1 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3 * emit-rtl.c (gen_reg_rtx): Assert that
4 crtl->emit.regno_pointer_align_length is non-zero.
6 2014-02-20 Richard Henderson <rth@redhat.com>
9 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
10 on failure the store back into EXPECT.
12 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
13 Sandra Loosemore <sandra@codesourcery.com>
15 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
16 * config/nios2/nios2.c (nios2_function_profiler):
17 Add -fPIC (flag_pic == 2) support.
18 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
19 (nios2_large_offset_p): New function.
20 (nios2_unspec_reloc_p): Move up position, update to use
22 (nios2_unspec_address): Remove function.
23 (nios2_unspec_offset): New function.
24 (nios2_large_got_address): New function.
25 (nios2_got_address): Add large offset support.
26 (nios2_legitimize_tls_address): Update usage of removed and new
28 (nios2_symbol_binds_local_p): New function.
29 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
30 (nios2_legitimize_address): Update to use nios2_large_offset_p.
31 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
32 (nios2_print_operand): Merge H/L processing, add hiadj/lo
33 processing for (const (unspec ...)).
34 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
36 2014-02-20 Richard Biener <rguenther@suse.de>
38 * tree-cfg.c (replace_uses_by): Mark altered BBs before
39 doing the substitution.
40 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
42 2014-02-20 Martin Jambor <mjambor@suse.cz>
45 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
46 info when checking whether lattices are bottom.
48 2014-02-20 Richard Biener <rguenther@suse.de>
51 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
54 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
57 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
58 parameter specifying the scaling.
59 (inline_call): Update.
60 (want_inline_recursively): Guard division by zero.
61 (recursive_inlining): Update.
62 * ipa-inline.h (clone_inlined_nodes): Update.
64 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
67 * config/i386/i386.c (classify_argument): Pass structures of size
68 64 bytes or less in register.
70 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
71 Kirill Yukhin <kirill.yukhin@intel.com>
73 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
74 (_mm_rcp28_round_ss): Ditto.
75 (_mm_rsqrt28_round_sd): Ditto.
76 (_mm_rsqrt28_round_ss): Ditto.
77 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
78 (_mm_rcp14_round_ss): Ditto.
79 (_mm_rsqrt14_round_sd): Ditto.
80 (_mm_rsqrt14_round_ss): Ditto.
81 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
82 the first input operand, get rid of match_dup.
83 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
85 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
87 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
88 operand as the first input operand, set type attribute.
89 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
91 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
92 operand as the first input operand, set type attribute.
94 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
96 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
99 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
102 * config/i386/i386.c (construct_container): Remove TFmode check
103 for X86_64_INTEGER_CLASS.
105 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
108 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
109 only when -Wpsabi is enabled.
111 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
114 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
115 passing arrays in registers are the same as for structs, so remove the
116 special case for them.
118 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
120 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
121 destination type, extract only the valid bits if the source type is not
122 integral and has a different mode.
124 2014-02-19 Richard Biener <rguenther@suse.de>
127 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
130 2014-02-19 Richard Biener <rguenther@suse.de>
133 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
134 (ipa_modify_call_arguments): Emit an argument load explicitely and
135 preserve virtual SSA form there and for the replacement call.
136 Do not update SSA form nor free dominance info.
138 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
140 * ipa.c (function_and_variable_visibility): Also clear WEAK
141 flag when disolving COMDAT_GROUP.
143 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
145 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
146 * ipa-prop.c (ipa_set_jf_known_type): Return early when
148 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
149 do more sanity checks.
150 (detect_type_change): Return true when giving up early.
151 (compute_complex_assign_jump_func): Fix type parameter of
153 (compute_complex_ancestor_jump_func): Likewise.
154 (update_jump_functions_after_inlining): Fix updating of
156 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
158 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
160 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
161 inline clones when edge disappears.
163 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
166 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
167 Split 64-bit moves into 2 patterns. Do not allow the use of
168 direct move for TDmode in little endian, since the decimal value
169 has little endian bytes within a word, but the 64-bit pieces are
170 ordered in a big endian fashion, and normal subreg's of TDmode are
172 (mov<mode>_64bit_dm): Likewise.
173 (movtd_64bit_nodm): Likewise.
175 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
177 PR tree-optimization/60174
178 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
179 statement of an SSA_NAME that occurs in an abnormal PHI node.
181 2014-02-18 Jakub Jelinek <jakub@redhat.com>
184 * final.c (SEEN_BB): Remove.
185 (SEEN_NOTE, SEEN_EMITTED): Renumber.
186 (final_scan_insn): Don't force_source_line on second
187 NOTE_INSN_BASIC_BLOCK.
189 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
192 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
193 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
194 (type_natural_mode): Warn ABI change when %zmm register is not
195 available for AVX512F vector value passing.
197 2014-02-18 Kai Tietz <ktietz@redhat.com>
200 * config/i386/i386.c (ix86_expand_prologue): Use value in
201 rax register as displacement when restoring %r10 or %rax.
202 Fix wrong offset when restoring both registers.
204 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
206 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
207 assertion with conditional return.
209 2014-02-18 Jakub Jelinek <jakub@redhat.com>
210 Uros Bizjak <ubizjak@gmail.com>
213 * config/i386/driver-i386.c (host_detect_local_cpu): If
214 YMM state is not saved by the OS, also clear has_f16c. Move
215 CPUID 0x80000001 handling before YMM state saving checking.
217 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
219 PR rtl-optimization/58960
220 * haifa-sched.c (alloc_global_sched_pressure_data): New,
221 factored out from ...
222 (sched_init): ... here.
223 (free_global_sched_pressure_data): New, factored out from ...
224 (sched_finish): ... here.
225 * sched-int.h (free_global_sched_pressure_data): Declare.
226 * sched-rgn.c (nr_regions_initial): New static global.
227 (haifa_find_rgns): Initialize it.
228 (schedule_region): Disable sched-pressure for the newly
231 2014-02-17 Richard Biener <rguenther@suse.de>
233 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
234 release SSA defs of pattern stmts.
236 2014-02-17 Richard Biener <rguenther@suse.de>
238 * tree-inline.c (expand_call_inline): Release the virtual
239 operand defined by the call we are about to inline.
241 2014-02-17 Richard Biener <rguenther@suse.de>
243 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
245 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
246 Ilya Tocar <ilya.tocar@intel.com>
248 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
249 arguments order in builtin.
250 (_mm512_permutexvar_epi64): Ditto.
251 (_mm512_mask_permutexvar_epi64): Ditto
252 (_mm512_maskz_permutexvar_epi32): Ditto
253 (_mm512_permutexvar_epi32): Ditto
254 (_mm512_mask_permutexvar_epi32): Ditto
256 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
258 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
259 (p8_vmrgow): Likewise.
261 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
263 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
266 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
269 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
270 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
271 into 64-bit and 32-bit moves. On 64-bit moves, add support for
272 using direct move instructions on ISA 2.07. Also adjust
273 instruction length for 64-bit.
274 (mov<mode>_64bit, TFmode/TDmode): Likewise.
275 (mov<mode>_32bit, TFmode/TDmode): Likewise.
277 2014-02-15 Alan Modra <amodra@gmail.com>
281 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
282 find_replacement on parts of insn rtl that might be reloaded.
284 2014-02-15 Richard Biener <rguenther@suse.de>
286 PR tree-optimization/60183
287 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
288 (tree_ssa_phiprop): Calculate and free post-dominators.
290 2014-02-14 Jeff Law <law@redhat.com>
292 PR rtl-optimization/60131
293 * ree.c (get_extended_src_reg): New function.
294 (combine_reaching_defs): Use it rather than assuming location of REG.
295 (find_and_remove_re): Verify first operand of extension is
296 a REG before adding the insns to the copy list.
298 2014-02-14 Roland McGrath <mcgrathr@google.com>
300 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
301 * configure: Regenerated.
302 * config.in: Regenerated.
303 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
304 instead of ASM_SHORT.
306 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
307 Richard Earnshaw <rearnsha@arm.com>
309 PR rtl-optimization/59535
310 * lra-constraints.c (process_alt_operands): Encourage alternative
311 when unassigned pseudo class is superset of the alternative class.
312 (inherit_reload_reg): Don't inherit when optimizing for code size.
313 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
314 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
315 modes not less than 4 for Thumb1.
317 2014-02-14 Kyle McMartin <kyle@redhat.com>
320 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
322 2014-02-14 Richard Biener <rguenther@suse.de>
324 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
325 (get_frame_arg): Drop the assert with langhook types_compatible_p.
326 Do not strip INDIRECT_REFs.
328 2014-02-14 Richard Biener <rguenther@suse.de>
331 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
332 DECL_FUNCTION_SPECIFIC_TARGET.
333 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
334 * tree-streamer-out.c (pack_ts_target_option): Remove.
335 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
336 (write_ts_function_decl_tree_pointers): Do not stream
337 DECL_FUNCTION_SPECIFIC_TARGET.
338 * tree-streamer-in.c (unpack_ts_target_option): Remove.
339 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
340 (lto_input_ts_function_decl_tree_pointers): Do not stream
341 DECL_FUNCTION_SPECIFIC_TARGET.
343 2014-02-14 Jakub Jelinek <jakub@redhat.com>
345 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
346 (get_initial_def_for_induction, vectorizable_induction): Ignore
347 debug stmts when looking for exit_phi.
348 (vectorizable_live_operation): Fix up condition.
350 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
352 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
353 nreverse() because it changes the content of original tree list.
355 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
357 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
358 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
360 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
362 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
363 GNU coding standards.
365 2014-02-13 Jakub Jelinek <jakub@redhat.com>
368 * dwarf2out.c (gen_subprogram_die): Don't call
369 add_calling_convention_attribute if subr_die is old_die.
371 2014-02-13 Sharad Singhai <singhai@google.com>
373 * doc/optinfo.texi: Fix order of nodes.
375 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
377 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
378 operands[2], not operands[3].
380 2014-02-13 Richard Biener <rguenther@suse.de>
383 * doc/install.texi (ISL): Update recommended version to 0.12.2,
384 mention the possibility of an in-tree build.
385 (CLooG): Update recommended version to 0.18.1, mention the
386 possibility of an in-tree build and clarify that the ISL
387 bundled with CLooG does not work.
389 2014-02-13 Jakub Jelinek <jakub@redhat.com>
392 * expr.c (compress_float_constant): If x is a hard register,
393 extend into a pseudo and then move to x.
395 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
397 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
398 caused by bad second argument to warning_at() with -mhotpatch and
399 nested functions (e.g. with gfortran).
401 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
403 * opts.c (option_name): Remove "enabled by default" rider.
405 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
407 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
409 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
410 Uros Bizjak <ubizjak@gmail.com>
413 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
414 * configure: Regenerated.
416 2014-02-12 Richard Biener <rguenther@suse.de>
418 * vec.c (vec_prefix::calculate_allocation): Move as
419 inline variant to vec.h.
420 (vec_prefix::calculate_allocation_1): New out-of-line version.
421 * vec.h (vec_prefix::calculate_allocation_1): Declare.
422 (vec_prefix::m_has_auto_buf): Rename to ...
423 (vec_prefix::m_using_auto_storage): ... this.
424 (vec_prefix::calculate_allocation): Inline the easy cases
425 and dispatch to calculate_allocation_1 which doesn't need the
427 (va_heap::reserve): Use gcc_checking_assert.
428 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
429 m_using_auto_storage.
430 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
432 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
433 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
434 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
436 2014-02-12 Richard Biener <rguenther@suse.de>
438 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
439 when we found a dependence.
441 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
443 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
445 (maybe_fold_stmt): ... into this new function.
446 * omp-low.c (lower_omp): Update comment.
448 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
451 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
454 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
456 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
457 identifiers in comments.
458 (cortexa53_extra_costs): Likewise.
459 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
460 (cortexa7_extra_costs): Likewise.
461 (cortexa12_extra_costs): Likewise.
462 (cortexa15_extra_costs): Likewise.
463 (v7m_extra_costs): Likewise.
465 2014-02-12 Richard Biener <rguenther@suse.de>
468 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
469 of posix_memalign being successful.
470 (lower_stmt): Restrict lowering of posix_memalign to when
471 -ftree-bit-ccp is enabled.
473 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
475 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
477 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
479 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
481 PR rtl-optimization/60116
482 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
483 other_insn once the combination has been validated.
485 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
488 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
490 * ipa-devirt.c: Include demangle.h
491 (odr_violation_reported): New static variable.
492 (add_type_duplicate): Update odr_violations.
493 (maybe_record_node): Add completep parameter; update it.
494 (record_target_from_binfo): Add COMPLETEP parameter;
496 (possible_polymorphic_call_targets_1): Likewise.
497 (struct polymorphic_call_target_d): Add nonconstruction_targets;
498 rename FINAL to COMPLETE.
499 (record_targets_from_bases): Sanity check we found the binfo;
500 fix COMPLETEP updating.
501 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
502 parameter, fix computing of COMPLETEP.
503 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
504 at LTO time do demangling.
505 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
506 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
508 (gimple_get_virt_method_for_binfo): Likewise.
509 * gimple-fold.h (gimple_get_virt_method_for_binfo,
510 gimple_get_virt_method_for_vtable): Update prototypes.
512 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
515 * genautomata.c (add_presence_absence): Fix typo with
516 {final_}presence_list.
518 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
521 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
522 for VSX/Altivec vectors that land in GPR registers.
524 2014-02-11 Richard Henderson <rth@redhat.com>
525 Jakub Jelinek <jakub@redhat.com>
528 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
529 around drhs if type conversion to lacc->type is not useless.
531 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
533 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
535 (cortex-a57.cortex-a53): Likewise.
536 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
538 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
540 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
543 2014-02-11 Renlin Li <Renlin.Li@arm.com>
545 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
546 add_options_for_arm_vfp3.
548 2014-02-11 Jeff Law <law@redhat.com>
551 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
552 object with an undesirable mode.
554 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
557 * config/i386/sol2-9.h: New file.
558 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
559 *-*-solaris2.9*): Use it.
561 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
563 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
564 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
566 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
568 * config/microblaze/microblaze.c: Extend mcpu version format
570 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
572 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
574 2014-02-10 Richard Henderson <rth@redhat.com>
577 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
578 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
579 ms-abi vs -mno-accumulate-outgoing-args.
580 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
581 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
584 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
587 * cfgexpand.c (expand_asm_operands): Attach source location to
588 ASM_INPUT rtx objects.
589 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
591 2014-02-10 Nick Clifton <nickc@redhat.com>
593 * config/mn10300/mn10300.c (popcount): New function.
594 (mn10300_expand_prologue): Include saved registers in stack usage
597 2014-02-10 Jeff Law <law@redhat.com>
600 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
601 when changing the SET_DEST of a prior insn to avoid an input reload.
603 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
605 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
606 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
607 -mcall-openbsd, or -mcall-linux.
608 (CC1_ENDIAN_BIG_SPEC): Remove.
609 (CC1_ENDIAN_LITTLE_SPEC): Remove.
610 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
611 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
612 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
613 and %cc1_endian_default.
614 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
616 2014-02-10 Richard Biener <rguenther@suse.de>
618 PR tree-optimization/60115
619 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
620 MEM_REF handling. Properly verify that the accesses are not
621 out of the objects bound.
623 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
625 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
628 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
630 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
631 proper constants and fix formatting.
632 (possible_polymorphic_call_targets): Fix formatting.
634 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
635 Ilya Tocar <ilya.tocar@intel.com>
637 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
638 (_mm512_loadu_epi32): Renamed into...
639 (_mm512_loadu_si512): This.
640 (_mm512_storeu_epi32): Renamed into...
641 (_mm512_storeu_si512): This.
642 (_mm512_maskz_ceil_ps): Removed.
643 (_mm512_maskz_ceil_pd): Ditto.
644 (_mm512_maskz_floor_ps): Ditto.
645 (_mm512_maskz_floor_pd): Ditto.
646 (_mm512_floor_round_ps): Ditto.
647 (_mm512_floor_round_pd): Ditto.
648 (_mm512_ceil_round_ps): Ditto.
649 (_mm512_ceil_round_pd): Ditto.
650 (_mm512_mask_floor_round_ps): Ditto.
651 (_mm512_mask_floor_round_pd): Ditto.
652 (_mm512_mask_ceil_round_ps): Ditto.
653 (_mm512_mask_ceil_round_pd): Ditto.
654 (_mm512_maskz_floor_round_ps): Ditto.
655 (_mm512_maskz_floor_round_pd): Ditto.
656 (_mm512_maskz_ceil_round_ps): Ditto.
657 (_mm512_maskz_ceil_round_pd): Ditto.
658 (_mm512_expand_pd): Ditto.
659 (_mm512_expand_ps): Ditto.
660 * config/i386/i386.c (ix86_builtins): Remove
661 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
663 * config/i386/predicates.md (const1256_operand): New.
664 (const_1_to_2_operand): Ditto.
665 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
666 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
667 (*avx512pf_gatherpf<mode>sf): Ditto.
668 (avx512pf_gatherpf<mode>df): Ditto.
669 (*avx512pf_gatherpf<mode>df_mask): Ditto.
670 (*avx512pf_gatherpf<mode>df): Ditto.
671 (avx512pf_scatterpf<mode>sf): Ditto.
672 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
673 (*avx512pf_scatterpf<mode>sf): Ditto.
674 (avx512pf_scatterpf<mode>df): Ditto.
675 (*avx512pf_scatterpf<mode>df_mask): Ditto.
676 (*avx512pf_scatterpf<mode>df): Ditto.
677 (avx512f_expand<mode>): Removed.
678 (<shift_insn><mode>3<mask_name>): Change predicate type.
680 2014-02-08 Jakub Jelinek <jakub@redhat.com>
682 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
683 not at the end of datarefs vector use ordered_remove to avoid
684 reordering datarefs vector.
687 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
688 mark local addressable non-static vars as GOVD_PRIVATE
689 instead of GOVD_LOCAL.
690 * omp-low.c (lower_omp_for): Move gimple_bind_vars
691 and BLOCK_VARS of gimple_bind_block to new_stmt rather
695 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
696 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
697 assume_aligned or alloc_align attributes.
698 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
699 arguments. Handle also assume_aligned and alloc_align attributes.
700 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
701 calls to functions with assume_aligned or alloc_align attributes.
702 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
704 2014-02-08 Terry Guo <terry.guo@arm.com>
706 * doc/invoke.texi: Document ARM -march=armv7e-m.
708 2014-02-08 Jakub Jelinek <jakub@redhat.com>
710 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
711 flag on __cilkrts_rethrow builtin.
714 * ipa-cp.c (determine_versionability): Fail at -O0
715 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
716 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
719 2014-02-04 Jakub Jelinek <jakub@redhat.com>
722 * tree-inline.c (copy_forbidden): Fail for
723 __attribute__((optimize (0))) functions.
725 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
727 * varpool.c: Include pointer-set.h.
728 (varpool_remove_unreferenced_decls): Variables in other partitions
729 will not be output; be however careful to not lose information
732 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
734 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
735 lookup in the vtable constructor.
737 2014-02-07 Jeff Law <law@redhat.com>
740 * config/m68k/m68k.md (ashldi_extsi): Turn into a
741 define_insn_and_split.
743 * ipa-inline.c (inline_small_functions): Fix typos.
745 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
747 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
748 (s390_can_use_return_insn): Declare.
749 * config/s390/s390.h (EPILOGUE_USES): Define.
750 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
752 (s390_chunkify_start): Handle return JUMP_LABELs.
753 (s390_early_mach): Emit a main_pool instruction on the entry edge.
754 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
755 (s390_can_use_return_insn): New functions.
756 (s390_fix_long_loop_prediction): Handle conditional returns.
757 (TARGET_SET_UP_BY_PROLOGUE): Define.
758 * config/s390/s390.md (ANY_RETURN): New code iterator.
759 (*creturn, *csimple_return, return, simple_return): New patterns.
761 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
763 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
764 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
765 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
766 REG_CFA_RESTORE list when deciding not to restore a register.
768 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
770 * config/s390/s390.c: Include tree-pass.h and context.h.
771 (s390_early_mach): New function, split out from...
772 (s390_emit_prologue): ...here.
773 (pass_data_s390_early_mach): New pass structure.
774 (pass_s390_early_mach): New class.
775 (s390_option_override): Create and register early_mach pass.
778 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
780 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
781 to match for the exit block.
783 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
785 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
786 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
787 Reject misaligned operands.
789 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
791 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
793 2014-02-07 Richard Biener <rguenther@suse.de>
796 * gimple-low.c (lower_builtin_posix_memalign): New function.
797 (lower_stmt): Call it to lower posix_memalign in a way
798 to make alignment info accessible.
800 2014-02-07 Jakub Jelinek <jakub@redhat.com>
803 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
804 __builtin_setjmp_receiver.
806 2014-02-07 Richard Biener <rguenther@suse.de>
809 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
810 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
811 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
812 Handle BUILT_IN_POSIX_MEMALIGN.
813 (find_func_clobbers): Likewise.
814 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
815 (call_may_clobber_ref_p_1): Likewise.
817 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
820 * ipa-devirt.c (record_target_from_binfo): Remove overactive
823 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
826 * lto-cgraph.c (lto_output_node): Use
827 symtab_get_symbol_partitioning_class.
828 (lto_output_varpool_node): likewise.
829 (symtab_get_symbol_partitioning_class): Move here from
831 * cgraph.h (symbol_partitioning_class): Likewise.
832 (symtab_get_symbol_partitioning_class): Declare.
834 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
836 * ggc.h (ggc_internal_cleared_alloc): New macro.
837 * vec.h (vec_safe_copy): Handle memory stats.
838 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
839 * target-globals.c (save_target_globals): Likewise.
841 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
844 * expr.c (emit_move_resolve_push): Export; be bit more selective
845 on when to clear alias set.
846 * expr.h (emit_move_resolve_push): Declare.
847 * function.h (struct function): Add tail_call_marked.
848 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
849 * config/i386/i386-protos.h (ix86_expand_push): Remove.
850 * config/i386/i386.md (TImode move expander): De not call
852 (FP push expanders): Preserve memory attributes.
853 * config/i386/sse.md (push<mode>1): Remove.
854 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
855 (ix86_expand_push): Remove.
856 * config/i386/mmx.md (push<mode>1): Remove.
858 2014-02-06 Jakub Jelinek <jakub@redhat.com>
860 PR rtl-optimization/60030
861 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
862 lopart with paradoxical subreg before shifting it up by hprec.
864 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
866 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
867 Remove extra newline at end of file.
868 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
869 (arm_issue_rate): Handle cortexa57.
870 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
871 (cortex-a57.cortex-a53): Likewise.
873 2014-02-06 Jakub Jelinek <jakub@redhat.com>
876 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
877 don't record in REG_FRAME_RELATED_EXPR registers not set in that
879 (arm_expand_prologue): Adjust all callers.
880 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
881 info, registers also at the lowest numbered registers side. Use
882 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
886 * var-tracking.c (adjust_mems): Before adding a SET to
887 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
889 2014-02-06 Alan Modra <amodra@gmail.com>
892 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
893 change SDmode to DDmode when lra_in_progress.
895 2014-02-06 Jakub Jelinek <jakub@redhat.com>
898 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
899 free_data_ref on the dr first, and before goto again also set dr
900 to the next dr. For simd_lane_access, free old datarefs[i] before
901 overwriting it. For get_vectype_for_scalar_type failure, don't
902 free_data_ref if simd_lane_access.
904 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
907 * tree.h (opts_for_fn): New inline function.
908 (opt_for_fn): Define.
909 * config/i386/i386.c (ix86_function_regparm): Use
910 opt_for_fn (decl, optimize) instead of optimize.
912 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
914 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
915 for SYMBOL_REF in large memory model.
917 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
919 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
921 (cortex-a57): Likewise.
922 (cortex-a57.cortex-a53): Likewise.
924 2014-02-06 Yury Gribov <y.gribov@samsung.com>
925 Kugan Vivekanandarajah <kuganv@linaro.org>
927 * config/arm/arm.c (arm_vector_alignment_reachable): Check
929 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
931 2014-02-06 Richard Biener <rguenther@suse.de>
933 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
934 set_loop_copy and initialize_original_copy_tables.
936 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
938 * config/aarch64/aarch64-simd.md
939 (aarch64_ashr_simddi): Change QI to SI.
941 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
942 Jakub Jelinek <jakub@redhat.com>
945 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
948 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
950 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
951 CODE_FOR_altivec_vpku[hw]um to
952 CODE_FOR_altivec_vpku[hw]um_direct.
953 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
954 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
955 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
956 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
958 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
960 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
961 generation for -maltivec=be.
962 (altivec_vsumsws): Simplify redundant test.
964 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
966 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
967 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
968 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
969 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
971 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
973 (altivec_vpks<VI_char>ss): Likewise.
974 (altivec_vpks<VI_char>us): Likewise.
975 (altivec_vpku<VI_char>us): Likewise.
976 (altivec_vpku<VI_char>um): Likewise.
977 (altivec_vpku<VI_char>um_direct): New (copy of
978 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
980 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
981 target is little endian and -maltivec=be is not specified.
982 (*altivec_vupkhs<VU_char>_direct): New (copy of
983 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
984 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
985 target is little endian and -maltivec=be is not specified.
986 (*altivec_vupkls<VU_char>_direct): New (copy of
987 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
988 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
989 little endian and -maltivec=be is not specified.
990 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
991 little endian and -maltivec=be is not specified.
993 2014-02-05 Richard Henderson <rth@redhat.com>
996 * combine-stack-adj.c: Revert r206943.
997 * sched-int.h (struct deps_desc): Add last_args_size.
998 * sched-deps.c (init_deps): Initialize it.
999 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1000 contain REG_ARGS_SIZE notes.
1002 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1004 * lto-cgraph.c (asm_nodes_output): Make global.
1005 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1006 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1007 (driver_handle_option): Handle OPT_fwpa.
1009 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1012 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1013 a comment typo and formatting issue. If odr_hash hasn't been
1014 created, return vNULL and set *completep to false.
1017 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1018 bb with no successors.
1020 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1023 * doc/invoke.texi (-march): Clarify documentation for ARM.
1027 2014-02-05 Richard Biener <rguenther@suse.de>
1029 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1030 when not vectorizing because of too many alias checks.
1031 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1032 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1034 2014-02-05 Nick Clifton <nickc@redhat.com>
1036 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1037 accept extended registers in any mode when compiling for the MN10300.
1039 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1041 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1042 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1043 sanitization attributes.
1044 (can_inline_edge_p): Likewise.
1045 (sanitize_attrs_match_for_inline_p): New function.
1047 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1049 * ipa-prop.c (detect_type_change): Shor circuit testing of
1050 type changes on THIS pointer.
1052 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1055 * config/pa/pa.c (legitimize_tls_address): Return original address
1056 if not passed a SYMBOL_REF rtx.
1057 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1059 (pa_emit_move_sequence): Simplify TLS source operands.
1060 (pa_legitimate_constant_p): Reject all TLS constants.
1061 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1062 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1064 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1066 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1067 groups when we know they are controlled by LTO.
1068 * varasm.c (default_binds_local_p_1): If object is in other partition,
1069 it will be resolved locally.
1071 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1073 * config/host-linux.c (linux_gt_pch_use_address): Don't
1074 use SSIZE_MAX because it is not always defined.
1076 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1079 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1080 threshold for pseudo splitting.
1081 (update_ebb_live_info): Process call argument hard registers and
1082 hard registers from insn definition too.
1083 (max_small_class_regs_num): New constant.
1084 (inherit_in_ebb): Update live hard regs through EBBs. Update
1085 reloads_num only for small register classes. Don't split for
1088 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1091 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1094 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1096 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1099 2014-02-04 Marek Polacek <polacek@redhat.com>
1101 * gdbinit.in (pel): Define.
1103 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1105 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1108 2014-02-04 Richard Biener <rguenther@suse.de>
1111 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1112 in function context local.
1113 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1114 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1115 similar to LTO_imported_decl_ref.
1117 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1119 PR tree-optimization/60002
1120 * cgraphclones.c (build_function_decl_skip_args): Clear
1123 PR tree-optimization/60023
1124 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1125 false to gsi_replace.
1126 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1127 has been in some EH region and vec_stmt could throw, add
1128 vec_stmt into the same EH region.
1129 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1130 has no lhs, ignore it.
1131 * internal-fn.c (expand_MASK_LOAD): Likewise.
1134 * tree-inline.c (copy_forbidden): Fail for
1135 __attribute__((optimize (0))) functions.
1138 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1139 is set, copy one less argument.
1140 (expand_simd_clones): Don't subtract clone_info->inbranch
1141 from simd_clone_struct_alloc argument.
1143 PR rtl-optimization/57915
1144 * recog.c (simplify_while_replacing): If all unary/binary/relational
1145 operation arguments are constant, attempt to simplify those.
1148 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1149 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1151 2014-02-04 Richard Biener <rguenther@suse.de>
1153 PR tree-optimization/60012
1154 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1155 TBAA disambiguation to all DDRs.
1157 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1160 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1161 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1163 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1166 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1168 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1170 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1171 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1173 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1176 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1177 to figure out targets of polymorphic calls with known decl.
1178 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1179 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1180 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1181 (get_polymorphic_call_info): ... here.
1182 (get_polymorphic_call_info_from_invariant): New function.
1184 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1186 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1187 lookup via vtable pointer; check for type consistency
1188 and turn inconsitent facts into UNREACHABLE.
1189 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1190 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1191 type inconsistent querries; return UNREACHABLE instead.
1193 2014-02-03 Richard Henderson <rth@twiddle.net>
1196 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1197 already processed this node.
1198 (normalize_one_pred_1): Pass along mark_set.
1199 (normalize_one_pred): Create and destroy a pointer_set_t.
1200 (normalize_one_pred_chain): Likewise.
1202 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1204 PR gcov-profile/58602
1205 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1207 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1210 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1211 -fno-devirtualize; try to devirtualize by the knowledge of
1212 virtual table pointer given by aggregate propagation.
1213 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1214 (ipa_print_node_jump_functions): Dump also offset that
1215 is relevant for polymorphic calls.
1216 (determine_known_aggregate_parts): Add arg_type parameter; use it
1217 instead of determining the type from pointer type.
1218 (ipa_compute_jump_functions_for_edge): Update call of
1219 determine_known_aggregate_parts.
1220 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1221 (gimple_get_virt_method_for_binfo): ... here; simplify using
1222 vtable_pointer_value_to_vtable.
1223 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1224 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1225 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1226 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1228 (vtable_pointer_value_to_binfo): ... here.
1229 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1231 2014-02-03 Teresa Johnson <tejohnson@google.com>
1233 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1234 redef of outer loop index variable.
1236 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1240 * doc/extend.texi (Function Attributes): Typo.
1242 2014-02-03 Cong Hou <congh@google.com>
1244 PR tree-optimization/60000
1245 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1246 if the vectorized statement is a store. A store statement can only
1247 appear at the end of pattern statements.
1249 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1251 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1252 (ix86_option_override_internal): Default long double to 64-bit for
1253 32-bit Bionic and to 128-bit for 64-bit Bionic.
1255 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1256 TARGET_LONG_DOUBLE_128 is true.
1257 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1259 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1260 (mlong-double-64): Negate -mlong-double-128.
1261 (mlong-double-128): New option.
1263 * config/i386/i386-c.c (ix86_target_macros): Define
1264 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1266 * doc/invoke.texi: Document -mlong-double-128.
1268 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1270 PR rtl-optimization/60024
1271 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1273 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1275 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1277 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1279 PR rtl-optimization/57662
1280 * sel-sched.c (code_motion_path_driver): Do not mark already not
1281 existing blocks in the visiting bitmap.
1283 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1285 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1286 on the insn being emitted.
1288 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1289 Will Deacon <will.deacon@arm.com>
1291 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1293 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1295 * config/arm/arm-tables.opt: Regenerate.
1297 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1299 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1300 for vector types other than V16QImode.
1301 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1302 define_expand, and call altivec_expand_vec_perm_le when producing
1303 code with little endian element order.
1304 (*altivec_vperm_<mode>_internal): New insn having previous
1305 behavior of altivec_vperm_<mode>.
1306 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1307 altivec_expand_vec_perm_le when producing code with little endian
1309 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1310 behavior of altivec_vperm_<mode>_uns.
1312 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1314 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1315 (altivec_vsumsws): Add handling for -maltivec=be with a little
1317 (altivec_vsumsws_direct): New.
1318 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1319 gen_altivec_vsumsws.
1321 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1323 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1324 vtable_pointer_value_to_binfo): New functions.
1325 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1326 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1328 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1330 * config/nios2/nios2.md (load_got_register): Initialize GOT
1331 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1332 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1334 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1336 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1337 preserverd by passthrough, do not propagate the type.
1339 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1341 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1342 (mips_atomic_assign_expand_fenv): New function.
1343 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1345 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1347 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1348 (__builtin_mips_set_fcsr): Likewise.
1349 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1350 MIPS_USI_FTYPE_VOID.
1351 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1352 (mips16_expand_set_fcsr): Likewise.
1353 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1354 (mips16_set_fcsr_stub): Likewise.
1355 (mips16_get_fcsr_one_only_stub): New class.
1356 (mips16_set_fcsr_one_only_stub): Likewise.
1357 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1358 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1359 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1360 (hard_float): New availability predicate.
1361 (mips_builtins): Add get_fcsr and set_fcsr.
1362 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1363 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1364 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1365 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1366 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1369 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1371 * config/mips/mips.c (mips_one_only_stub): New class.
1372 (mips_need_mips16_rdhwr_p): Replace with...
1373 (mips16_rdhwr_stub): ...this new variable.
1374 (mips16_stub_call_address): New function.
1375 (mips16_rdhwr_one_only_stub): New class.
1376 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1377 (mips_output_mips16_rdhwr): Delete.
1378 (mips_finish_stub): New function.
1379 (mips_code_end): Use it to handle rdhwr stubs.
1381 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1384 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1385 when calculating size of integer atomic types.
1387 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1389 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1391 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1393 PR tree-optimization/60003
1394 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1395 * profile.c (branch_prob): Use gimple_call_builtin_p
1396 to check for BUILT_IN_SETJMP_RECEIVER.
1397 * tree-inline.c (copy_bb): Call notice_special_calls.
1399 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1402 * lra-constraints.c (process_alt_operands): Update reload_sum only
1405 2014-01-31 Richard Henderson <rth@redhat.com>
1408 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1409 until after else_eh is processed.
1411 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1413 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1414 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1415 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1416 in smmintrin.h, remove them.
1417 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1418 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1419 * config/i386/i386.md (ROUND_SAE): Fix value.
1420 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1421 (const48_operand): New.
1422 * config/i386/subst.md (round), (round_expand): Use
1423 const_4_or_8_to_11_operand.
1424 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1426 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1428 * config/i386/constraints.md (Yk): Swap meaning with k.
1429 * config/i386/i386.md (movhi_internal): Change Yk to k.
1430 (movqi_internal): Ditto.
1431 (*k<logic><mode>): Ditto.
1434 (kandn<mode>): Ditto.
1435 (*<code>hi_1): Ditto.
1436 (*<code>qi_1): Ditto.
1437 (kxnor<mode>): Ditto.
1438 (kortestzhi): Ditto.
1439 (kortestchi): Ditto.
1441 (*one_cmplhi2_1): Ditto.
1442 (*one_cmplqi2_1): Ditto.
1443 * config/i386/sse.md (): Change k to Yk.
1444 (avx512f_load<mode>_mask): Ditto.
1445 (avx512f_blendm<mode>): Ditto.
1446 (avx512f_store<mode>_mask): Ditto.
1447 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1448 (avx512f_storedqu<mode>_mask): Ditto.
1449 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1451 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1452 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1453 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1454 (avx512f_maskcmp<mode>3): Ditto.
1455 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1456 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1457 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1458 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1459 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1460 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1461 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1462 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1463 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1464 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1465 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1466 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1467 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1468 (vec_extract_lo_<mode>_maskm): Ditto.
1469 (vec_extract_hi_<mode>_maskm): Ditto.
1470 (avx512f_vternlog<mode>_mask): Ditto.
1471 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1472 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1473 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1474 (avx512f_<code>v8div16qi2_mask): Ditto.
1475 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1476 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1477 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1478 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1479 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1480 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1481 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1482 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1483 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1484 (avx512cd_maskb_vec_dupv8di): Ditto.
1485 (avx512cd_maskw_vec_dupv16si): Ditto.
1486 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1487 (avx512f_vpermi2var<mode>3_mask): Ditto.
1488 (avx512f_vpermi2var<mode>3_mask): Ditto.
1489 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1490 (*avx512f_gathersi<mode>): Ditto.
1491 (*avx512f_gathersi<mode>_2): Ditto.
1492 (*avx512f_gatherdi<mode>): Ditto.
1493 (*avx512f_gatherdi<mode>_2): Ditto.
1494 (*avx512f_scattersi<mode>): Ditto.
1495 (*avx512f_scatterdi<mode>): Ditto.
1496 (avx512f_compress<mode>_mask): Ditto.
1497 (avx512f_compressstore<mode>_mask): Ditto.
1498 (avx512f_expand<mode>_mask): Ditto.
1499 * config/i386/subst.md (mask): Change k to Yk.
1500 (mask_scalar_merge): Ditto.
1503 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1505 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1507 2014-01-31 Richard Biener <rguenther@suse.de>
1510 * builtins.c (fold_builtin_memory_op): Make sure to not
1511 use a floating-point mode or a boolean or enumeral type for
1514 2014-01-30 DJ Delorie <dj@redhat.com>
1516 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1517 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1518 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1519 whenever main() has an epilogue.
1521 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1523 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1524 unused variable "field".
1525 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1526 (vsx_mergeh_<mode>): Likewise.
1527 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1528 (altivec_vmrghh): Likewise.
1529 (altivec_vmrghw): Likewise.
1530 (altivec_vmrglb): Likewise.
1531 (altivec_vmrglh): Likewise.
1532 (altivec_vmrglw): Likewise.
1533 (altivec_vspltb): Add missing uses.
1534 (altivec_vsplth): Likewise.
1535 (altivec_vspltw): Likewise.
1536 (altivec_vspltsf): Likewise.
1538 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1541 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1542 frame related instructions.
1544 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1546 PR rtl-optimization/59959
1547 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1548 any reload of register whose subreg is invalid.
1550 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1552 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1553 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1554 Add missing return type - void.
1556 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1558 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1559 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1560 remove element index adjustment for endian (now handled in vsx.md
1562 (altivec_expand_vec_perm_const): Use
1563 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1564 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1565 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1566 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1567 define_expand and a new define_insn *altivec_vspltb_internal;
1568 adjust for -maltivec=be on a little endian target.
1569 (altivec_vspltb_direct): New.
1570 (altivec_vsplth): Divide into a define_expand and a new
1571 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1572 little endian target.
1573 (altivec_vsplth_direct): New.
1574 (altivec_vspltw): Divide into a define_expand and a new
1575 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1576 little endian target.
1577 (altivec_vspltw_direct): New.
1578 (altivec_vspltsf): Divide into a define_expand and a new
1579 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1580 a little endian target.
1582 2014-01-30 Richard Biener <rguenther@suse.de>
1584 PR tree-optimization/59993
1585 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1586 can propagate form the earlier stmt and avoid the transform
1587 when the intermediate result is needed.
1589 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1591 * README.Portability: Fix typo.
1593 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1595 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1596 comparison_operator with ordered_comparison_operator.
1598 2014-01-30 Nick Clifton <nickc@redhat.com>
1600 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1601 Rename to mn10300_store_multiple_regs.
1602 * config/mn10300/mn10300.c: Likewise.
1603 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1604 store_multiple_regs.
1605 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1606 Call mn10300_store_multiple_regs.
1608 2014-01-30 Nick Clifton <nickc@redhat.com>
1609 DJ Delorie <dj@redhat.com>
1611 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1612 %fp 2 to keep registers after it properly word-aligned.
1613 (rl78_alloc_physical_registers_umul): Handle the case where both
1614 input operands are the same.
1616 2014-01-30 Richard Biener <rguenther@suse.de>
1618 PR tree-optimization/59903
1619 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1622 2014-01-30 Jason Merrill <jason@redhat.com>
1625 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1628 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1630 2014-01-30 Richard Biener <rguenther@suse.de>
1632 PR tree-optimization/59951
1633 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1635 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1638 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1639 SFmode to DFmode case.
1641 2014-01-29 DJ Delorie <dj@redhat.com>
1643 * config/msp430/msp430.opt (-minrt): New.
1644 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1646 (ENDFILE_SPEC): Likewise.
1648 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1650 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1651 (estimate_function_body_sizes): Use it.
1653 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1656 * dwarf2out.c (is_cxx_auto): New.
1657 (is_base_type): Use it.
1658 (gen_type_die_with_usage): Likewise.
1660 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1662 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1663 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1664 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1665 -maltivec=be with LE targets.
1666 (vsx_mergeh_<mode>): Likewise.
1667 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1668 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1669 (altivec_vmrghb): Replace with define_expand and new
1670 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1671 (altivec_vmrghb_direct): New define_insn.
1672 (altivec_vmrghh): Replace with define_expand and new
1673 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1674 (altivec_vmrghh_direct): New define_insn.
1675 (altivec_vmrghw): Replace with define_expand and new
1676 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1677 (altivec_vmrghw_direct): New define_insn.
1678 (*altivec_vmrghsf): Adjust for endianness.
1679 (altivec_vmrglb): Replace with define_expand and new
1680 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1681 (altivec_vmrglb_direct): New define_insn.
1682 (altivec_vmrglh): Replace with define_expand and new
1683 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1684 (altivec_vmrglh_direct): New define_insn.
1685 (altivec_vmrglw): Replace with define_expand and new
1686 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1687 (altivec_vmrglw_direct): New define_insn.
1688 (*altivec_vmrglsf): Adjust for endianness.
1689 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1690 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1691 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1692 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1693 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1694 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1695 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1696 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1698 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1700 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1701 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1704 2014-01-29 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/58742
1707 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1708 associate_pointerplus_align.
1709 (associate_pointerplus_diff): New function.
1710 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1711 and associate_pointerplus_diff.
1713 2014-01-29 Richard Biener <rguenther@suse.de>
1715 * lto-streamer.h (LTO_major_version): Bump to 3.
1716 (LTO_minor_version): Reset to 0.
1718 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1720 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1721 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1722 (arm_file_start): Generate correct asm header for armv7ve.
1723 * config/arm/bpabi.h: Add multilib support for armv7ve.
1724 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1725 and cortex-a15 to armv7ve.
1726 * config/arm/t-aprofile: Add multilib support for armv7ve.
1727 * doc/invoke.texi: Document -march=armv7ve.
1729 2014-01-29 Richard Biener <rguenther@suse.de>
1731 PR tree-optimization/58742
1732 * tree-ssa-forwprop.c (associate_plusminus): Return true
1733 if we changed sth, defer EH cleanup to ...
1734 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1735 (simplify_mult): New function.
1737 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1740 PR tree-optimization/59920
1741 * tree.c (build_common_builtin_nodes): Remove
1742 __builtin_setjmp_dispatcher initialization.
1743 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1744 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1745 instead of gsi_after_labels + manually skipping debug stmts.
1746 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1747 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1748 * tree-inline.c (copy_edges_for_bb): Remove
1749 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1750 argument. Ignore computed_goto_p stmts. Don't call
1751 make_abnormal_goto_edges. If a call might need abnormal edges
1752 for non-local gotos, see if it already has an edge to
1753 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1754 with true argument, don't do anything then, otherwise add
1755 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1756 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1758 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1759 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1760 (lower_stmt): Don't set data->calls_builtin_setjmp.
1761 (lower_builtin_setjmp): Adjust comment.
1762 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1763 * tree-cfg.c (found_computed_goto): Remove.
1764 (factor_computed_gotos): Remove.
1765 (make_goto_expr_edges): Return bool, true for computed gotos.
1766 Don't call make_abnormal_goto_edges.
1767 (build_gimple_cfg): Don't set found_computed_goto, don't call
1768 factor_computed_gotos.
1769 (computed_goto_p): No longer static.
1770 (make_blocks): Don't set found_computed_goto.
1771 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1772 (make_edges): If make_goto_expr_edges returns true, push bb
1773 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1774 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1775 vector. Record mapping between bbs and OpenMP regions if there
1776 are any, adjust make_gimple_omp_edges caller. Call
1777 handle_abnormal_edges.
1778 (make_abnormal_goto_edges): Remove.
1779 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1780 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1781 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1782 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1783 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1784 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1785 filling *region also set *region_idx to (*region)->entry->index.
1788 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1789 For REGs set ORIGINAL_REGNO.
1791 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1793 * doc/md.texi: Mention that a target shouldn't implement
1794 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1797 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1799 PR tree-optimization/59594
1800 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1801 a copy of the datarefs vector rather than the vector itself.
1803 2014-01-28 Jason Merrill <jason@redhat.com>
1806 * dwarf2out.c (auto_die): New static.
1807 (gen_type_die_with_usage): Handle C++1y 'auto'.
1808 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1811 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1814 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1815 (SPEC_X32): Likewise.
1816 (SPEC_64): Likewise.
1817 * config/i386/i386.c (ix86_option_override_internal): Turn off
1818 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1820 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1821 * config/i386/i386.h (TARGET_16BIT): New macro.
1822 (TARGET_16BIT_P): Likewise.
1823 * config/i386/i386.opt: Add m16.
1824 * doc/invoke.texi: Document -m16.
1826 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1828 PR preprocessor/59935
1829 * input.c (location_get_source_line): Bail out on when line number
1830 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1832 2014-01-28 Richard Biener <rguenther@suse.de>
1834 PR tree-optimization/58742
1835 * tree-ssa-forwprop.c (associate_plusminus): Handle
1836 pointer subtraction of the form (T)(P + A) - (T)P.
1838 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1840 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1843 2014-01-28 Richard Biener <rguenther@suse.de>
1846 2014-01-28 Richard Biener <rguenther@suse.de>
1848 PR rtl-optimization/45364
1849 PR rtl-optimization/59890
1850 * var-tracking.c (local_get_addr_clear_given_value): Handle
1851 already cleared slot.
1852 (val_reset): Handle not allocated local_get_addr_cache.
1853 (vt_find_locations): Use post-order on the inverted CFG.
1855 2014-01-28 Richard Biener <rguenther@suse.de>
1857 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1859 2014-01-28 Richard Biener <rguenther@suse.de>
1861 PR rtl-optimization/45364
1862 PR rtl-optimization/59890
1863 * var-tracking.c (local_get_addr_clear_given_value): Handle
1864 already cleared slot.
1865 (val_reset): Handle not allocated local_get_addr_cache.
1866 (vt_find_locations): Use post-order on the inverted CFG.
1868 2014-01-28 Alan Modra <amodra@gmail.com>
1870 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1871 * configure.ac <recursive call for build != host>: Define
1872 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1873 and LD_FOR_BUILD too.
1874 * configure: Regenerate.
1876 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1878 * config/i386/i386.c (get_builtin_code_for_version): Separate
1879 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1880 Broadwell from Haswell.
1882 2014-01-27 Steve Ellcey <sellcey@mips.com>
1884 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1885 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1886 * config/mips/mips.c (mips_option_override): Change setting
1888 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1889 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1890 Change from Mask to Var.
1892 2014-01-27 Jeff Law <law@redhat.com>
1894 * ipa-inline.c (inline_small_functions): Fix typo.
1896 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1898 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1899 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1900 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1901 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1902 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1903 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1904 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1905 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1906 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1907 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1908 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1909 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1910 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1911 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1912 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1913 (_mm512_storeu_epi64): Ditto.
1914 (_mm512_cmpge_epi32_mask): Ditto.
1915 (_mm512_cmpge_epu32_mask): Ditto.
1916 (_mm512_cmpge_epi64_mask): Ditto.
1917 (_mm512_cmpge_epu64_mask): Ditto.
1918 (_mm512_cmple_epi32_mask): Ditto.
1919 (_mm512_cmple_epu32_mask): Ditto.
1920 (_mm512_cmple_epi64_mask): Ditto.
1921 (_mm512_cmple_epu64_mask): Ditto.
1922 (_mm512_cmplt_epi32_mask): Ditto.
1923 (_mm512_cmplt_epu32_mask): Ditto.
1924 (_mm512_cmplt_epi64_mask): Ditto.
1925 (_mm512_cmplt_epu64_mask): Ditto.
1926 (_mm512_cmpneq_epi32_mask): Ditto.
1927 (_mm512_cmpneq_epu32_mask): Ditto.
1928 (_mm512_cmpneq_epi64_mask): Ditto.
1929 (_mm512_cmpneq_epu64_mask): Ditto.
1930 (_mm512_expand_pd): Ditto.
1931 (_mm512_expand_ps): Ditto.
1932 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1933 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1934 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1935 * config/i386/i386.c (ix86_builtins): Add
1936 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1937 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1938 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1939 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1940 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1941 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1942 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1943 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1944 IX86_BUILTIN_PMOVUSQW512_MEM.
1945 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1946 __builtin_ia32_pmovsqd512mem_mask,
1947 __builtin_ia32_pmovqd512mem_mask,
1948 __builtin_ia32_pmovusqw512mem_mask,
1949 __builtin_ia32_pmovsqw512mem_mask,
1950 __builtin_ia32_pmovqw512mem_mask,
1951 __builtin_ia32_pmovusdw512mem_mask,
1952 __builtin_ia32_pmovsdw512mem_mask,
1953 __builtin_ia32_pmovdw512mem_mask,
1954 __builtin_ia32_pmovqb512mem_mask,
1955 __builtin_ia32_pmovusqb512mem_mask,
1956 __builtin_ia32_pmovsqb512mem_mask,
1957 __builtin_ia32_pmovusdb512mem_mask,
1958 __builtin_ia32_pmovsdb512mem_mask,
1959 __builtin_ia32_pmovdb512mem_mask.
1960 (bdesc_args): Add __builtin_ia32_expanddf512,
1961 __builtin_ia32_expandsf512.
1962 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1963 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1964 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1965 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1966 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1967 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1968 (avx512f_<code>v8div16qi2_mask_store): This.
1969 (avx512f_expand<mode>): New.
1971 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1973 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1975 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1976 (_mm512_prefetch_i32scatter_pd): Ditto.
1977 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1978 (_mm512_prefetch_i64scatter_pd): Ditto.
1979 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1980 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1981 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1982 (_mm512_prefetch_i32scatter_ps): Ditto.
1983 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1984 (_mm512_prefetch_i64scatter_ps): Ditto.
1985 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1986 * config/i386/i386-builtin-types.def: Define
1987 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1988 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1989 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1990 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1991 IX86_BUILTIN_SCATTERPFQPD.
1992 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1993 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1994 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1995 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1996 __builtin_ia32_scatterpfqps.
1997 (ix86_expand_builtin): Expand new built-ins.
1998 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1999 fix memory access data type.
2000 (*avx512pf_gatherpf<mode>_mask): Ditto.
2001 (*avx512pf_gatherpf<mode>): Ditto.
2002 (avx512pf_scatterpf<mode>): Ditto.
2003 (*avx512pf_scatterpf<mode>_mask): Ditto.
2004 (*avx512pf_scatterpf<mode>): Ditto.
2005 (GATHER_SCATTER_SF_MEM_MODE): New.
2006 (avx512pf_gatherpf<mode>df): Ditto.
2007 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2008 (*avx512pf_scatterpf<mode>df): Ditto.
2010 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2013 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2014 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2017 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2019 * common/config/arm/arm-common.c
2020 (arm_rewrite_mcpu): Handle multiple names.
2022 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2024 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2026 * gimple-builder.h (create_gimple_tmp): Delete.
2028 2014-01-27 Christian Bruel <christian.bruel@st.com>
2030 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2033 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2035 * config/pa/pa.md (call): Generate indirect long calls to non-local
2036 functions when outputing 32-bit code.
2037 (call_value): Likewise except for special call to buggy powf function.
2039 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2040 portable runtime and PIC indirect calls.
2041 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2042 and PIC call sequences. Use ldo instead of blr to set return register
2043 in PIC call sequence.
2045 2014-01-25 Walter Lee <walt@tilera.com>
2047 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2048 avoid clobbering a live register.
2050 2014-01-25 Walter Lee <walt@tilera.com>
2052 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2053 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2054 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2055 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2057 2014-01-25 Walter Lee <walt@tilera.com>
2059 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2060 arguments on even registers.
2061 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2063 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2064 (BIGGEST_ALIGNMENT): Ditto.
2065 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2067 2014-01-25 Walter Lee <walt@tilera.com>
2069 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2070 insns before bundling.
2071 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2073 2014-01-25 Walter Lee <walt@tilera.com>
2075 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2076 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2077 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2079 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2081 * config/mips/constraints.md (kl): Delete.
2082 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2083 define expands, using...
2084 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2085 instructions for MIPS16.
2086 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2087 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2089 2014-01-25 Walter Lee <walt@tilera.com>
2091 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2095 2014-01-25 Walter Lee <walt@tilera.com>
2097 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2098 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2100 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2102 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2105 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2107 * print-rtl.c (in_call_function_usage): New var.
2108 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2109 EXPR_LIST mode as mode and not as reg note name.
2112 * cfgloopmanip.c (copy_loop_info): If
2113 loop->warned_aggressive_loop_optimizations, make sure
2114 the flag is set in target loop too.
2116 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2118 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2120 * builtins.def: Likewise.
2121 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2122 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2123 * ira.c (ira_setup_eliminable_regset): Likewise.
2124 * omp-low.c (gate_expand_omp): Likewise.
2125 (execute_lower_omp): Likewise.
2126 (diagnose_sb_0): Likewise.
2127 (gate_diagnose_omp_blocks): Likewise.
2128 (simd_clone_clauses_extract): Likewise.
2131 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2133 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2134 correction for little endian...
2135 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2138 2014-01-24 Jeff Law <law@redhat.com>
2140 PR tree-optimization/59919
2141 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2142 for non-returning calls.
2144 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2146 * common/config/aarch64/aarch64-common.c
2147 (aarch64_rewrite_mcpu): Handle multiple names.
2148 * config/aarch64/aarch64.h
2149 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2151 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2153 * input.c (add_file_to_cache_tab): Handle the case where fopen
2156 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2159 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2160 from push operand if code of push isn't PRE_DEC.
2162 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2165 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2166 -mquad-memory-atomic. Update -mquad-memory documentation to say
2167 it is only used for non-atomic loads/stores.
2169 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2170 -mquad-memory or -mquad-memory-atomic switches.
2172 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2173 -mquad-memory-atomic to ISA 2.07 support.
2175 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2176 to separate support of normal quad word memory operations (ldq, stq)
2177 from the atomic quad word memory operations.
2179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2180 support to separate non-atomic quad word operations from atomic
2181 quad word operations. Disable non-atomic quad word operations in
2182 little endian mode so that we don't have to swap words after the
2183 load and before the store.
2184 (quad_load_store_p): Add comment about atomic quad word support.
2185 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2186 options printed with -mdebug=reg.
2188 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2189 -mquad-memory-atomic as the test for whether we have quad word
2190 atomic instructions.
2191 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2192 or -mp8-vector are used, allow byte/half-word atomic operations.
2194 * config/rs6000/sync.md (load_lockedti): Insure that the address
2195 is a proper indexed or indirect address for the lqarx instruction.
2196 On little endian systems, swap the hi/lo registers after the lqarx
2198 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2199 insure the address is valid for the lqarx instruction.
2200 (store_conditionalti): Insure that the address is a proper indexed
2201 or indirect address for the stqcrx. instruction. On little endian
2202 systems, swap the hi/lo registers before doing the stqcrx.
2204 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2205 insure the address is valid for the stqcrx. instruction.
2207 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2208 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2209 type of quad memory support is available.
2211 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2214 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2215 there is a danger of looping.
2217 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2219 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2220 force flag_ira_loop_pressure if set via command line.
2222 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2224 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2225 (ashr_simd): New builtin handling DI mode.
2226 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2227 (aarch64_sshr_simddi): New match pattern.
2228 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2229 (vshrd_n_s64): Likewise.
2230 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2232 2014-01-23 Nick Clifton <nickc@redhat.com>
2234 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2235 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2236 favour of mcu specific scripts.
2237 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2240 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2241 Alex Velenko <Alex.Velenko@arm.com>
2243 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2244 (vaddv_s16): Likewise.
2245 (vaddv_s32): Likewise.
2246 (vaddv_u8): Likewise.
2247 (vaddv_u16): Likewise.
2248 (vaddv_u32): Likewise.
2249 (vaddvq_s8): Likewise.
2250 (vaddvq_s16): Likewise.
2251 (vaddvq_s32): Likewise.
2252 (vaddvq_s64): Likewise.
2253 (vaddvq_u8): Likewise.
2254 (vaddvq_u16): Likewise.
2255 (vaddvq_u32): Likewise.
2256 (vaddvq_u64): Likewise.
2257 (vaddv_f32): Likewise.
2258 (vaddvq_f32): Likewise.
2259 (vaddvq_f64): Likewise.
2260 (vmaxv_f32): Likewise.
2261 (vmaxv_s8): Likewise.
2262 (vmaxv_s16): Likewise.
2263 (vmaxv_s32): Likewise.
2264 (vmaxv_u8): Likewise.
2265 (vmaxv_u16): Likewise.
2266 (vmaxv_u32): Likewise.
2267 (vmaxvq_f32): Likewise.
2268 (vmaxvq_f64): Likewise.
2269 (vmaxvq_s8): Likewise.
2270 (vmaxvq_s16): Likewise.
2271 (vmaxvq_s32): Likewise.
2272 (vmaxvq_u8): Likewise.
2273 (vmaxvq_u16): Likewise.
2274 (vmaxvq_u32): Likewise.
2275 (vmaxnmv_f32): Likewise.
2276 (vmaxnmvq_f32): Likewise.
2277 (vmaxnmvq_f64): Likewise.
2278 (vminv_f32): Likewise.
2279 (vminv_s8): Likewise.
2280 (vminv_s16): Likewise.
2281 (vminv_s32): Likewise.
2282 (vminv_u8): Likewise.
2283 (vminv_u16): Likewise.
2284 (vminv_u32): Likewise.
2285 (vminvq_f32): Likewise.
2286 (vminvq_f64): Likewise.
2287 (vminvq_s8): Likewise.
2288 (vminvq_s16): Likewise.
2289 (vminvq_s32): Likewise.
2290 (vminvq_u8): Likewise.
2291 (vminvq_u16): Likewise.
2292 (vminvq_u32): Likewise.
2293 (vminnmv_f32): Likewise.
2294 (vminnmvq_f32): Likewise.
2295 (vminnmvq_f64): Likewise.
2297 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2299 * config/aarch64/aarch64-simd.md
2300 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2301 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2302 (*aarch64_mul3_elt<mode>): Likewise.
2303 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2304 (*aarch64_mul3_elt_to_64v2df): Likewise.
2305 (*aarch64_mla_elt<mode>): Likewise.
2306 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2307 (*aarch64_mls_elt<mode>): Likewise.
2308 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2309 (*aarch64_fma4_elt<mode>): Likewise.
2310 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2311 (*aarch64_fma4_elt_to_64v2df): Likewise.
2312 (*aarch64_fnma4_elt<mode>): Likewise.
2313 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2314 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2315 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2316 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2317 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2318 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2319 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2320 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2321 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2323 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2325 * config/aarch64/aarch64-simd.md
2326 (aarch64_be_checked_get_lane<mode>): New define_expand.
2327 * config/aarch64/aarch64-simd-builtins.def
2328 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2329 New builtin definition.
2330 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2331 Use new safe be builtin.
2333 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2335 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2337 (aarch64_be_st1<mode>): Likewise.
2338 (aarch_ld1<VALL:mode>): Define_expand modified.
2339 (aarch_st1<VALL:mode>): Likewise.
2340 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2341 (UNSPEC_ST1): Likewise.
2343 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2345 * config/microblaze/microblaze.md: Add trap insn and attribute
2347 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2349 PR preprocessor/58580
2350 * input.h (location_get_source_line): Take an additional line_size
2352 (void diagnostics_file_cache_fini): Declare new function.
2353 * input.c (struct fcache): New type.
2354 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2355 New static constants.
2356 (diagnostic_file_cache_init, total_lines_num)
2357 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2358 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2359 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2360 (get_next_line, read_next_line, goto_next_line, read_line_num):
2361 New static function definitions.
2362 (diagnostic_file_cache_fini): New function.
2363 (location_get_source_line): Take an additional output line_len
2364 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2366 * diagnostic.c (diagnostic_finish): Call
2367 diagnostic_file_cache_fini.
2368 (adjust_line): Take an additional input parameter for the length
2369 of the line, rather than calculating it with strlen.
2370 (diagnostic_show_locus): Adjust the use of
2371 location_get_source_line and adjust_line with respect to their new
2372 signature. While displaying a line now, do not stop at the first
2373 null byte. Rather, display the zero byte as a space and keep
2374 going until we reach the size of the line.
2375 * Makefile.in: Add vec.o to OBJS-libcommon
2377 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2378 Ilya Tocar <ilya.tocar@intel.com>
2380 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2381 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2382 (__builtin_ia32_kmov16): Ditto.
2383 * config/i386/i386.md (UNSPEC_KMOV): New.
2386 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2388 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2389 (_mm512_storeu_si512): Ditto.
2391 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2394 * rtl.h (get_referenced_operands): Declare.
2395 * recog.c (get_referenced_operands): New function.
2396 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2397 operands have been referenced when recording LO_SUM references.
2399 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2401 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2403 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2405 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2406 Enable for generic and recent AMD targets.
2408 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2410 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2411 ARG_SIZE note when adjustment was eliminated.
2413 2014-01-22 Jeff Law <law@redhat.com>
2415 PR tree-optimization/59597
2416 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2417 in file. Accept new argument REGISTERING and use it to modify
2418 dump output appropriately.
2419 (register_jump_thread): Corresponding changes.
2420 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2421 thread paths involving joiner blocks. Add code to dump cancelled
2422 jump threading paths.
2424 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2426 PR rtl-optimization/59477
2427 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2428 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2430 2014-01-22 Tom Tromey <tromey@redhat.com>
2432 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2434 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2436 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2438 PR rtl-optimization/59896
2439 * lra-constraints.c (process_alt_operands): Check unused note for
2440 matched operands of insn with no output reloads.
2442 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2444 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2445 (mips_move_from_gpr_cost): Likewise.
2447 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2449 PR rtl-optimization/59858
2450 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2451 ira_class_hard_regs_num.
2452 (process_alt_operands): Increase reject for dying matched operand.
2454 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2457 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2458 smaller than size, perform several stores or loads and stores
2459 at dst + count - size to store or copy all of size bytes, rather
2460 than just last modesize bytes.
2462 2014-01-20 DJ Delorie <dj@redhat.com>
2464 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2465 that CLOBBERs are REGs before propogating their values.
2467 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2470 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2471 (cgraph_inline_failed_type): New function.
2472 * cgraph.h (DEFCIFCODE): Add type.
2473 (cgraph_inline_failed_type_t): New enum.
2474 (cgraph_inline_failed_type): New prototype.
2475 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2476 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2477 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2478 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2479 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2480 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2481 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2482 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2483 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2484 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2485 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2486 OPTIMIZATION_MISMATCH.
2487 * tree-inline.c (expand_call_inline): Emit errors during
2488 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2490 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2493 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2494 mode attribute in insn output.
2496 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2498 * output.h (output_constant): Delete.
2499 * varasm.c (output_constant): Make private.
2501 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2503 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2505 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2508 * tree.h (fold_builtin_strcat): New prototype.
2509 * builtins.c (fold_builtin_strcat): No longer static. Add len
2510 argument, if non-NULL, don't call c_strlen. Optimize
2511 directly into __builtin_memcpy instead of __builtin_strcpy.
2512 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2513 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2515 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2517 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2518 for SImode_address_operand operands, having only a REG argument.
2520 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2522 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2523 loader name using mbig-endian.
2524 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2526 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2528 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2532 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2534 * config/aarch64/aarch64-protos.h
2535 (aarch64_cannot_change_mode_class_ptr): Declare.
2536 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2537 aarch64_cannot_change_mode_class_ptr): New.
2538 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2539 backend hook aarch64_cannot_change_mode_class.
2541 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2543 * common/config/aarch64/aarch64-common.c
2544 (aarch64_handle_option): Don't handle any option order logic here.
2545 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2546 selected_cpu, warn on architecture version mismatch.
2547 (aarch64_override_options): Fix parsing order for option strings.
2549 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2550 Iain Sandoe <iain@codesourcery.com>
2553 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2554 warning. Amend comment to reflect current functionality.
2556 2014-01-20 Richard Biener <rguenther@suse.de>
2559 * builtins.c (fold_builtin_strcat): Remove case better handled
2560 by tree-ssa-strlen.c.
2562 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2564 * config/aarch64/aarch64.opt
2565 (mcpu, march, mtune): Make case-insensitive.
2567 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2570 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2571 if operands[1] is a REG or ZERO_EXTEND of a REG.
2573 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2575 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2577 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2579 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2580 long non-pic millicode calls.
2582 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2584 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2586 2014-01-19 Kito Cheng <kito@0xlab.org>
2588 * builtins.c (expand_movstr): Check movstr expand done or fail.
2590 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2591 H.J. Lu <hongjiu.lu@intel.com>
2594 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2595 to DImode for zero-extended addresses.
2597 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2599 PR rtl-optimization/57763
2600 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2601 on the new indirect jump_insn and increment LABEL_NUSES (label).
2603 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2607 * config.gcc (x86_archs): New variable.
2608 (x86_64_archs): Likewise.
2609 (x86_cpus): Likewise.
2610 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2611 --with-arch/--with-cpu= options.
2612 Support --with-arch=/--with-cpu={nehalem,westmere,
2613 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2615 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2617 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2618 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2620 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2622 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2624 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2627 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2628 clear cpp_get_options (parse_in)->warn_unused_macros for
2629 ix86_target_macros_internal with cpp_define.
2631 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2633 * jump.c (delete_related_insns): Keep (use (insn))s.
2634 * reorg.c (redundant_insn): Check for barriers too.
2636 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2638 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2640 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2642 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2643 call to $$dyncall when TARGET_LONG_CALLS is true.
2645 2014-01-17 Jeff Law <law@redhat.com>
2647 * ree.c (combine_set_extension): Temporarily disable test for
2648 changing number of hard registers.
2650 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2653 * ipa-inline-analysis.c (inline_free_summary):
2654 Do not free summary of aliases.
2656 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2659 * gimplify.c (gimplify_expr): Use create_tmp_var
2660 instead of create_tmp_var_raw. If cond doesn't have
2661 integral type, don't add the IFN_ANNOTATE builtin at all.
2663 2014-01-17 Martin Jambor <mjambor@suse.cz>
2666 * ipa-cp.c (prev_edge_clone): New variable.
2667 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2668 Also resize prev_edge_clone vector.
2669 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2670 (ipcp_edge_removal_hook): New function.
2671 (ipcp_driver): Register ipcp_edge_removal_hook.
2673 2014-01-17 Andrew Pinski <apinski@cavium.com>
2674 Steve Ellcey <sellcey@mips.com>
2677 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2680 2014-01-17 Jeff Law <law@redhat.com>
2683 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2684 so that pass_ccp runs first.
2686 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2688 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2689 (ix86_adjust_cost): Use !TARGET_XXX.
2690 (do_reorder_for_imul): Likewise.
2691 (swap_top_of_ready_list): Likewise.
2692 (ix86_sched_reorder): Likewise.
2694 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2696 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2697 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2698 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2699 (intel_memset): New. Duplicate slm_memset.
2700 (intel_cost): New. Duplicate slm_cost.
2701 (m_INTEL): New macro.
2702 (processor_target_table): Add "intel".
2703 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2704 with PROCESSOR_INTEL for "intel".
2705 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2706 PROCESSOR_SILVERMONT.
2707 (ix86_issue_rate): Likewise.
2708 (ix86_adjust_cost): Likewise.
2709 (ia32_multipass_dfa_lookahead): Likewise.
2710 (swap_top_of_ready_list): Likewise.
2711 (ix86_sched_reorder): Likewise.
2712 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2713 instead of TARGET_OPT_AGU.
2714 * config/i386/i386.h (TARGET_INTEL): New.
2715 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2716 (processor_type): Add PROCESSOR_INTEL.
2717 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2718 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2720 2014-01-17 Marek Polacek <polacek@redhat.com>
2723 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2726 2014-01-17 Richard Biener <rguenther@suse.de>
2728 PR tree-optimization/46590
2729 * opts.c (default_options_table): Add entries for
2730 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2731 all enabled at -O1 but not for -Og.
2732 * common.opt (fbranch-count-reg): Remove Init(1).
2733 (fmove-loop-invariants): Likewise.
2734 (ftree-pta): Likewise.
2736 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2738 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2739 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2740 decls to at least the GCC 4.8 used alignments.
2743 * tree-nested.c (convert_nonlocal_reference_stmt,
2744 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2745 of GIMPLE_BIND stmts, adjust associated decls.
2747 2014-01-17 Richard Biener <rguenther@suse.de>
2749 PR tree-optimization/46590
2750 * vec.h (vec<>::bseach): New member function implementing
2751 binary search according to C89 bsearch.
2752 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2753 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2754 bitmap pointer again. Make accesses_in_loop a flat array.
2755 (mem_ref_obstack): New global.
2756 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2757 (mark_ref_stored): Likewise.
2758 (ref_indep_loop_p_2): Likewise.
2759 (set_ref_stored_in_loop): New helper function.
2760 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2761 (memref_free): Adjust.
2762 (record_mem_ref_loc): Simplify.
2763 (gather_mem_refs_stmt): Adjust.
2764 (sort_locs_in_loop_postorder_cmp): New function.
2765 (analyze_memory_references): Sort accesses_in_loop after
2766 loop postorder number.
2767 (find_ref_loc_in_loop_cmp): New function.
2768 (for_all_locs_in_loop): Find relevant cluster of locs in
2769 accesses_in_loop and iterate without recursion.
2770 (execute_sm): Avoid uninit warning.
2771 (struct ref_always_accessed): Simplify.
2772 (ref_always_accessed::operator ()): Likewise.
2773 (ref_always_accessed_p): Likewise.
2774 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2775 loop postorder numbers here.
2776 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2779 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2782 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2783 on decls for which assemble_alias has been called.
2785 2014-01-17 Nick Clifton <nickc@redhat.com>
2787 * config/msp430/msp430.opt: (mcpu): New option.
2788 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2789 (msp430_option_override): Parse target_cpu. If the MCU name
2790 matches a generic string, clear target_mcu.
2791 (msp430_attr): Allow numeric interrupt values up to 63.
2792 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2793 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2795 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2797 * config/msp430/msp430.md (popm): Use %J rather than %I.
2798 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2799 (addhi_cy_i): Use immediate_operand for operand 2.
2800 * doc/invoke.texi: Document -mcpu option.
2802 2014-01-17 Richard Biener <rguenther@suse.de>
2804 PR rtl-optimization/38518
2805 * df.h (df_analyze_loop): Declare.
2806 * df-core.c: Include cfgloop.h.
2807 (df_analyze_1): Split out main part of df_analyze.
2808 (df_analyze): Adjust.
2809 (loop_inverted_post_order_compute): New function.
2810 (loop_post_order_compute): Likewise.
2811 (df_analyze_loop): New function avoiding whole-function
2813 * loop-invariant.c (find_defs): Use df_analyze_loop.
2814 (find_invariants): Adjust.
2815 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2817 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2819 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2820 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2822 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2824 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2825 traversal when removing references.
2827 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2830 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2832 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2835 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2836 pushing a reload for an autoinc when we had previously reloaded an
2837 inner part of the address.
2839 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2841 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2843 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2844 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2845 when not giving up or versioning for alias only because of
2847 (vect_analyze_data_ref_dependences): Set to true.
2848 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2850 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2851 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2855 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2858 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2859 operand 0 predicate for gathers, use a new pseudo as subtarget.
2861 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2864 * lra-constraints.c (process_alt_operands): Add printing debug info.
2865 Check absence of input/output reloads for matched operands too.
2867 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2869 PR rtl-optimization/59835
2870 * ira.c (ira_init_register_move_cost): Increase cost for
2873 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2875 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2877 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2880 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2881 non-register objects. Use gen_(high/low)part more consistently.
2884 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2887 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2888 endian support, remove tests for WORDS_BIG_ENDIAN.
2889 (p8_mfvsrd_3_<mode>): Likewise.
2890 (reload_gpr_from_vsx<mode>): Likewise.
2891 (reload_gpr_from_vsxsf): Likewise.
2892 (p8_mfvsrd_4_disf): Likewise.
2894 2014-01-16 Richard Biener <rguenther@suse.de>
2896 PR rtl-optimization/46590
2897 * lcm.c (compute_antinout_edge): Use postorder iteration.
2898 (compute_laterin): Use inverted postorder iteration.
2900 2014-01-16 Nick Clifton <nickc@redhat.com>
2903 * varasm.c (output_constant): Return the number of bytes actually
2905 (output_constructor_array_range): Update the field size with the
2906 number of bytes emitted by output_constant.
2907 (output_constructor_regular_field): Likewise. Also do not
2908 complain if the total number of bytes emitted is now greater
2909 than the expected fieldpos.
2910 * output.h (output_constant): Update prototype and descriptive comment.
2912 2014-01-16 Marek Polacek <polacek@redhat.com>
2915 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2916 it is error_mark_node.
2918 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2920 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2921 VALID_AVX256_REG_OR_OI_MODE.
2923 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2925 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2926 current procedure should be profiled.
2928 2014-01-15 Andrew Pinski <apinski@cavium.com>
2930 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2931 of moving from/to the STACK_REG register class.
2933 2014-01-15 Richard Henderson <rth@redhat.com>
2936 * reginfo.c (global_regs_decl): Globalize.
2937 * rtl.h (global_regs_decl): Declare.
2938 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2939 reserved via global_regs.
2941 2014-01-15 Teresa Johnson <tejohnson@google.com>
2943 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2945 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2947 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2948 and vmulosh rather than call gen_vec_widen_smult_*.
2949 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2950 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2951 (vec_widen_smult_even_v16qi): Likewise.
2952 (vec_widen_umult_even_v8hi): Likewise.
2953 (vec_widen_smult_even_v8hi): Likewise.
2954 (vec_widen_umult_odd_v16qi): Likewise.
2955 (vec_widen_smult_odd_v16qi): Likewise.
2956 (vec_widen_umult_odd_v8hi): Likewise.
2957 (vec_widen_smult_odd_v8hi): Likewise.
2958 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2959 vmuloub rather than call gen_vec_widen_umult_*.
2960 (vec_widen_umult_lo_v16qi): Likewise.
2961 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2962 vmulosb rather than call gen_vec_widen_smult_*.
2963 (vec_widen_smult_lo_v16qi): Likewise.
2964 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2965 rather than call gen_vec_widen_umult_*.
2966 (vec_widen_umult_lo_v8hi): Likewise.
2967 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2968 rather than call gen_vec_widen_smult_*.
2969 (vec_widen_smult_lo_v8hi): Likewise.
2971 2014-01-15 Jeff Law <law@redhat.com>
2973 PR tree-optimization/59747
2974 * ree.c (find_and_remove_re): Properly handle case where a second
2975 eliminated extension requires widening a copy created for elimination
2976 of a prior extension.
2977 (combine_set_extension): Ensure that the number of hard regs needed
2978 for a destination register does not change when we widen it.
2980 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2982 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2983 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2984 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2985 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2986 (avr-*-rtems*): Likewise.
2987 (bfin*-rtems*): Likewise.
2988 (moxie-*-rtems*): Likewise.
2989 (h8300-*-rtems*): Likewise.
2990 (i[34567]86-*-rtems*): Likewise.
2991 (lm32-*-rtems*): Likewise.
2992 (m32r-*-rtems*): Likewise.
2993 (m68k-*-rtems*): Likewise.
2994 (microblaze*-*-rtems*): Likewise.
2995 (mips*-*-rtems*): Likewise.
2996 (powerpc-*-rtems*): Likewise.
2997 (sh-*-rtems*): Likewise.
2998 (sparc-*-rtems*): Likewise.
2999 (sparc64-*-rtems*): Likewise.
3000 (v850-*-rtems*): Likewise.
3001 (m32c-*-rtems*): Likewise.
3003 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3005 PR rtl-optimization/59511
3006 * ira.c (ira_init_register_move_cost): Use memory costs for some
3007 cases of register move cost calculations.
3008 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3009 instead of BB frequency.
3010 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3011 * lra-assigns.c (find_hard_regno_for): Ditto.
3013 2014-01-15 Richard Biener <rguenther@suse.de>
3015 PR tree-optimization/59822
3016 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3017 (vectorizable_load): Use it to hoist defs of uses of invariant
3018 loads out of the loop.
3020 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3021 Kugan Vivekanandarajah <kuganv@linaro.org>
3024 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3027 2014-01-15 Richard Biener <rguenther@suse.de>
3029 PR rtl-optimization/59802
3030 * lcm.c (compute_available): Use inverted postorder to seed
3031 the initial worklist.
3033 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3036 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3037 ADDR_REGS for invalid symrefs in non-PIC code.
3039 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3042 * builtins.c (determine_block_size): Initialize *probable_max_size
3043 even if len_rtx is CONST_INT.
3045 2014-01-14 Andrew Pinski <apinski@cavium.com>
3047 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3048 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3049 (cortexa53_tunings): Likewise.
3050 (aarch64_sched_issue_rate): New function.
3051 (TARGET_SCHED_ISSUE_RATE): Define.
3053 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3055 * ira-costs.c (find_costs_and_classes): Add missed
3056 ira_init_register_move_cost_if_necessary.
3058 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3061 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3063 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3066 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3067 to indicate if type is used for function return value. Warn ABI
3068 change if the vector mode isn't available for function return value.
3069 (ix86_function_arg_advance): Pass false to type_natural_mode.
3070 (ix86_function_arg): Likewise.
3071 (ix86_gimplify_va_arg): Likewise.
3072 (function_arg_32): Don't warn ABI change.
3073 (ix86_function_value): Pass true to type_natural_mode.
3074 (ix86_return_in_memory): Likewise.
3075 (ix86_struct_value_rtx): Removed.
3076 (TARGET_STRUCT_VALUE_RTX): Likewise.
3078 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3080 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3081 converting a conditional jump into a conditional return.
3083 2014-01-14 Richard Biener <rguenther@suse.de>
3085 PR tree-optimization/58921
3086 PR tree-optimization/59006
3087 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3088 hoisting invariant stmts.
3089 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3090 invariant loads on the preheader edge if possible.
3092 2014-01-14 Joey Ye <joey.ye@arm.com>
3094 * doc/plugin.texi (Building GCC plugins): Update to C++.
3096 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3098 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3099 (_mm_rcp28_round_ss): Ditto.
3100 (_mm_rsqrt28_round_sd): Ditto.
3101 (_mm_rsqrt28_round_ss): Ditto.
3102 (_mm_rcp28_sd): Ditto.
3103 (_mm_rcp28_ss): Ditto.
3104 (_mm_rsqrt28_sd): Ditto.
3105 (_mm_rsqrt28_ss): Ditto.
3106 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3107 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3108 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3109 (IX86_BUILTIN_RCP28SD): Ditto.
3110 (IX86_BUILTIN_RCP28SS): Ditto.
3111 (IX86_BUILTIN_RSQRT28SD): Ditto.
3112 (IX86_BUILTIN_RSQRT28SS): Ditto.
3113 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3114 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3115 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3116 (ix86_expand_special_args_builtin): Expand new FTYPE.
3117 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3118 (srcp14<mode>): Make insn unary.
3119 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3120 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3121 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3122 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3123 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3124 Fix rounding: make it SAE only.
3125 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3127 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3129 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3130 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3131 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3132 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3133 (round_saeonly_mask_scalar_operand4): Ditto.
3134 (round_saeonly_mask_scalar_op3): Ditto.
3135 (round_saeonly_mask_scalar_op4): Ditto.
3137 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3139 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3140 Implement -maltivec=be for vec_insert and vec_extract.
3142 2014-01-10 DJ Delorie <dj@redhat.com>
3144 * config/msp430/msp430.md (call_internal): Don't allow memory
3145 references with SP as the base register.
3146 (call_value_internal): Likewise.
3147 * config/msp430/constraints.md (Yc): New. For memory references
3148 that don't use SP as a base register.
3150 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3151 "an integer without a # prefix"
3152 * config/msp430/msp430.md (epilogue_helper): Use it.
3154 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3157 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3158 AVX512F gather builtins.
3159 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3160 on gather decls with INTEGER_TYPE masktype.
3161 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3162 directly into the builtin rather than hoisting it before loop.
3164 PR tree-optimization/59387
3165 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3166 (scev_const_prop): If folded_casts and type has undefined overflow,
3167 use force_gimple_operand instead of force_gimple_operand_gsi and
3168 for each added stmt if it is assign with
3169 arith_code_with_undefined_signed_overflow, call
3170 rewrite_to_defined_overflow.
3171 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3172 gimple-fold.h instead.
3173 (arith_code_with_undefined_signed_overflow,
3174 rewrite_to_defined_overflow): Moved to ...
3175 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3176 rewrite_to_defined_overflow): ... here. No longer static.
3177 Include gimplify-me.h.
3178 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3179 rewrite_to_defined_overflow): New prototypes.
3181 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3183 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3185 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3187 * builtins.c (get_object_alignment_2): Minor tweak.
3188 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3190 2014-01-13 Christian Bruel <christian.bruel@st.com>
3192 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3193 optimized non constant lengths.
3195 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3198 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3199 load as __atomic_load_N if possible.
3201 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3203 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3205 (rs6000_expand_builtin): Adjust call.
3207 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3210 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3211 * config/rs6000/rs6000.c: Include target-globals.h.
3212 (rs6000_set_current_function): Instead of doing target_reinit
3213 unconditionally, use save_target_globals_default_opts and
3214 restore_target_globals.
3216 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3218 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3219 (rs6000_expand_builtin): Handle mffs and mtfsf.
3220 (rs6000_init_builtins): Define mffs and mtfsf.
3221 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3222 (rs6000_mffs): New pattern.
3223 (rs6000_mtfsf): New pattern.
3225 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3227 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3228 Start narrowing with START. Apply candidate-use pair
3229 and check overall cost in narrowing.
3230 (iv_ca_prune): Pass new argument.
3232 2014-01-10 Jeff Law <law@redhat.com>
3235 * ree.c (combine_reaching_defs): Ensure the defining statement
3236 occurs before the extension when optimizing extensions with
3237 different source and destination hard registers.
3239 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3242 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3243 vtables into the type inheritance graph.
3245 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3247 PR rtl-optimization/59754
3248 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3249 modes in the REGNO != REGNO case.
3251 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3253 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3255 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3257 PR tree-optimization/59745
3258 * tree-predcom.c (tree_predictive_commoning_loop): Call
3259 free_affine_expand_cache if giving up because components is NULL.
3261 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3262 GC in payload of target_globals struct instead of allocating them on
3263 the heap and the larger structs separately using GC.
3264 * target-globals.h (struct target_globals): Make regs, hard_regs,
3265 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3266 of GTY((skip)) and change type to void *.
3267 (reset_target_globals): Cast loads from those fields to corresponding
3270 2014-01-10 Steve Ellcey <sellcey@mips.com>
3273 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3274 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3275 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3277 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3280 * aarch64-modes.def (CC_Zmode): New flags mode.
3281 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3282 represents an equality.
3283 (aarch64_get_condition_code): Handle CC_Zmode.
3284 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3286 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3288 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3289 extraction in good case.
3291 2014-01-10 Richard Biener <rguenther@suse.de>
3293 PR tree-optimization/59374
3294 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3295 checking after SLP discovery. Mark stmts not participating
3296 in any SLP instance properly.
3298 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3300 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3301 when handling a SET rtx.
3303 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3305 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3306 (cortex-a57): Likewise.
3307 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3309 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3311 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3312 non-iwmmxt builtins.
3314 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3318 * ipa-devirt.c record_target_from_binfo): Take as argument
3319 stack of binfos and lookup matching one for virtual inheritance.
3320 (possible_polymorphic_call_targets_1): Update.
3322 2014-01-10 Huacai Chen <chenhc@lemote.com>
3324 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3325 kernel strings for Loongson-2E/2F/3A.
3327 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3330 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3331 is_gimple_call before calling gimple_call_internal_p.
3333 2014-01-09 Steve Ellcey <sellcey@mips.com>
3335 * Makefile.in (TREE_FLOW_H): Remove.
3336 (TREE_SSA_H): Add file names from tree-flow.h.
3337 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3338 * tree.h: Remove tree-flow.h reference.
3339 * hash-table.h: Remove tree-flow.h reference.
3340 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3341 reference with tree-ssa-loop.h.
3343 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3345 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3346 default element-order behavior for -maltivec.
3347 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3348 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3349 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3350 when targeting big endian, at least for now.
3351 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3353 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3356 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3357 var satisfies use_register_for_decl, just take into account type
3358 alignment, rather than decl alignment.
3360 PR tree-optimization/59622
3361 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3362 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3363 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3364 Don't devirtualize for inplace at all. For targets.length () == 1,
3365 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3367 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3369 * config/i386/i386.md (cpu): Remove the unused btver1.
3371 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3373 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3375 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3378 * tree-core.h (struct target_globals): New forward declaration.
3379 (struct tree_target_option): Add globals field.
3380 * tree.h (TREE_TARGET_GLOBALS): Define.
3381 (prepare_target_option_nodes_for_pch): New prototype.
3382 * target-globals.h (struct target_globals): Define even if
3384 * tree.c (prepare_target_option_node_for_pch,
3385 prepare_target_option_nodes_for_pch): New functions.
3386 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3387 * config/i386/i386.c: Include target-globals.h.
3388 (ix86_set_current_function): Instead of doing target_reinit
3389 unconditionally, use save_target_globals_default_opts and
3390 restore_target_globals.
3392 2014-01-09 Richard Biener <rguenther@suse.de>
3394 PR tree-optimization/59715
3395 * tree-cfg.h (split_critical_edges): Declare.
3396 * tree-cfg.c (split_critical_edges): Export.
3397 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3399 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3401 * cfgexpand.c (expand_stack_vars): Optionally disable
3402 asan stack protection.
3403 (expand_used_vars): Likewise.
3404 (partition_stack_vars): Likewise.
3405 * asan.c (asan_emit_stack_protection): Optionally disable
3406 after return stack usage.
3407 (instrument_derefs): Optionally disable memory access instrumentation.
3408 (instrument_builtin_call): Likewise.
3409 (instrument_strlen_call): Likewise.
3410 (asan_protect_global): Optionally disable global variables protection.
3411 * doc/invoke.texi: Added doc for new options.
3412 * params.def: Added new options.
3413 * params.h: Likewise.
3415 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3417 PR rtl-optimization/59724
3418 * ifcvt.c (cond_exec_process_if_block): Don't call
3419 flow_find_head_matching_sequence with 0 longest_match.
3420 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3421 non-active insns if !stop_after.
3422 (try_head_merge_bb): Revert 2014-01-07 changes.
3424 2014-01-08 Jeff Law <law@redhat.com>
3426 * ree.c (get_sub_rtx): New function, extracted from...
3427 (merge_def_and_ext): Here.
3428 (combine_reaching_defs): Use get_sub_rtx.
3430 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3432 * cgraph.h (varpool_variable_node): Do not choke on null node.
3434 2014-01-08 Catherine Moore <clm@codesourcery.com>
3436 * config/mips/mips.md (simple_return): Attempt to use JRC
3438 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3440 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3442 PR rtl-optimization/59137
3443 * reorg.c (steal_delay_list_from_target): Call update_block for
3445 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3447 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3449 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3450 two duplicate entries.
3452 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3455 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3457 * config/mips/mips.c (mips_truncated_op_cost): New function.
3458 (mips_rtx_costs): Adjust test for BADDU.
3459 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3461 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3463 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3464 (*baddu_si): ...this new pattern.
3466 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3469 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3471 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3474 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3476 (expand_expr, expand_normal): Adjust.
3477 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3478 inner_reference_p. Use inner_reference_p to expand inner references.
3479 (store_expr): Adjust.
3480 * cfgexpand.c (expand_call_stmt): Adjust.
3482 2014-01-08 Rong Xu <xur@google.com>
3484 * gcov-io.c (gcov_var): Move from gcov-io.h.
3485 (gcov_position): Ditto.
3486 (gcov_is_error): Ditto.
3487 (gcov_rewrite): Ditto.
3488 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3489 only part to libgcc/libgcov.h.
3491 2014-01-08 Marek Polacek <polacek@redhat.com>
3494 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3496 2014-01-08 Marek Polacek <polacek@redhat.com>
3499 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3501 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3503 PR rtl-optimization/59649
3504 * stor-layout.c (get_mode_bounds): For BImode return
3505 0 and STORE_FLAG_VALUE.
3507 2014-01-08 Richard Biener <rguenther@suse.de>
3510 * gimple.h (is_gimple_builtin_call): Remove.
3511 (gimple_builtin_call_types_compatible_p): New.
3512 (gimple_call_builtin_p): New overload.
3513 * gimple.c (is_gimple_builtin_call): Remove.
3514 (validate_call): Rename to ...
3515 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3517 (validate_type): New static function.
3518 (gimple_call_builtin_p): New overload and adjust.
3519 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3520 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3521 (gimple_fold_stmt_to_constant_1): Likewise.
3522 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3524 2014-01-08 Richard Biener <rguenther@suse.de>
3527 * gimplify.c (gimplify_expr): Gimplify register-register type
3528 VIEW_CONVERT_EXPRs to separate stmts.
3530 2014-01-07 Jeff Law <law@redhat.com>
3533 * ree.c (combine_set_extension): Handle case where source
3534 and destination registers in an extension insn are different.
3535 (combine_reaching_defs): Allow source and destination registers
3536 in extension to be different under limited circumstances.
3537 (add_removable_extension): Remove restriction that the
3538 source and destination registers in the extension are the same.
3539 (find_and_remove_re): Emit a copy from the extension's
3540 destination to its source after the defining insn if
3541 the source and destination registers are different.
3544 * ifcvt.c (merge_if_block): If we are merging a block with more than
3545 one successor with a block with no successors, remove any BARRIER
3546 after the second block.
3548 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3550 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3552 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3555 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3556 for 14-bit register offsets when INT14_OK_STRICT is false.
3558 2014-01-07 Roland Stigge <stigge@antcom.de>
3559 Michael Meissner <meissner@linux.vnet.ibm.com>
3562 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3563 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3565 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3567 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3570 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3572 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3573 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3574 rtx is const0_rtx or not.
3576 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3579 * target-globals.c (save_target_globals): Remove this_fn_optab
3581 * toplev.c: Include optabs.h.
3582 (target_reinit): Temporarily restore the global options if another
3583 set of options are in force.
3585 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3587 PR rtl-optimization/58668
3588 * cfgcleanup.c (flow_find_cross_jump): Don't count
3589 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3590 to determine what is counted.
3591 (flow_find_head_matching_sequence): Use active_insn_p to determine
3593 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3595 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3596 determine what is counted.
3598 PR tree-optimization/59643
3599 * tree-predcom.c (split_data_refs_to_components): If one dr is
3600 read and one write, determine_offset fails and the write isn't
3601 in the bad component, just put the read into the bad component.
3603 2014-01-07 Mike Stump <mikestump@comcast.net>
3604 Jakub Jelinek <jakub@redhat.com>
3607 * tree-core.h (struct tree_optimization_option): Change optabs
3608 type from unsigned char * to void *.
3609 * optabs.c (init_tree_optimization_optabs): Adjust
3610 TREE_OPTIMIZATION_OPTABS initialization.
3612 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3615 * config/i386/i386.h (struct machine_function): Add
3616 no_drap_save_restore field.
3617 * config/i386/i386.c (ix86_save_reg): Use
3618 !cfun->machine->no_drap_save_restore instead of
3619 crtl->stack_realign_needed.
3620 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3621 this function clears frame_pointer_needed. Set
3622 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3623 and DRAP reg is needed.
3625 2014-01-06 Marek Polacek <polacek@redhat.com>
3628 * doc/implement-c.texi: Mention that other integer types are
3629 permitted as bit-field types in strictly conforming mode.
3631 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3633 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3636 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3638 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3640 2014-01-06 Martin Jambor <mjambor@suse.cz>
3643 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3645 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3647 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3651 * var-tracking.c (add_stores): Preserve the value of the source even if
3652 we don't record the store.
3654 2014-01-06 Terry Guo <terry.guo@arm.com>
3656 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3658 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3661 * config/darwin.c (darwin_function_section): Adjust return values to
3662 correspond to optimisation changes made in r206070.
3664 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3666 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3667 from prefetch_block tune setting.
3668 (nocona_cost): Correct size of prefetch block to 64.
3670 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3672 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3673 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3674 used to save the static chain register in the computation of the offset
3675 from which the FP registers need to be restored.
3677 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3679 PR tree-optimization/59519
3680 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3681 ICE if get_current_def (current_new_name) is already non-NULL, as long
3682 as it is a phi result of some other phi in *new_exit_bb that has
3685 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3686 or vmovdqu* for misaligned_operand.
3687 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3688 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3689 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3690 aligned_mem for AVX512F masked aligned load and store builtins and for
3693 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3695 PR tree-optimization/59651
3696 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3697 Address range for negative step should be added by TYPE_SIZE_UNIT.
3699 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3701 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3702 overlapping registers also for registers other than the stack pointer.
3704 2014-01-03 Marek Polacek <polacek@redhat.com>
3707 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3710 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3713 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3716 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3717 (push splitter): Use <P:MODE_SIZE> instead of
3718 GET_MODE_SIZE (<P:MODE>mode).
3719 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3720 (mov -1, reg peephole2): Likewise.
3721 * config/i386/sse.md (*mov<mode>_internal,
3722 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3723 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3724 *<code><mode>3, *andnot<mode>3<mask_name>,
3725 <mask_codefor><code><mode>3<mask_name>): Likewise.
3726 * config/i386/subst.md (mask_mode512bit_condition,
3727 sd_mask_mode512bit_condition): Likewise.
3729 2014-01-02 Xinliang David Li <davidxl@google.com>
3731 PR tree-optimization/59303
3732 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3733 (dump_predicates): Better output format.
3734 (pred_equal_p): New function.
3735 (is_neq_relop_p): Ditto.
3736 (is_neq_zero_form_p): Ditto.
3737 (pred_expr_equal_p): Ditto.
3738 (pred_neg_p): Ditto.
3739 (simplify_pred): Ditto.
3740 (simplify_preds_2): Ditto.
3741 (simplify_preds_3): Ditto.
3742 (simplify_preds_4): Ditto.
3743 (simplify_preds): Ditto.
3745 (push_to_worklist): Ditto.
3746 (get_pred_info_from_cmp): Ditto.
3747 (is_degenerated_phi): Ditto.
3748 (normalize_one_pred_1): Ditto.
3749 (normalize_one_pred): Ditto.
3750 (normalize_one_pred_chain): Ditto.
3751 (normalize_preds): Ditto.
3752 (normalize_cond_1): Remove function.
3753 (normalize_cond): Ditto.
3754 (is_gcond_subset_of): Ditto.
3755 (is_subset_of_any): Ditto.
3756 (is_or_set_subset_of): Ditto.
3757 (is_and_set_subset_of): Ditto.
3758 (is_norm_cond_subset_of): Ditto.
3759 (pred_chain_length_cmp): Ditto.
3760 (convert_control_dep_chain_into_preds): Type change.
3761 (find_predicates): Ditto.
3762 (find_def_preds): Ditto.
3763 (destroy_predicates_vecs): Ditto.
3764 (find_matching_predicates_in_rest_chains): Ditto.
3765 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3766 (is_pred_expr_subset): Ditto.
3767 (is_pred_chain_subset_of): Ditto.
3768 (is_included_in): Ditto.
3769 (is_superset_of): Ditto.
3771 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3773 Update copyright years.
3775 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3777 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3778 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3779 config/arc/arc.md, config/arc/arc.opt,
3780 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3781 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3782 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3783 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3784 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3785 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3786 vtable-verify.c, vtable-verify.h: Use the standard form for the
3789 2014-01-02 Tobias Burnus <burnus@net-b.de>
3791 * gcc.c (process_command): Update copyright notice dates.
3792 * gcov-dump.c: Ditto.
3794 * doc/cpp.texi: Bump @copying's copyright year.
3795 * doc/cppinternals.texi: Ditto.
3796 * doc/gcc.texi: Ditto.
3797 * doc/gccint.texi: Ditto.
3798 * doc/gcov.texi: Ditto.
3799 * doc/install.texi: Ditto.
3800 * doc/invoke.texi: Ditto.
3802 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3804 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3806 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3808 * config/i386/sse.md (*mov<mode>_internal): Guard
3809 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3811 PR rtl-optimization/59647
3812 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3813 new_rtx into UNSIGNED_FLOAT rtxes.
3815 Copyright (C) 2014 Free Software Foundation, Inc.
3817 Copying and distribution of this file, with or without modification,
3818 are permitted in any medium without royalty provided the copyright
3819 notice and this notice are preserved.