1 2014-02-20 Richard Henderson <rth@redhat.com>
4 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
5 on failure the store back into EXPECT.
7 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
8 Sandra Loosemore <sandra@codesourcery.com>
10 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
11 * config/nios2/nios2.c (nios2_function_profiler):
12 Add -fPIC (flag_pic == 2) support.
13 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
14 (nios2_large_offset_p): New function.
15 (nios2_unspec_reloc_p): Move up position, update to use
17 (nios2_unspec_address): Remove function.
18 (nios2_unspec_offset): New function.
19 (nios2_large_got_address): New function.
20 (nios2_got_address): Add large offset support.
21 (nios2_legitimize_tls_address): Update usage of removed and new
23 (nios2_symbol_binds_local_p): New function.
24 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
25 (nios2_legitimize_address): Update to use nios2_large_offset_p.
26 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
27 (nios2_print_operand): Merge H/L processing, add hiadj/lo
28 processing for (const (unspec ...)).
29 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
31 2014-02-20 Richard Biener <rguenther@suse.de>
33 * tree-cfg.c (replace_uses_by): Mark altered BBs before
34 doing the substitution.
35 (verify_gimple_assign_single): Also verify bare MEM_REFs
38 2014-02-20 Martin Jambor <mjambor@suse.cz>
41 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
42 info when checking whether lattices are bottom.
44 2014-02-20 Richard Biener <rguenther@suse.de>
47 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
50 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
53 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale parameter
54 specifying the scaling.
55 (inline_call): Update.
56 (want_inline_recursively): Guard division by zero.
57 (recursive_inlining): Update.
58 * ipa-inline.h (clone_inlined_nodes): Update.
60 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
63 * config/i386/i386.c (classify_argument): Pass structures of size
64 64 bytes or less in register.
66 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
67 Kirill Yukhin <kirill.yukhin@intel.com>
69 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
70 (_mm_rcp28_round_ss): Ditto.
71 (_mm_rsqrt28_round_sd): Ditto.
72 (_mm_rsqrt28_round_ss): Ditto.
73 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
74 (_mm_rcp14_round_ss): Ditto.
75 (_mm_rsqrt14_round_sd): Ditto.
76 (_mm_rsqrt14_round_ss): Ditto.
77 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
78 the first input operand, get rid of match_dup.
79 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
81 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
83 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
84 operand as the first input operand, set type attribute.
85 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
87 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
88 operand as the first input operand, set type attribute.
90 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
92 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
95 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
98 * config/i386/i386.c (construct_container): Remove TFmode check
99 for X86_64_INTEGER_CLASS.
101 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
104 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
105 only when -Wpsabi is enabled.
107 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
110 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
111 passing arrays in registers are the same as for structs, so remove the
112 special case for them.
114 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
116 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
117 destination type, extract only the valid bits if the source type is not
118 integral and has a different mode.
120 2014-02-19 Richard Biener <rguenther@suse.de>
123 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
126 2014-02-19 Richard Biener <rguenther@suse.de>
129 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
130 (ipa_modify_call_arguments): Emit an argument load explicitely and
131 preserve virtual SSA form there and for the replacement call.
132 Do not update SSA form nor free dominance info.
134 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
136 * ipa.c (function_and_variable_visibility): Also clear WEAK
137 flag when disolving COMDAT_GROUP.
139 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
141 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
142 * ipa-prop.c (ipa_set_jf_known_type): Return early when
144 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
145 do more sanity checks.
146 (detect_type_change): Return true when giving up early.
147 (compute_complex_assign_jump_func): Fix type parameter of
149 (compute_complex_ancestor_jump_func): Likewise.
150 (update_jump_functions_after_inlining): Fix updating of
152 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
154 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
156 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
157 inline clones when edge disappears.
159 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
162 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
163 Split 64-bit moves into 2 patterns. Do not allow the use of
164 direct move for TDmode in little endian, since the decimal value
165 has little endian bytes within a word, but the 64-bit pieces are
166 ordered in a big endian fashion, and normal subreg's of TDmode are
168 (mov<mode>_64bit_dm): Likewise.
169 (movtd_64bit_nodm): Likewise.
171 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
173 PR tree-optimization/60174
174 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
175 statement of an SSA_NAME that occurs in an abnormal PHI node.
177 2014-02-18 Jakub Jelinek <jakub@redhat.com>
180 * final.c (SEEN_BB): Remove.
181 (SEEN_NOTE, SEEN_EMITTED): Renumber.
182 (final_scan_insn): Don't force_source_line on second
183 NOTE_INSN_BASIC_BLOCK.
185 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
188 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
189 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
190 (type_natural_mode): Warn ABI change when %zmm register is not
191 available for AVX512F vector value passing.
193 2014-02-18 Kai Tietz <ktietz@redhat.com>
196 * config/i386/i386.c (ix86_expand_prologue): Use value in
197 rax register as displacement when restoring %r10 or %rax.
198 Fix wrong offset when restoring both registers.
200 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
202 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
203 assertion with conditional return.
205 2014-02-18 Jakub Jelinek <jakub@redhat.com>
206 Uros Bizjak <ubizjak@gmail.com>
209 * config/i386/driver-i386.c (host_detect_local_cpu): If
210 YMM state is not saved by the OS, also clear has_f16c. Move
211 CPUID 0x80000001 handling before YMM state saving checking.
213 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
215 PR rtl-optimization/58960
216 * haifa-sched.c (alloc_global_sched_pressure_data): New,
217 factored out from ...
218 (sched_init): ... here.
219 (free_global_sched_pressure_data): New, factored out from ...
220 (sched_finish): ... here.
221 * sched-int.h (free_global_sched_pressure_data): Declare.
222 * sched-rgn.c (nr_regions_initial): New static global.
223 (haifa_find_rgns): Initialize it.
224 (schedule_region): Disable sched-pressure for the newly
227 2014-02-17 Richard Biener <rguenther@suse.de>
229 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
230 release SSA defs of pattern stmts.
232 2014-02-17 Richard Biener <rguenther@suse.de>
234 * tree-inline.c (expand_call_inline): Release the virtual
235 operand defined by the call we are about to inline.
237 2014-02-17 Richard Biener <rguenther@suse.de>
239 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
241 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
242 Ilya Tocar <ilya.tocar@intel.com>
244 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
245 arguments order in builtin.
246 (_mm512_permutexvar_epi64): Ditto.
247 (_mm512_mask_permutexvar_epi64): Ditto
248 (_mm512_maskz_permutexvar_epi32): Ditto
249 (_mm512_permutexvar_epi32): Ditto
250 (_mm512_mask_permutexvar_epi32): Ditto
252 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
254 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
255 (p8_vmrgow): Likewise.
257 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
259 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
262 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
265 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
266 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
267 into 64-bit and 32-bit moves. On 64-bit moves, add support for
268 using direct move instructions on ISA 2.07. Also adjust
269 instruction length for 64-bit.
270 (mov<mode>_64bit, TFmode/TDmode): Likewise.
271 (mov<mode>_32bit, TFmode/TDmode): Likewise.
273 2014-02-15 Alan Modra <amodra@gmail.com>
277 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
278 find_replacement on parts of insn rtl that might be reloaded.
280 2014-02-15 Richard Biener <rguenther@suse.de>
282 PR tree-optimization/60183
283 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
284 (tree_ssa_phiprop): Calculate and free post-dominators.
286 2014-02-14 Jeff Law <law@redhat.com>
288 PR rtl-optimization/60131
289 * ree.c (get_extended_src_reg): New function.
290 (combine_reaching_defs): Use it rather than assuming location of REG.
291 (find_and_remove_re): Verify first operand of extension is
292 a REG before adding the insns to the copy list.
294 2014-02-14 Roland McGrath <mcgrathr@google.com>
296 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
297 * configure: Regenerated.
298 * config.in: Regenerated.
299 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
300 instead of ASM_SHORT.
302 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
303 Richard Earnshaw <rearnsha@arm.com>
305 PR rtl-optimization/59535
306 * lra-constraints.c (process_alt_operands): Encourage alternative
307 when unassigned pseudo class is superset of the alternative class.
308 (inherit_reload_reg): Don't inherit when optimizing for code size.
309 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
310 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
311 modes not less than 4 for Thumb1.
313 2014-02-14 Kyle McMartin <kyle@redhat.com>
316 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
318 2014-02-14 Richard Biener <rguenther@suse.de>
320 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
321 (get_frame_arg): Drop the assert with langhook types_compatible_p.
322 Do not strip INDIRECT_REFs.
324 2014-02-14 Richard Biener <rguenther@suse.de>
327 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
328 DECL_FUNCTION_SPECIFIC_TARGET.
329 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
330 * tree-streamer-out.c (pack_ts_target_option): Remove.
331 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
332 (write_ts_function_decl_tree_pointers): Do not stream
333 DECL_FUNCTION_SPECIFIC_TARGET.
334 * tree-streamer-in.c (unpack_ts_target_option): Remove.
335 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
336 (lto_input_ts_function_decl_tree_pointers): Do not stream
337 DECL_FUNCTION_SPECIFIC_TARGET.
339 2014-02-14 Jakub Jelinek <jakub@redhat.com>
341 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
342 (get_initial_def_for_induction, vectorizable_induction): Ignore
343 debug stmts when looking for exit_phi.
344 (vectorizable_live_operation): Fix up condition.
346 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
348 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
349 nreverse() because it changes the content of original tree list.
351 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
353 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
354 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
356 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
358 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
359 GNU coding standards.
361 2014-02-13 Jakub Jelinek <jakub@redhat.com>
364 * dwarf2out.c (gen_subprogram_die): Don't call
365 add_calling_convention_attribute if subr_die is old_die.
367 2014-02-13 Sharad Singhai <singhai@google.com>
369 * doc/optinfo.texi: Fix order of nodes.
371 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
373 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
374 operands[2], not operands[3].
376 2014-02-13 Richard Biener <rguenther@suse.de>
379 * doc/install.texi (ISL): Update recommended version to 0.12.2,
380 mention the possibility of an in-tree build.
381 (CLooG): Update recommended version to 0.18.1, mention the
382 possibility of an in-tree build and clarify that the ISL
383 bundled with CLooG does not work.
385 2014-02-13 Jakub Jelinek <jakub@redhat.com>
388 * expr.c (compress_float_constant): If x is a hard register,
389 extend into a pseudo and then move to x.
391 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
393 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
394 caused by bad second argument to warning_at() with -mhotpatch and
395 nested functions (e.g. with gfortran).
397 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
399 * opts.c (option_name): Remove "enabled by default" rider.
401 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
403 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
405 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
406 Uros Bizjak <ubizjak@gmail.com>
409 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
410 * configure: Regenerated.
412 2014-02-12 Richard Biener <rguenther@suse.de>
414 * vec.c (vec_prefix::calculate_allocation): Move as
415 inline variant to vec.h.
416 (vec_prefix::calculate_allocation_1): New out-of-line version.
417 * vec.h (vec_prefix::calculate_allocation_1): Declare.
418 (vec_prefix::m_has_auto_buf): Rename to ...
419 (vec_prefix::m_using_auto_storage): ... this.
420 (vec_prefix::calculate_allocation): Inline the easy cases
421 and dispatch to calculate_allocation_1 which doesn't need the
423 (va_heap::reserve): Use gcc_checking_assert.
424 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
425 m_using_auto_storage.
426 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
428 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
429 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
430 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
432 2014-02-12 Richard Biener <rguenther@suse.de>
434 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
435 when we found a dependence.
437 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
439 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
441 (maybe_fold_stmt): ... into this new function.
442 * omp-low.c (lower_omp): Update comment.
444 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
447 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
450 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
452 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
453 identifiers in comments.
454 (cortexa53_extra_costs): Likewise.
455 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
456 (cortexa7_extra_costs): Likewise.
457 (cortexa12_extra_costs): Likewise.
458 (cortexa15_extra_costs): Likewise.
459 (v7m_extra_costs): Likewise.
461 2014-02-12 Richard Biener <rguenther@suse.de>
464 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
465 of posix_memalign being successful.
466 (lower_stmt): Restrict lowering of posix_memalign to when
467 -ftree-bit-ccp is enabled.
469 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
471 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
473 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
475 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
477 PR rtl-optimization/60116
478 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
479 other_insn once the combination has been validated.
481 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
484 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
486 * ipa-devirt.c: Include demangle.h
487 (odr_violation_reported): New static variable.
488 (add_type_duplicate): Update odr_violations.
489 (maybe_record_node): Add completep parameter; update it.
490 (record_target_from_binfo): Add COMPLETEP parameter;
492 (possible_polymorphic_call_targets_1): Likewise.
493 (struct polymorphic_call_target_d): Add nonconstruction_targets;
494 rename FINAL to COMPLETE.
495 (record_targets_from_bases): Sanity check we found the binfo;
496 fix COMPLETEP updating.
497 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
498 parameter, fix computing of COMPLETEP.
499 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
500 at LTO time do demangling.
501 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
502 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
504 (gimple_get_virt_method_for_binfo): Likewise.
505 * gimple-fold.h (gimple_get_virt_method_for_binfo,
506 gimple_get_virt_method_for_vtable): Update prototypes.
508 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
511 * genautomata.c (add_presence_absence): Fix typo with
512 {final_}presence_list.
514 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
517 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
518 for VSX/Altivec vectors that land in GPR registers.
520 2014-02-11 Richard Henderson <rth@redhat.com>
521 Jakub Jelinek <jakub@redhat.com>
524 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
525 around drhs if type conversion to lacc->type is not useless.
527 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
529 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
531 (cortex-a57.cortex-a53): Likewise.
532 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
534 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
536 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
539 2014-02-11 Renlin Li <Renlin.Li@arm.com>
541 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
542 add_options_for_arm_vfp3.
544 2014-02-11 Jeff Law <law@redhat.com>
547 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
548 object with an undesirable mode.
550 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
553 * config/i386/sol2-9.h: New file.
554 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
555 *-*-solaris2.9*): Use it.
557 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
559 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
560 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
562 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
564 * config/microblaze/microblaze.c: Extend mcpu version format
566 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
568 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
570 2014-02-10 Richard Henderson <rth@redhat.com>
573 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
574 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
575 ms-abi vs -mno-accumulate-outgoing-args.
576 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
577 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
580 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
583 * cfgexpand.c (expand_asm_operands): Attach source location to
584 ASM_INPUT rtx objects.
585 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
587 2014-02-10 Nick Clifton <nickc@redhat.com>
589 * config/mn10300/mn10300.c (popcount): New function.
590 (mn10300_expand_prologue): Include saved registers in stack usage
593 2014-02-10 Jeff Law <law@redhat.com>
596 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
597 when changing the SET_DEST of a prior insn to avoid an input reload.
599 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
601 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
602 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
603 -mcall-openbsd, or -mcall-linux.
604 (CC1_ENDIAN_BIG_SPEC): Remove.
605 (CC1_ENDIAN_LITTLE_SPEC): Remove.
606 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
607 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
608 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
609 and %cc1_endian_default.
610 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
612 2014-02-10 Richard Biener <rguenther@suse.de>
614 PR tree-optimization/60115
615 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
616 MEM_REF handling. Properly verify that the accesses are not
617 out of the objects bound.
619 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
621 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
624 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
626 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
627 proper constants and fix formatting.
628 (possible_polymorphic_call_targets): Fix formatting.
630 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
631 Ilya Tocar <ilya.tocar@intel.com>
633 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
634 (_mm512_loadu_epi32): Renamed into...
635 (_mm512_loadu_si512): This.
636 (_mm512_storeu_epi32): Renamed into...
637 (_mm512_storeu_si512): This.
638 (_mm512_maskz_ceil_ps): Removed.
639 (_mm512_maskz_ceil_pd): Ditto.
640 (_mm512_maskz_floor_ps): Ditto.
641 (_mm512_maskz_floor_pd): Ditto.
642 (_mm512_floor_round_ps): Ditto.
643 (_mm512_floor_round_pd): Ditto.
644 (_mm512_ceil_round_ps): Ditto.
645 (_mm512_ceil_round_pd): Ditto.
646 (_mm512_mask_floor_round_ps): Ditto.
647 (_mm512_mask_floor_round_pd): Ditto.
648 (_mm512_mask_ceil_round_ps): Ditto.
649 (_mm512_mask_ceil_round_pd): Ditto.
650 (_mm512_maskz_floor_round_ps): Ditto.
651 (_mm512_maskz_floor_round_pd): Ditto.
652 (_mm512_maskz_ceil_round_ps): Ditto.
653 (_mm512_maskz_ceil_round_pd): Ditto.
654 (_mm512_expand_pd): Ditto.
655 (_mm512_expand_ps): Ditto.
656 * config/i386/i386.c (ix86_builtins): Remove
657 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
659 * config/i386/predicates.md (const1256_operand): New.
660 (const_1_to_2_operand): Ditto.
661 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
662 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
663 (*avx512pf_gatherpf<mode>sf): Ditto.
664 (avx512pf_gatherpf<mode>df): Ditto.
665 (*avx512pf_gatherpf<mode>df_mask): Ditto.
666 (*avx512pf_gatherpf<mode>df): Ditto.
667 (avx512pf_scatterpf<mode>sf): Ditto.
668 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
669 (*avx512pf_scatterpf<mode>sf): Ditto.
670 (avx512pf_scatterpf<mode>df): Ditto.
671 (*avx512pf_scatterpf<mode>df_mask): Ditto.
672 (*avx512pf_scatterpf<mode>df): Ditto.
673 (avx512f_expand<mode>): Removed.
674 (<shift_insn><mode>3<mask_name>): Change predicate type.
676 2014-02-08 Jakub Jelinek <jakub@redhat.com>
678 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
679 not at the end of datarefs vector use ordered_remove to avoid
680 reordering datarefs vector.
683 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
684 mark local addressable non-static vars as GOVD_PRIVATE
685 instead of GOVD_LOCAL.
686 * omp-low.c (lower_omp_for): Move gimple_bind_vars
687 and BLOCK_VARS of gimple_bind_block to new_stmt rather
691 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
692 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
693 assume_aligned or alloc_align attributes.
694 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
695 arguments. Handle also assume_aligned and alloc_align attributes.
696 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
697 calls to functions with assume_aligned or alloc_align attributes.
698 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
700 2014-02-08 Terry Guo <terry.guo@arm.com>
702 * doc/invoke.texi: Document ARM -march=armv7e-m.
704 2014-02-08 Jakub Jelinek <jakub@redhat.com>
706 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
707 flag on __cilkrts_rethrow builtin.
710 * ipa-cp.c (determine_versionability): Fail at -O0
711 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
712 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
715 2014-02-04 Jakub Jelinek <jakub@redhat.com>
718 * tree-inline.c (copy_forbidden): Fail for
719 __attribute__((optimize (0))) functions.
721 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
723 * varpool.c: Include pointer-set.h.
724 (varpool_remove_unreferenced_decls): Variables in other partitions
725 will not be output; be however careful to not lose information
728 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
730 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
731 lookup in the vtable constructor.
733 2014-02-07 Jeff Law <law@redhat.com>
736 * config/m68k/m68k.md (ashldi_extsi): Turn into a
737 define_insn_and_split.
739 * ipa-inline.c (inline_small_functions): Fix typos.
741 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
743 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
744 (s390_can_use_return_insn): Declare.
745 * config/s390/s390.h (EPILOGUE_USES): Define.
746 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
748 (s390_chunkify_start): Handle return JUMP_LABELs.
749 (s390_early_mach): Emit a main_pool instruction on the entry edge.
750 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
751 (s390_can_use_return_insn): New functions.
752 (s390_fix_long_loop_prediction): Handle conditional returns.
753 (TARGET_SET_UP_BY_PROLOGUE): Define.
754 * config/s390/s390.md (ANY_RETURN): New code iterator.
755 (*creturn, *csimple_return, return, simple_return): New patterns.
757 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
759 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
760 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
761 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
762 REG_CFA_RESTORE list when deciding not to restore a register.
764 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
766 * config/s390/s390.c: Include tree-pass.h and context.h.
767 (s390_early_mach): New function, split out from...
768 (s390_emit_prologue): ...here.
769 (pass_data_s390_early_mach): New pass structure.
770 (pass_s390_early_mach): New class.
771 (s390_option_override): Create and register early_mach pass.
774 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
776 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
777 to match for the exit block.
779 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
781 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
782 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
783 Reject misaligned operands.
785 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
787 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
789 2014-02-07 Richard Biener <rguenther@suse.de>
792 * gimple-low.c (lower_builtin_posix_memalign): New function.
793 (lower_stmt): Call it to lower posix_memalign in a way
794 to make alignment info accessible.
796 2014-02-07 Jakub Jelinek <jakub@redhat.com>
799 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
800 __builtin_setjmp_receiver.
802 2014-02-07 Richard Biener <rguenther@suse.de>
805 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
806 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
807 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
808 Handle BUILT_IN_POSIX_MEMALIGN.
809 (find_func_clobbers): Likewise.
810 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
811 (call_may_clobber_ref_p_1): Likewise.
813 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
816 * ipa-devirt.c (record_target_from_binfo): Remove overactive
819 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
822 * lto-cgraph.c (lto_output_node): Use
823 symtab_get_symbol_partitioning_class.
824 (lto_output_varpool_node): likewise.
825 (symtab_get_symbol_partitioning_class): Move here from
827 * cgraph.h (symbol_partitioning_class): Likewise.
828 (symtab_get_symbol_partitioning_class): Declare.
830 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
832 * ggc.h (ggc_internal_cleared_alloc): New macro.
833 * vec.h (vec_safe_copy): Handle memory stats.
834 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
835 * target-globals.c (save_target_globals): Likewise.
837 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
840 * expr.c (emit_move_resolve_push): Export; be bit more selective
841 on when to clear alias set.
842 * expr.h (emit_move_resolve_push): Declare.
843 * function.h (struct function): Add tail_call_marked.
844 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
845 * config/i386/i386-protos.h (ix86_expand_push): Remove.
846 * config/i386/i386.md (TImode move expander): De not call
848 (FP push expanders): Preserve memory attributes.
849 * config/i386/sse.md (push<mode>1): Remove.
850 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
851 (ix86_expand_push): Remove.
852 * config/i386/mmx.md (push<mode>1): Remove.
854 2014-02-06 Jakub Jelinek <jakub@redhat.com>
856 PR rtl-optimization/60030
857 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
858 lopart with paradoxical subreg before shifting it up by hprec.
860 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
862 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
863 Remove extra newline at end of file.
864 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
865 (arm_issue_rate): Handle cortexa57.
866 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
867 (cortex-a57.cortex-a53): Likewise.
869 2014-02-06 Jakub Jelinek <jakub@redhat.com>
872 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
873 don't record in REG_FRAME_RELATED_EXPR registers not set in that
875 (arm_expand_prologue): Adjust all callers.
876 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
877 info, registers also at the lowest numbered registers side. Use
878 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
882 * var-tracking.c (adjust_mems): Before adding a SET to
883 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
885 2014-02-06 Alan Modra <amodra@gmail.com>
888 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
889 change SDmode to DDmode when lra_in_progress.
891 2014-02-06 Jakub Jelinek <jakub@redhat.com>
894 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
895 free_data_ref on the dr first, and before goto again also set dr
896 to the next dr. For simd_lane_access, free old datarefs[i] before
897 overwriting it. For get_vectype_for_scalar_type failure, don't
898 free_data_ref if simd_lane_access.
900 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
903 * tree.h (opts_for_fn): New inline function.
904 (opt_for_fn): Define.
905 * config/i386/i386.c (ix86_function_regparm): Use
906 opt_for_fn (decl, optimize) instead of optimize.
908 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
910 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
911 for SYMBOL_REF in large memory model.
913 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
915 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
917 (cortex-a57): Likewise.
918 (cortex-a57.cortex-a53): Likewise.
920 2014-02-06 Yury Gribov <y.gribov@samsung.com>
921 Kugan Vivekanandarajah <kuganv@linaro.org>
923 * config/arm/arm.c (arm_vector_alignment_reachable): Check
925 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
927 2014-02-06 Richard Biener <rguenther@suse.de>
929 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
930 set_loop_copy and initialize_original_copy_tables.
932 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
934 * config/aarch64/aarch64-simd.md
935 (aarch64_ashr_simddi): Change QI to SI.
937 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
938 Jakub Jelinek <jakub@redhat.com>
941 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
944 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
946 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
947 CODE_FOR_altivec_vpku[hw]um to
948 CODE_FOR_altivec_vpku[hw]um_direct.
949 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
950 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
951 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
952 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
954 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
956 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
957 generation for -maltivec=be.
958 (altivec_vsumsws): Simplify redundant test.
960 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
962 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
963 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
964 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
965 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
967 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
969 (altivec_vpks<VI_char>ss): Likewise.
970 (altivec_vpks<VI_char>us): Likewise.
971 (altivec_vpku<VI_char>us): Likewise.
972 (altivec_vpku<VI_char>um): Likewise.
973 (altivec_vpku<VI_char>um_direct): New (copy of
974 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
976 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
977 target is little endian and -maltivec=be is not specified.
978 (*altivec_vupkhs<VU_char>_direct): New (copy of
979 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
980 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
981 target is little endian and -maltivec=be is not specified.
982 (*altivec_vupkls<VU_char>_direct): New (copy of
983 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
984 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
985 little endian and -maltivec=be is not specified.
986 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
987 little endian and -maltivec=be is not specified.
989 2014-02-05 Richard Henderson <rth@redhat.com>
992 * combine-stack-adj.c: Revert r206943.
993 * sched-int.h (struct deps_desc): Add last_args_size.
994 * sched-deps.c (init_deps): Initialize it.
995 (sched_analyze_insn): Add OUTPUT dependencies between insns that
996 contain REG_ARGS_SIZE notes.
998 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1000 * lto-cgraph.c (asm_nodes_output): Make global.
1001 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1002 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1003 (driver_handle_option): Handle OPT_fwpa.
1005 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1008 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1009 a comment typo and formatting issue. If odr_hash hasn't been
1010 created, return vNULL and set *completep to false.
1013 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1014 bb with no successors.
1016 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1019 * doc/invoke.texi (-march): Clarify documentation for ARM.
1023 2014-02-05 Richard Biener <rguenther@suse.de>
1025 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1026 when not vectorizing because of too many alias checks.
1027 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1028 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1030 2014-02-05 Nick Clifton <nickc@redhat.com>
1032 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1033 accept extended registers in any mode when compiling for the MN10300.
1035 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1037 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1038 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1039 sanitization attributes.
1040 (can_inline_edge_p): Likewise.
1041 (sanitize_attrs_match_for_inline_p): New function.
1043 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1045 * ipa-prop.c (detect_type_change): Shor circuit testing of
1046 type changes on THIS pointer.
1048 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1051 * config/pa/pa.c (legitimize_tls_address): Return original address
1052 if not passed a SYMBOL_REF rtx.
1053 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1055 (pa_emit_move_sequence): Simplify TLS source operands.
1056 (pa_legitimate_constant_p): Reject all TLS constants.
1057 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1058 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1060 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1062 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1063 groups when we know they are controlled by LTO.
1064 * varasm.c (default_binds_local_p_1): If object is in other partition,
1065 it will be resolved locally.
1067 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1069 * config/host-linux.c (linux_gt_pch_use_address): Don't
1070 use SSIZE_MAX because it is not always defined.
1072 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1075 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1076 threshold for pseudo splitting.
1077 (update_ebb_live_info): Process call argument hard registers and
1078 hard registers from insn definition too.
1079 (max_small_class_regs_num): New constant.
1080 (inherit_in_ebb): Update live hard regs through EBBs. Update
1081 reloads_num only for small register classes. Don't split for
1084 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1087 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1090 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1092 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1095 2014-02-04 Marek Polacek <polacek@redhat.com>
1097 * gdbinit.in (pel): Define.
1099 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1101 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1104 2014-02-04 Richard Biener <rguenther@suse.de>
1107 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1108 in function context local.
1109 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1110 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1111 similar to LTO_imported_decl_ref.
1113 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1115 PR tree-optimization/60002
1116 * cgraphclones.c (build_function_decl_skip_args): Clear
1119 PR tree-optimization/60023
1120 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1121 false to gsi_replace.
1122 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1123 has been in some EH region and vec_stmt could throw, add
1124 vec_stmt into the same EH region.
1125 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1126 has no lhs, ignore it.
1127 * internal-fn.c (expand_MASK_LOAD): Likewise.
1130 * tree-inline.c (copy_forbidden): Fail for
1131 __attribute__((optimize (0))) functions.
1134 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1135 is set, copy one less argument.
1136 (expand_simd_clones): Don't subtract clone_info->inbranch
1137 from simd_clone_struct_alloc argument.
1139 PR rtl-optimization/57915
1140 * recog.c (simplify_while_replacing): If all unary/binary/relational
1141 operation arguments are constant, attempt to simplify those.
1144 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1145 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1147 2014-02-04 Richard Biener <rguenther@suse.de>
1149 PR tree-optimization/60012
1150 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1151 TBAA disambiguation to all DDRs.
1153 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1156 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1157 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1159 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1162 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1164 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1166 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1167 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1169 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1172 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1173 to figure out targets of polymorphic calls with known decl.
1174 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1175 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1176 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1177 (get_polymorphic_call_info): ... here.
1178 (get_polymorphic_call_info_from_invariant): New function.
1180 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1182 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1183 lookup via vtable pointer; check for type consistency
1184 and turn inconsitent facts into UNREACHABLE.
1185 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1186 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1187 type inconsistent querries; return UNREACHABLE instead.
1189 2014-02-03 Richard Henderson <rth@twiddle.net>
1192 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1193 already processed this node.
1194 (normalize_one_pred_1): Pass along mark_set.
1195 (normalize_one_pred): Create and destroy a pointer_set_t.
1196 (normalize_one_pred_chain): Likewise.
1198 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1200 PR gcov-profile/58602
1201 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1203 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1206 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1207 -fno-devirtualize; try to devirtualize by the knowledge of
1208 virtual table pointer given by aggregate propagation.
1209 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1210 (ipa_print_node_jump_functions): Dump also offset that
1211 is relevant for polymorphic calls.
1212 (determine_known_aggregate_parts): Add arg_type parameter; use it
1213 instead of determining the type from pointer type.
1214 (ipa_compute_jump_functions_for_edge): Update call of
1215 determine_known_aggregate_parts.
1216 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1217 (gimple_get_virt_method_for_binfo): ... here; simplify using
1218 vtable_pointer_value_to_vtable.
1219 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1220 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1221 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1222 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1224 (vtable_pointer_value_to_binfo): ... here.
1225 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1227 2014-02-03 Teresa Johnson <tejohnson@google.com>
1229 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1230 redef of outer loop index variable.
1232 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1236 * doc/extend.texi (Function Attributes): Typo.
1238 2014-02-03 Cong Hou <congh@google.com>
1240 PR tree-optimization/60000
1241 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1242 if the vectorized statement is a store. A store statement can only
1243 appear at the end of pattern statements.
1245 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1247 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1248 (ix86_option_override_internal): Default long double to 64-bit for
1249 32-bit Bionic and to 128-bit for 64-bit Bionic.
1251 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1252 TARGET_LONG_DOUBLE_128 is true.
1253 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1255 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1256 (mlong-double-64): Negate -mlong-double-128.
1257 (mlong-double-128): New option.
1259 * config/i386/i386-c.c (ix86_target_macros): Define
1260 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1262 * doc/invoke.texi: Document -mlong-double-128.
1264 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1266 PR rtl-optimization/60024
1267 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1269 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1271 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1273 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1275 PR rtl-optimization/57662
1276 * sel-sched.c (code_motion_path_driver): Do not mark already not
1277 existing blocks in the visiting bitmap.
1279 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1281 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1282 on the insn being emitted.
1284 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1285 Will Deacon <will.deacon@arm.com>
1287 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1289 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1291 * config/arm/arm-tables.opt: Regenerate.
1293 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1295 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1296 for vector types other than V16QImode.
1297 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1298 define_expand, and call altivec_expand_vec_perm_le when producing
1299 code with little endian element order.
1300 (*altivec_vperm_<mode>_internal): New insn having previous
1301 behavior of altivec_vperm_<mode>.
1302 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1303 altivec_expand_vec_perm_le when producing code with little endian
1305 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1306 behavior of altivec_vperm_<mode>_uns.
1308 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1310 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1311 (altivec_vsumsws): Add handling for -maltivec=be with a little
1313 (altivec_vsumsws_direct): New.
1314 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1315 gen_altivec_vsumsws.
1317 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1319 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1320 vtable_pointer_value_to_binfo): New functions.
1321 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1322 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1324 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1326 * config/nios2/nios2.md (load_got_register): Initialize GOT
1327 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1328 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1330 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1332 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1333 preserverd by passthrough, do not propagate the type.
1335 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1337 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1338 (mips_atomic_assign_expand_fenv): New function.
1339 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1341 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1343 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1344 (__builtin_mips_set_fcsr): Likewise.
1345 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1346 MIPS_USI_FTYPE_VOID.
1347 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1348 (mips16_expand_set_fcsr): Likewise.
1349 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1350 (mips16_set_fcsr_stub): Likewise.
1351 (mips16_get_fcsr_one_only_stub): New class.
1352 (mips16_set_fcsr_one_only_stub): Likewise.
1353 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1354 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1355 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1356 (hard_float): New availability predicate.
1357 (mips_builtins): Add get_fcsr and set_fcsr.
1358 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1359 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1360 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1361 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1362 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1365 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1367 * config/mips/mips.c (mips_one_only_stub): New class.
1368 (mips_need_mips16_rdhwr_p): Replace with...
1369 (mips16_rdhwr_stub): ...this new variable.
1370 (mips16_stub_call_address): New function.
1371 (mips16_rdhwr_one_only_stub): New class.
1372 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1373 (mips_output_mips16_rdhwr): Delete.
1374 (mips_finish_stub): New function.
1375 (mips_code_end): Use it to handle rdhwr stubs.
1377 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1380 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1381 when calculating size of integer atomic types.
1383 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1385 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1387 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1389 PR tree-optimization/60003
1390 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1391 * profile.c (branch_prob): Use gimple_call_builtin_p
1392 to check for BUILT_IN_SETJMP_RECEIVER.
1393 * tree-inline.c (copy_bb): Call notice_special_calls.
1395 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1398 * lra-constraints.c (process_alt_operands): Update reload_sum only
1401 2014-01-31 Richard Henderson <rth@redhat.com>
1404 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1405 until after else_eh is processed.
1407 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1409 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1410 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1411 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1412 in smmintrin.h, remove them.
1413 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1414 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1415 * config/i386/i386.md (ROUND_SAE): Fix value.
1416 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1417 (const48_operand): New.
1418 * config/i386/subst.md (round), (round_expand): Use
1419 const_4_or_8_to_11_operand.
1420 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1422 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1424 * config/i386/constraints.md (Yk): Swap meaning with k.
1425 * config/i386/i386.md (movhi_internal): Change Yk to k.
1426 (movqi_internal): Ditto.
1427 (*k<logic><mode>): Ditto.
1430 (kandn<mode>): Ditto.
1431 (*<code>hi_1): Ditto.
1432 (*<code>qi_1): Ditto.
1433 (kxnor<mode>): Ditto.
1434 (kortestzhi): Ditto.
1435 (kortestchi): Ditto.
1437 (*one_cmplhi2_1): Ditto.
1438 (*one_cmplqi2_1): Ditto.
1439 * config/i386/sse.md (): Change k to Yk.
1440 (avx512f_load<mode>_mask): Ditto.
1441 (avx512f_blendm<mode>): Ditto.
1442 (avx512f_store<mode>_mask): Ditto.
1443 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1444 (avx512f_storedqu<mode>_mask): Ditto.
1445 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1447 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1448 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1449 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1450 (avx512f_maskcmp<mode>3): Ditto.
1451 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1452 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1453 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1454 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1455 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1456 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1457 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1458 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1459 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1460 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1461 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1462 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1463 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1464 (vec_extract_lo_<mode>_maskm): Ditto.
1465 (vec_extract_hi_<mode>_maskm): Ditto.
1466 (avx512f_vternlog<mode>_mask): Ditto.
1467 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1468 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1469 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1470 (avx512f_<code>v8div16qi2_mask): Ditto.
1471 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1472 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1473 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1474 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1475 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1476 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1477 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1478 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1479 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1480 (avx512cd_maskb_vec_dupv8di): Ditto.
1481 (avx512cd_maskw_vec_dupv16si): Ditto.
1482 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1483 (avx512f_vpermi2var<mode>3_mask): Ditto.
1484 (avx512f_vpermi2var<mode>3_mask): Ditto.
1485 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1486 (*avx512f_gathersi<mode>): Ditto.
1487 (*avx512f_gathersi<mode>_2): Ditto.
1488 (*avx512f_gatherdi<mode>): Ditto.
1489 (*avx512f_gatherdi<mode>_2): Ditto.
1490 (*avx512f_scattersi<mode>): Ditto.
1491 (*avx512f_scatterdi<mode>): Ditto.
1492 (avx512f_compress<mode>_mask): Ditto.
1493 (avx512f_compressstore<mode>_mask): Ditto.
1494 (avx512f_expand<mode>_mask): Ditto.
1495 * config/i386/subst.md (mask): Change k to Yk.
1496 (mask_scalar_merge): Ditto.
1499 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1501 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1503 2014-01-31 Richard Biener <rguenther@suse.de>
1506 * builtins.c (fold_builtin_memory_op): Make sure to not
1507 use a floating-point mode or a boolean or enumeral type for
1510 2014-01-30 DJ Delorie <dj@redhat.com>
1512 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1513 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1514 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1515 whenever main() has an epilogue.
1517 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1519 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1520 unused variable "field".
1521 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1522 (vsx_mergeh_<mode>): Likewise.
1523 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1524 (altivec_vmrghh): Likewise.
1525 (altivec_vmrghw): Likewise.
1526 (altivec_vmrglb): Likewise.
1527 (altivec_vmrglh): Likewise.
1528 (altivec_vmrglw): Likewise.
1529 (altivec_vspltb): Add missing uses.
1530 (altivec_vsplth): Likewise.
1531 (altivec_vspltw): Likewise.
1532 (altivec_vspltsf): Likewise.
1534 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1537 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1538 frame related instructions.
1540 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1542 PR rtl-optimization/59959
1543 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1544 any reload of register whose subreg is invalid.
1546 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1548 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1549 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1550 Add missing return type - void.
1552 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1554 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1555 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1556 remove element index adjustment for endian (now handled in vsx.md
1558 (altivec_expand_vec_perm_const): Use
1559 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1560 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1561 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1562 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1563 define_expand and a new define_insn *altivec_vspltb_internal;
1564 adjust for -maltivec=be on a little endian target.
1565 (altivec_vspltb_direct): New.
1566 (altivec_vsplth): Divide into a define_expand and a new
1567 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1568 little endian target.
1569 (altivec_vsplth_direct): New.
1570 (altivec_vspltw): Divide into a define_expand and a new
1571 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1572 little endian target.
1573 (altivec_vspltw_direct): New.
1574 (altivec_vspltsf): Divide into a define_expand and a new
1575 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1576 a little endian target.
1578 2014-01-30 Richard Biener <rguenther@suse.de>
1580 PR tree-optimization/59993
1581 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1582 can propagate form the earlier stmt and avoid the transform
1583 when the intermediate result is needed.
1585 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1587 * README.Portability: Fix typo.
1589 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1591 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1592 comparison_operator with ordered_comparison_operator.
1594 2014-01-30 Nick Clifton <nickc@redhat.com>
1596 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1597 Rename to mn10300_store_multiple_regs.
1598 * config/mn10300/mn10300.c: Likewise.
1599 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1600 store_multiple_regs.
1601 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1602 Call mn10300_store_multiple_regs.
1604 2014-01-30 Nick Clifton <nickc@redhat.com>
1605 DJ Delorie <dj@redhat.com>
1607 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1608 %fp 2 to keep registers after it properly word-aligned.
1609 (rl78_alloc_physical_registers_umul): Handle the case where both
1610 input operands are the same.
1612 2014-01-30 Richard Biener <rguenther@suse.de>
1614 PR tree-optimization/59903
1615 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1618 2014-01-30 Jason Merrill <jason@redhat.com>
1621 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1624 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1626 2014-01-30 Richard Biener <rguenther@suse.de>
1628 PR tree-optimization/59951
1629 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1631 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1634 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1635 SFmode to DFmode case.
1637 2014-01-29 DJ Delorie <dj@redhat.com>
1639 * config/msp430/msp430.opt (-minrt): New.
1640 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1642 (ENDFILE_SPEC): Likewise.
1644 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1646 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1647 (estimate_function_body_sizes): Use it.
1649 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1652 * dwarf2out.c (is_cxx_auto): New.
1653 (is_base_type): Use it.
1654 (gen_type_die_with_usage): Likewise.
1656 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1658 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1659 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1660 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1661 -maltivec=be with LE targets.
1662 (vsx_mergeh_<mode>): Likewise.
1663 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1664 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1665 (altivec_vmrghb): Replace with define_expand and new
1666 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1667 (altivec_vmrghb_direct): New define_insn.
1668 (altivec_vmrghh): Replace with define_expand and new
1669 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1670 (altivec_vmrghh_direct): New define_insn.
1671 (altivec_vmrghw): Replace with define_expand and new
1672 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1673 (altivec_vmrghw_direct): New define_insn.
1674 (*altivec_vmrghsf): Adjust for endianness.
1675 (altivec_vmrglb): Replace with define_expand and new
1676 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1677 (altivec_vmrglb_direct): New define_insn.
1678 (altivec_vmrglh): Replace with define_expand and new
1679 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1680 (altivec_vmrglh_direct): New define_insn.
1681 (altivec_vmrglw): Replace with define_expand and new
1682 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1683 (altivec_vmrglw_direct): New define_insn.
1684 (*altivec_vmrglsf): Adjust for endianness.
1685 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1686 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1687 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1688 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1689 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1690 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1691 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1692 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1694 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1696 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1697 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1700 2014-01-29 Richard Biener <rguenther@suse.de>
1702 PR tree-optimization/58742
1703 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1704 associate_pointerplus_align.
1705 (associate_pointerplus_diff): New function.
1706 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1707 and associate_pointerplus_diff.
1709 2014-01-29 Richard Biener <rguenther@suse.de>
1711 * lto-streamer.h (LTO_major_version): Bump to 3.
1712 (LTO_minor_version): Reset to 0.
1714 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1716 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1717 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1718 (arm_file_start): Generate correct asm header for armv7ve.
1719 * config/arm/bpabi.h: Add multilib support for armv7ve.
1720 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1721 and cortex-a15 to armv7ve.
1722 * config/arm/t-aprofile: Add multilib support for armv7ve.
1723 * doc/invoke.texi: Document -march=armv7ve.
1725 2014-01-29 Richard Biener <rguenther@suse.de>
1727 PR tree-optimization/58742
1728 * tree-ssa-forwprop.c (associate_plusminus): Return true
1729 if we changed sth, defer EH cleanup to ...
1730 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1731 (simplify_mult): New function.
1733 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1736 PR tree-optimization/59920
1737 * tree.c (build_common_builtin_nodes): Remove
1738 __builtin_setjmp_dispatcher initialization.
1739 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1740 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1741 instead of gsi_after_labels + manually skipping debug stmts.
1742 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1743 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1744 * tree-inline.c (copy_edges_for_bb): Remove
1745 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1746 argument. Ignore computed_goto_p stmts. Don't call
1747 make_abnormal_goto_edges. If a call might need abnormal edges
1748 for non-local gotos, see if it already has an edge to
1749 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1750 with true argument, don't do anything then, otherwise add
1751 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1752 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1754 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1755 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1756 (lower_stmt): Don't set data->calls_builtin_setjmp.
1757 (lower_builtin_setjmp): Adjust comment.
1758 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1759 * tree-cfg.c (found_computed_goto): Remove.
1760 (factor_computed_gotos): Remove.
1761 (make_goto_expr_edges): Return bool, true for computed gotos.
1762 Don't call make_abnormal_goto_edges.
1763 (build_gimple_cfg): Don't set found_computed_goto, don't call
1764 factor_computed_gotos.
1765 (computed_goto_p): No longer static.
1766 (make_blocks): Don't set found_computed_goto.
1767 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1768 (make_edges): If make_goto_expr_edges returns true, push bb
1769 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1770 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1771 vector. Record mapping between bbs and OpenMP regions if there
1772 are any, adjust make_gimple_omp_edges caller. Call
1773 handle_abnormal_edges.
1774 (make_abnormal_goto_edges): Remove.
1775 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1776 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1777 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1778 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1779 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1780 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1781 filling *region also set *region_idx to (*region)->entry->index.
1784 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1785 For REGs set ORIGINAL_REGNO.
1787 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1789 * doc/md.texi: Mention that a target shouldn't implement
1790 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1793 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1795 PR tree-optimization/59594
1796 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1797 a copy of the datarefs vector rather than the vector itself.
1799 2014-01-28 Jason Merrill <jason@redhat.com>
1802 * dwarf2out.c (auto_die): New static.
1803 (gen_type_die_with_usage): Handle C++1y 'auto'.
1804 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1807 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1810 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1811 (SPEC_X32): Likewise.
1812 (SPEC_64): Likewise.
1813 * config/i386/i386.c (ix86_option_override_internal): Turn off
1814 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1816 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1817 * config/i386/i386.h (TARGET_16BIT): New macro.
1818 (TARGET_16BIT_P): Likewise.
1819 * config/i386/i386.opt: Add m16.
1820 * doc/invoke.texi: Document -m16.
1822 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1824 PR preprocessor/59935
1825 * input.c (location_get_source_line): Bail out on when line number
1826 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1828 2014-01-28 Richard Biener <rguenther@suse.de>
1830 PR tree-optimization/58742
1831 * tree-ssa-forwprop.c (associate_plusminus): Handle
1832 pointer subtraction of the form (T)(P + A) - (T)P.
1834 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1836 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1839 2014-01-28 Richard Biener <rguenther@suse.de>
1842 2014-01-28 Richard Biener <rguenther@suse.de>
1844 PR rtl-optimization/45364
1845 PR rtl-optimization/59890
1846 * var-tracking.c (local_get_addr_clear_given_value): Handle
1847 already cleared slot.
1848 (val_reset): Handle not allocated local_get_addr_cache.
1849 (vt_find_locations): Use post-order on the inverted CFG.
1851 2014-01-28 Richard Biener <rguenther@suse.de>
1853 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1855 2014-01-28 Richard Biener <rguenther@suse.de>
1857 PR rtl-optimization/45364
1858 PR rtl-optimization/59890
1859 * var-tracking.c (local_get_addr_clear_given_value): Handle
1860 already cleared slot.
1861 (val_reset): Handle not allocated local_get_addr_cache.
1862 (vt_find_locations): Use post-order on the inverted CFG.
1864 2014-01-28 Alan Modra <amodra@gmail.com>
1866 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1867 * configure.ac <recursive call for build != host>: Define
1868 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1869 and LD_FOR_BUILD too.
1870 * configure: Regenerate.
1872 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1874 * config/i386/i386.c (get_builtin_code_for_version): Separate
1875 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1876 Broadwell from Haswell.
1878 2014-01-27 Steve Ellcey <sellcey@mips.com>
1880 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1881 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1882 * config/mips/mips.c (mips_option_override): Change setting
1884 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1885 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1886 Change from Mask to Var.
1888 2014-01-27 Jeff Law <law@redhat.com>
1890 * ipa-inline.c (inline_small_functions): Fix typo.
1892 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1894 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1895 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1896 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1897 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1898 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1899 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1900 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1901 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1902 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1903 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1904 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1905 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1906 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1907 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1908 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1909 (_mm512_storeu_epi64): Ditto.
1910 (_mm512_cmpge_epi32_mask): Ditto.
1911 (_mm512_cmpge_epu32_mask): Ditto.
1912 (_mm512_cmpge_epi64_mask): Ditto.
1913 (_mm512_cmpge_epu64_mask): Ditto.
1914 (_mm512_cmple_epi32_mask): Ditto.
1915 (_mm512_cmple_epu32_mask): Ditto.
1916 (_mm512_cmple_epi64_mask): Ditto.
1917 (_mm512_cmple_epu64_mask): Ditto.
1918 (_mm512_cmplt_epi32_mask): Ditto.
1919 (_mm512_cmplt_epu32_mask): Ditto.
1920 (_mm512_cmplt_epi64_mask): Ditto.
1921 (_mm512_cmplt_epu64_mask): Ditto.
1922 (_mm512_cmpneq_epi32_mask): Ditto.
1923 (_mm512_cmpneq_epu32_mask): Ditto.
1924 (_mm512_cmpneq_epi64_mask): Ditto.
1925 (_mm512_cmpneq_epu64_mask): Ditto.
1926 (_mm512_expand_pd): Ditto.
1927 (_mm512_expand_ps): Ditto.
1928 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1929 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1930 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1931 * config/i386/i386.c (ix86_builtins): Add
1932 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1933 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1934 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1935 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1936 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1937 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1938 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1939 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1940 IX86_BUILTIN_PMOVUSQW512_MEM.
1941 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1942 __builtin_ia32_pmovsqd512mem_mask,
1943 __builtin_ia32_pmovqd512mem_mask,
1944 __builtin_ia32_pmovusqw512mem_mask,
1945 __builtin_ia32_pmovsqw512mem_mask,
1946 __builtin_ia32_pmovqw512mem_mask,
1947 __builtin_ia32_pmovusdw512mem_mask,
1948 __builtin_ia32_pmovsdw512mem_mask,
1949 __builtin_ia32_pmovdw512mem_mask,
1950 __builtin_ia32_pmovqb512mem_mask,
1951 __builtin_ia32_pmovusqb512mem_mask,
1952 __builtin_ia32_pmovsqb512mem_mask,
1953 __builtin_ia32_pmovusdb512mem_mask,
1954 __builtin_ia32_pmovsdb512mem_mask,
1955 __builtin_ia32_pmovdb512mem_mask.
1956 (bdesc_args): Add __builtin_ia32_expanddf512,
1957 __builtin_ia32_expandsf512.
1958 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1959 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1960 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1961 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1962 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1963 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1964 (avx512f_<code>v8div16qi2_mask_store): This.
1965 (avx512f_expand<mode>): New.
1967 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1969 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1971 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1972 (_mm512_prefetch_i32scatter_pd): Ditto.
1973 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1974 (_mm512_prefetch_i64scatter_pd): Ditto.
1975 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1976 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1977 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1978 (_mm512_prefetch_i32scatter_ps): Ditto.
1979 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1980 (_mm512_prefetch_i64scatter_ps): Ditto.
1981 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1982 * config/i386/i386-builtin-types.def: Define
1983 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1984 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1985 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1986 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1987 IX86_BUILTIN_SCATTERPFQPD.
1988 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1989 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1990 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1991 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1992 __builtin_ia32_scatterpfqps.
1993 (ix86_expand_builtin): Expand new built-ins.
1994 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1995 fix memory access data type.
1996 (*avx512pf_gatherpf<mode>_mask): Ditto.
1997 (*avx512pf_gatherpf<mode>): Ditto.
1998 (avx512pf_scatterpf<mode>): Ditto.
1999 (*avx512pf_scatterpf<mode>_mask): Ditto.
2000 (*avx512pf_scatterpf<mode>): Ditto.
2001 (GATHER_SCATTER_SF_MEM_MODE): New.
2002 (avx512pf_gatherpf<mode>df): Ditto.
2003 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2004 (*avx512pf_scatterpf<mode>df): Ditto.
2006 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2009 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2010 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2013 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2015 * common/config/arm/arm-common.c
2016 (arm_rewrite_mcpu): Handle multiple names.
2018 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2020 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2022 * gimple-builder.h (create_gimple_tmp): Delete.
2024 2014-01-27 Christian Bruel <christian.bruel@st.com>
2026 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2029 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2031 * config/pa/pa.md (call): Generate indirect long calls to non-local
2032 functions when outputing 32-bit code.
2033 (call_value): Likewise except for special call to buggy powf function.
2035 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2036 portable runtime and PIC indirect calls.
2037 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2038 and PIC call sequences. Use ldo instead of blr to set return register
2039 in PIC call sequence.
2041 2014-01-25 Walter Lee <walt@tilera.com>
2043 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2044 avoid clobbering a live register.
2046 2014-01-25 Walter Lee <walt@tilera.com>
2048 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2049 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2050 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2051 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2053 2014-01-25 Walter Lee <walt@tilera.com>
2055 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2056 arguments on even registers.
2057 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2059 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2060 (BIGGEST_ALIGNMENT): Ditto.
2061 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2063 2014-01-25 Walter Lee <walt@tilera.com>
2065 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2066 insns before bundling.
2067 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2069 2014-01-25 Walter Lee <walt@tilera.com>
2071 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2072 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2073 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2075 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2077 * config/mips/constraints.md (kl): Delete.
2078 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2079 define expands, using...
2080 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2081 instructions for MIPS16.
2082 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2083 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2085 2014-01-25 Walter Lee <walt@tilera.com>
2087 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2091 2014-01-25 Walter Lee <walt@tilera.com>
2093 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2094 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2096 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2098 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2101 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2103 * print-rtl.c (in_call_function_usage): New var.
2104 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2105 EXPR_LIST mode as mode and not as reg note name.
2108 * cfgloopmanip.c (copy_loop_info): If
2109 loop->warned_aggressive_loop_optimizations, make sure
2110 the flag is set in target loop too.
2112 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2114 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2116 * builtins.def: Likewise.
2117 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2118 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2119 * ira.c (ira_setup_eliminable_regset): Likewise.
2120 * omp-low.c (gate_expand_omp): Likewise.
2121 (execute_lower_omp): Likewise.
2122 (diagnose_sb_0): Likewise.
2123 (gate_diagnose_omp_blocks): Likewise.
2124 (simd_clone_clauses_extract): Likewise.
2127 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2129 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2130 correction for little endian...
2131 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2134 2014-01-24 Jeff Law <law@redhat.com>
2136 PR tree-optimization/59919
2137 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2138 for non-returning calls.
2140 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2142 * common/config/aarch64/aarch64-common.c
2143 (aarch64_rewrite_mcpu): Handle multiple names.
2144 * config/aarch64/aarch64.h
2145 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2147 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2149 * input.c (add_file_to_cache_tab): Handle the case where fopen
2152 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2155 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2156 from push operand if code of push isn't PRE_DEC.
2158 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2161 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2162 -mquad-memory-atomic. Update -mquad-memory documentation to say
2163 it is only used for non-atomic loads/stores.
2165 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2166 -mquad-memory or -mquad-memory-atomic switches.
2168 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2169 -mquad-memory-atomic to ISA 2.07 support.
2171 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2172 to separate support of normal quad word memory operations (ldq, stq)
2173 from the atomic quad word memory operations.
2175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2176 support to separate non-atomic quad word operations from atomic
2177 quad word operations. Disable non-atomic quad word operations in
2178 little endian mode so that we don't have to swap words after the
2179 load and before the store.
2180 (quad_load_store_p): Add comment about atomic quad word support.
2181 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2182 options printed with -mdebug=reg.
2184 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2185 -mquad-memory-atomic as the test for whether we have quad word
2186 atomic instructions.
2187 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2188 or -mp8-vector are used, allow byte/half-word atomic operations.
2190 * config/rs6000/sync.md (load_lockedti): Insure that the address
2191 is a proper indexed or indirect address for the lqarx instruction.
2192 On little endian systems, swap the hi/lo registers after the lqarx
2194 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2195 insure the address is valid for the lqarx instruction.
2196 (store_conditionalti): Insure that the address is a proper indexed
2197 or indirect address for the stqcrx. instruction. On little endian
2198 systems, swap the hi/lo registers before doing the stqcrx.
2200 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2201 insure the address is valid for the stqcrx. instruction.
2203 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2204 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2205 type of quad memory support is available.
2207 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2210 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2211 there is a danger of looping.
2213 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2215 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2216 force flag_ira_loop_pressure if set via command line.
2218 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2220 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2221 (ashr_simd): New builtin handling DI mode.
2222 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2223 (aarch64_sshr_simddi): New match pattern.
2224 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2225 (vshrd_n_s64): Likewise.
2226 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2228 2014-01-23 Nick Clifton <nickc@redhat.com>
2230 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2231 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2232 favour of mcu specific scripts.
2233 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2236 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2237 Alex Velenko <Alex.Velenko@arm.com>
2239 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2240 (vaddv_s16): Likewise.
2241 (vaddv_s32): Likewise.
2242 (vaddv_u8): Likewise.
2243 (vaddv_u16): Likewise.
2244 (vaddv_u32): Likewise.
2245 (vaddvq_s8): Likewise.
2246 (vaddvq_s16): Likewise.
2247 (vaddvq_s32): Likewise.
2248 (vaddvq_s64): Likewise.
2249 (vaddvq_u8): Likewise.
2250 (vaddvq_u16): Likewise.
2251 (vaddvq_u32): Likewise.
2252 (vaddvq_u64): Likewise.
2253 (vaddv_f32): Likewise.
2254 (vaddvq_f32): Likewise.
2255 (vaddvq_f64): Likewise.
2256 (vmaxv_f32): Likewise.
2257 (vmaxv_s8): Likewise.
2258 (vmaxv_s16): Likewise.
2259 (vmaxv_s32): Likewise.
2260 (vmaxv_u8): Likewise.
2261 (vmaxv_u16): Likewise.
2262 (vmaxv_u32): Likewise.
2263 (vmaxvq_f32): Likewise.
2264 (vmaxvq_f64): Likewise.
2265 (vmaxvq_s8): Likewise.
2266 (vmaxvq_s16): Likewise.
2267 (vmaxvq_s32): Likewise.
2268 (vmaxvq_u8): Likewise.
2269 (vmaxvq_u16): Likewise.
2270 (vmaxvq_u32): Likewise.
2271 (vmaxnmv_f32): Likewise.
2272 (vmaxnmvq_f32): Likewise.
2273 (vmaxnmvq_f64): Likewise.
2274 (vminv_f32): Likewise.
2275 (vminv_s8): Likewise.
2276 (vminv_s16): Likewise.
2277 (vminv_s32): Likewise.
2278 (vminv_u8): Likewise.
2279 (vminv_u16): Likewise.
2280 (vminv_u32): Likewise.
2281 (vminvq_f32): Likewise.
2282 (vminvq_f64): Likewise.
2283 (vminvq_s8): Likewise.
2284 (vminvq_s16): Likewise.
2285 (vminvq_s32): Likewise.
2286 (vminvq_u8): Likewise.
2287 (vminvq_u16): Likewise.
2288 (vminvq_u32): Likewise.
2289 (vminnmv_f32): Likewise.
2290 (vminnmvq_f32): Likewise.
2291 (vminnmvq_f64): Likewise.
2293 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2295 * config/aarch64/aarch64-simd.md
2296 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2297 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2298 (*aarch64_mul3_elt<mode>): Likewise.
2299 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2300 (*aarch64_mul3_elt_to_64v2df): Likewise.
2301 (*aarch64_mla_elt<mode>): Likewise.
2302 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2303 (*aarch64_mls_elt<mode>): Likewise.
2304 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2305 (*aarch64_fma4_elt<mode>): Likewise.
2306 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2307 (*aarch64_fma4_elt_to_64v2df): Likewise.
2308 (*aarch64_fnma4_elt<mode>): Likewise.
2309 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2310 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2311 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2312 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2313 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2314 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2315 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2316 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2317 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2319 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2321 * config/aarch64/aarch64-simd.md
2322 (aarch64_be_checked_get_lane<mode>): New define_expand.
2323 * config/aarch64/aarch64-simd-builtins.def
2324 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2325 New builtin definition.
2326 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2327 Use new safe be builtin.
2329 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2331 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2333 (aarch64_be_st1<mode>): Likewise.
2334 (aarch_ld1<VALL:mode>): Define_expand modified.
2335 (aarch_st1<VALL:mode>): Likewise.
2336 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2337 (UNSPEC_ST1): Likewise.
2339 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2341 * config/microblaze/microblaze.md: Add trap insn and attribute
2343 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2345 PR preprocessor/58580
2346 * input.h (location_get_source_line): Take an additional line_size
2348 (void diagnostics_file_cache_fini): Declare new function.
2349 * input.c (struct fcache): New type.
2350 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2351 New static constants.
2352 (diagnostic_file_cache_init, total_lines_num)
2353 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2354 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2355 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2356 (get_next_line, read_next_line, goto_next_line, read_line_num):
2357 New static function definitions.
2358 (diagnostic_file_cache_fini): New function.
2359 (location_get_source_line): Take an additional output line_len
2360 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2362 * diagnostic.c (diagnostic_finish): Call
2363 diagnostic_file_cache_fini.
2364 (adjust_line): Take an additional input parameter for the length
2365 of the line, rather than calculating it with strlen.
2366 (diagnostic_show_locus): Adjust the use of
2367 location_get_source_line and adjust_line with respect to their new
2368 signature. While displaying a line now, do not stop at the first
2369 null byte. Rather, display the zero byte as a space and keep
2370 going until we reach the size of the line.
2371 * Makefile.in: Add vec.o to OBJS-libcommon
2373 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2374 Ilya Tocar <ilya.tocar@intel.com>
2376 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2377 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2378 (__builtin_ia32_kmov16): Ditto.
2379 * config/i386/i386.md (UNSPEC_KMOV): New.
2382 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2384 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2385 (_mm512_storeu_si512): Ditto.
2387 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2390 * rtl.h (get_referenced_operands): Declare.
2391 * recog.c (get_referenced_operands): New function.
2392 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2393 operands have been referenced when recording LO_SUM references.
2395 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2397 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2399 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2401 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2402 Enable for generic and recent AMD targets.
2404 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2406 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2407 ARG_SIZE note when adjustment was eliminated.
2409 2014-01-22 Jeff Law <law@redhat.com>
2411 PR tree-optimization/59597
2412 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2413 in file. Accept new argument REGISTERING and use it to modify
2414 dump output appropriately.
2415 (register_jump_thread): Corresponding changes.
2416 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2417 thread paths involving joiner blocks. Add code to dump cancelled
2418 jump threading paths.
2420 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2422 PR rtl-optimization/59477
2423 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2424 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2426 2014-01-22 Tom Tromey <tromey@redhat.com>
2428 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2430 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2432 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2434 PR rtl-optimization/59896
2435 * lra-constraints.c (process_alt_operands): Check unused note for
2436 matched operands of insn with no output reloads.
2438 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2440 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2441 (mips_move_from_gpr_cost): Likewise.
2443 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2445 PR rtl-optimization/59858
2446 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2447 ira_class_hard_regs_num.
2448 (process_alt_operands): Increase reject for dying matched operand.
2450 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2453 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2454 smaller than size, perform several stores or loads and stores
2455 at dst + count - size to store or copy all of size bytes, rather
2456 than just last modesize bytes.
2458 2014-01-20 DJ Delorie <dj@redhat.com>
2460 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2461 that CLOBBERs are REGs before propogating their values.
2463 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2466 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2467 (cgraph_inline_failed_type): New function.
2468 * cgraph.h (DEFCIFCODE): Add type.
2469 (cgraph_inline_failed_type_t): New enum.
2470 (cgraph_inline_failed_type): New prototype.
2471 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2472 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2473 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2474 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2475 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2476 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2477 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2478 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2479 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2480 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2481 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2482 OPTIMIZATION_MISMATCH.
2483 * tree-inline.c (expand_call_inline): Emit errors during
2484 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2486 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2489 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2490 mode attribute in insn output.
2492 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2494 * output.h (output_constant): Delete.
2495 * varasm.c (output_constant): Make private.
2497 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2499 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2501 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2504 * tree.h (fold_builtin_strcat): New prototype.
2505 * builtins.c (fold_builtin_strcat): No longer static. Add len
2506 argument, if non-NULL, don't call c_strlen. Optimize
2507 directly into __builtin_memcpy instead of __builtin_strcpy.
2508 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2509 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2511 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2513 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2514 for SImode_address_operand operands, having only a REG argument.
2516 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2518 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2519 loader name using mbig-endian.
2520 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2522 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2524 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2528 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2530 * config/aarch64/aarch64-protos.h
2531 (aarch64_cannot_change_mode_class_ptr): Declare.
2532 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2533 aarch64_cannot_change_mode_class_ptr): New.
2534 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2535 backend hook aarch64_cannot_change_mode_class.
2537 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2539 * common/config/aarch64/aarch64-common.c
2540 (aarch64_handle_option): Don't handle any option order logic here.
2541 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2542 selected_cpu, warn on architecture version mismatch.
2543 (aarch64_override_options): Fix parsing order for option strings.
2545 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2546 Iain Sandoe <iain@codesourcery.com>
2549 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2550 warning. Amend comment to reflect current functionality.
2552 2014-01-20 Richard Biener <rguenther@suse.de>
2555 * builtins.c (fold_builtin_strcat): Remove case better handled
2556 by tree-ssa-strlen.c.
2558 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2560 * config/aarch64/aarch64.opt
2561 (mcpu, march, mtune): Make case-insensitive.
2563 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2566 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2567 if operands[1] is a REG or ZERO_EXTEND of a REG.
2569 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2571 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2573 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2575 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2576 long non-pic millicode calls.
2578 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2580 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2582 2014-01-19 Kito Cheng <kito@0xlab.org>
2584 * builtins.c (expand_movstr): Check movstr expand done or fail.
2586 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2587 H.J. Lu <hongjiu.lu@intel.com>
2590 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2591 to DImode for zero-extended addresses.
2593 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2595 PR rtl-optimization/57763
2596 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2597 on the new indirect jump_insn and increment LABEL_NUSES (label).
2599 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2603 * config.gcc (x86_archs): New variable.
2604 (x86_64_archs): Likewise.
2605 (x86_cpus): Likewise.
2606 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2607 --with-arch/--with-cpu= options.
2608 Support --with-arch=/--with-cpu={nehalem,westmere,
2609 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2611 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2613 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2614 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2616 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2618 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2620 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2623 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2624 clear cpp_get_options (parse_in)->warn_unused_macros for
2625 ix86_target_macros_internal with cpp_define.
2627 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2629 * jump.c (delete_related_insns): Keep (use (insn))s.
2630 * reorg.c (redundant_insn): Check for barriers too.
2632 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2634 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2636 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2638 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2639 call to $$dyncall when TARGET_LONG_CALLS is true.
2641 2014-01-17 Jeff Law <law@redhat.com>
2643 * ree.c (combine_set_extension): Temporarily disable test for
2644 changing number of hard registers.
2646 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2649 * ipa-inline-analysis.c (inline_free_summary):
2650 Do not free summary of aliases.
2652 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2655 * gimplify.c (gimplify_expr): Use create_tmp_var
2656 instead of create_tmp_var_raw. If cond doesn't have
2657 integral type, don't add the IFN_ANNOTATE builtin at all.
2659 2014-01-17 Martin Jambor <mjambor@suse.cz>
2662 * ipa-cp.c (prev_edge_clone): New variable.
2663 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2664 Also resize prev_edge_clone vector.
2665 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2666 (ipcp_edge_removal_hook): New function.
2667 (ipcp_driver): Register ipcp_edge_removal_hook.
2669 2014-01-17 Andrew Pinski <apinski@cavium.com>
2670 Steve Ellcey <sellcey@mips.com>
2673 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2676 2014-01-17 Jeff Law <law@redhat.com>
2679 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2680 so that pass_ccp runs first.
2682 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2684 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2685 (ix86_adjust_cost): Use !TARGET_XXX.
2686 (do_reorder_for_imul): Likewise.
2687 (swap_top_of_ready_list): Likewise.
2688 (ix86_sched_reorder): Likewise.
2690 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2692 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2693 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2694 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2695 (intel_memset): New. Duplicate slm_memset.
2696 (intel_cost): New. Duplicate slm_cost.
2697 (m_INTEL): New macro.
2698 (processor_target_table): Add "intel".
2699 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2700 with PROCESSOR_INTEL for "intel".
2701 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2702 PROCESSOR_SILVERMONT.
2703 (ix86_issue_rate): Likewise.
2704 (ix86_adjust_cost): Likewise.
2705 (ia32_multipass_dfa_lookahead): Likewise.
2706 (swap_top_of_ready_list): Likewise.
2707 (ix86_sched_reorder): Likewise.
2708 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2709 instead of TARGET_OPT_AGU.
2710 * config/i386/i386.h (TARGET_INTEL): New.
2711 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2712 (processor_type): Add PROCESSOR_INTEL.
2713 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2714 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2716 2014-01-17 Marek Polacek <polacek@redhat.com>
2719 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2722 2014-01-17 Richard Biener <rguenther@suse.de>
2724 PR tree-optimization/46590
2725 * opts.c (default_options_table): Add entries for
2726 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2727 all enabled at -O1 but not for -Og.
2728 * common.opt (fbranch-count-reg): Remove Init(1).
2729 (fmove-loop-invariants): Likewise.
2730 (ftree-pta): Likewise.
2732 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2734 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2735 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2736 decls to at least the GCC 4.8 used alignments.
2739 * tree-nested.c (convert_nonlocal_reference_stmt,
2740 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2741 of GIMPLE_BIND stmts, adjust associated decls.
2743 2014-01-17 Richard Biener <rguenther@suse.de>
2745 PR tree-optimization/46590
2746 * vec.h (vec<>::bseach): New member function implementing
2747 binary search according to C89 bsearch.
2748 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2749 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2750 bitmap pointer again. Make accesses_in_loop a flat array.
2751 (mem_ref_obstack): New global.
2752 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2753 (mark_ref_stored): Likewise.
2754 (ref_indep_loop_p_2): Likewise.
2755 (set_ref_stored_in_loop): New helper function.
2756 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2757 (memref_free): Adjust.
2758 (record_mem_ref_loc): Simplify.
2759 (gather_mem_refs_stmt): Adjust.
2760 (sort_locs_in_loop_postorder_cmp): New function.
2761 (analyze_memory_references): Sort accesses_in_loop after
2762 loop postorder number.
2763 (find_ref_loc_in_loop_cmp): New function.
2764 (for_all_locs_in_loop): Find relevant cluster of locs in
2765 accesses_in_loop and iterate without recursion.
2766 (execute_sm): Avoid uninit warning.
2767 (struct ref_always_accessed): Simplify.
2768 (ref_always_accessed::operator ()): Likewise.
2769 (ref_always_accessed_p): Likewise.
2770 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2771 loop postorder numbers here.
2772 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2775 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2778 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2779 on decls for which assemble_alias has been called.
2781 2014-01-17 Nick Clifton <nickc@redhat.com>
2783 * config/msp430/msp430.opt: (mcpu): New option.
2784 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2785 (msp430_option_override): Parse target_cpu. If the MCU name
2786 matches a generic string, clear target_mcu.
2787 (msp430_attr): Allow numeric interrupt values up to 63.
2788 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2789 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2791 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2793 * config/msp430/msp430.md (popm): Use %J rather than %I.
2794 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2795 (addhi_cy_i): Use immediate_operand for operand 2.
2796 * doc/invoke.texi: Document -mcpu option.
2798 2014-01-17 Richard Biener <rguenther@suse.de>
2800 PR rtl-optimization/38518
2801 * df.h (df_analyze_loop): Declare.
2802 * df-core.c: Include cfgloop.h.
2803 (df_analyze_1): Split out main part of df_analyze.
2804 (df_analyze): Adjust.
2805 (loop_inverted_post_order_compute): New function.
2806 (loop_post_order_compute): Likewise.
2807 (df_analyze_loop): New function avoiding whole-function
2809 * loop-invariant.c (find_defs): Use df_analyze_loop.
2810 (find_invariants): Adjust.
2811 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2813 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2815 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2816 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2818 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2820 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2821 traversal when removing references.
2823 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2826 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2828 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2831 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2832 pushing a reload for an autoinc when we had previously reloaded an
2833 inner part of the address.
2835 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2837 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2839 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2840 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2841 when not giving up or versioning for alias only because of
2843 (vect_analyze_data_ref_dependences): Set to true.
2844 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2846 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2847 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2851 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2854 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2855 operand 0 predicate for gathers, use a new pseudo as subtarget.
2857 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2860 * lra-constraints.c (process_alt_operands): Add printing debug info.
2861 Check absence of input/output reloads for matched operands too.
2863 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2865 PR rtl-optimization/59835
2866 * ira.c (ira_init_register_move_cost): Increase cost for
2869 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2871 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2873 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2876 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2877 non-register objects. Use gen_(high/low)part more consistently.
2880 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2883 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2884 endian support, remove tests for WORDS_BIG_ENDIAN.
2885 (p8_mfvsrd_3_<mode>): Likewise.
2886 (reload_gpr_from_vsx<mode>): Likewise.
2887 (reload_gpr_from_vsxsf): Likewise.
2888 (p8_mfvsrd_4_disf): Likewise.
2890 2014-01-16 Richard Biener <rguenther@suse.de>
2892 PR rtl-optimization/46590
2893 * lcm.c (compute_antinout_edge): Use postorder iteration.
2894 (compute_laterin): Use inverted postorder iteration.
2896 2014-01-16 Nick Clifton <nickc@redhat.com>
2899 * varasm.c (output_constant): Return the number of bytes actually
2901 (output_constructor_array_range): Update the field size with the
2902 number of bytes emitted by output_constant.
2903 (output_constructor_regular_field): Likewise. Also do not
2904 complain if the total number of bytes emitted is now greater
2905 than the expected fieldpos.
2906 * output.h (output_constant): Update prototype and descriptive comment.
2908 2014-01-16 Marek Polacek <polacek@redhat.com>
2911 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2912 it is error_mark_node.
2914 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2916 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2917 VALID_AVX256_REG_OR_OI_MODE.
2919 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2921 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2922 current procedure should be profiled.
2924 2014-01-15 Andrew Pinski <apinski@cavium.com>
2926 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2927 of moving from/to the STACK_REG register class.
2929 2014-01-15 Richard Henderson <rth@redhat.com>
2932 * reginfo.c (global_regs_decl): Globalize.
2933 * rtl.h (global_regs_decl): Declare.
2934 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2935 reserved via global_regs.
2937 2014-01-15 Teresa Johnson <tejohnson@google.com>
2939 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2941 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2943 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2944 and vmulosh rather than call gen_vec_widen_smult_*.
2945 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2946 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2947 (vec_widen_smult_even_v16qi): Likewise.
2948 (vec_widen_umult_even_v8hi): Likewise.
2949 (vec_widen_smult_even_v8hi): Likewise.
2950 (vec_widen_umult_odd_v16qi): Likewise.
2951 (vec_widen_smult_odd_v16qi): Likewise.
2952 (vec_widen_umult_odd_v8hi): Likewise.
2953 (vec_widen_smult_odd_v8hi): Likewise.
2954 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2955 vmuloub rather than call gen_vec_widen_umult_*.
2956 (vec_widen_umult_lo_v16qi): Likewise.
2957 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2958 vmulosb rather than call gen_vec_widen_smult_*.
2959 (vec_widen_smult_lo_v16qi): Likewise.
2960 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2961 rather than call gen_vec_widen_umult_*.
2962 (vec_widen_umult_lo_v8hi): Likewise.
2963 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2964 rather than call gen_vec_widen_smult_*.
2965 (vec_widen_smult_lo_v8hi): Likewise.
2967 2014-01-15 Jeff Law <law@redhat.com>
2969 PR tree-optimization/59747
2970 * ree.c (find_and_remove_re): Properly handle case where a second
2971 eliminated extension requires widening a copy created for elimination
2972 of a prior extension.
2973 (combine_set_extension): Ensure that the number of hard regs needed
2974 for a destination register does not change when we widen it.
2976 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2978 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2979 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2980 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2981 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2982 (avr-*-rtems*): Likewise.
2983 (bfin*-rtems*): Likewise.
2984 (moxie-*-rtems*): Likewise.
2985 (h8300-*-rtems*): Likewise.
2986 (i[34567]86-*-rtems*): Likewise.
2987 (lm32-*-rtems*): Likewise.
2988 (m32r-*-rtems*): Likewise.
2989 (m68k-*-rtems*): Likewise.
2990 (microblaze*-*-rtems*): Likewise.
2991 (mips*-*-rtems*): Likewise.
2992 (powerpc-*-rtems*): Likewise.
2993 (sh-*-rtems*): Likewise.
2994 (sparc-*-rtems*): Likewise.
2995 (sparc64-*-rtems*): Likewise.
2996 (v850-*-rtems*): Likewise.
2997 (m32c-*-rtems*): Likewise.
2999 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3001 PR rtl-optimization/59511
3002 * ira.c (ira_init_register_move_cost): Use memory costs for some
3003 cases of register move cost calculations.
3004 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3005 instead of BB frequency.
3006 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3007 * lra-assigns.c (find_hard_regno_for): Ditto.
3009 2014-01-15 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/59822
3012 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3013 (vectorizable_load): Use it to hoist defs of uses of invariant
3014 loads out of the loop.
3016 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3017 Kugan Vivekanandarajah <kuganv@linaro.org>
3020 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3023 2014-01-15 Richard Biener <rguenther@suse.de>
3025 PR rtl-optimization/59802
3026 * lcm.c (compute_available): Use inverted postorder to seed
3027 the initial worklist.
3029 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3032 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3033 ADDR_REGS for invalid symrefs in non-PIC code.
3035 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3038 * builtins.c (determine_block_size): Initialize *probable_max_size
3039 even if len_rtx is CONST_INT.
3041 2014-01-14 Andrew Pinski <apinski@cavium.com>
3043 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3044 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3045 (cortexa53_tunings): Likewise.
3046 (aarch64_sched_issue_rate): New function.
3047 (TARGET_SCHED_ISSUE_RATE): Define.
3049 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3051 * ira-costs.c (find_costs_and_classes): Add missed
3052 ira_init_register_move_cost_if_necessary.
3054 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3057 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3059 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3062 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3063 to indicate if type is used for function return value. Warn ABI
3064 change if the vector mode isn't available for function return value.
3065 (ix86_function_arg_advance): Pass false to type_natural_mode.
3066 (ix86_function_arg): Likewise.
3067 (ix86_gimplify_va_arg): Likewise.
3068 (function_arg_32): Don't warn ABI change.
3069 (ix86_function_value): Pass true to type_natural_mode.
3070 (ix86_return_in_memory): Likewise.
3071 (ix86_struct_value_rtx): Removed.
3072 (TARGET_STRUCT_VALUE_RTX): Likewise.
3074 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3076 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3077 converting a conditional jump into a conditional return.
3079 2014-01-14 Richard Biener <rguenther@suse.de>
3081 PR tree-optimization/58921
3082 PR tree-optimization/59006
3083 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3084 hoisting invariant stmts.
3085 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3086 invariant loads on the preheader edge if possible.
3088 2014-01-14 Joey Ye <joey.ye@arm.com>
3090 * doc/plugin.texi (Building GCC plugins): Update to C++.
3092 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3094 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3095 (_mm_rcp28_round_ss): Ditto.
3096 (_mm_rsqrt28_round_sd): Ditto.
3097 (_mm_rsqrt28_round_ss): Ditto.
3098 (_mm_rcp28_sd): Ditto.
3099 (_mm_rcp28_ss): Ditto.
3100 (_mm_rsqrt28_sd): Ditto.
3101 (_mm_rsqrt28_ss): Ditto.
3102 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3103 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3104 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3105 (IX86_BUILTIN_RCP28SD): Ditto.
3106 (IX86_BUILTIN_RCP28SS): Ditto.
3107 (IX86_BUILTIN_RSQRT28SD): Ditto.
3108 (IX86_BUILTIN_RSQRT28SS): Ditto.
3109 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3110 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3111 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3112 (ix86_expand_special_args_builtin): Expand new FTYPE.
3113 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3114 (srcp14<mode>): Make insn unary.
3115 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3116 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3117 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3118 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3119 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3120 Fix rounding: make it SAE only.
3121 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3123 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3125 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3126 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3127 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3128 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3129 (round_saeonly_mask_scalar_operand4): Ditto.
3130 (round_saeonly_mask_scalar_op3): Ditto.
3131 (round_saeonly_mask_scalar_op4): Ditto.
3133 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3135 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3136 Implement -maltivec=be for vec_insert and vec_extract.
3138 2014-01-10 DJ Delorie <dj@redhat.com>
3140 * config/msp430/msp430.md (call_internal): Don't allow memory
3141 references with SP as the base register.
3142 (call_value_internal): Likewise.
3143 * config/msp430/constraints.md (Yc): New. For memory references
3144 that don't use SP as a base register.
3146 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3147 "an integer without a # prefix"
3148 * config/msp430/msp430.md (epilogue_helper): Use it.
3150 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3153 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3154 AVX512F gather builtins.
3155 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3156 on gather decls with INTEGER_TYPE masktype.
3157 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3158 directly into the builtin rather than hoisting it before loop.
3160 PR tree-optimization/59387
3161 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3162 (scev_const_prop): If folded_casts and type has undefined overflow,
3163 use force_gimple_operand instead of force_gimple_operand_gsi and
3164 for each added stmt if it is assign with
3165 arith_code_with_undefined_signed_overflow, call
3166 rewrite_to_defined_overflow.
3167 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3168 gimple-fold.h instead.
3169 (arith_code_with_undefined_signed_overflow,
3170 rewrite_to_defined_overflow): Moved to ...
3171 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3172 rewrite_to_defined_overflow): ... here. No longer static.
3173 Include gimplify-me.h.
3174 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3175 rewrite_to_defined_overflow): New prototypes.
3177 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3179 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3181 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3183 * builtins.c (get_object_alignment_2): Minor tweak.
3184 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3186 2014-01-13 Christian Bruel <christian.bruel@st.com>
3188 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3189 optimized non constant lengths.
3191 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3194 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3195 load as __atomic_load_N if possible.
3197 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3199 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3201 (rs6000_expand_builtin): Adjust call.
3203 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3206 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3207 * config/rs6000/rs6000.c: Include target-globals.h.
3208 (rs6000_set_current_function): Instead of doing target_reinit
3209 unconditionally, use save_target_globals_default_opts and
3210 restore_target_globals.
3212 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3214 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3215 (rs6000_expand_builtin): Handle mffs and mtfsf.
3216 (rs6000_init_builtins): Define mffs and mtfsf.
3217 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3218 (rs6000_mffs): New pattern.
3219 (rs6000_mtfsf): New pattern.
3221 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3223 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3224 Start narrowing with START. Apply candidate-use pair
3225 and check overall cost in narrowing.
3226 (iv_ca_prune): Pass new argument.
3228 2014-01-10 Jeff Law <law@redhat.com>
3231 * ree.c (combine_reaching_defs): Ensure the defining statement
3232 occurs before the extension when optimizing extensions with
3233 different source and destination hard registers.
3235 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3238 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3239 vtables into the type inheritance graph.
3241 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3243 PR rtl-optimization/59754
3244 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3245 modes in the REGNO != REGNO case.
3247 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3249 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3251 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3253 PR tree-optimization/59745
3254 * tree-predcom.c (tree_predictive_commoning_loop): Call
3255 free_affine_expand_cache if giving up because components is NULL.
3257 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3258 GC in payload of target_globals struct instead of allocating them on
3259 the heap and the larger structs separately using GC.
3260 * target-globals.h (struct target_globals): Make regs, hard_regs,
3261 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3262 of GTY((skip)) and change type to void *.
3263 (reset_target_globals): Cast loads from those fields to corresponding
3266 2014-01-10 Steve Ellcey <sellcey@mips.com>
3269 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3270 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3271 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3273 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3276 * aarch64-modes.def (CC_Zmode): New flags mode.
3277 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3278 represents an equality.
3279 (aarch64_get_condition_code): Handle CC_Zmode.
3280 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3282 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3284 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3285 extraction in good case.
3287 2014-01-10 Richard Biener <rguenther@suse.de>
3289 PR tree-optimization/59374
3290 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3291 checking after SLP discovery. Mark stmts not participating
3292 in any SLP instance properly.
3294 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3296 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3297 when handling a SET rtx.
3299 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3301 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3302 (cortex-a57): Likewise.
3303 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3305 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3307 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3308 non-iwmmxt builtins.
3310 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3314 * ipa-devirt.c record_target_from_binfo): Take as argument
3315 stack of binfos and lookup matching one for virtual inheritance.
3316 (possible_polymorphic_call_targets_1): Update.
3318 2014-01-10 Huacai Chen <chenhc@lemote.com>
3320 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3321 kernel strings for Loongson-2E/2F/3A.
3323 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3326 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3327 is_gimple_call before calling gimple_call_internal_p.
3329 2014-01-09 Steve Ellcey <sellcey@mips.com>
3331 * Makefile.in (TREE_FLOW_H): Remove.
3332 (TREE_SSA_H): Add file names from tree-flow.h.
3333 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3334 * tree.h: Remove tree-flow.h reference.
3335 * hash-table.h: Remove tree-flow.h reference.
3336 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3337 reference with tree-ssa-loop.h.
3339 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3341 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3342 default element-order behavior for -maltivec.
3343 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3345 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3346 when targeting big endian, at least for now.
3347 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3349 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3352 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3353 var satisfies use_register_for_decl, just take into account type
3354 alignment, rather than decl alignment.
3356 PR tree-optimization/59622
3357 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3358 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3359 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3360 Don't devirtualize for inplace at all. For targets.length () == 1,
3361 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3363 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3365 * config/i386/i386.md (cpu): Remove the unused btver1.
3367 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3369 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3371 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3374 * tree-core.h (struct target_globals): New forward declaration.
3375 (struct tree_target_option): Add globals field.
3376 * tree.h (TREE_TARGET_GLOBALS): Define.
3377 (prepare_target_option_nodes_for_pch): New prototype.
3378 * target-globals.h (struct target_globals): Define even if
3380 * tree.c (prepare_target_option_node_for_pch,
3381 prepare_target_option_nodes_for_pch): New functions.
3382 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3383 * config/i386/i386.c: Include target-globals.h.
3384 (ix86_set_current_function): Instead of doing target_reinit
3385 unconditionally, use save_target_globals_default_opts and
3386 restore_target_globals.
3388 2014-01-09 Richard Biener <rguenther@suse.de>
3390 PR tree-optimization/59715
3391 * tree-cfg.h (split_critical_edges): Declare.
3392 * tree-cfg.c (split_critical_edges): Export.
3393 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3395 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3397 * cfgexpand.c (expand_stack_vars): Optionally disable
3398 asan stack protection.
3399 (expand_used_vars): Likewise.
3400 (partition_stack_vars): Likewise.
3401 * asan.c (asan_emit_stack_protection): Optionally disable
3402 after return stack usage.
3403 (instrument_derefs): Optionally disable memory access instrumentation.
3404 (instrument_builtin_call): Likewise.
3405 (instrument_strlen_call): Likewise.
3406 (asan_protect_global): Optionally disable global variables protection.
3407 * doc/invoke.texi: Added doc for new options.
3408 * params.def: Added new options.
3409 * params.h: Likewise.
3411 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3413 PR rtl-optimization/59724
3414 * ifcvt.c (cond_exec_process_if_block): Don't call
3415 flow_find_head_matching_sequence with 0 longest_match.
3416 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3417 non-active insns if !stop_after.
3418 (try_head_merge_bb): Revert 2014-01-07 changes.
3420 2014-01-08 Jeff Law <law@redhat.com>
3422 * ree.c (get_sub_rtx): New function, extracted from...
3423 (merge_def_and_ext): Here.
3424 (combine_reaching_defs): Use get_sub_rtx.
3426 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3428 * cgraph.h (varpool_variable_node): Do not choke on null node.
3430 2014-01-08 Catherine Moore <clm@codesourcery.com>
3432 * config/mips/mips.md (simple_return): Attempt to use JRC
3434 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3436 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3438 PR rtl-optimization/59137
3439 * reorg.c (steal_delay_list_from_target): Call update_block for
3441 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3443 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3446 two duplicate entries.
3448 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3451 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3453 * config/mips/mips.c (mips_truncated_op_cost): New function.
3454 (mips_rtx_costs): Adjust test for BADDU.
3455 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3457 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3459 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3460 (*baddu_si): ...this new pattern.
3462 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3465 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3467 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3470 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3472 (expand_expr, expand_normal): Adjust.
3473 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3474 inner_reference_p. Use inner_reference_p to expand inner references.
3475 (store_expr): Adjust.
3476 * cfgexpand.c (expand_call_stmt): Adjust.
3478 2014-01-08 Rong Xu <xur@google.com>
3480 * gcov-io.c (gcov_var): Move from gcov-io.h.
3481 (gcov_position): Ditto.
3482 (gcov_is_error): Ditto.
3483 (gcov_rewrite): Ditto.
3484 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3485 only part to libgcc/libgcov.h.
3487 2014-01-08 Marek Polacek <polacek@redhat.com>
3490 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3492 2014-01-08 Marek Polacek <polacek@redhat.com>
3495 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3497 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3499 PR rtl-optimization/59649
3500 * stor-layout.c (get_mode_bounds): For BImode return
3501 0 and STORE_FLAG_VALUE.
3503 2014-01-08 Richard Biener <rguenther@suse.de>
3506 * gimple.h (is_gimple_builtin_call): Remove.
3507 (gimple_builtin_call_types_compatible_p): New.
3508 (gimple_call_builtin_p): New overload.
3509 * gimple.c (is_gimple_builtin_call): Remove.
3510 (validate_call): Rename to ...
3511 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3513 (validate_type): New static function.
3514 (gimple_call_builtin_p): New overload and adjust.
3515 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3516 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3517 (gimple_fold_stmt_to_constant_1): Likewise.
3518 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3520 2014-01-08 Richard Biener <rguenther@suse.de>
3523 * gimplify.c (gimplify_expr): Gimplify register-register type
3524 VIEW_CONVERT_EXPRs to separate stmts.
3526 2014-01-07 Jeff Law <law@redhat.com>
3529 * ree.c (combine_set_extension): Handle case where source
3530 and destination registers in an extension insn are different.
3531 (combine_reaching_defs): Allow source and destination registers
3532 in extension to be different under limited circumstances.
3533 (add_removable_extension): Remove restriction that the
3534 source and destination registers in the extension are the same.
3535 (find_and_remove_re): Emit a copy from the extension's
3536 destination to its source after the defining insn if
3537 the source and destination registers are different.
3540 * ifcvt.c (merge_if_block): If we are merging a block with more than
3541 one successor with a block with no successors, remove any BARRIER
3542 after the second block.
3544 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3546 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3548 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3551 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3552 for 14-bit register offsets when INT14_OK_STRICT is false.
3554 2014-01-07 Roland Stigge <stigge@antcom.de>
3555 Michael Meissner <meissner@linux.vnet.ibm.com>
3558 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3559 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3561 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3563 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3566 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3568 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3569 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3570 rtx is const0_rtx or not.
3572 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3575 * target-globals.c (save_target_globals): Remove this_fn_optab
3577 * toplev.c: Include optabs.h.
3578 (target_reinit): Temporarily restore the global options if another
3579 set of options are in force.
3581 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3583 PR rtl-optimization/58668
3584 * cfgcleanup.c (flow_find_cross_jump): Don't count
3585 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3586 to determine what is counted.
3587 (flow_find_head_matching_sequence): Use active_insn_p to determine
3589 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3591 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3592 determine what is counted.
3594 PR tree-optimization/59643
3595 * tree-predcom.c (split_data_refs_to_components): If one dr is
3596 read and one write, determine_offset fails and the write isn't
3597 in the bad component, just put the read into the bad component.
3599 2014-01-07 Mike Stump <mikestump@comcast.net>
3600 Jakub Jelinek <jakub@redhat.com>
3603 * tree-core.h (struct tree_optimization_option): Change optabs
3604 type from unsigned char * to void *.
3605 * optabs.c (init_tree_optimization_optabs): Adjust
3606 TREE_OPTIMIZATION_OPTABS initialization.
3608 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3611 * config/i386/i386.h (struct machine_function): Add
3612 no_drap_save_restore field.
3613 * config/i386/i386.c (ix86_save_reg): Use
3614 !cfun->machine->no_drap_save_restore instead of
3615 crtl->stack_realign_needed.
3616 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3617 this function clears frame_pointer_needed. Set
3618 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3619 and DRAP reg is needed.
3621 2014-01-06 Marek Polacek <polacek@redhat.com>
3624 * doc/implement-c.texi: Mention that other integer types are
3625 permitted as bit-field types in strictly conforming mode.
3627 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3629 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3632 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3634 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3636 2014-01-06 Martin Jambor <mjambor@suse.cz>
3639 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3641 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3643 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3647 * var-tracking.c (add_stores): Preserve the value of the source even if
3648 we don't record the store.
3650 2014-01-06 Terry Guo <terry.guo@arm.com>
3652 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3654 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3657 * config/darwin.c (darwin_function_section): Adjust return values to
3658 correspond to optimisation changes made in r206070.
3660 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3662 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3663 from prefetch_block tune setting.
3664 (nocona_cost): Correct size of prefetch block to 64.
3666 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3668 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3669 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3670 used to save the static chain register in the computation of the offset
3671 from which the FP registers need to be restored.
3673 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3675 PR tree-optimization/59519
3676 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3677 ICE if get_current_def (current_new_name) is already non-NULL, as long
3678 as it is a phi result of some other phi in *new_exit_bb that has
3681 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3682 or vmovdqu* for misaligned_operand.
3683 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3684 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3685 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3686 aligned_mem for AVX512F masked aligned load and store builtins and for
3689 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3691 PR tree-optimization/59651
3692 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3693 Address range for negative step should be added by TYPE_SIZE_UNIT.
3695 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3697 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3698 overlapping registers also for registers other than the stack pointer.
3700 2014-01-03 Marek Polacek <polacek@redhat.com>
3703 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3706 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3709 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3712 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3713 (push splitter): Use <P:MODE_SIZE> instead of
3714 GET_MODE_SIZE (<P:MODE>mode).
3715 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3716 (mov -1, reg peephole2): Likewise.
3717 * config/i386/sse.md (*mov<mode>_internal,
3718 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3719 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3720 *<code><mode>3, *andnot<mode>3<mask_name>,
3721 <mask_codefor><code><mode>3<mask_name>): Likewise.
3722 * config/i386/subst.md (mask_mode512bit_condition,
3723 sd_mask_mode512bit_condition): Likewise.
3725 2014-01-02 Xinliang David Li <davidxl@google.com>
3727 PR tree-optimization/59303
3728 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3729 (dump_predicates): Better output format.
3730 (pred_equal_p): New function.
3731 (is_neq_relop_p): Ditto.
3732 (is_neq_zero_form_p): Ditto.
3733 (pred_expr_equal_p): Ditto.
3734 (pred_neg_p): Ditto.
3735 (simplify_pred): Ditto.
3736 (simplify_preds_2): Ditto.
3737 (simplify_preds_3): Ditto.
3738 (simplify_preds_4): Ditto.
3739 (simplify_preds): Ditto.
3741 (push_to_worklist): Ditto.
3742 (get_pred_info_from_cmp): Ditto.
3743 (is_degenerated_phi): Ditto.
3744 (normalize_one_pred_1): Ditto.
3745 (normalize_one_pred): Ditto.
3746 (normalize_one_pred_chain): Ditto.
3747 (normalize_preds): Ditto.
3748 (normalize_cond_1): Remove function.
3749 (normalize_cond): Ditto.
3750 (is_gcond_subset_of): Ditto.
3751 (is_subset_of_any): Ditto.
3752 (is_or_set_subset_of): Ditto.
3753 (is_and_set_subset_of): Ditto.
3754 (is_norm_cond_subset_of): Ditto.
3755 (pred_chain_length_cmp): Ditto.
3756 (convert_control_dep_chain_into_preds): Type change.
3757 (find_predicates): Ditto.
3758 (find_def_preds): Ditto.
3759 (destroy_predicates_vecs): Ditto.
3760 (find_matching_predicates_in_rest_chains): Ditto.
3761 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3762 (is_pred_expr_subset): Ditto.
3763 (is_pred_chain_subset_of): Ditto.
3764 (is_included_in): Ditto.
3765 (is_superset_of): Ditto.
3767 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3769 Update copyright years.
3771 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3773 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3774 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3775 config/arc/arc.md, config/arc/arc.opt,
3776 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3777 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3778 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3779 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3780 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3781 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3782 vtable-verify.c, vtable-verify.h: Use the standard form for the
3785 2014-01-02 Tobias Burnus <burnus@net-b.de>
3787 * gcc.c (process_command): Update copyright notice dates.
3788 * gcov-dump.c: Ditto.
3790 * doc/cpp.texi: Bump @copying's copyright year.
3791 * doc/cppinternals.texi: Ditto.
3792 * doc/gcc.texi: Ditto.
3793 * doc/gccint.texi: Ditto.
3794 * doc/gcov.texi: Ditto.
3795 * doc/install.texi: Ditto.
3796 * doc/invoke.texi: Ditto.
3798 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3800 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3802 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3804 * config/i386/sse.md (*mov<mode>_internal): Guard
3805 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3807 PR rtl-optimization/59647
3808 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3809 new_rtx into UNSIGNED_FLOAT rtxes.
3811 Copyright (C) 2014 Free Software Foundation, Inc.
3813 Copying and distribution of this file, with or without modification,
3814 are permitted in any medium without royalty provided the copyright
3815 notice and this notice are preserved.