_1): Ditto.
1470 (*one_cmplqi2_1): Ditto.
1471 * config/i386/sse.md (): Change k to Yk.
1472 (avx512f_load<mode>_mask): Ditto.
1473 (avx512f_blendm<mode>): Ditto.
1474 (avx512f_store<mode>_mask): Ditto.
1475 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1476 (avx512f_storedqu<mode>_mask): Ditto.
1477 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1479 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1480 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1481 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1482 (avx512f_maskcmp<mode>3): Ditto.
1483 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1484 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1485 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1486 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1487 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1488 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1489 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1490 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1491 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1492 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1493 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1494 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1495 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1496 (vec_extract_lo_<mode>_maskm): Ditto.
1497 (vec_extract_hi_<mode>_maskm): Ditto.
1498 (avx512f_vternlog<mode>_mask): Ditto.
1499 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1500 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1501 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1502 (avx512f_<code>v8div16qi2_mask): Ditto.
1503 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1504 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1505 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1506 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1507 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1508 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1509 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1510 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1511 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1512 (avx512cd_maskb_vec_dupv8di): Ditto.
1513 (avx512cd_maskw_vec_dupv16si): Ditto.
1514 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1515 (avx512f_vpermi2var<mode>3_mask): Ditto.
1516 (avx512f_vpermi2var<mode>3_mask): Ditto.
1517 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1518 (*avx512f_gathersi<mode>): Ditto.
1519 (*avx512f_gathersi<mode>_2): Ditto.
1520 (*avx512f_gatherdi<mode>): Ditto.
1521 (*avx512f_gatherdi<mode>_2): Ditto.
1522 (*avx512f_scattersi<mode>): Ditto.
1523 (*avx512f_scatterdi<mode>): Ditto.
1524 (avx512f_compress<mode>_mask): Ditto.
1525 (avx512f_compressstore<mode>_mask): Ditto.
1526 (avx512f_expand<mode>_mask): Ditto.
1527 * config/i386/subst.md (mask): Change k to Yk.
1528 (mask_scalar_merge): Ditto.
1531 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1533 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1535 2014-01-31 Richard Biener <rguenther@suse.de>
1538 * builtins.c (fold_builtin_memory_op): Make sure to not
1539 use a floating-point mode or a boolean or enumeral type for
1542 2014-01-30 DJ Delorie <dj@redhat.com>
1544 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1545 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1546 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1547 whenever main() has an epilogue.
1549 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1551 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1552 unused variable "field".
1553 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1554 (vsx_mergeh_<mode>): Likewise.
1555 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1556 (altivec_vmrghh): Likewise.
1557 (altivec_vmrghw): Likewise.
1558 (altivec_vmrglb): Likewise.
1559 (altivec_vmrglh): Likewise.
1560 (altivec_vmrglw): Likewise.
1561 (altivec_vspltb): Add missing uses.
1562 (altivec_vsplth): Likewise.
1563 (altivec_vspltw): Likewise.
1564 (altivec_vspltsf): Likewise.
1566 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1569 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1570 frame related instructions.
1572 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1574 PR rtl-optimization/59959
1575 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1576 any reload of register whose subreg is invalid.
1578 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1580 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1581 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1582 Add missing return type - void.
1584 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1586 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1587 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1588 remove element index adjustment for endian (now handled in vsx.md
1590 (altivec_expand_vec_perm_const): Use
1591 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1592 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1593 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1594 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1595 define_expand and a new define_insn *altivec_vspltb_internal;
1596 adjust for -maltivec=be on a little endian target.
1597 (altivec_vspltb_direct): New.
1598 (altivec_vsplth): Divide into a define_expand and a new
1599 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1600 little endian target.
1601 (altivec_vsplth_direct): New.
1602 (altivec_vspltw): Divide into a define_expand and a new
1603 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1604 little endian target.
1605 (altivec_vspltw_direct): New.
1606 (altivec_vspltsf): Divide into a define_expand and a new
1607 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1608 a little endian target.
1610 2014-01-30 Richard Biener <rguenther@suse.de>
1612 PR tree-optimization/59993
1613 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1614 can propagate form the earlier stmt and avoid the transform
1615 when the intermediate result is needed.
1617 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1619 * README.Portability: Fix typo.
1621 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1623 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1624 comparison_operator with ordered_comparison_operator.
1626 2014-01-30 Nick Clifton <nickc@redhat.com>
1628 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1629 Rename to mn10300_store_multiple_regs.
1630 * config/mn10300/mn10300.c: Likewise.
1631 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1632 store_multiple_regs.
1633 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1634 Call mn10300_store_multiple_regs.
1636 2014-01-30 Nick Clifton <nickc@redhat.com>
1637 DJ Delorie <dj@redhat.com>
1639 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1640 %fp 2 to keep registers after it properly word-aligned.
1641 (rl78_alloc_physical_registers_umul): Handle the case where both
1642 input operands are the same.
1644 2014-01-30 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/59903
1647 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1650 2014-01-30 Jason Merrill <jason@redhat.com>
1653 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1656 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1658 2014-01-30 Richard Biener <rguenther@suse.de>
1660 PR tree-optimization/59951
1661 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1663 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1666 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1667 SFmode to DFmode case.
1669 2014-01-29 DJ Delorie <dj@redhat.com>
1671 * config/msp430/msp430.opt (-minrt): New.
1672 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1674 (ENDFILE_SPEC): Likewise.
1676 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1678 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1679 (estimate_function_body_sizes): Use it.
1681 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1684 * dwarf2out.c (is_cxx_auto): New.
1685 (is_base_type): Use it.
1686 (gen_type_die_with_usage): Likewise.
1688 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1690 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1691 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1692 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1693 -maltivec=be with LE targets.
1694 (vsx_mergeh_<mode>): Likewise.
1695 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1696 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1697 (altivec_vmrghb): Replace with define_expand and new
1698 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1699 (altivec_vmrghb_direct): New define_insn.
1700 (altivec_vmrghh): Replace with define_expand and new
1701 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1702 (altivec_vmrghh_direct): New define_insn.
1703 (altivec_vmrghw): Replace with define_expand and new
1704 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1705 (altivec_vmrghw_direct): New define_insn.
1706 (*altivec_vmrghsf): Adjust for endianness.
1707 (altivec_vmrglb): Replace with define_expand and new
1708 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1709 (altivec_vmrglb_direct): New define_insn.
1710 (altivec_vmrglh): Replace with define_expand and new
1711 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1712 (altivec_vmrglh_direct): New define_insn.
1713 (altivec_vmrglw): Replace with define_expand and new
1714 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1715 (altivec_vmrglw_direct): New define_insn.
1716 (*altivec_vmrglsf): Adjust for endianness.
1717 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1718 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1719 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1720 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1721 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1722 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1723 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1724 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1726 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1728 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1729 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1732 2014-01-29 Richard Biener <rguenther@suse.de>
1734 PR tree-optimization/58742
1735 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1736 associate_pointerplus_align.
1737 (associate_pointerplus_diff): New function.
1738 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1739 and associate_pointerplus_diff.
1741 2014-01-29 Richard Biener <rguenther@suse.de>
1743 * lto-streamer.h (LTO_major_version): Bump to 3.
1744 (LTO_minor_version): Reset to 0.
1746 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1748 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1749 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1750 (arm_file_start): Generate correct asm header for armv7ve.
1751 * config/arm/bpabi.h: Add multilib support for armv7ve.
1752 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1753 and cortex-a15 to armv7ve.
1754 * config/arm/t-aprofile: Add multilib support for armv7ve.
1755 * doc/invoke.texi: Document -march=armv7ve.
1757 2014-01-29 Richard Biener <rguenther@suse.de>
1759 PR tree-optimization/58742
1760 * tree-ssa-forwprop.c (associate_plusminus): Return true
1761 if we changed sth, defer EH cleanup to ...
1762 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1763 (simplify_mult): New function.
1765 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1768 PR tree-optimization/59920
1769 * tree.c (build_common_builtin_nodes): Remove
1770 __builtin_setjmp_dispatcher initialization.
1771 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1772 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1773 instead of gsi_after_labels + manually skipping debug stmts.
1774 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1775 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1776 * tree-inline.c (copy_edges_for_bb): Remove
1777 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1778 argument. Ignore computed_goto_p stmts. Don't call
1779 make_abnormal_goto_edges. If a call might need abnormal edges
1780 for non-local gotos, see if it already has an edge to
1781 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1782 with true argument, don't do anything then, otherwise add
1783 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1784 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1786 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1787 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1788 (lower_stmt): Don't set data->calls_builtin_setjmp.
1789 (lower_builtin_setjmp): Adjust comment.
1790 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1791 * tree-cfg.c (found_computed_goto): Remove.
1792 (factor_computed_gotos): Remove.
1793 (make_goto_expr_edges): Return bool, true for computed gotos.
1794 Don't call make_abnormal_goto_edges.
1795 (build_gimple_cfg): Don't set found_computed_goto, don't call
1796 factor_computed_gotos.
1797 (computed_goto_p): No longer static.
1798 (make_blocks): Don't set found_computed_goto.
1799 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1800 (make_edges): If make_goto_expr_edges returns true, push bb
1801 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1802 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1803 vector. Record mapping between bbs and OpenMP regions if there
1804 are any, adjust make_gimple_omp_edges caller. Call
1805 handle_abnormal_edges.
1806 (make_abnormal_goto_edges): Remove.
1807 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1808 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1809 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1810 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1811 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1812 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1813 filling *region also set *region_idx to (*region)->entry->index.
1816 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1817 For REGs set ORIGINAL_REGNO.
1819 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1821 * doc/md.texi: Mention that a target shouldn't implement
1822 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1825 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1827 PR tree-optimization/59594
1828 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1829 a copy of the datarefs vector rather than the vector itself.
1831 2014-01-28 Jason Merrill <jason@redhat.com>
1834 * dwarf2out.c (auto_die): New static.
1835 (gen_type_die_with_usage): Handle C++1y 'auto'.
1836 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1839 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1842 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1843 (SPEC_X32): Likewise.
1844 (SPEC_64): Likewise.
1845 * config/i386/i386.c (ix86_option_override_internal): Turn off
1846 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1848 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1849 * config/i386/i386.h (TARGET_16BIT): New macro.
1850 (TARGET_16BIT_P): Likewise.
1851 * config/i386/i386.opt: Add m16.
1852 * doc/invoke.texi: Document -m16.
1854 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1856 PR preprocessor/59935
1857 * input.c (location_get_source_line): Bail out on when line number
1858 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1860 2014-01-28 Richard Biener <rguenther@suse.de>
1862 PR tree-optimization/58742
1863 * tree-ssa-forwprop.c (associate_plusminus): Handle
1864 pointer subtraction of the form (T)(P + A) - (T)P.
1866 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1868 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1871 2014-01-28 Richard Biener <rguenther@suse.de>
1874 2014-01-28 Richard Biener <rguenther@suse.de>
1876 PR rtl-optimization/45364
1877 PR rtl-optimization/59890
1878 * var-tracking.c (local_get_addr_clear_given_value): Handle
1879 already cleared slot.
1880 (val_reset): Handle not allocated local_get_addr_cache.
1881 (vt_find_locations): Use post-order on the inverted CFG.
1883 2014-01-28 Richard Biener <rguenther@suse.de>
1885 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1887 2014-01-28 Richard Biener <rguenther@suse.de>
1889 PR rtl-optimization/45364
1890 PR rtl-optimization/59890
1891 * var-tracking.c (local_get_addr_clear_given_value): Handle
1892 already cleared slot.
1893 (val_reset): Handle not allocated local_get_addr_cache.
1894 (vt_find_locations): Use post-order on the inverted CFG.
1896 2014-01-28 Alan Modra <amodra@gmail.com>
1898 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1899 * configure.ac <recursive call for build != host>: Define
1900 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1901 and LD_FOR_BUILD too.
1902 * configure: Regenerate.
1904 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1906 * config/i386/i386.c (get_builtin_code_for_version): Separate
1907 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1908 Broadwell from Haswell.
1910 2014-01-27 Steve Ellcey <sellcey@mips.com>
1912 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1913 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1914 * config/mips/mips.c (mips_option_override): Change setting
1916 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1917 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1918 Change from Mask to Var.
1920 2014-01-27 Jeff Law <law@redhat.com>
1922 * ipa-inline.c (inline_small_functions): Fix typo.
1924 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1926 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1927 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1928 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1929 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1930 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1931 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1932 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1933 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1934 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1935 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1936 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1937 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1938 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1939 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1940 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1941 (_mm512_storeu_epi64): Ditto.
1942 (_mm512_cmpge_epi32_mask): Ditto.
1943 (_mm512_cmpge_epu32_mask): Ditto.
1944 (_mm512_cmpge_epi64_mask): Ditto.
1945 (_mm512_cmpge_epu64_mask): Ditto.
1946 (_mm512_cmple_epi32_mask): Ditto.
1947 (_mm512_cmple_epu32_mask): Ditto.
1948 (_mm512_cmple_epi64_mask): Ditto.
1949 (_mm512_cmple_epu64_mask): Ditto.
1950 (_mm512_cmplt_epi32_mask): Ditto.
1951 (_mm512_cmplt_epu32_mask): Ditto.
1952 (_mm512_cmplt_epi64_mask): Ditto.
1953 (_mm512_cmplt_epu64_mask): Ditto.
1954 (_mm512_cmpneq_epi32_mask): Ditto.
1955 (_mm512_cmpneq_epu32_mask): Ditto.
1956 (_mm512_cmpneq_epi64_mask): Ditto.
1957 (_mm512_cmpneq_epu64_mask): Ditto.
1958 (_mm512_expand_pd): Ditto.
1959 (_mm512_expand_ps): Ditto.
1960 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1961 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1962 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1963 * config/i386/i386.c (ix86_builtins): Add
1964 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1965 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1966 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1967 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1968 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1969 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1970 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1971 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1972 IX86_BUILTIN_PMOVUSQW512_MEM.
1973 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1974 __builtin_ia32_pmovsqd512mem_mask,
1975 __builtin_ia32_pmovqd512mem_mask,
1976 __builtin_ia32_pmovusqw512mem_mask,
1977 __builtin_ia32_pmovsqw512mem_mask,
1978 __builtin_ia32_pmovqw512mem_mask,
1979 __builtin_ia32_pmovusdw512mem_mask,
1980 __builtin_ia32_pmovsdw512mem_mask,
1981 __builtin_ia32_pmovdw512mem_mask,
1982 __builtin_ia32_pmovqb512mem_mask,
1983 __builtin_ia32_pmovusqb512mem_mask,
1984 __builtin_ia32_pmovsqb512mem_mask,
1985 __builtin_ia32_pmovusdb512mem_mask,
1986 __builtin_ia32_pmovsdb512mem_mask,
1987 __builtin_ia32_pmovdb512mem_mask.
1988 (bdesc_args): Add __builtin_ia32_expanddf512,
1989 __builtin_ia32_expandsf512.
1990 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1991 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1992 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1993 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1994 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1995 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1996 (avx512f_<code>v8div16qi2_mask_store): This.
1997 (avx512f_expand<mode>): New.
1999 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2001 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2003 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2004 (_mm512_prefetch_i32scatter_pd): Ditto.
2005 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2006 (_mm512_prefetch_i64scatter_pd): Ditto.
2007 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2008 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2009 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2010 (_mm512_prefetch_i32scatter_ps): Ditto.
2011 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2012 (_mm512_prefetch_i64scatter_ps): Ditto.
2013 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2014 * config/i386/i386-builtin-types.def: Define
2015 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2016 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2017 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2018 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2019 IX86_BUILTIN_SCATTERPFQPD.
2020 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2021 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2022 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2023 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2024 __builtin_ia32_scatterpfqps.
2025 (ix86_expand_builtin): Expand new built-ins.
2026 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2027 fix memory access data type.
2028 (*avx512pf_gatherpf<mode>_mask): Ditto.
2029 (*avx512pf_gatherpf<mode>): Ditto.
2030 (avx512pf_scatterpf<mode>): Ditto.
2031 (*avx512pf_scatterpf<mode>_mask): Ditto.
2032 (*avx512pf_scatterpf<mode>): Ditto.
2033 (GATHER_SCATTER_SF_MEM_MODE): New.
2034 (avx512pf_gatherpf<mode>df): Ditto.
2035 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2036 (*avx512pf_scatterpf<mode>df): Ditto.
2038 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2041 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2042 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2045 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2047 * common/config/arm/arm-common.c
2048 (arm_rewrite_mcpu): Handle multiple names.
2050 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2052 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2054 * gimple-builder.h (create_gimple_tmp): Delete.
2056 2014-01-27 Christian Bruel <christian.bruel@st.com>
2058 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2061 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2063 * config/pa/pa.md (call): Generate indirect long calls to non-local
2064 functions when outputing 32-bit code.
2065 (call_value): Likewise except for special call to buggy powf function.
2067 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2068 portable runtime and PIC indirect calls.
2069 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2070 and PIC call sequences. Use ldo instead of blr to set return register
2071 in PIC call sequence.
2073 2014-01-25 Walter Lee <walt@tilera.com>
2075 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2076 avoid clobbering a live register.
2078 2014-01-25 Walter Lee <walt@tilera.com>
2080 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2081 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2082 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2083 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2085 2014-01-25 Walter Lee <walt@tilera.com>
2087 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2088 arguments on even registers.
2089 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2091 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2092 (BIGGEST_ALIGNMENT): Ditto.
2093 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2095 2014-01-25 Walter Lee <walt@tilera.com>
2097 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2098 insns before bundling.
2099 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2101 2014-01-25 Walter Lee <walt@tilera.com>
2103 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2104 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2105 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2107 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2109 * config/mips/constraints.md (kl): Delete.
2110 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2111 define expands, using...
2112 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2113 instructions for MIPS16.
2114 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2115 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2117 2014-01-25 Walter Lee <walt@tilera.com>
2119 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2123 2014-01-25 Walter Lee <walt@tilera.com>
2125 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2126 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2128 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2130 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2133 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2135 * print-rtl.c (in_call_function_usage): New var.
2136 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2137 EXPR_LIST mode as mode and not as reg note name.
2140 * cfgloopmanip.c (copy_loop_info): If
2141 loop->warned_aggressive_loop_optimizations, make sure
2142 the flag is set in target loop too.
2144 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2146 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2148 * builtins.def: Likewise.
2149 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2150 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2151 * ira.c (ira_setup_eliminable_regset): Likewise.
2152 * omp-low.c (gate_expand_omp): Likewise.
2153 (execute_lower_omp): Likewise.
2154 (diagnose_sb_0): Likewise.
2155 (gate_diagnose_omp_blocks): Likewise.
2156 (simd_clone_clauses_extract): Likewise.
2159 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2161 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2162 correction for little endian...
2163 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2166 2014-01-24 Jeff Law <law@redhat.com>
2168 PR tree-optimization/59919
2169 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2170 for non-returning calls.
2172 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2174 * common/config/aarch64/aarch64-common.c
2175 (aarch64_rewrite_mcpu): Handle multiple names.
2176 * config/aarch64/aarch64.h
2177 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2179 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2181 * input.c (add_file_to_cache_tab): Handle the case where fopen
2184 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2187 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2188 from push operand if code of push isn't PRE_DEC.
2190 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2193 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2194 -mquad-memory-atomic. Update -mquad-memory documentation to say
2195 it is only used for non-atomic loads/stores.
2197 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2198 -mquad-memory or -mquad-memory-atomic switches.
2200 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2201 -mquad-memory-atomic to ISA 2.07 support.
2203 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2204 to separate support of normal quad word memory operations (ldq, stq)
2205 from the atomic quad word memory operations.
2207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2208 support to separate non-atomic quad word operations from atomic
2209 quad word operations. Disable non-atomic quad word operations in
2210 little endian mode so that we don't have to swap words after the
2211 load and before the store.
2212 (quad_load_store_p): Add comment about atomic quad word support.
2213 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2214 options printed with -mdebug=reg.
2216 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2217 -mquad-memory-atomic as the test for whether we have quad word
2218 atomic instructions.
2219 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2220 or -mp8-vector are used, allow byte/half-word atomic operations.
2222 * config/rs6000/sync.md (load_lockedti): Insure that the address
2223 is a proper indexed or indirect address for the lqarx instruction.
2224 On little endian systems, swap the hi/lo registers after the lqarx
2226 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2227 insure the address is valid for the lqarx instruction.
2228 (store_conditionalti): Insure that the address is a proper indexed
2229 or indirect address for the stqcrx. instruction. On little endian
2230 systems, swap the hi/lo registers before doing the stqcrx.
2232 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2233 insure the address is valid for the stqcrx. instruction.
2235 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2236 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2237 type of quad memory support is available.
2239 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2242 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2243 there is a danger of looping.
2245 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2248 force flag_ira_loop_pressure if set via command line.
2250 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2252 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2253 (ashr_simd): New builtin handling DI mode.
2254 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2255 (aarch64_sshr_simddi): New match pattern.
2256 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2257 (vshrd_n_s64): Likewise.
2258 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2260 2014-01-23 Nick Clifton <nickc@redhat.com>
2262 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2263 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2264 favour of mcu specific scripts.
2265 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2268 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2269 Alex Velenko <Alex.Velenko@arm.com>
2271 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2272 (vaddv_s16): Likewise.
2273 (vaddv_s32): Likewise.
2274 (vaddv_u8): Likewise.
2275 (vaddv_u16): Likewise.
2276 (vaddv_u32): Likewise.
2277 (vaddvq_s8): Likewise.
2278 (vaddvq_s16): Likewise.
2279 (vaddvq_s32): Likewise.
2280 (vaddvq_s64): Likewise.
2281 (vaddvq_u8): Likewise.
2282 (vaddvq_u16): Likewise.
2283 (vaddvq_u32): Likewise.
2284 (vaddvq_u64): Likewise.
2285 (vaddv_f32): Likewise.
2286 (vaddvq_f32): Likewise.
2287 (vaddvq_f64): Likewise.
2288 (vmaxv_f32): Likewise.
2289 (vmaxv_s8): Likewise.
2290 (vmaxv_s16): Likewise.
2291 (vmaxv_s32): Likewise.
2292 (vmaxv_u8): Likewise.
2293 (vmaxv_u16): Likewise.
2294 (vmaxv_u32): Likewise.
2295 (vmaxvq_f32): Likewise.
2296 (vmaxvq_f64): Likewise.
2297 (vmaxvq_s8): Likewise.
2298 (vmaxvq_s16): Likewise.
2299 (vmaxvq_s32): Likewise.
2300 (vmaxvq_u8): Likewise.
2301 (vmaxvq_u16): Likewise.
2302 (vmaxvq_u32): Likewise.
2303 (vmaxnmv_f32): Likewise.
2304 (vmaxnmvq_f32): Likewise.
2305 (vmaxnmvq_f64): Likewise.
2306 (vminv_f32): Likewise.
2307 (vminv_s8): Likewise.
2308 (vminv_s16): Likewise.
2309 (vminv_s32): Likewise.
2310 (vminv_u8): Likewise.
2311 (vminv_u16): Likewise.
2312 (vminv_u32): Likewise.
2313 (vminvq_f32): Likewise.
2314 (vminvq_f64): Likewise.
2315 (vminvq_s8): Likewise.
2316 (vminvq_s16): Likewise.
2317 (vminvq_s32): Likewise.
2318 (vminvq_u8): Likewise.
2319 (vminvq_u16): Likewise.
2320 (vminvq_u32): Likewise.
2321 (vminnmv_f32): Likewise.
2322 (vminnmvq_f32): Likewise.
2323 (vminnmvq_f64): Likewise.
2325 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2327 * config/aarch64/aarch64-simd.md
2328 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2329 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2330 (*aarch64_mul3_elt<mode>): Likewise.
2331 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2332 (*aarch64_mul3_elt_to_64v2df): Likewise.
2333 (*aarch64_mla_elt<mode>): Likewise.
2334 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2335 (*aarch64_mls_elt<mode>): Likewise.
2336 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2337 (*aarch64_fma4_elt<mode>): Likewise.
2338 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2339 (*aarch64_fma4_elt_to_64v2df): Likewise.
2340 (*aarch64_fnma4_elt<mode>): Likewise.
2341 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2342 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2343 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2344 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2345 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2346 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2347 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2348 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2349 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2351 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2353 * config/aarch64/aarch64-simd.md
2354 (aarch64_be_checked_get_lane<mode>): New define_expand.
2355 * config/aarch64/aarch64-simd-builtins.def
2356 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2357 New builtin definition.
2358 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2359 Use new safe be builtin.
2361 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2363 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2365 (aarch64_be_st1<mode>): Likewise.
2366 (aarch_ld1<VALL:mode>): Define_expand modified.
2367 (aarch_st1<VALL:mode>): Likewise.
2368 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2369 (UNSPEC_ST1): Likewise.
2371 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2373 * config/microblaze/microblaze.md: Add trap insn and attribute
2375 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2377 PR preprocessor/58580
2378 * input.h (location_get_source_line): Take an additional line_size
2380 (void diagnostics_file_cache_fini): Declare new function.
2381 * input.c (struct fcache): New type.
2382 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2383 New static constants.
2384 (diagnostic_file_cache_init, total_lines_num)
2385 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2386 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2387 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2388 (get_next_line, read_next_line, goto_next_line, read_line_num):
2389 New static function definitions.
2390 (diagnostic_file_cache_fini): New function.
2391 (location_get_source_line): Take an additional output line_len
2392 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2394 * diagnostic.c (diagnostic_finish): Call
2395 diagnostic_file_cache_fini.
2396 (adjust_line): Take an additional input parameter for the length
2397 of the line, rather than calculating it with strlen.
2398 (diagnostic_show_locus): Adjust the use of
2399 location_get_source_line and adjust_line with respect to their new
2400 signature. While displaying a line now, do not stop at the first
2401 null byte. Rather, display the zero byte as a space and keep
2402 going until we reach the size of the line.
2403 * Makefile.in: Add vec.o to OBJS-libcommon
2405 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2406 Ilya Tocar <ilya.tocar@intel.com>
2408 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2409 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2410 (__builtin_ia32_kmov16): Ditto.
2411 * config/i386/i386.md (UNSPEC_KMOV): New.
2414 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2416 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2417 (_mm512_storeu_si512): Ditto.
2419 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2422 * rtl.h (get_referenced_operands): Declare.
2423 * recog.c (get_referenced_operands): New function.
2424 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2425 operands have been referenced when recording LO_SUM references.
2427 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2429 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2431 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2433 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2434 Enable for generic and recent AMD targets.
2436 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2438 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2439 ARG_SIZE note when adjustment was eliminated.
2441 2014-01-22 Jeff Law <law@redhat.com>
2443 PR tree-optimization/59597
2444 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2445 in file. Accept new argument REGISTERING and use it to modify
2446 dump output appropriately.
2447 (register_jump_thread): Corresponding changes.
2448 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2449 thread paths involving joiner blocks. Add code to dump cancelled
2450 jump threading paths.
2452 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2454 PR rtl-optimization/59477
2455 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2456 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2458 2014-01-22 Tom Tromey <tromey@redhat.com>
2460 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2462 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2464 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2466 PR rtl-optimization/59896
2467 * lra-constraints.c (process_alt_operands): Check unused note for
2468 matched operands of insn with no output reloads.
2470 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2472 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2473 (mips_move_from_gpr_cost): Likewise.
2475 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2477 PR rtl-optimization/59858
2478 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2479 ira_class_hard_regs_num.
2480 (process_alt_operands): Increase reject for dying matched operand.
2482 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2485 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2486 smaller than size, perform several stores or loads and stores
2487 at dst + count - size to store or copy all of size bytes, rather
2488 than just last modesize bytes.
2490 2014-01-20 DJ Delorie <dj@redhat.com>
2492 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2493 that CLOBBERs are REGs before propogating their values.
2495 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2498 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2499 (cgraph_inline_failed_type): New function.
2500 * cgraph.h (DEFCIFCODE): Add type.
2501 (cgraph_inline_failed_type_t): New enum.
2502 (cgraph_inline_failed_type): New prototype.
2503 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2504 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2505 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2506 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2507 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2508 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2509 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2510 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2511 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2512 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2513 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2514 OPTIMIZATION_MISMATCH.
2515 * tree-inline.c (expand_call_inline): Emit errors during
2516 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2518 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2521 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2522 mode attribute in insn output.
2524 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2526 * output.h (output_constant): Delete.
2527 * varasm.c (output_constant): Make private.
2529 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2531 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2533 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2536 * tree.h (fold_builtin_strcat): New prototype.
2537 * builtins.c (fold_builtin_strcat): No longer static. Add len
2538 argument, if non-NULL, don't call c_strlen. Optimize
2539 directly into __builtin_memcpy instead of __builtin_strcpy.
2540 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2541 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2543 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2545 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2546 for SImode_address_operand operands, having only a REG argument.
2548 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2550 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2551 loader name using mbig-endian.
2552 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2554 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2556 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2560 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2562 * config/aarch64/aarch64-protos.h
2563 (aarch64_cannot_change_mode_class_ptr): Declare.
2564 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2565 aarch64_cannot_change_mode_class_ptr): New.
2566 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2567 backend hook aarch64_cannot_change_mode_class.
2569 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2571 * common/config/aarch64/aarch64-common.c
2572 (aarch64_handle_option): Don't handle any option order logic here.
2573 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2574 selected_cpu, warn on architecture version mismatch.
2575 (aarch64_override_options): Fix parsing order for option strings.
2577 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2578 Iain Sandoe <iain@codesourcery.com>
2581 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2582 warning. Amend comment to reflect current functionality.
2584 2014-01-20 Richard Biener <rguenther@suse.de>
2587 * builtins.c (fold_builtin_strcat): Remove case better handled
2588 by tree-ssa-strlen.c.
2590 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2592 * config/aarch64/aarch64.opt
2593 (mcpu, march, mtune): Make case-insensitive.
2595 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2598 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2599 if operands[1] is a REG or ZERO_EXTEND of a REG.
2601 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2603 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2605 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2607 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2608 long non-pic millicode calls.
2610 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2612 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2614 2014-01-19 Kito Cheng <kito@0xlab.org>
2616 * builtins.c (expand_movstr): Check movstr expand done or fail.
2618 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2619 H.J. Lu <hongjiu.lu@intel.com>
2622 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2623 to DImode for zero-extended addresses.
2625 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2627 PR rtl-optimization/57763
2628 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2629 on the new indirect jump_insn and increment LABEL_NUSES (label).
2631 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2635 * config.gcc (x86_archs): New variable.
2636 (x86_64_archs): Likewise.
2637 (x86_cpus): Likewise.
2638 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2639 --with-arch/--with-cpu= options.
2640 Support --with-arch=/--with-cpu={nehalem,westmere,
2641 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2643 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2645 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2646 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2648 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2650 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2652 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2655 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2656 clear cpp_get_options (parse_in)->warn_unused_macros for
2657 ix86_target_macros_internal with cpp_define.
2659 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2661 * jump.c (delete_related_insns): Keep (use (insn))s.
2662 * reorg.c (redundant_insn): Check for barriers too.
2664 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2666 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2668 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2670 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2671 call to $$dyncall when TARGET_LONG_CALLS is true.
2673 2014-01-17 Jeff Law <law@redhat.com>
2675 * ree.c (combine_set_extension): Temporarily disable test for
2676 changing number of hard registers.
2678 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2681 * ipa-inline-analysis.c (inline_free_summary):
2682 Do not free summary of aliases.
2684 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2687 * gimplify.c (gimplify_expr): Use create_tmp_var
2688 instead of create_tmp_var_raw. If cond doesn't have
2689 integral type, don't add the IFN_ANNOTATE builtin at all.
2691 2014-01-17 Martin Jambor <mjambor@suse.cz>
2694 * ipa-cp.c (prev_edge_clone): New variable.
2695 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2696 Also resize prev_edge_clone vector.
2697 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2698 (ipcp_edge_removal_hook): New function.
2699 (ipcp_driver): Register ipcp_edge_removal_hook.
2701 2014-01-17 Andrew Pinski <apinski@cavium.com>
2702 Steve Ellcey <sellcey@mips.com>
2705 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2708 2014-01-17 Jeff Law <law@redhat.com>
2711 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2712 so that pass_ccp runs first.
2714 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2716 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2717 (ix86_adjust_cost): Use !TARGET_XXX.
2718 (do_reorder_for_imul): Likewise.
2719 (swap_top_of_ready_list): Likewise.
2720 (ix86_sched_reorder): Likewise.
2722 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2724 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2725 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2726 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2727 (intel_memset): New. Duplicate slm_memset.
2728 (intel_cost): New. Duplicate slm_cost.
2729 (m_INTEL): New macro.
2730 (processor_target_table): Add "intel".
2731 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2732 with PROCESSOR_INTEL for "intel".
2733 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2734 PROCESSOR_SILVERMONT.
2735 (ix86_issue_rate): Likewise.
2736 (ix86_adjust_cost): Likewise.
2737 (ia32_multipass_dfa_lookahead): Likewise.
2738 (swap_top_of_ready_list): Likewise.
2739 (ix86_sched_reorder): Likewise.
2740 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2741 instead of TARGET_OPT_AGU.
2742 * config/i386/i386.h (TARGET_INTEL): New.
2743 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2744 (processor_type): Add PROCESSOR_INTEL.
2745 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2746 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2748 2014-01-17 Marek Polacek <polacek@redhat.com>
2751 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2754 2014-01-17 Richard Biener <rguenther@suse.de>
2756 PR tree-optimization/46590
2757 * opts.c (default_options_table): Add entries for
2758 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2759 all enabled at -O1 but not for -Og.
2760 * common.opt (fbranch-count-reg): Remove Init(1).
2761 (fmove-loop-invariants): Likewise.
2762 (ftree-pta): Likewise.
2764 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2766 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2767 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2768 decls to at least the GCC 4.8 used alignments.
2771 * tree-nested.c (convert_nonlocal_reference_stmt,
2772 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2773 of GIMPLE_BIND stmts, adjust associated decls.
2775 2014-01-17 Richard Biener <rguenther@suse.de>
2777 PR tree-optimization/46590
2778 * vec.h (vec<>::bseach): New member function implementing
2779 binary search according to C89 bsearch.
2780 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2781 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2782 bitmap pointer again. Make accesses_in_loop a flat array.
2783 (mem_ref_obstack): New global.
2784 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2785 (mark_ref_stored): Likewise.
2786 (ref_indep_loop_p_2): Likewise.
2787 (set_ref_stored_in_loop): New helper function.
2788 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2789 (memref_free): Adjust.
2790 (record_mem_ref_loc): Simplify.
2791 (gather_mem_refs_stmt): Adjust.
2792 (sort_locs_in_loop_postorder_cmp): New function.
2793 (analyze_memory_references): Sort accesses_in_loop after
2794 loop postorder number.
2795 (find_ref_loc_in_loop_cmp): New function.
2796 (for_all_locs_in_loop): Find relevant cluster of locs in
2797 accesses_in_loop and iterate without recursion.
2798 (execute_sm): Avoid uninit warning.
2799 (struct ref_always_accessed): Simplify.
2800 (ref_always_accessed::operator ()): Likewise.
2801 (ref_always_accessed_p): Likewise.
2802 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2803 loop postorder numbers here.
2804 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2807 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2810 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2811 on decls for which assemble_alias has been called.
2813 2014-01-17 Nick Clifton <nickc@redhat.com>
2815 * config/msp430/msp430.opt: (mcpu): New option.
2816 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2817 (msp430_option_override): Parse target_cpu. If the MCU name
2818 matches a generic string, clear target_mcu.
2819 (msp430_attr): Allow numeric interrupt values up to 63.
2820 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2821 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2823 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2825 * config/msp430/msp430.md (popm): Use %J rather than %I.
2826 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2827 (addhi_cy_i): Use immediate_operand for operand 2.
2828 * doc/invoke.texi: Document -mcpu option.
2830 2014-01-17 Richard Biener <rguenther@suse.de>
2832 PR rtl-optimization/38518
2833 * df.h (df_analyze_loop): Declare.
2834 * df-core.c: Include cfgloop.h.
2835 (df_analyze_1): Split out main part of df_analyze.
2836 (df_analyze): Adjust.
2837 (loop_inverted_post_order_compute): New function.
2838 (loop_post_order_compute): Likewise.
2839 (df_analyze_loop): New function avoiding whole-function
2841 * loop-invariant.c (find_defs): Use df_analyze_loop.
2842 (find_invariants): Adjust.
2843 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2845 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2847 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2848 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2850 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2852 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2853 traversal when removing references.
2855 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2858 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2860 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2863 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2864 pushing a reload for an autoinc when we had previously reloaded an
2865 inner part of the address.
2867 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2869 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2871 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2872 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2873 when not giving up or versioning for alias only because of
2875 (vect_analyze_data_ref_dependences): Set to true.
2876 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2878 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2879 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2883 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2886 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2887 operand 0 predicate for gathers, use a new pseudo as subtarget.
2889 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2892 * lra-constraints.c (process_alt_operands): Add printing debug info.
2893 Check absence of input/output reloads for matched operands too.
2895 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2897 PR rtl-optimization/59835
2898 * ira.c (ira_init_register_move_cost): Increase cost for
2901 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2903 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2905 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2908 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2909 non-register objects. Use gen_(high/low)part more consistently.
2912 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2915 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2916 endian support, remove tests for WORDS_BIG_ENDIAN.
2917 (p8_mfvsrd_3_<mode>): Likewise.
2918 (reload_gpr_from_vsx<mode>): Likewise.
2919 (reload_gpr_from_vsxsf): Likewise.
2920 (p8_mfvsrd_4_disf): Likewise.
2922 2014-01-16 Richard Biener <rguenther@suse.de>
2924 PR rtl-optimization/46590
2925 * lcm.c (compute_antinout_edge): Use postorder iteration.
2926 (compute_laterin): Use inverted postorder iteration.
2928 2014-01-16 Nick Clifton <nickc@redhat.com>
2931 * varasm.c (output_constant): Return the number of bytes actually
2933 (output_constructor_array_range): Update the field size with the
2934 number of bytes emitted by output_constant.
2935 (output_constructor_regular_field): Likewise. Also do not
2936 complain if the total number of bytes emitted is now greater
2937 than the expected fieldpos.
2938 * output.h (output_constant): Update prototype and descriptive comment.
2940 2014-01-16 Marek Polacek <polacek@redhat.com>
2943 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2944 it is error_mark_node.
2946 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2948 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2949 VALID_AVX256_REG_OR_OI_MODE.
2951 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2953 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2954 current procedure should be profiled.
2956 2014-01-15 Andrew Pinski <apinski@cavium.com>
2958 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2959 of moving from/to the STACK_REG register class.
2961 2014-01-15 Richard Henderson <rth@redhat.com>
2964 * reginfo.c (global_regs_decl): Globalize.
2965 * rtl.h (global_regs_decl): Declare.
2966 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2967 reserved via global_regs.
2969 2014-01-15 Teresa Johnson <tejohnson@google.com>
2971 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2973 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2975 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2976 and vmulosh rather than call gen_vec_widen_smult_*.
2977 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2978 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2979 (vec_widen_smult_even_v16qi): Likewise.
2980 (vec_widen_umult_even_v8hi): Likewise.
2981 (vec_widen_smult_even_v8hi): Likewise.
2982 (vec_widen_umult_odd_v16qi): Likewise.
2983 (vec_widen_smult_odd_v16qi): Likewise.
2984 (vec_widen_umult_odd_v8hi): Likewise.
2985 (vec_widen_smult_odd_v8hi): Likewise.
2986 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2987 vmuloub rather than call gen_vec_widen_umult_*.
2988 (vec_widen_umult_lo_v16qi): Likewise.
2989 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2990 vmulosb rather than call gen_vec_widen_smult_*.
2991 (vec_widen_smult_lo_v16qi): Likewise.
2992 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2993 rather than call gen_vec_widen_umult_*.
2994 (vec_widen_umult_lo_v8hi): Likewise.
2995 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2996 rather than call gen_vec_widen_smult_*.
2997 (vec_widen_smult_lo_v8hi): Likewise.
2999 2014-01-15 Jeff Law <law@redhat.com>
3001 PR tree-optimization/59747
3002 * ree.c (find_and_remove_re): Properly handle case where a second
3003 eliminated extension requires widening a copy created for elimination
3004 of a prior extension.
3005 (combine_set_extension): Ensure that the number of hard regs needed
3006 for a destination register does not change when we widen it.
3008 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3010 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3011 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3012 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3013 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3014 (avr-*-rtems*): Likewise.
3015 (bfin*-rtems*): Likewise.
3016 (moxie-*-rtems*): Likewise.
3017 (h8300-*-rtems*): Likewise.
3018 (i[34567]86-*-rtems*): Likewise.
3019 (lm32-*-rtems*): Likewise.
3020 (m32r-*-rtems*): Likewise.
3021 (m68k-*-rtems*): Likewise.
3022 (microblaze*-*-rtems*): Likewise.
3023 (mips*-*-rtems*): Likewise.
3024 (powerpc-*-rtems*): Likewise.
3025 (sh-*-rtems*): Likewise.
3026 (sparc-*-rtems*): Likewise.
3027 (sparc64-*-rtems*): Likewise.
3028 (v850-*-rtems*): Likewise.
3029 (m32c-*-rtems*): Likewise.
3031 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3033 PR rtl-optimization/59511
3034 * ira.c (ira_init_register_move_cost): Use memory costs for some
3035 cases of register move cost calculations.
3036 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3037 instead of BB frequency.
3038 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3039 * lra-assigns.c (find_hard_regno_for): Ditto.
3041 2014-01-15 Richard Biener <rguenther@suse.de>
3043 PR tree-optimization/59822
3044 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3045 (vectorizable_load): Use it to hoist defs of uses of invariant
3046 loads out of the loop.
3048 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3049 Kugan Vivekanandarajah <kuganv@linaro.org>
3052 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3055 2014-01-15 Richard Biener <rguenther@suse.de>
3057 PR rtl-optimization/59802
3058 * lcm.c (compute_available): Use inverted postorder to seed
3059 the initial worklist.
3061 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3064 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3065 ADDR_REGS for invalid symrefs in non-PIC code.
3067 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3070 * builtins.c (determine_block_size): Initialize *probable_max_size
3071 even if len_rtx is CONST_INT.
3073 2014-01-14 Andrew Pinski <apinski@cavium.com>
3075 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3076 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3077 (cortexa53_tunings): Likewise.
3078 (aarch64_sched_issue_rate): New function.
3079 (TARGET_SCHED_ISSUE_RATE): Define.
3081 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3083 * ira-costs.c (find_costs_and_classes): Add missed
3084 ira_init_register_move_cost_if_necessary.
3086 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3089 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3091 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3094 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3095 to indicate if type is used for function return value. Warn ABI
3096 change if the vector mode isn't available for function return value.
3097 (ix86_function_arg_advance): Pass false to type_natural_mode.
3098 (ix86_function_arg): Likewise.
3099 (ix86_gimplify_va_arg): Likewise.
3100 (function_arg_32): Don't warn ABI change.
3101 (ix86_function_value): Pass true to type_natural_mode.
3102 (ix86_return_in_memory): Likewise.
3103 (ix86_struct_value_rtx): Removed.
3104 (TARGET_STRUCT_VALUE_RTX): Likewise.
3106 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3108 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3109 converting a conditional jump into a conditional return.
3111 2014-01-14 Richard Biener <rguenther@suse.de>
3113 PR tree-optimization/58921
3114 PR tree-optimization/59006
3115 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3116 hoisting invariant stmts.
3117 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3118 invariant loads on the preheader edge if possible.
3120 2014-01-14 Joey Ye <joey.ye@arm.com>
3122 * doc/plugin.texi (Building GCC plugins): Update to C++.
3124 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3126 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3127 (_mm_rcp28_round_ss): Ditto.
3128 (_mm_rsqrt28_round_sd): Ditto.
3129 (_mm_rsqrt28_round_ss): Ditto.
3130 (_mm_rcp28_sd): Ditto.
3131 (_mm_rcp28_ss): Ditto.
3132 (_mm_rsqrt28_sd): Ditto.
3133 (_mm_rsqrt28_ss): Ditto.
3134 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3135 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3136 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3137 (IX86_BUILTIN_RCP28SD): Ditto.
3138 (IX86_BUILTIN_RCP28SS): Ditto.
3139 (IX86_BUILTIN_RSQRT28SD): Ditto.
3140 (IX86_BUILTIN_RSQRT28SS): Ditto.
3141 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3142 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3143 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3144 (ix86_expand_special_args_builtin): Expand new FTYPE.
3145 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3146 (srcp14<mode>): Make insn unary.
3147 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3148 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3149 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3150 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3151 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3152 Fix rounding: make it SAE only.
3153 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3155 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3157 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3158 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3159 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3160 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3161 (round_saeonly_mask_scalar_operand4): Ditto.
3162 (round_saeonly_mask_scalar_op3): Ditto.
3163 (round_saeonly_mask_scalar_op4): Ditto.
3165 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3167 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3168 Implement -maltivec=be for vec_insert and vec_extract.
3170 2014-01-10 DJ Delorie <dj@redhat.com>
3172 * config/msp430/msp430.md (call_internal): Don't allow memory
3173 references with SP as the base register.
3174 (call_value_internal): Likewise.
3175 * config/msp430/constraints.md (Yc): New. For memory references
3176 that don't use SP as a base register.
3178 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3179 "an integer without a # prefix"
3180 * config/msp430/msp430.md (epilogue_helper): Use it.
3182 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3185 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3186 AVX512F gather builtins.
3187 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3188 on gather decls with INTEGER_TYPE masktype.
3189 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3190 directly into the builtin rather than hoisting it before loop.
3192 PR tree-optimization/59387
3193 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3194 (scev_const_prop): If folded_casts and type has undefined overflow,
3195 use force_gimple_operand instead of force_gimple_operand_gsi and
3196 for each added stmt if it is assign with
3197 arith_code_with_undefined_signed_overflow, call
3198 rewrite_to_defined_overflow.
3199 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3200 gimple-fold.h instead.
3201 (arith_code_with_undefined_signed_overflow,
3202 rewrite_to_defined_overflow): Moved to ...
3203 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3204 rewrite_to_defined_overflow): ... here. No longer static.
3205 Include gimplify-me.h.
3206 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3207 rewrite_to_defined_overflow): New prototypes.
3209 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3211 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3213 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3215 * builtins.c (get_object_alignment_2): Minor tweak.
3216 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3218 2014-01-13 Christian Bruel <christian.bruel@st.com>
3220 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3221 optimized non constant lengths.
3223 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3226 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3227 load as __atomic_load_N if possible.
3229 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3231 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3233 (rs6000_expand_builtin): Adjust call.
3235 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3238 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3239 * config/rs6000/rs6000.c: Include target-globals.h.
3240 (rs6000_set_current_function): Instead of doing target_reinit
3241 unconditionally, use save_target_globals_default_opts and
3242 restore_target_globals.
3244 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3246 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3247 (rs6000_expand_builtin): Handle mffs and mtfsf.
3248 (rs6000_init_builtins): Define mffs and mtfsf.
3249 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3250 (rs6000_mffs): New pattern.
3251 (rs6000_mtfsf): New pattern.
3253 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3255 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3256 Start narrowing with START. Apply candidate-use pair
3257 and check overall cost in narrowing.
3258 (iv_ca_prune): Pass new argument.
3260 2014-01-10 Jeff Law <law@redhat.com>
3263 * ree.c (combine_reaching_defs): Ensure the defining statement
3264 occurs before the extension when optimizing extensions with
3265 different source and destination hard registers.
3267 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3270 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3271 vtables into the type inheritance graph.
3273 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3275 PR rtl-optimization/59754
3276 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3277 modes in the REGNO != REGNO case.
3279 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3281 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3283 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3285 PR tree-optimization/59745
3286 * tree-predcom.c (tree_predictive_commoning_loop): Call
3287 free_affine_expand_cache if giving up because components is NULL.
3289 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3290 GC in payload of target_globals struct instead of allocating them on
3291 the heap and the larger structs separately using GC.
3292 * target-globals.h (struct target_globals): Make regs, hard_regs,
3293 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3294 of GTY((skip)) and change type to void *.
3295 (reset_target_globals): Cast loads from those fields to corresponding
3298 2014-01-10 Steve Ellcey <sellcey@mips.com>
3301 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3302 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3303 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3305 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3308 * aarch64-modes.def (CC_Zmode): New flags mode.
3309 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3310 represents an equality.
3311 (aarch64_get_condition_code): Handle CC_Zmode.
3312 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3314 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3316 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3317 extraction in good case.
3319 2014-01-10 Richard Biener <rguenther@suse.de>
3321 PR tree-optimization/59374
3322 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3323 checking after SLP discovery. Mark stmts not participating
3324 in any SLP instance properly.
3326 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3328 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3329 when handling a SET rtx.
3331 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3333 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3334 (cortex-a57): Likewise.
3335 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3337 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3339 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3340 non-iwmmxt builtins.
3342 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3346 * ipa-devirt.c record_target_from_binfo): Take as argument
3347 stack of binfos and lookup matching one for virtual inheritance.
3348 (possible_polymorphic_call_targets_1): Update.
3350 2014-01-10 Huacai Chen <chenhc@lemote.com>
3352 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3353 kernel strings for Loongson-2E/2F/3A.
3355 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3358 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3359 is_gimple_call before calling gimple_call_internal_p.
3361 2014-01-09 Steve Ellcey <sellcey@mips.com>
3363 * Makefile.in (TREE_FLOW_H): Remove.
3364 (TREE_SSA_H): Add file names from tree-flow.h.
3365 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3366 * tree.h: Remove tree-flow.h reference.
3367 * hash-table.h: Remove tree-flow.h reference.
3368 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3369 reference with tree-ssa-loop.h.
3371 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3373 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3374 default element-order behavior for -maltivec.
3375 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3376 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3377 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3378 when targeting big endian, at least for now.
3379 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3381 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3384 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3385 var satisfies use_register_for_decl, just take into account type
3386 alignment, rather than decl alignment.
3388 PR tree-optimization/59622
3389 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3390 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3391 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3392 Don't devirtualize for inplace at all. For targets.length () == 1,
3393 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3395 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3397 * config/i386/i386.md (cpu): Remove the unused btver1.
3399 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3401 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3403 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3406 * tree-core.h (struct target_globals): New forward declaration.
3407 (struct tree_target_option): Add globals field.
3408 * tree.h (TREE_TARGET_GLOBALS): Define.
3409 (prepare_target_option_nodes_for_pch): New prototype.
3410 * target-globals.h (struct target_globals): Define even if
3412 * tree.c (prepare_target_option_node_for_pch,
3413 prepare_target_option_nodes_for_pch): New functions.
3414 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3415 * config/i386/i386.c: Include target-globals.h.
3416 (ix86_set_current_function): Instead of doing target_reinit
3417 unconditionally, use save_target_globals_default_opts and
3418 restore_target_globals.
3420 2014-01-09 Richard Biener <rguenther@suse.de>
3422 PR tree-optimization/59715
3423 * tree-cfg.h (split_critical_edges): Declare.
3424 * tree-cfg.c (split_critical_edges): Export.
3425 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3427 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3429 * cfgexpand.c (expand_stack_vars): Optionally disable
3430 asan stack protection.
3431 (expand_used_vars): Likewise.
3432 (partition_stack_vars): Likewise.
3433 * asan.c (asan_emit_stack_protection): Optionally disable
3434 after return stack usage.
3435 (instrument_derefs): Optionally disable memory access instrumentation.
3436 (instrument_builtin_call): Likewise.
3437 (instrument_strlen_call): Likewise.
3438 (asan_protect_global): Optionally disable global variables protection.
3439 * doc/invoke.texi: Added doc for new options.
3440 * params.def: Added new options.
3441 * params.h: Likewise.
3443 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3445 PR rtl-optimization/59724
3446 * ifcvt.c (cond_exec_process_if_block): Don't call
3447 flow_find_head_matching_sequence with 0 longest_match.
3448 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3449 non-active insns if !stop_after.
3450 (try_head_merge_bb): Revert 2014-01-07 changes.
3452 2014-01-08 Jeff Law <law@redhat.com>
3454 * ree.c (get_sub_rtx): New function, extracted from...
3455 (merge_def_and_ext): Here.
3456 (combine_reaching_defs): Use get_sub_rtx.
3458 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3460 * cgraph.h (varpool_variable_node): Do not choke on null node.
3462 2014-01-08 Catherine Moore <clm@codesourcery.com>
3464 * config/mips/mips.md (simple_return): Attempt to use JRC
3466 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3468 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3470 PR rtl-optimization/59137
3471 * reorg.c (steal_delay_list_from_target): Call update_block for
3473 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3475 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3478 two duplicate entries.
3480 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3483 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3485 * config/mips/mips.c (mips_truncated_op_cost): New function.
3486 (mips_rtx_costs): Adjust test for BADDU.
3487 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3489 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3491 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3492 (*baddu_si): ...this new pattern.
3494 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3497 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3499 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3502 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3504 (expand_expr, expand_normal): Adjust.
3505 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3506 inner_reference_p. Use inner_reference_p to expand inner references.
3507 (store_expr): Adjust.
3508 * cfgexpand.c (expand_call_stmt): Adjust.
3510 2014-01-08 Rong Xu <xur@google.com>
3512 * gcov-io.c (gcov_var): Move from gcov-io.h.
3513 (gcov_position): Ditto.
3514 (gcov_is_error): Ditto.
3515 (gcov_rewrite): Ditto.
3516 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3517 only part to libgcc/libgcov.h.
3519 2014-01-08 Marek Polacek <polacek@redhat.com>
3522 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3524 2014-01-08 Marek Polacek <polacek@redhat.com>
3527 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3529 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3531 PR rtl-optimization/59649
3532 * stor-layout.c (get_mode_bounds): For BImode return
3533 0 and STORE_FLAG_VALUE.
3535 2014-01-08 Richard Biener <rguenther@suse.de>
3538 * gimple.h (is_gimple_builtin_call): Remove.
3539 (gimple_builtin_call_types_compatible_p): New.
3540 (gimple_call_builtin_p): New overload.
3541 * gimple.c (is_gimple_builtin_call): Remove.
3542 (validate_call): Rename to ...
3543 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3545 (validate_type): New static function.
3546 (gimple_call_builtin_p): New overload and adjust.
3547 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3548 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3549 (gimple_fold_stmt_to_constant_1): Likewise.
3550 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3552 2014-01-08 Richard Biener <rguenther@suse.de>
3555 * gimplify.c (gimplify_expr): Gimplify register-register type
3556 VIEW_CONVERT_EXPRs to separate stmts.
3558 2014-01-07 Jeff Law <law@redhat.com>
3561 * ree.c (combine_set_extension): Handle case where source
3562 and destination registers in an extension insn are different.
3563 (combine_reaching_defs): Allow source and destination registers
3564 in extension to be different under limited circumstances.
3565 (add_removable_extension): Remove restriction that the
3566 source and destination registers in the extension are the same.
3567 (find_and_remove_re): Emit a copy from the extension's
3568 destination to its source after the defining insn if
3569 the source and destination registers are different.
3572 * ifcvt.c (merge_if_block): If we are merging a block with more than
3573 one successor with a block with no successors, remove any BARRIER
3574 after the second block.
3576 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3578 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3580 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3583 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3584 for 14-bit register offsets when INT14_OK_STRICT is false.
3586 2014-01-07 Roland Stigge <stigge@antcom.de>
3587 Michael Meissner <meissner@linux.vnet.ibm.com>
3590 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3591 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3593 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3595 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3598 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3600 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3601 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3602 rtx is const0_rtx or not.
3604 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3607 * target-globals.c (save_target_globals): Remove this_fn_optab
3609 * toplev.c: Include optabs.h.
3610 (target_reinit): Temporarily restore the global options if another
3611 set of options are in force.
3613 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3615 PR rtl-optimization/58668
3616 * cfgcleanup.c (flow_find_cross_jump): Don't count
3617 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3618 to determine what is counted.
3619 (flow_find_head_matching_sequence): Use active_insn_p to determine
3621 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3623 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3624 determine what is counted.
3626 PR tree-optimization/59643
3627 * tree-predcom.c (split_data_refs_to_components): If one dr is
3628 read and one write, determine_offset fails and the write isn't
3629 in the bad component, just put the read into the bad component.
3631 2014-01-07 Mike Stump <mikestump@comcast.net>
3632 Jakub Jelinek <jakub@redhat.com>
3635 * tree-core.h (struct tree_optimization_option): Change optabs
3636 type from unsigned char * to void *.
3637 * optabs.c (init_tree_optimization_optabs): Adjust
3638 TREE_OPTIMIZATION_OPTABS initialization.
3640 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3643 * config/i386/i386.h (struct machine_function): Add
3644 no_drap_save_restore field.
3645 * config/i386/i386.c (ix86_save_reg): Use
3646 !cfun->machine->no_drap_save_restore instead of
3647 crtl->stack_realign_needed.
3648 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3649 this function clears frame_pointer_needed. Set
3650 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3651 and DRAP reg is needed.
3653 2014-01-06 Marek Polacek <polacek@redhat.com>
3656 * doc/implement-c.texi: Mention that other integer types are
3657 permitted as bit-field types in strictly conforming mode.
3659 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3661 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3664 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3666 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3668 2014-01-06 Martin Jambor <mjambor@suse.cz>
3671 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3673 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3675 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3679 * var-tracking.c (add_stores): Preserve the value of the source even if
3680 we don't record the store.
3682 2014-01-06 Terry Guo <terry.guo@arm.com>
3684 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3686 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3689 * config/darwin.c (darwin_function_section): Adjust return values to
3690 correspond to optimisation changes made in r206070.
3692 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3694 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3695 from prefetch_block tune setting.
3696 (nocona_cost): Correct size of prefetch block to 64.
3698 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3700 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3701 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3702 used to save the static chain register in the computation of the offset
3703 from which the FP registers need to be restored.
3705 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3707 PR tree-optimization/59519
3708 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3709 ICE if get_current_def (current_new_name) is already non-NULL, as long
3710 as it is a phi result of some other phi in *new_exit_bb that has
3713 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3714 or vmovdqu* for misaligned_operand.
3715 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3716 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3717 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3718 aligned_mem for AVX512F masked aligned load and store builtins and for
3721 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3723 PR tree-optimization/59651
3724 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3725 Address range for negative step should be added by TYPE_SIZE_UNIT.
3727 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3729 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3730 overlapping registers also for registers other than the stack pointer.
3732 2014-01-03 Marek Polacek <polacek@redhat.com>
3735 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3738 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3741 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3744 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3745 (push splitter): Use <P:MODE_SIZE> instead of
3746 GET_MODE_SIZE (<P:MODE>mode).
3747 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3748 (mov -1, reg peephole2): Likewise.
3749 * config/i386/sse.md (*mov<mode>_internal,
3750 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3751 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3752 *<code><mode>3, *andnot<mode>3<mask_name>,
3753 <mask_codefor><code><mode>3<mask_name>): Likewise.
3754 * config/i386/subst.md (mask_mode512bit_condition,
3755 sd_mask_mode512bit_condition): Likewise.
3757 2014-01-02 Xinliang David Li <davidxl@google.com>
3759 PR tree-optimization/59303
3760 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3761 (dump_predicates): Better output format.
3762 (pred_equal_p): New function.
3763 (is_neq_relop_p): Ditto.
3764 (is_neq_zero_form_p): Ditto.
3765 (pred_expr_equal_p): Ditto.
3766 (pred_neg_p): Ditto.
3767 (simplify_pred): Ditto.
3768 (simplify_preds_2): Ditto.
3769 (simplify_preds_3): Ditto.
3770 (simplify_preds_4): Ditto.
3771 (simplify_preds): Ditto.
3773 (push_to_worklist): Ditto.
3774 (get_pred_info_from_cmp): Ditto.
3775 (is_degenerated_phi): Ditto.
3776 (normalize_one_pred_1): Ditto.
3777 (normalize_one_pred): Ditto.
3778 (normalize_one_pred_chain): Ditto.
3779 (normalize_preds): Ditto.
3780 (normalize_cond_1): Remove function.
3781 (normalize_cond): Ditto.
3782 (is_gcond_subset_of): Ditto.
3783 (is_subset_of_any): Ditto.
3784 (is_or_set_subset_of): Ditto.
3785 (is_and_set_subset_of): Ditto.
3786 (is_norm_cond_subset_of): Ditto.
3787 (pred_chain_length_cmp): Ditto.
3788 (convert_control_dep_chain_into_preds): Type change.
3789 (find_predicates): Ditto.
3790 (find_def_preds): Ditto.
3791 (destroy_predicates_vecs): Ditto.
3792 (find_matching_predicates_in_rest_chains): Ditto.
3793 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3794 (is_pred_expr_subset): Ditto.
3795 (is_pred_chain_subset_of): Ditto.
3796 (is_included_in): Ditto.
3797 (is_superset_of): Ditto.
3799 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3801 Update copyright years.
3803 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3805 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3806 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3807 config/arc/arc.md, config/arc/arc.opt,
3808 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3809 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3810 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3811 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3812 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3813 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3814 vtable-verify.c, vtable-verify.h: Use the standard form for the
3817 2014-01-02 Tobias Burnus <burnus@net-b.de>
3819 * gcc.c (process_command): Update copyright notice dates.
3820 * gcov-dump.c: Ditto.
3822 * doc/cpp.texi: Bump @copying's copyright year.
3823 * doc/cppinternals.texi: Ditto.
3824 * doc/gcc.texi: Ditto.
3825 * doc/gccint.texi: Ditto.
3826 * doc/gcov.texi: Ditto.
3827 * doc/install.texi: Ditto.
3828 * doc/invoke.texi: Ditto.
3830 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3832 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3834 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3836 * config/i386/sse.md (*mov<mode>_internal): Guard
3837 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3839 PR rtl-optimization/59647
3840 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3841 new_rtx into UNSIGNED_FLOAT rtxes.
3843 Copyright (C) 2014 Free Software Foundation, Inc.
3845 Copying and distribution of this file, with or without modification,
3846 are permitted in any medium without royalty provided the copyright
3847 notice and this notice are preserved.