1 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
3 * common/config/epiphany/epiphany-common.c
4 (epiphany_option_optimization_table): Enable section anchors by
5 default at -O1 or higher.
6 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
7 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
8 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
10 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
11 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
12 * config/epiphany/predicates.md (memclob_operand): New predicate.
13 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
14 Use memclob_operand predicate and X constraint for operand 3.
16 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
18 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
19 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
22 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
24 PR rtl-optimization/60651
25 * mode-switching.c (optimize_mode_switching): Make sure to emit
26 sets of a lower numbered entity before sets of a higher numbered
27 entity to a mode of the same or lower priority.
28 When creating a seginfo for a basic block that starts with a code
29 label, move the insertion point past the code label.
30 (new_seginfo): Document and enforce requirement that
31 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
32 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
33 * doc/tm.texi: Regenerate.
35 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
38 * config/arc/arc.c (arc_save_restore): Fix assert typo.
40 2013-04-11 Jakub Jelinek <jakub@redhat.com>
42 * BASE-VER: Set to 4.10.0.
44 2014-04-11 Tobias Burnus <burnus@net-b.de>
47 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
49 * doc/gcc.texi (Service): Update description in the @menu
50 * doc/invoke.texi (Option Summary): Remove misplaced and
53 2014-04-11 Steve Ellcey <sellcey@mips.com>
54 Jakub Jelinek <jakub@redhat.com>
57 * expr.c (convert_move): Use emit_store_flag_force instead of
58 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
61 2014-04-11 Richard Biener <rguenther@suse.de>
64 * varasm.c (assemble_alias): Avoid endless error reporting
65 recursion by setting TREE_ASM_WRITTEN.
67 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
69 * config/s390/s390.md: Add a splitter for NOT rtx.
71 2014-04-11 Jakub Jelinek <jakub@redhat.com>
73 PR rtl-optimization/60663
74 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
77 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
78 Jakub Jelinek <jakub@redhat.com>
81 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
84 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
87 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
88 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
89 ameliorating the cases where it can be.
91 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
94 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
96 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
97 (loadsync_<mode>): Change mode.
98 (load_quadpti, store_quadpti): New.
99 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
100 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
101 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
103 2014-04-09 Cong Hou <congh@google.com>
106 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
109 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
111 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
112 instead of vnor to exploit possible fusion opportunity in the
114 (altivec_expand_vec_perm_const_le): Likewise.
116 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
118 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
119 (loadsync_<mode>): Change mode.
120 (load_quadpti, store_quadpti): New.
121 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
122 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
124 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
127 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
128 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
129 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
131 2014-04-08 Richard Biener <rguenther@suse.de>
134 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
135 a 64bit widest int print double-int similar to on HWI64 hosts.
137 2014-04-08 Richard Biener <rguenther@suse.de>
139 PR tree-optimization/60785
140 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
141 default defs properly.
143 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
145 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
148 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
150 * ipa-devirt.c (maybe_record_node): When node is not recorded,
151 set completep to false rather than true.
153 2014-04-07 Douglas B Rupp <rupp@adacore.com>
156 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
157 ARM_TARGET2_DWARF_FORMAT.
159 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
162 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
163 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
166 2014-04-07 Richard Biener <rguenther@suse.de>
168 PR tree-optimization/60766
169 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
170 (may_eliminate_iv): Convert cand_value_at result to desired type.
172 2014-04-07 Jason Merrill <jason@redhat.com>
175 * common.opt (-fno-gnu-unique): Add.
176 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
178 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
180 * haifa-sched.c: Fix outdated function reference and minor
181 grammar errors in introductory comment.
183 2014-04-07 Richard Biener <rguenther@suse.de>
186 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
188 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
190 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
193 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
194 size accounting for thunks.
195 (pa_asm_output_mi_thunk): Use final_start_function() and
196 final_end_function() to output function start and end directives.
198 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
200 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
201 specific ISA/ feature information. Remove short_sp and errata_skip ds.
202 Add avr_device_specific_features enum to have device specific info.
203 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
204 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
205 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
206 updated device specific info.
207 * config/avr/avr-mcus.def: Merge device specific details to
209 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
211 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
212 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
213 assembler if RMW isa supported by current device.
214 * config/avr/genmultilib.awk: Update as device info structure changed.
215 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
217 2014-04-04 Cong Hou <congh@google.com>
219 PR tree-optimization/60656
220 * tree-vect-stmts.c (supportable_widening_operation):
221 Fix a bug that elements in a vector with vect_used_by_reduction
222 property are incorrectly reordered when the operation on it is not
223 consistant with the one in reduction operation.
225 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
227 PR rtl-optimization/60155
228 * gcse.c (record_set_data): New function.
229 (single_set_gcse): New function.
230 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
231 (hoist_code): Likewise.
232 (get_pressure_class_and_nregs): Likewise.
234 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
236 * explow.c (probe_stack_range): Emit a final optimization blockage.
238 2014-04-04 Anthony Green <green@moxielogic.com>
240 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
243 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
246 * lto-cgraph.c (input_overwrite_node): Check that partitioning
247 flags are set only during streaming.
248 * ipa.c (process_references, walk_polymorphic_call_targets,
249 symtab_remove_unreachable_nodes): Drop bodies of always inline
250 after early inlining.
251 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
253 2014-04-04 Jakub Jelinek <jakub@redhat.com>
254 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
257 * dwarf2out.c (const_ok_for_output_1): Reject expressions
260 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
263 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
265 (cortex_a53_fdivd): Likewise.
267 2014-04-04 Martin Jambor <mjambor@suse.cz>
270 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
272 * cgraph.c (clone_of_p): Also return true if thunks match.
273 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
274 cgraph_function_or_thunk_node and an obsolete comment.
275 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
277 (build_function_decl_skip_args): Likewise.
278 (set_new_clone_decl_and_node_flags): New function.
279 (duplicate_thunk_for_node): Likewise.
280 (redirect_edge_duplicating_thunks): Likewise.
281 (cgraph_clone_node): New parameter args_to_skip, pass it to
282 redirect_edge_duplicating_thunks which is called instead of
283 cgraph_redirect_edge_callee.
284 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
285 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
287 2014-04-04 Jeff Law <law@redhat.com>
290 * config/arm/predicates.md (const_int_I_operand): New predicate.
291 (const_int_M_operand): Similarly.
292 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
294 (insv_t2, extv_reg, extzv_t2): Likewise.
295 (load_multiple_with_writeback): Similarly for const_int_I_operand.
296 (pop_multiple_with_writeback_and_return): Likewise.
297 (vfp_pop_multiple_with_writeback): Likewise
299 2014-04-04 Richard Biener <rguenther@suse.de>
302 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
303 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
305 * gimplify.h (gimple_add_tmp_var_fn): Declare.
306 * gimplify.c (gimple_add_tmp_var_fn): New function.
307 * gimple-expr.h (create_tmp_reg_fn): Declare.
308 * gimple-expr.c (create_tmp_reg_fn): New function.
309 * gimple-low.c (record_vars_into): Don't change cfun.
310 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
311 code generation without cfun.
313 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
316 * Makefile.in (install-driver): Fix shell scripting.
318 2014-04-03 Cong Hou <congh@google.com>
320 PR tree-optimization/60505
321 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
322 threshold of number of iterations below which no vectorization will be
324 * tree-vect-loop.c (new_loop_vec_info):
325 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
326 * tree-vect-loop.c (vect_analyze_loop_operations):
327 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
328 * tree-vect-loop.c (vect_transform_loop):
329 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
330 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
331 of iterations of the loop and see if we should build the epilogue.
333 2014-04-03 Richard Biener <rguenther@suse.de>
335 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
337 (streamer_tree_cache_create): Adjust.
338 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
339 to allow optional nodes array.
340 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
341 (streamer_tree_cache_append): Likewise.
342 (streamer_tree_cache_create): Create nodes array optionally
343 as specified by parameter.
344 * lto-streamer-out.c (create_output_block): Avoid maintaining
345 the node array in the writer cache.
346 (DFS_write_tree): Remove assertion.
347 (produce_asm_for_decls): Free the out decl state hash table
349 * lto-streamer-in.c (lto_data_in_create): Adjust for
350 streamer_tree_cache_create prototype change.
352 2014-04-03 Richard Biener <rguenther@suse.de>
354 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
355 set TREE_CHAIN to NULL_TREE.
357 2014-04-03 Richard Biener <rguenther@suse.de>
359 PR tree-optimization/60740
360 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
361 over all GIMPLE_COND operands.
363 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
365 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
366 (Weffc++): Remove Scott's numbering, merge lists and reference
372 * class.c (accessible_nvdtor_p): New.
373 (check_bases): Don't check base destructor here ...
374 (check_bases_and_members): ... check them here. Trigger on
375 Wnon-virtual-dtor flag.
376 (finish_struct_1): Use accessible_nvdtor_p.
379 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
380 * g++.dg/warn/Wnvdtor-2.C: New.
381 * g++.dg/warn/Wnvdtor-3.C: New.
382 * g++.dg/warn/Wnvdtor-4.C: New.
383 * g++.dg/warn/Weff1.C: Delete.
384 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
385 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
387 2014-04-03 Nick Clifton <nickc@redhat.com>
389 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
392 2014-04-03 Martin Jambor <mjambor@suse.cz>
394 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
395 mention gcc_unreachable before failing.
396 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
399 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
402 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
403 code and instead mark the context inconsistent.
404 (possible_polymorphic_call_targets): For inconsistent contexts
405 return empty complete list.
407 2014-04-02 Anthony Green <green@moxielogic.com>
409 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
410 (extendqisi2, extendhisi2): Define.
411 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
412 (WCHAR_TYPE): Change to unsigned int.
414 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
416 PR tree-optimization/60733
417 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
418 insertion point for PHI candidates to be the end of the feeding
419 block for the PHI argument.
421 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
423 PR rtl-optimization/60650
424 * lra-constraints.c (process_alt_operands): Decrease reject for
425 earlyclobber matching.
427 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
429 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
431 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
433 * config/spu/spu.c (pad_bb): Do not crash when the last
434 insn is CODE_FOR_blockage.
436 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
438 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
439 lies outside the target mode.
441 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
444 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
445 software floating point or no floating point registers, do not
446 allow any type in the FPRs. Eliminate a test for SPE SIMD types
447 in GPRs that occurs after we tested for GPRs that would never be
450 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
451 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
452 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
453 specifically allow DDmode, since that does not use the SPE SIMD
456 2014-04-02 Richard Biener <rguenther@suse.de>
459 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
460 MODE_INTs. Properly use negv_optab.
461 (expand_abs): Likewise.
463 2014-04-02 Richard Biener <rguenther@suse.de>
466 * Makefile.in (install-driver): Guard extra installs with special
469 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
471 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
474 2014-04-01 Richard Henderson <rth@redhat.com>
477 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
478 alternative enabled before register allocation.
480 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
482 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
483 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
485 (nios2_large_got_address): Remove unneeded 'sym' parameter.
486 (nios2_got_address): Update nios2_large_got_address call site.
487 (nios2_delegitimize_address): New function.
488 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
489 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
490 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
492 2014-04-01 Martin Husemann <martin@duskware.de>
494 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
497 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
499 PR rtl-optimization/60604
500 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
501 check from register_operand.
502 (register_operand): Redefine in terms of general_operand.
503 (nonmemory_operand): Use register_operand for the non-constant cases.
505 2014-04-01 Richard Biener <rguenther@suse.de>
507 * gimple.h (struct gimple_statement_base): Align subcode to
510 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
512 * doc/invoke.texi (mapp-regs): Clarify.
514 2014-03-31 Ulrich Drepper <drepper@gmail.com>
516 * config/i386/avx512fintrin.h (__v32hi): Define type.
518 (_mm512_set1_epi8): Define.
519 (_mm512_set1_epi16): Define.
520 (_mm512_set4_epi32): Define.
521 (_mm512_set4_epi64): Define.
522 (_mm512_set4_pd): Define.
523 (_mm512_set4_ps): Define.
524 (_mm512_setr4_epi64): Define.
525 (_mm512_setr4_epi32): Define.
526 (_mm512_setr4_pd): Define.
527 (_mm512_setr4_ps): Define.
528 (_mm512_setzero_epi32): Define.
530 2014-03-31 Martin Jambor <mjambor@suse.cz>
533 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
534 callsite_arguments_match_p. Updated all callers. Also check types of
535 corresponding formal parameters and actual arguments.
536 (not_all_callers_have_enough_arguments_p) Renamed to
537 some_callers_have_mismatched_arguments_p.
539 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
541 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
543 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
546 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
549 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
551 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
552 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
554 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
555 Use FMAMODE_NOVF512 mode iterator.
556 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
557 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
558 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
559 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
561 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
562 Use VF_128_256 mode iterator.
563 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
566 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
568 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
569 static chain if needed.
571 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
574 * lra-constraints.c (index_part_to_reg): New.
575 (process_address): Use it.
577 2014-03-27 Jeff Law <law@redhat.com>
578 Jakub Jelinek <jakub@redhat.com>
581 * expr.c (do_tablejump): Use simplify_gen_binary rather than
582 gen_rtx_{PLUS,MULT} to build up the address expression.
584 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
585 creating non-canonical RTL.
587 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
590 * ipa-inline.c (want_inline_small_function_p): Short circuit large
591 functions; reorganize to make cheap checks first.
592 (inline_small_functions): Do not estimate growth when dumping;
594 * ipa-inline.h (inline_summary): Add min_size.
595 (growth_likely_positive): New function.
596 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
597 (set_cond_stmt_execution_predicate): Cleanup.
598 (estimate_edge_size_and_time): Compute min_size.
599 (estimate_calls_size_and_time): Likewise.
600 (estimate_node_size_and_time): Likewise.
601 (inline_update_overall_summary): Update min_size.
602 (do_estimate_edge_time): Likewise.
603 (do_estimate_edge_size): Update.
604 (do_estimate_edge_hints): Update.
605 (growth_likely_positive): New function.
607 2014-03-28 Jakub Jelinek <jakub@redhat.com>
610 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
611 also if addr has VOIDmode.
613 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
615 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
616 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
618 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
619 instructions as well as AdvancedSIMD loads.
621 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
623 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
624 Use crypto_aese type.
625 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
626 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
627 crypto_aese, crypto_aesmc. Move to types.md.
628 * config/arm/types.md (crypto_aes): Split into crypto_aese,
630 * config/arm/iterators.md (crypto_type): Likewise.
632 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
634 * cgraph.c: Include expr.h and tree-dfa.h.
635 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
638 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
641 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
642 regs from checking multi-reg pseudos.
644 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
646 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
648 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
650 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
651 if it would clobber the stack pointer, even temporarily.
653 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
655 * mode-switching.c: Make small adjustments to the top comment.
657 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
659 * config/rs6000/constraints.md (wD constraint): New constraint to
660 match the constant integer to get the top DImode/DFmode out of a
661 vector in a VSX register.
663 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
664 match the constant integer to get the top DImode/DFmode out of a
665 vector in a VSX register.
667 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
670 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
673 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
674 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
676 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
677 Optimize vec_extract of 64-bit values, where the value being
678 extracted is in the top word, where we can use scalar
679 instructions. Add direct move and store support. Combine the big
680 endian/little endian vector select load support into a single insn.
681 (vsx_extract_<mode>_internal1): Likewise.
682 (vsx_extract_<mode>_internal2): Likewise.
683 (vsx_extract_<mode>_load): Likewise.
684 (vsx_extract_<mode>_store): Likewise.
685 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
686 combined into vsx_extract_<mode>_load.
687 (vsx_extract_<mode>_one_le): Likewise.
689 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
690 define the top 64-bit vector element.
692 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
695 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
696 Document vec_vbpermq builtin.
699 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
700 enable use of xxsldwi and xxpermdi builtin functions.
701 (vec_xxpermdi): Likewise.
703 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
704 Document use of vec_xxsldwi and vec_xxpermdi builtins.
706 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
708 PR rtl-optimization/60650
709 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
711 (find_spills_for): New.
712 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
713 Spill all pseudos on the second iteration.
715 2014-03-27 Marek Polacek <polacek@redhat.com>
718 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
721 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
723 * config/s390/s390.c (s390_can_use_return_insn): Check for
724 call-saved FPRs on 31 bit.
726 2014-03-27 Jakub Jelinek <jakub@redhat.com>
729 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
730 if they need regimplification, just drop them instead of
731 calling gimple_regimplify_operands on them.
733 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
736 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
737 (aarch64_frame_pointer_required): Adjust logic.
738 (aarch64_can_eliminate): Adjust logic.
739 (aarch64_override_options_after_change): Adjust logic.
741 2014-03-27 Dehao Chen <dehao@google.com>
743 * ipa-inline.c (early_inliner): Update node's inline info.
745 2014-03-26 Dehao Chen <dehao@google.com>
747 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
748 compiler inserted conditional jumps for NAN float check.
750 2014-03-26 Jakub Jelinek <jakub@redhat.com>
752 * ubsan.h (ubsan_create_data): Change second argument's type
753 to const location_t *.
754 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
756 (ubsan_create_data): Change second argument to const location_t *PLOC.
757 Create Loc field whenever PLOC is non-NULL.
758 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
759 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
763 * real.c (real_to_integer2): Change type of low to UHWI.
765 2014-03-26 Tobias Burnus <burnus@net-b.de>
767 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
768 (CILK_SELF_SPECS): New define.
769 (driver_self_specs): Use it.
771 2014-03-26 Richard Biener <rguenther@suse.de>
773 * tree-pretty-print.c (percent_K_format): Implement special
774 case for LTO and its stripped down BLOCK tree.
776 2014-03-26 Jakub Jelinek <jakub@redhat.com>
779 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
781 * tree-vrp.c (simplify_internal_call_using_ranges): If only
782 one range is range_int_cst_p, but not both, at least optimize
783 addition/subtraction of 0 and multiplication by 0 or 1.
784 * gimple-fold.c (gimple_fold_call): Fold
785 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
786 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
787 INTEGER_CSTs, try to fold at least x * 0 and y - y.
789 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
791 PR rtl-optimization/60452
792 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
793 <case REG>: Return 1 for invalid offsets from the frame pointer.
795 2014-03-26 Marek Polacek <polacek@redhat.com>
798 * doc/extend.texi (C Extensions): Mention variable-length arrays in
801 2014-03-26 Marek Polacek <polacek@redhat.com>
804 * doc/extend.texi (Designated Inits): Describe what happens to omitted
807 2014-03-26 Marek Polacek <polacek@redhat.com>
810 * ira-color.c (update_conflict_hard_regno_costs): Perform the
811 multiplication in unsigned type.
813 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
815 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
817 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
819 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
821 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
824 * cif-code.def (UNREACHABLE) New code.
825 * ipa-inline.c (inline_small_functions): Skip edges to
826 __builtlin_unreachable.
827 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
828 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
829 predicate to __bulitin_unreachable.
830 (set_cond_stmt_execution_predicate): Fix issue when
831 invert_tree_comparison returns ERROR_MARK.
832 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
833 propagate to inline clones.
834 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
836 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
837 * cgraphclones.c (cgraph_clone_node): If call destination is already
838 ureachable, do not redirect it back.
839 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
842 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
844 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
845 Do not modify inline clones.
847 2014-03-25 Jakub Jelinek <jakub@redhat.com>
849 * config/i386/i386.md (general_sext_operand): New mode attr.
850 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
851 don't generate (sign_extend (const_int)).
852 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
853 operands[2]. Use We constraint instead of <i> and
854 <general_sext_operand> predicate instead of <general_operand>.
855 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
856 * config/i386/constraints.md (We): New constraint.
857 * config/i386/predicates.md (x86_64_sext_operand,
858 sext_operand): New predicates.
860 2014-03-25 Martin Jambor <mjambor@suse.cz>
863 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
864 inconsistent devirtualizations to __builtin_unreachable.
866 2014-03-25 Marek Polacek <polacek@redhat.com>
869 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
871 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
873 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
874 order of elements for big-endian.
876 2014-03-25 Richard Biener <rguenther@suse.de>
879 * gimplify-me.c (gimple_regimplify_operands): Update the
882 2014-03-25 Martin Jambor <mjambor@suse.cz>
885 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
886 (lto_output_varpool_node): Likewise.
887 (input_overwrite_node): Likewise.
888 (input_varpool_node): Likewise.
890 2014-03-25 Richard Biener <rguenther@suse.de>
892 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
895 2014-03-25 Jakub Jelinek <jakub@redhat.com>
897 * combine.c (simplify_compare_const): Add MODE argument.
898 Handle mode_width 0 as very large mode_width.
899 (try_combine, simplify_comparison): Adjust callers.
901 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
902 type to avoid signed integer overflow.
903 * explow.c (plus_constant): Likewise.
905 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
907 * doc/generic.texi: Correct typos.
909 2014-03-24 Tobias Burnus <burnus@net-b.de>
911 * doc/invoke.texi (-flto): Expand section about
912 using static libraries with LTO.
914 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
916 PR rtl-optimization/60501
917 * optabs.def (addptr3_optab): New optab.
918 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
919 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
920 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
922 * lra.c (emit_add3_insn): Use the addptr pattern if available.
924 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
926 2014-03-24 Ulrich Drepper <drepper@gmail.com>
928 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
931 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
932 (_mm256_undefined_ps): Define.
933 (_mm256_undefined_pd): Define.
934 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
935 (_mm_undefined_pd): Define.
936 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
937 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
938 (_mm512_undefined_ps): Define.
939 (_mm512_undefined_pd): Define.
940 Use _mm*_undefined_*.
941 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
943 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
945 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
946 (lshr_simd): DI mode added.
947 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
948 (aarch64_ushr_simddi): Likewise.
949 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
950 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
951 (vshrd_n_u64): Likewise.
953 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
955 * Makefile.in (s-macro_list): Depend on cc1.
957 2014-03-23 Teresa Johnson <tejohnson@google.com>
959 * ipa-utils.c (ipa_print_order): Use specified dump file.
961 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
963 PR rtl-optimization/60601
964 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
966 * gcc.c (eval_spec_function): Initialize save_growing_value.
968 2014-03-22 Jakub Jelinek <jakub@redhat.com>
971 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
972 code == MINUS_EXPR, never swap op0 with op1.
974 * toplev.c (init_local_tick): Avoid signed integer multiplication
976 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
977 shift by first operand's bitsize.
979 2014-03-21 Jakub Jelinek <jakub@redhat.com>
982 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
984 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
985 TARGET_ISA_64BIT_P(x).
987 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
989 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
990 pattern for vector nor instead of subtract from splat(-1).
991 (altivec_expand_vec_perm_const_le): Likewise.
993 2014-03-21 Richard Henderson <rth@twiddle.net>
996 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
997 related insns after epilogue_completed.
999 2014-03-21 Martin Jambor <mjambor@suse.cz>
1002 * cgraph.h (symtab_node): New flag body_removed.
1003 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1004 when removing bodies.
1005 * symtab.c (dump_symtab_base): Dump body_removed flag.
1006 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1007 had their bodies removed.
1009 2014-03-21 Martin Jambor <mjambor@suse.cz>
1012 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1015 2014-03-21 Richard Biener <rguenther@suse.de>
1017 PR tree-optimization/60577
1018 * tree-core.h (struct tree_base): Document nothrow_flag use
1019 in VAR_DECL_NONALIASED.
1020 * tree.h (VAR_DECL_NONALIASED): New.
1021 (may_be_aliased): Adjust.
1022 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1024 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1026 * expr.c (expand_expr_real_1): Remove outdated comment.
1028 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1031 * ira.c (adjust_cleared_regs): Call copy_rtx on
1032 *reg_equiv[REGNO (loc)].src_p before passing it to
1033 simplify_replace_fn_rtx.
1036 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1037 into CONST, put pic register as first operand of PLUS. Use
1038 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1040 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1042 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1044 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1046 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1047 around for store forwarding issue in the FPU on the UT699.
1048 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1049 loads and operations if -mfix-ut699 is specified.
1050 (divtf3_hq): Tweak attribute.
1051 (sqrttf2_hq): Likewise.
1053 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1055 * calls.c (store_one_arg): Remove incorrect const qualification on the
1056 type of the temporary.
1057 * cfgexpand.c (expand_return): Likewise.
1058 * expr.c (expand_constructor): Likewise.
1059 (expand_expr_real_1): Likewise.
1061 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1063 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1066 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1069 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1071 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1073 * config/arm/aarch-common-protos.h
1074 (alu_cost_table): Fix spelling of "extend".
1075 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1077 2014-03-19 Richard Biener <rguenther@suse.de>
1080 * tree-core.h (tree_type_common): Re-order pointer members
1081 to reduce recursion depth during GC walks.
1083 2014-03-19 Marek Polacek <polacek@redhat.com>
1086 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1087 before accessing it.
1089 2014-03-19 Richard Biener <rguenther@suse.de>
1092 * lto-streamer-in.c (input_function): In WPA stage do not drop
1095 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1097 PR tree-optimization/60559
1098 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1099 with build_zero_cst assignment.
1101 2014-03-18 Kai Tietz <ktietz@redhat.com>
1103 PR rtl-optimization/56356
1104 * sdbout.c (sdbout_parms): Verify that parms'
1105 incoming argument is valid.
1106 (sdbout_reg_parms): Likewise.
1108 2014-03-18 Richard Henderson <rth@redhat.com>
1111 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1112 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1113 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1115 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1117 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1118 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1119 Italicize plugin event names in description. Explain that
1120 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1121 Remind that no GCC functions should be called after PLUGIN_FINISH.
1122 Explain what pragmas with expansion are.
1124 2014-03-18 Martin Liska <mliska@suse.cz>
1126 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1127 gimple call statement is update.
1128 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1129 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1131 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1134 * ubsan.c (ubsan_instrument_unreachable): Call
1135 initialize_sanitizer_builtins.
1136 (ubsan_pass): Likewise.
1139 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1140 varpool_finalize_decl instead of rest_of_decl_compilation.
1142 2014-03-18 Richard Biener <rguenther@suse.de>
1144 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1145 by using bitmap_and_compl instead of bitmap_and_compl_into.
1146 (df_rd_transfer_function): Likewise.
1148 2014-03-18 Richard Biener <rguenther@suse.de>
1150 * doc/lto.texi (fresolution): Fix typo.
1152 2014-03-18 Richard Biener <rguenther@suse.de>
1154 * doc/invoke.texi (flto): Update for changes in 4.9.
1156 2014-03-18 Richard Biener <rguenther@suse.de>
1158 * doc/loop.texi: Remove section on the removed lambda framework.
1159 Update loop docs with recent changes in preserving loop structure.
1161 2014-03-18 Richard Biener <rguenther@suse.de>
1163 * doc/lto.texi (-fresolution): Document.
1165 2014-03-18 Richard Biener <rguenther@suse.de>
1167 * doc/contrib.texi: Adjust my name.
1169 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1172 * internal-fn.c: Include diagnostic-core.h.
1173 (expand_BUILTIN_EXPECT): New function.
1174 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1175 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1176 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1177 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1179 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1180 Revert 3 argument __builtin_expect code.
1181 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1182 * gimple-fold.c (gimple_fold_call): Likewise.
1183 * tree.h (fold_builtin_expect): New prototype.
1184 * builtins.c (build_builtin_expect_predicate): Add predictor
1185 argument, if non-NULL, create 3 argument __builtin_expect.
1186 (fold_builtin_expect): No longer static. Add ARG2 argument,
1187 pass it through to build_builtin_expect_predicate.
1188 (fold_builtin_2): Adjust caller.
1189 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1190 * internal-fn.def (BUILTIN_EXPECT): New.
1192 2014-03-18 Tobias Burnus <burnus@net-b.de>
1195 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1196 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1197 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1199 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1202 * predict.c (combine_predictions_for_bb): Fix up formatting.
1203 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1204 fill what it points to if non-NULL.
1205 (tree_predict_by_opcode): Adjust caller, use the predictor.
1206 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1208 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1210 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1211 proper constant for the store mode.
1213 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1215 * symtab.c (change_decl_assembler_name): Fix transparent alias
1218 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1220 * config/aarch64/aarch64.c: Correct the comments about the
1221 aarch64 stack layout.
1223 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1225 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1226 check for GF_OMP_FOR_KIND_FOR.
1228 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1230 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1231 ymm and zmm register names.
1233 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1236 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1237 note creation for the 2010-08-31 changes.
1239 2014-03-17 Marek Polacek <polacek@redhat.com>
1242 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1243 as -fno-tree-loop-vectorize.
1244 (expand_omp_simd): Likewise.
1246 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1248 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1249 (eligible_for_call_delay): New prototype.
1250 * config/sparc/sparc.c (tls_call_delay): Rename into...
1251 (eligible_for_call_delay): ...this. Return false if the instruction
1252 cannot be put in the delay slot of a branch.
1253 (eligible_for_restore_insn): Simplify.
1254 (eligible_for_return_delay): Return false if the instruction cannot be
1255 put in the delay slot of a branch and simplify.
1256 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1257 put in the delay slot of a branch.
1258 * config/sparc/sparc.md (fix_ut699): New attribute.
1259 (tls_call_delay): Delete.
1260 (in_call_delay): Reimplement.
1261 (eligible_for_sibcall_delay): Rename into...
1262 (in_sibcall_delay): ...this.
1263 (eligible_for_return_delay): Rename into...
1264 (in_return_delay): ...this.
1265 (in_branch_delay): Reimplement.
1266 (in_uncond_branch_delay): Delete.
1267 (in_annul_branch_delay): Delete.
1269 2014-03-14 Richard Henderson <rth@redhat.com>
1272 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1273 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1274 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1275 (floathi splitters): Remove.
1276 (float<SWI48x>xf2): New pattern.
1277 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1278 code that tried to handle DImode for 32-bit, but which was excluded
1279 by the pattern's condition. Drop allocation of stack temporary.
1280 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1281 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1282 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1283 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1284 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1285 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1286 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1287 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1288 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1289 (*float<SWI48x><X87MODEF>2_i387): Remove.
1290 (all float _with_temp splitters): Remove.
1291 (*float<SWI48x><MODEF>2_i387): New pattern.
1292 (*float<SWI48><MODEF>2_sse): New pattern.
1293 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1294 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1296 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1297 Marek Polacek <polacek@redhat.com>
1300 * common.opt (dump_base_name_prefixed): New Variable.
1301 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1302 if x_dump_base_name_prefixed is already set, set it at the end.
1304 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1306 PR rtl-optimization/60508
1307 * lra-constraints.c (get_reload_reg): Add new parameter
1309 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1310 Pass the new parameter values.
1312 2014-03-14 Richard Biener <rguenther@suse.de>
1314 * common.opt: Revert unintented changes from r205065.
1317 2014-03-14 Richard Biener <rguenther@suse.de>
1320 * cfghooks.c (split_block): Properly adjust all loops the
1321 block was a latch of.
1323 2014-03-14 Martin Jambor <mjambor@suse.cz>
1326 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1329 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1332 * config/avr/avr.c (avr_set_current_function): Pass function name
1333 through default_strip_name_encoding before sanity checking instead
1334 of skipping the first char of the assembler name.
1336 2014-03-13 Richard Henderson <rth@redhat.com>
1339 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1340 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1341 * config/i386/i386-protos.h: Likewise.
1342 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1343 in the expander instead of a splitter.
1344 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1345 any possibility of requiring a memory.
1346 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1347 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1348 (fp branch splitters): Update for ix86_split_fp_branch.
1349 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1350 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1351 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1352 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1353 (*fop_<MODEF>_3_i387): Likewise.
1354 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1355 (splitters for the fop_* register patterns): Remove.
1356 (fscalexf4_i387): Rename from *fscalexf4_i387.
1357 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1359 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1361 PR tree-optimization/59779
1362 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1363 type for bitsize and maxsize instead of HOST_WIDE_INT.
1365 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1367 PR rtl-optimization/57320
1368 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1369 the CFG after thread_prologue_and_epilogue_insns.
1371 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1373 PR rtl-optimization/57189
1374 * lra-constraints.c (process_alt_operands): Disfavor spilling
1377 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1379 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1381 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1383 PR tree-optimization/59025
1385 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1386 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1388 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1391 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1392 calls of avr_out_plus_1.
1394 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1396 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1397 BB's single pred and update the father loop's latch info later.
1399 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1401 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1405 (VEC_base): Likewise.
1406 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1407 registers, we need to swap double words in little endian mode.
1409 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1410 to be a container mode for 128-bit integer operations added in ISA
1411 2.07. Unlike TImode and PTImode, the preferred register set is
1412 the Altivec/VMX registers for the 128-bit operations.
1414 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1416 (rs6000_split_128bit_ok_p): Likewise.
1418 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1419 macros for creating ISA 2.07 normal and overloaded builtin
1420 functions with 3 arguments.
1421 (BU_P8V_OVERLOAD_3): Likewise.
1422 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1423 for use as overloaded functions.
1424 (VPERM_1TI_UNS): Likewise.
1425 (VSEL_1TI): Likewise.
1426 (VSEL_1TI_UNS): Likewise.
1427 (ST_INTERNAL_1ti): Likewise.
1428 (LD_INTERNAL_1ti): Likewise.
1429 (XXSEL_1TI): Likewise.
1430 (XXSEL_1TI_UNS): Likewise.
1431 (VPERM_1TI): Likewise.
1432 (VPERM_1TI_UNS): Likewise.
1433 (XXPERMDI_1TI): Likewise.
1434 (SET_1TI): Likewise.
1435 (LXVD2X_V1TI): Likewise.
1436 (STXVD2X_V1TI): Likewise.
1437 (VEC_INIT_V1TI): Likewise.
1438 (VEC_SET_V1TI): Likewise.
1439 (VEC_EXT_V1TI): Likewise.
1440 (EQV_V1TI): Likewise.
1441 (NAND_V1TI): Likewise.
1442 (ORC_V1TI): Likewise.
1443 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1444 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1446 (VADDUQM): Likewise.
1447 (VSUBCUQ): Likewise.
1448 (VADDEUQM): Likewise.
1449 (VADDECUQ): Likewise.
1450 (VSUBEUQM): Likewise.
1451 (VSUBECUQ): Likewise.
1453 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1454 __int128_t and __uint128_t types.
1455 (__uint128_type): Likewise.
1456 (altivec_categorize_keyword): Add support for vector __int128_t,
1457 vector __uint128_t, vector __int128, and vector unsigned __int128
1458 as a container type for TImode operations that need to be done in
1459 VSX/Altivec registers.
1460 (rs6000_macro_to_expand): Likewise.
1461 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1462 to support 128-bit integer instructions vaddcuq, vadduqm,
1463 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1464 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1466 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1467 for V1TImode, and set up preferences to use VSX/Altivec registers.
1468 Setup VSX reload handlers.
1469 (rs6000_debug_reg_global): Likewise.
1470 (rs6000_init_hard_regno_mode_ok): Likewise.
1471 (rs6000_preferred_simd_mode): Likewise.
1472 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1473 (easy_altivec_constant): Likewise.
1474 (output_vec_const_move): Likewise.
1475 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1477 (rs6000_expand_vector_extract): Likewise.
1478 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1480 (rs6000_const_vec): Add support for V1TImode.
1481 (rs6000_emit_le_vsx_load): Swap double words when loading or
1482 storing TImode/V1TImode.
1483 (rs6000_emit_le_vsx_store): Likewise.
1484 (rs6000_emit_le_vsx_move): Likewise.
1485 (rs6000_emit_move): Add support for V1TImode.
1486 (altivec_expand_ld_builtin): Likewise.
1487 (altivec_expand_st_builtin): Likewise.
1488 (altivec_expand_vec_init_builtin): Likewise.
1489 (altivec_expand_builtin): Likewise.
1490 (rs6000_init_builtins): Add support for V1TImode type. Add
1491 support for ISA 2.07 128-bit integer builtins. Define type names
1492 for the VSX/Altivec vector types.
1493 (altivec_init_builtins): Add support for overloaded vector
1494 functions with V1TImode type.
1495 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1496 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1498 (rs6000_split_128bit_ok_p): Likewise.
1499 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1500 __int128_t and vector __uint128_t.
1502 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1503 and mode attributes.
1510 (VS_scalar): Likewise.
1511 (VS_double): Likewise.
1512 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1514 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1515 we support the ISA 2.07 128-bit integer arithmetic instructions.
1516 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1517 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1518 and TImode types for use with the builtin functions.
1519 (V1TI_type_node): Likewise.
1520 (unsigned_V1TI_type_node): Likewise.
1521 (intTI_type_internal_node): Likewise.
1522 (uintTI_type_internal_node): Likewise.
1524 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1525 128-bit builtin functions.
1526 (UNSPEC_VADDEUQM): Likewise.
1527 (UNSPEC_VADDECUQ): Likewise.
1528 (UNSPEC_VSUBCUQ): Likewise.
1529 (UNSPEC_VSUBEUQM): Likewise.
1530 (UNSPEC_VSUBECUQ): Likewise.
1531 (VM): Add V1TImode to vector mode iterators.
1533 (VI_unit): Likewise.
1534 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1535 (altivec_vaddcuq): Likewise.
1536 (altivec_vsubuqm): Likewise.
1537 (altivec_vsubcuq): Likewise.
1538 (altivec_vaddeuqm): Likewise.
1539 (altivec_vaddecuq): Likewise.
1540 (altivec_vsubeuqm): Likewise.
1541 (altivec_vsubecuq): Likewise.
1543 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1545 (BOOL_128): Likewise.
1546 (BOOL_REGS_OUTPUT): Likewise.
1547 (BOOL_REGS_OP1): Likewise.
1548 (BOOL_REGS_OP2): Likewise.
1549 (BOOL_REGS_UNARY): Likewise.
1550 (BOOL_REGS_AND_CR0): Likewise.
1552 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1553 128-bit integer builtin support.
1554 (vec_vadduqm): Likewise.
1555 (vec_vaddecuq): Likewise.
1556 (vec_vaddeuqm): Likewise.
1557 (vec_vsubecuq): Likewise.
1558 (vec_vsubeuqm): Likewise.
1559 (vec_vsubcuq): Likewise.
1560 (vec_vsubuqm): Likewise.
1562 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1563 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1564 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1565 128-bit integer add/subtract to ISA 2.07.
1567 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1569 * config/arc/arc.c (arc_predicate_delay_insns):
1570 Fix third argument passed to conditionalize_nonjump.
1572 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1574 * config/aarch64/aarch64-builtins.c
1575 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1576 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1577 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1578 instead of __builtin_lfloor.
1579 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1581 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1583 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1584 (tree_ssa_ifcombine_bb_1): New function.
1585 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1586 is an empty forwarder block to then_bb or vice versa and then_bb
1587 and else_bb are effectively swapped.
1589 2014-03-12 Christian Bruel <christian.bruel@st.com>
1592 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1593 REG_CFA_DEF_CFA note.
1594 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1595 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1597 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1599 PR tree-optimization/60454
1600 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1602 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1604 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1605 Do not define target_cpu_default2 to generic.
1606 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1607 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1608 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1610 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1611 Marc Glisse <marc.glisse@inria.fr>
1613 PR tree-optimization/60502
1614 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1615 instead of build_low_bits_mask.
1617 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1620 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1621 if there are multiple uses, but op doesn't live on E edge.
1622 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1623 clobber stmts before __builtin_unreachable.
1625 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1627 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1628 hard_frame_pointer_rtx.
1629 * cse.c (cse_insn): Remove volatile check.
1630 * cselib.c (cselib_process_insn): Likewise.
1631 * dse.c (scan_insn): Likewise.
1633 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1635 * config/arc/arc.c (conditionalize_nonjump): New function,
1637 (arc_ifcvt): ... this.
1638 (arc_predicate_delay_insns): Use it.
1640 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1642 * config/arc/predicates.md (extend_operand): During/after reload,
1643 allow const_int_operand.
1644 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1645 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1646 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1648 (umulsi3_highpart_i): Likewise.
1650 2014-03-11 Richard Biener <rguenther@suse.de>
1652 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1653 Add asserts to guard possible wrong-code bugs.
1655 2014-03-11 Richard Biener <rguenther@suse.de>
1657 PR tree-optimization/60429
1658 PR tree-optimization/60485
1659 * tree-ssa-structalias.c (set_union_with_increment): Properly
1660 take into account all fields that overlap the shifted vars.
1661 (do_sd_constraint): Likewise.
1662 (do_ds_constraint): Likewise.
1663 (get_constraint_for_ptr_offset): Likewise.
1665 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1667 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1668 (nios2_compute_frame_layout):
1669 Add calculation of cfun->machine->fp_save_offset.
1670 (nios2_expand_prologue): Correct setting of frame pointer register
1672 (nios2_expand_epilogue): Update recovery of stack pointer from
1673 frame pointer accordingly.
1674 (nios2_initial_elimination_offset): Update calculation of offset
1675 for eliminating to HARD_FRAME_POINTER_REGNUM.
1677 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1680 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1681 cgraph_get_create_node on VAR_DECLs.
1683 2014-03-10 Richard Biener <rguenther@suse.de>
1686 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1688 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1690 * config/vms/vms.opt (vms_float_format): New variable.
1692 2014-03-08 Tobias Burnus <burnus@net-b.de>
1694 * doc/invoke.texi (-fcilkplus): Update implementation status.
1696 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1697 Richard Biener <rguenther@suse.de>
1699 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1700 consistently accross all TUs.
1701 (run_gcc): Enable -fshort-double automatically at link at link-time
1702 and disallow override.
1704 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1707 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1708 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1709 if they can't be used.
1711 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1713 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1714 for Solaris 11/x86 ld.
1715 * configure: Regenerate.
1717 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1719 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1720 (LIB_TLS_SPEC): Save as ld_tls_libs.
1721 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1722 (HAVE_AS_IX86_TLSLDM): New test.
1723 * configure, config.in: Regenerate.
1724 * config/i386/i386.c (legitimize_tls_address): Fall back to
1725 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1726 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1727 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1728 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1730 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1732 * common.opt (fira-loop-pressure): Mark as optimization.
1734 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1736 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1737 an OpenMP mappable type.
1739 2014-03-06 Matthias Klose <doko@ubuntu.com>
1741 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1742 MULTILIB_OSDIRNAMES is not defined.
1744 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1745 Meador Inge <meadori@codesourcery.com>
1748 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1749 (arm_legitimize_address): Call legitimize_tls_address for any
1750 arm_tls_referenced_p expression, handle constant addend. Call it
1751 before testing for !TARGET_ARM.
1752 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1754 2014-03-06 Richard Biener <rguenther@suse.de>
1760 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1762 * tree-streamer.c (record_common_node): Assert we don't record
1763 nodes with type double.
1764 (preload_common_node): Skip type double, complex double and double
1765 pointer since it is now frontend dependent due to fshort-double option.
1767 2014-03-06 Richard Biener <rguenther@suse.de>
1769 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1770 or -fno-lto is specified and the linker has full plugin support.
1771 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1772 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1773 * lto-wrapper.c (merge_and_complain): Merge compile-time
1774 optimization levels.
1775 (run_gcc): And pass it through to the link options.
1777 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1781 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1783 * cselib.c (remove_useless_values): Skip to avoid quadratic
1784 behavior if the condition moved from...
1785 (cselib_process_insn): ... here holds.
1787 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1790 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1791 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1794 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1795 (TM_H): Add x86-tune.def.
1797 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1799 * config/aarch64/aarch64.c (generic_tunings):
1800 Use cortexa57_extra_costs.
1802 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1805 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1806 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1807 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1810 2014-03-04 Heiher <r@hev.cc>
1812 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1813 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1815 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1817 * config/i386/predicates.md (const2356_operand): Change to ...
1818 (const2367_operand): ... this.
1819 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1821 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1822 (*avx512pf_scatterpf<mode>sf): Ditto.
1823 (avx512pf_scatterpf<mode>df): Ditto.
1824 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1825 (*avx512pf_scatterpf<mode>df): Ditto.
1826 * config/i386/i386.c (ix86_expand_builtin): Update
1827 incorrect hint operand error message.
1829 2014-03-04 Richard Biener <rguenther@suse.de>
1831 * lto-section-in.c (lto_get_section_data): Fix const cast.
1833 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1835 * tree-streamer.c (record_common_node): Assert we don't record
1836 nodes with type double.
1837 (preload_common_node): Skip type double, complex double and double
1838 pointer since it is now frontend dependent due to fshort-double option.
1840 2014-03-04 Richard Biener <rguenther@suse.de>
1843 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1844 (lto_input_toplevel_asms): Likewise.
1845 * lto-section-in.c (lto_get_section_data): Instead do it here
1848 2014-03-04 Richard Biener <rguenther@suse.de>
1850 PR tree-optimization/60382
1851 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1852 dead PHIs a reduction.
1854 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1856 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1858 (_mm_prefetch): Move out of GCC target("sse") pragma.
1859 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1860 GCC target("prfchw") pragma.
1861 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1863 * config/i386/i386.c (ix86_option_override_internal): Enable
1864 -mprfchw with -mprefetchwt1.
1866 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1868 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1871 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1873 * opts.h (CL_PCH_IGNORE): Define.
1874 * targhooks.c (option_affects_pch_p):
1875 Return false for options that have CL_PCH_IGNORE set.
1876 * opt-functions.awk: Process PchIgnore.
1877 * doc/options.texi: Document PchIgnore.
1879 * config/arc/arc.opt (misize): Add PchIgnore property.
1881 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1883 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1884 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1885 constraint on constants to permit them being loaded into
1886 GENERAL_REGS or BASE_REGS.
1888 2014-03-03 Nick Clifton <nickc@redhat.com>
1890 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1891 anti-cacnonical alternatives.
1892 (negandhi3_real): New pattern.
1893 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1895 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1897 * config/avr/avr-mcus.def: Remove atxmega16x1.
1898 * config/avr/avr-tables.opt: Regenerate.
1899 * config/avr/t-multilib: Regenerate.
1900 * doc/avr-mmcu.texi: Regenerate.
1902 2014-03-03 Tobias Grosser <tobias@grosser.es>
1903 Mircea Namolaru <mircea.namolaru@inria.fr>
1905 PR tree-optimization/58028
1906 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1909 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1911 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1912 not handled by recognizers.
1914 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1917 * function.c (expand_function_end): Don't emit
1918 clobber_return_register sequence if clobber_after is a BARRIER.
1919 * cfgexpand.c (construct_exit_block): Append instructions before
1920 return_label to prev_bb.
1922 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1924 * config/rs6000/constraints.md: Document reserved use of "wc".
1926 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1929 * ipa.c (function_and_variable_visibility): When dissolving comdat
1930 group, also set all symbols to local.
1932 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1937 2013-12-14 Jan Hubicka <jh@suse.cz>
1939 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1941 2014-03-02 Jon Beniston <jon@beniston.com>
1947 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1948 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1949 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1950 (simple_return, *simple_return): New patterns
1951 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1952 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1954 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1956 * dwarf2out.c (gen_subprogram_die): Tidy.
1958 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1961 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1962 (*mov_t_msb_neg_negc): ... this new insn.
1964 2014-02-28 Jason Merrill <jason@redhat.com>
1967 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1970 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1973 * dwarf2out.c (decltype_auto_die): New static.
1974 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1975 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1976 (is_cxx_auto): Likewise.
1978 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1980 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1981 we are not using general regs only.
1983 2014-02-28 Richard Biener <rguenther@suse.de>
1986 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1987 previous fix and only allow to remove trivial pre-headers
1988 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1989 (remove_forwarder_block): Properly update the latch of a loop.
1991 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1994 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1995 (cselib_preserved_hash_table): New.
1996 (preserve_constants_and_equivs): Move preserved vals to it.
1997 (cselib_find_slot): Look it up first.
1998 (cselib_init): Initialize it.
1999 (cselib_finish): Release it.
2000 (dump_cselib_table): Dump it.
2002 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2005 * cselib.c (remove_useless_values): Skip to avoid quadratic
2006 behavior if the condition moved from...
2007 (cselib_process_insn): ... here holds.
2009 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2012 * var-tracking.c (vt_initialize): Apply the same condition to
2013 preserve the CFA base value.
2015 2014-02-28 Joey Ye <joey.ye@arm.com>
2018 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2019 if reload in progress or completed.
2021 2014-02-28 Tobias Burnus <burnus@net-b.de>
2024 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2027 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2029 * doc/tm.texi.in (Condition Code Status): Update documention for
2030 relative locations of cc0-setter and cc0-user.
2032 2014-02-27 Jeff Law <law@redhat.com>
2034 PR rtl-optimization/52714
2035 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2036 into two independent simple sets, if I3 is a jump, ensure the
2037 pattern we place into I3 is a (set (pc) ...).
2039 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2040 Jeff Law <law@redhat.com>
2042 PR rtl-optimization/49847
2043 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2044 are in different blocks.
2045 * doc/tm.texi (Condition Code Status): Update documention for
2046 relative locations of cc0-setter and cc0-user.
2048 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2051 * lra.c (lra_emit_add): Check SUBREG too.
2053 2014-02-27 Andreas Schwab <schwab@suse.de>
2055 * config/m68k/m68k.c (m68k_option_override): Disable
2056 -flive-range-shrinkage for classic m68k.
2057 (m68k_override_options_after_change): Likewise.
2059 2014-02-27 Marek Polacek <polacek@redhat.com>
2062 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2063 -Wmaybe-uninitialized.
2065 2014-02-27 Alan Modra <amodra@gmail.com>
2068 * reload1.c (emit_input_reload_insns): When reload_override_in,
2069 set old to rl->in_reg when rl->in_reg is a subreg.
2071 2014-02-26 Richard Biener <rguenther@suse.de>
2074 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2076 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2078 * common/config/i386/predicates.md (const1256_operand): Remove.
2079 (const2356_operand): New.
2080 (const_1_to_2_operand): Remove.
2081 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2082 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2083 (*avx512pf_gatherpf<mode>sf): Ditto.
2084 (avx512pf_gatherpf<mode>df): Ditto.
2085 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2086 (*avx512pf_gatherpf<mode>df): Ditto.
2087 (avx512pf_scatterpf<mode>sf): Ditto.
2088 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2089 (*avx512pf_scatterpf<mode>sf): Ditto.
2090 (avx512pf_scatterpf<mode>df): Ditto.
2091 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2092 (*avx512pf_scatterpf<mode>df): Ditto.
2093 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2095 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2097 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2098 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2099 (_mm512_mask_testn_epi64_mask): Move to ...
2100 * config/i386/avx512cdintrin.h: Here.
2101 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2102 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2103 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2104 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2105 TARGET_AVX512F from TARGET_AVX512CD.
2107 2014-02-26 Richard Biener <rguenther@suse.de>
2110 * ipa.c (walk_polymorphic_call_targets): Properly guard
2111 call to inline_update_overall_summary.
2113 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2116 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2117 and latches only if requested. Fix latch if it is removed.
2118 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2119 LOOPS_HAVE_PREHEADERS.
2121 2014-02-25 Andrew Pinski <apinski@cavium.com>
2123 * builtins.c (expand_builtin_thread_pointer): Create a new target
2124 when the target is NULL.
2126 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2128 PR rtl-optimization/60317
2129 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2130 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2131 * lra-assigns.c: Include params.h.
2132 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2133 other reload pseudos considerations.
2135 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2137 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2138 to use canonical form for nor<mode>3.
2140 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2143 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2146 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2148 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2149 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2150 (ix86_handle_option): Handle OPT_mprefetchwt1.
2151 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2152 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2154 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2155 OPTION_MASK_ISA_PREFETCHWT1.
2156 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2157 (PTA_PREFETCHWT1): New.
2158 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2159 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2160 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2161 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2162 (*prefetch_avx512pf_<mode>_: Change into ...
2163 (*prefetch_prefetchwt1_<mode>: This.
2164 * config/i386/i386.opt (mprefetchwt1): New.
2165 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2166 (_mm_prefetch): Handle intent to write.
2167 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2169 2014-02-25 Richard Biener <rguenther@suse.de>
2172 * emit-rtl.c (mem_attrs_htab): Remove.
2173 (mem_attrs_htab_hash): Likewise.
2174 (mem_attrs_htab_eq): Likewise.
2175 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2176 (init_emit_once): Do not allocate mem_attrs_htab.
2178 2014-02-25 Richard Biener <rguenther@suse.de>
2181 * lto-opts.c (lto_write_options): Output non-explicit conservative
2182 -fwrapv, -fno-trapv and -fno-strict-overflow.
2183 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2184 (run_gcc): And pass them through.
2186 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2188 * sel-sched.c (calculate_new_fences): New parameter ptime.
2189 Calculate it as a maximum over all fence cycles.
2190 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2191 Print the final schedule timing when sched_verbose.
2193 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2195 PR rtl-optimization/60292
2196 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2197 bit fot the fence instruction.
2199 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2201 * calls.h: Fix typo in comment.
2203 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2205 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2206 adjusting offsetable addresses.
2208 2014-02-24 Guozhi Wei <carrot@google.com>
2210 * sparseset.h (sparseset_pop): Fix the wrong index.
2212 2014-02-24 Walter Lee <walt@tilera.com>
2214 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2215 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2217 * common/config/tilegx/tilegx-common.c
2218 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2219 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2221 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2222 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2223 (tilegx_gimplify_va_arg_expr): Handle big endian.
2224 (tilegx_expand_unaligned_load): Ditto.
2225 (tilegx_expand_unaligned_store): Ditto.
2226 (TARGET_RETURN_IN_MSB): New.
2227 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2228 (TARGET_ENDIAN_DEFAULT): New.
2229 (TARGET_BIG_ENDIAN): Handle big endian.
2230 (BYTES_BIG_ENDIAN): Ditto.
2231 (WORDS_BIG_ENDIAN): Ditto.
2232 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2235 * config/tilegx/tilegx.md (extv): Handle big endian.
2237 (insn_st<n>): Ditto.
2238 (insn_st<n>_add<bitsuffix>): Ditto.
2239 (insn_stnt<n>): Ditto.
2240 (insn_stnt<n>_add<bitsuffix>):Ditto.
2241 (vec_interleave_highv8qi): Handle big endian.
2242 (vec_interleave_highv8qi_be): New.
2243 (vec_interleave_highv8qi_le): New.
2244 (insn_v1int_h): Handle big endian.
2245 (vec_interleave_lowv8qi): Handle big endian.
2246 (vec_interleave_lowv8qi_be): New.
2247 (vec_interleave_lowv8qi_le): New.
2248 (insn_v1int_l): Handle big endian.
2249 (vec_interleave_highv4hi): Handle big endian.
2250 (vec_interleave_highv4hi_be): New.
2251 (vec_interleave_highv4hi_le): New.
2252 (insn_v2int_h): Handle big endian.
2253 (vec_interleave_lowv4hi): Handle big endian.
2254 (vec_interleave_lowv4hi_be): New.
2255 (vec_interleave_lowv4hi_le): New.
2256 (insn_v2int_l): Handle big endian.
2257 (vec_interleave_highv2si): Handle big endian.
2258 (vec_interleave_highv2si_be): New.
2259 (vec_interleave_highv2si_le): New.
2260 (insn_v4int_h): Handle big endian.
2261 (vec_interleave_lowv2si): Handle big endian.
2262 (vec_interleave_lowv2si_be): New.
2263 (vec_interleave_lowv2si_le): New.
2264 (insn_v4int_l): Handle big endian.
2265 * config/tilegx/tilegx.opt (mbig-endian): New option.
2266 (mlittle-endian): New option.
2267 * doc/install.texi: Document tilegxbe-linux.
2268 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2270 2014-02-24 Martin Jambor <mjambor@suse.cz>
2273 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2274 there are no parameter descriptors.
2276 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2278 PR rtl-optimization/60268
2279 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2280 initialization to ...
2281 (sched_rgn_init): ... here.
2282 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2284 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2286 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2289 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2291 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2294 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2296 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2297 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2299 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2301 * config/microblaze/predicates.md: Add cmp_op predicate.
2302 * config/microblaze/microblaze.md: Add branch_compare instruction
2303 which uses cmp_op predicate and emits cmp insn before branch.
2304 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2305 to microblaze_expand_conditional_branch and consolidate logic.
2306 (microblaze_expand_conditional_branch): emit branch_compare
2307 insn instead of handling cmp op separate from branch insn.
2309 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2311 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2314 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2316 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2317 define_insn with define_expand and new define_insn
2318 *altivec_lve<VI_char>x_internal.
2319 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2320 and new define_insn *altivec_stve<VI_char>x_internal.
2321 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2323 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2325 (altivec_expand_stvex_be): New function.
2327 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2329 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2330 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2331 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2332 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2334 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2337 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2338 instead of emit_move_insn.
2340 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2342 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2344 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2345 gen_altivec_vsumsws.
2347 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2349 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2350 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2351 (altivec_lvxl_<mode>): New define_expand incorporating
2352 -maltivec=be semantics where needed.
2353 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2354 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2355 semantics where needed.
2356 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2357 (altivec_stvx_<mode>): New define_expand incorporating
2358 -maltivec=be semantics where needed.
2359 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2360 VM2 iterator instead of V4SI.
2361 (altivec_stvxl_<mode>): New define_expand incorporating
2362 -maltivec=be semantics where needed.
2363 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2364 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2365 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2366 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2367 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2368 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2369 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2370 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2371 ALTIVEC_BUILTIN_STVXL.
2372 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2373 (altivec_expand_stvx_be): Likewise.
2374 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2375 (altivec_expand_lvx_be): Likewise.
2376 (altivec_expand_stvx_be): Likewise.
2377 (altivec_expand_builtin): Add cases for
2378 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2379 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2380 (altivec_init_builtins): Add definitions for
2381 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2382 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2384 2014-02-21 Catherine Moore <clm@codesourcery.com>
2386 * doc/invoke.texi (mvirt, mno-virt): Document.
2387 * config/mips/mips.opt (mvirt): New option.
2388 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2390 2014-02-21 Richard Biener <rguenther@suse.de>
2392 PR tree-optimization/60276
2393 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2394 (STMT_VINFO_MIN_NEG_DIST): New macro.
2395 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2396 STMT_VINFO_MIN_NEG_DIST.
2397 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2398 made for negative dependence distances still hold.
2400 2014-02-21 Richard Biener <rguenther@suse.de>
2403 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2404 DECL_INITIAL for globals not in the current function context.
2406 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2408 PR tree-optimization/56490
2409 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2410 * tree-ssa-uninit.c: Include params.h.
2411 (compute_control_dep_chain): Add num_calls argument, return false
2412 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2413 num_calls to recursive call.
2414 (find_predicates): Change dep_chain into normal array,
2415 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2416 variable and adjust compute_control_dep_chain caller.
2417 (find_def_preds): Likewise.
2419 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2421 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2422 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2424 2014-02-21 Nick Clifton <nickc@redhat.com>
2426 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2427 (pushhi1): Likewise.
2428 (popqi1): Add mode to pre_dec.
2431 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2433 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2434 mode for mask of V8SFmode permutation.
2436 2014-02-20 Richard Henderson <rth@redhat.com>
2439 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2440 a new pseudo for OLDVAL.
2442 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2445 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2446 gen_reg_rtx if d->testing_p.
2447 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2448 if d->testing_p and we will certainly return true.
2449 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2452 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2454 * emit-rtl.c (gen_reg_rtx): Assert that
2455 crtl->emit.regno_pointer_align_length is non-zero.
2457 2014-02-20 Richard Henderson <rth@redhat.com>
2460 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2461 on failure the store back into EXPECT.
2463 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2464 Sandra Loosemore <sandra@codesourcery.com>
2466 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2467 * config/nios2/nios2.c (nios2_function_profiler): Add
2468 -fPIC (flag_pic == 2) support.
2469 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2470 (nios2_large_offset_p): New function.
2471 (nios2_unspec_reloc_p): Move up position, update to use
2472 nios2_large_offset_p.
2473 (nios2_unspec_address): Remove function.
2474 (nios2_unspec_offset): New function.
2475 (nios2_large_got_address): New function.
2476 (nios2_got_address): Add large offset support.
2477 (nios2_legitimize_tls_address): Update usage of removed and new
2479 (nios2_symbol_binds_local_p): New function.
2480 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2481 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2482 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2483 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2484 processing for (const (unspec ...)).
2485 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2487 2014-02-20 Richard Biener <rguenther@suse.de>
2489 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2490 doing the substitution.
2491 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2493 2014-02-20 Martin Jambor <mjambor@suse.cz>
2496 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2497 info when checking whether lattices are bottom.
2499 2014-02-20 Richard Biener <rguenther@suse.de>
2502 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2505 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2508 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2509 parameter specifying the scaling.
2510 (inline_call): Update.
2511 (want_inline_recursively): Guard division by zero.
2512 (recursive_inlining): Update.
2513 * ipa-inline.h (clone_inlined_nodes): Update.
2515 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2518 * config/i386/i386.c (classify_argument): Pass structures of size
2519 64 bytes or less in register.
2521 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2522 Kirill Yukhin <kirill.yukhin@intel.com>
2524 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2525 (_mm_rcp28_round_ss): Ditto.
2526 (_mm_rsqrt28_round_sd): Ditto.
2527 (_mm_rsqrt28_round_ss): Ditto.
2528 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2529 (_mm_rcp14_round_ss): Ditto.
2530 (_mm_rsqrt14_round_sd): Ditto.
2531 (_mm_rsqrt14_round_ss): Ditto.
2532 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2533 the first input operand, get rid of match_dup.
2534 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2536 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2538 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2539 operand as the first input operand, set type attribute.
2540 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2542 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2543 operand as the first input operand, set type attribute.
2545 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2547 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2550 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2553 * config/i386/i386.c (construct_container): Remove TFmode check
2554 for X86_64_INTEGER_CLASS.
2556 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2559 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2560 only when -Wpsabi is enabled.
2562 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2565 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2566 passing arrays in registers are the same as for structs, so remove the
2567 special case for them.
2569 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2571 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2572 destination type, extract only the valid bits if the source type is not
2573 integral and has a different mode.
2575 2014-02-19 Richard Biener <rguenther@suse.de>
2578 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2581 2014-02-19 Richard Biener <rguenther@suse.de>
2584 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2585 (ipa_modify_call_arguments): Emit an argument load explicitely and
2586 preserve virtual SSA form there and for the replacement call.
2587 Do not update SSA form nor free dominance info.
2589 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2591 * ipa.c (function_and_variable_visibility): Also clear WEAK
2592 flag when disolving COMDAT_GROUP.
2594 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2596 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2597 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2599 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2600 do more sanity checks.
2601 (detect_type_change): Return true when giving up early.
2602 (compute_complex_assign_jump_func): Fix type parameter of
2603 ipa_set_ancestor_jf.
2604 (compute_complex_ancestor_jump_func): Likewise.
2605 (update_jump_functions_after_inlining): Fix updating of
2607 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2609 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2611 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2612 inline clones when edge disappears.
2614 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2617 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2618 Split 64-bit moves into 2 patterns. Do not allow the use of
2619 direct move for TDmode in little endian, since the decimal value
2620 has little endian bytes within a word, but the 64-bit pieces are
2621 ordered in a big endian fashion, and normal subreg's of TDmode are
2623 (mov<mode>_64bit_dm): Likewise.
2624 (movtd_64bit_nodm): Likewise.
2626 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2628 PR tree-optimization/60174
2629 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2630 statement of an SSA_NAME that occurs in an abnormal PHI node.
2632 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2635 * final.c (SEEN_BB): Remove.
2636 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2637 (final_scan_insn): Don't force_source_line on second
2638 NOTE_INSN_BASIC_BLOCK.
2640 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2643 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2644 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2645 (type_natural_mode): Warn ABI change when %zmm register is not
2646 available for AVX512F vector value passing.
2648 2014-02-18 Kai Tietz <ktietz@redhat.com>
2651 * config/i386/i386.c (ix86_expand_prologue): Use value in
2652 rax register as displacement when restoring %r10 or %rax.
2653 Fix wrong offset when restoring both registers.
2655 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2657 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2658 assertion with conditional return.
2660 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2661 Uros Bizjak <ubizjak@gmail.com>
2664 * config/i386/driver-i386.c (host_detect_local_cpu): If
2665 YMM state is not saved by the OS, also clear has_f16c. Move
2666 CPUID 0x80000001 handling before YMM state saving checking.
2668 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2670 PR rtl-optimization/58960
2671 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2672 factored out from ...
2673 (sched_init): ... here.
2674 (free_global_sched_pressure_data): New, factored out from ...
2675 (sched_finish): ... here.
2676 * sched-int.h (free_global_sched_pressure_data): Declare.
2677 * sched-rgn.c (nr_regions_initial): New static global.
2678 (haifa_find_rgns): Initialize it.
2679 (schedule_region): Disable sched-pressure for the newly
2682 2014-02-17 Richard Biener <rguenther@suse.de>
2684 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2685 release SSA defs of pattern stmts.
2687 2014-02-17 Richard Biener <rguenther@suse.de>
2689 * tree-inline.c (expand_call_inline): Release the virtual
2690 operand defined by the call we are about to inline.
2692 2014-02-17 Richard Biener <rguenther@suse.de>
2694 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2696 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2697 Ilya Tocar <ilya.tocar@intel.com>
2699 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2700 arguments order in builtin.
2701 (_mm512_permutexvar_epi64): Ditto.
2702 (_mm512_mask_permutexvar_epi64): Ditto
2703 (_mm512_maskz_permutexvar_epi32): Ditto
2704 (_mm512_permutexvar_epi32): Ditto
2705 (_mm512_mask_permutexvar_epi32): Ditto
2707 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2709 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2710 (p8_vmrgow): Likewise.
2712 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2714 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2717 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2720 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2721 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2722 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2723 using direct move instructions on ISA 2.07. Also adjust
2724 instruction length for 64-bit.
2725 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2726 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2728 2014-02-15 Alan Modra <amodra@gmail.com>
2732 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2733 find_replacement on parts of insn rtl that might be reloaded.
2735 2014-02-15 Richard Biener <rguenther@suse.de>
2737 PR tree-optimization/60183
2738 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2739 (tree_ssa_phiprop): Calculate and free post-dominators.
2741 2014-02-14 Jeff Law <law@redhat.com>
2743 PR rtl-optimization/60131
2744 * ree.c (get_extended_src_reg): New function.
2745 (combine_reaching_defs): Use it rather than assuming location of REG.
2746 (find_and_remove_re): Verify first operand of extension is
2747 a REG before adding the insns to the copy list.
2749 2014-02-14 Roland McGrath <mcgrathr@google.com>
2751 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2752 * configure: Regenerated.
2753 * config.in: Regenerated.
2754 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2755 instead of ASM_SHORT.
2757 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2758 Richard Earnshaw <rearnsha@arm.com>
2760 PR rtl-optimization/59535
2761 * lra-constraints.c (process_alt_operands): Encourage alternative
2762 when unassigned pseudo class is superset of the alternative class.
2763 (inherit_reload_reg): Don't inherit when optimizing for code size.
2764 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2765 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2766 modes not less than 4 for Thumb1.
2768 2014-02-14 Kyle McMartin <kyle@redhat.com>
2771 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2773 2014-02-14 Richard Biener <rguenther@suse.de>
2775 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2776 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2777 Do not strip INDIRECT_REFs.
2779 2014-02-14 Richard Biener <rguenther@suse.de>
2782 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2783 DECL_FUNCTION_SPECIFIC_TARGET.
2784 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2785 * tree-streamer-out.c (pack_ts_target_option): Remove.
2786 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2787 (write_ts_function_decl_tree_pointers): Do not stream
2788 DECL_FUNCTION_SPECIFIC_TARGET.
2789 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2790 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2791 (lto_input_ts_function_decl_tree_pointers): Do not stream
2792 DECL_FUNCTION_SPECIFIC_TARGET.
2794 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2796 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2797 (get_initial_def_for_induction, vectorizable_induction): Ignore
2798 debug stmts when looking for exit_phi.
2799 (vectorizable_live_operation): Fix up condition.
2801 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2803 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2804 nreverse() because it changes the content of original tree list.
2806 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2808 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2809 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2811 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2813 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2814 GNU coding standards.
2816 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2819 * dwarf2out.c (gen_subprogram_die): Don't call
2820 add_calling_convention_attribute if subr_die is old_die.
2822 2014-02-13 Sharad Singhai <singhai@google.com>
2824 * doc/optinfo.texi: Fix order of nodes.
2826 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2828 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2829 operands[2], not operands[3].
2831 2014-02-13 Richard Biener <rguenther@suse.de>
2834 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2835 mention the possibility of an in-tree build.
2836 (CLooG): Update recommended version to 0.18.1, mention the
2837 possibility of an in-tree build and clarify that the ISL
2838 bundled with CLooG does not work.
2840 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2843 * expr.c (compress_float_constant): If x is a hard register,
2844 extend into a pseudo and then move to x.
2846 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2848 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2849 caused by bad second argument to warning_at() with -mhotpatch and
2850 nested functions (e.g. with gfortran).
2852 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2854 * opts.c (option_name): Remove "enabled by default" rider.
2856 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2858 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2860 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2861 Uros Bizjak <ubizjak@gmail.com>
2864 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2865 * configure: Regenerated.
2867 2014-02-12 Richard Biener <rguenther@suse.de>
2869 * vec.c (vec_prefix::calculate_allocation): Move as
2870 inline variant to vec.h.
2871 (vec_prefix::calculate_allocation_1): New out-of-line version.
2872 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2873 (vec_prefix::m_has_auto_buf): Rename to ...
2874 (vec_prefix::m_using_auto_storage): ... this.
2875 (vec_prefix::calculate_allocation): Inline the easy cases
2876 and dispatch to calculate_allocation_1 which doesn't need the
2878 (va_heap::reserve): Use gcc_checking_assert.
2879 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2880 m_using_auto_storage.
2881 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2883 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2884 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2885 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2887 2014-02-12 Richard Biener <rguenther@suse.de>
2889 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2890 when we found a dependence.
2892 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2894 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2896 (maybe_fold_stmt): ... into this new function.
2897 * omp-low.c (lower_omp): Update comment.
2899 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2902 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2905 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2907 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2908 identifiers in comments.
2909 (cortexa53_extra_costs): Likewise.
2910 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2911 (cortexa7_extra_costs): Likewise.
2912 (cortexa12_extra_costs): Likewise.
2913 (cortexa15_extra_costs): Likewise.
2914 (v7m_extra_costs): Likewise.
2916 2014-02-12 Richard Biener <rguenther@suse.de>
2919 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2920 of posix_memalign being successful.
2921 (lower_stmt): Restrict lowering of posix_memalign to when
2922 -ftree-bit-ccp is enabled.
2924 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2926 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2928 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2930 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2932 PR rtl-optimization/60116
2933 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2934 other_insn once the combination has been validated.
2936 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2939 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2941 * ipa-devirt.c: Include demangle.h
2942 (odr_violation_reported): New static variable.
2943 (add_type_duplicate): Update odr_violations.
2944 (maybe_record_node): Add completep parameter; update it.
2945 (record_target_from_binfo): Add COMPLETEP parameter;
2946 update it as needed.
2947 (possible_polymorphic_call_targets_1): Likewise.
2948 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2949 rename FINAL to COMPLETE.
2950 (record_targets_from_bases): Sanity check we found the binfo;
2951 fix COMPLETEP updating.
2952 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2953 parameter, fix computing of COMPLETEP.
2954 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2955 at LTO time do demangling.
2956 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2957 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2959 (gimple_get_virt_method_for_binfo): Likewise.
2960 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2961 gimple_get_virt_method_for_vtable): Update prototypes.
2963 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2966 * genautomata.c (add_presence_absence): Fix typo with
2967 {final_}presence_list.
2969 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2972 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2973 for VSX/Altivec vectors that land in GPR registers.
2975 2014-02-11 Richard Henderson <rth@redhat.com>
2976 Jakub Jelinek <jakub@redhat.com>
2979 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2980 around drhs if type conversion to lacc->type is not useless.
2982 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2984 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2986 (cortex-a57.cortex-a53): Likewise.
2987 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2989 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2991 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2994 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2996 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2997 add_options_for_arm_vfp3.
2999 2014-02-11 Jeff Law <law@redhat.com>
3002 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3003 object with an undesirable mode.
3005 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3008 * config/i386/sol2-9.h: New file.
3009 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3010 *-*-solaris2.9*): Use it.
3012 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3014 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3015 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3017 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3019 * config/microblaze/microblaze.c: Extend mcpu version format
3021 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3023 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3025 2014-02-10 Richard Henderson <rth@redhat.com>
3028 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3029 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3030 ms-abi vs -mno-accumulate-outgoing-args.
3031 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3032 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3035 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3038 * cfgexpand.c (expand_asm_operands): Attach source location to
3039 ASM_INPUT rtx objects.
3040 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3042 2014-02-10 Nick Clifton <nickc@redhat.com>
3044 * config/mn10300/mn10300.c (popcount): New function.
3045 (mn10300_expand_prologue): Include saved registers in stack usage
3048 2014-02-10 Jeff Law <law@redhat.com>
3051 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3052 when changing the SET_DEST of a prior insn to avoid an input reload.
3054 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3056 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3057 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3058 -mcall-openbsd, or -mcall-linux.
3059 (CC1_ENDIAN_BIG_SPEC): Remove.
3060 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3061 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3062 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3063 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3064 and %cc1_endian_default.
3065 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3067 2014-02-10 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/60115
3070 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3071 MEM_REF handling. Properly verify that the accesses are not
3072 out of the objects bound.
3074 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3076 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3079 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3081 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3082 proper constants and fix formatting.
3083 (possible_polymorphic_call_targets): Fix formatting.
3085 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3086 Ilya Tocar <ilya.tocar@intel.com>
3088 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3089 (_mm512_loadu_epi32): Renamed into...
3090 (_mm512_loadu_si512): This.
3091 (_mm512_storeu_epi32): Renamed into...
3092 (_mm512_storeu_si512): This.
3093 (_mm512_maskz_ceil_ps): Removed.
3094 (_mm512_maskz_ceil_pd): Ditto.
3095 (_mm512_maskz_floor_ps): Ditto.
3096 (_mm512_maskz_floor_pd): Ditto.
3097 (_mm512_floor_round_ps): Ditto.
3098 (_mm512_floor_round_pd): Ditto.
3099 (_mm512_ceil_round_ps): Ditto.
3100 (_mm512_ceil_round_pd): Ditto.
3101 (_mm512_mask_floor_round_ps): Ditto.
3102 (_mm512_mask_floor_round_pd): Ditto.
3103 (_mm512_mask_ceil_round_ps): Ditto.
3104 (_mm512_mask_ceil_round_pd): Ditto.
3105 (_mm512_maskz_floor_round_ps): Ditto.
3106 (_mm512_maskz_floor_round_pd): Ditto.
3107 (_mm512_maskz_ceil_round_ps): Ditto.
3108 (_mm512_maskz_ceil_round_pd): Ditto.
3109 (_mm512_expand_pd): Ditto.
3110 (_mm512_expand_ps): Ditto.
3111 * config/i386/i386.c (ix86_builtins): Remove
3112 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3113 (bdesc_args): Ditto.
3114 * config/i386/predicates.md (const1256_operand): New.
3115 (const_1_to_2_operand): Ditto.
3116 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3117 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3118 (*avx512pf_gatherpf<mode>sf): Ditto.
3119 (avx512pf_gatherpf<mode>df): Ditto.
3120 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3121 (*avx512pf_gatherpf<mode>df): Ditto.
3122 (avx512pf_scatterpf<mode>sf): Ditto.
3123 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3124 (*avx512pf_scatterpf<mode>sf): Ditto.
3125 (avx512pf_scatterpf<mode>df): Ditto.
3126 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3127 (*avx512pf_scatterpf<mode>df): Ditto.
3128 (avx512f_expand<mode>): Removed.
3129 (<shift_insn><mode>3<mask_name>): Change predicate type.
3131 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3133 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3134 not at the end of datarefs vector use ordered_remove to avoid
3135 reordering datarefs vector.
3138 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3139 mark local addressable non-static vars as GOVD_PRIVATE
3140 instead of GOVD_LOCAL.
3141 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3142 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3146 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3147 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3148 assume_aligned or alloc_align attributes.
3149 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3150 arguments. Handle also assume_aligned and alloc_align attributes.
3151 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3152 calls to functions with assume_aligned or alloc_align attributes.
3153 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3155 2014-02-08 Terry Guo <terry.guo@arm.com>
3157 * doc/invoke.texi: Document ARM -march=armv7e-m.
3159 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3161 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3162 flag on __cilkrts_rethrow builtin.
3165 * ipa-cp.c (determine_versionability): Fail at -O0
3166 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3167 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3170 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3173 * tree-inline.c (copy_forbidden): Fail for
3174 __attribute__((optimize (0))) functions.
3176 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3178 * varpool.c: Include pointer-set.h.
3179 (varpool_remove_unreferenced_decls): Variables in other partitions
3180 will not be output; be however careful to not lose information
3183 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3185 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3186 lookup in the vtable constructor.
3188 2014-02-07 Jeff Law <law@redhat.com>
3191 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3192 define_insn_and_split.
3194 * ipa-inline.c (inline_small_functions): Fix typos.
3196 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3198 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3199 (s390_can_use_return_insn): Declare.
3200 * config/s390/s390.h (EPILOGUE_USES): Define.
3201 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3203 (s390_chunkify_start): Handle return JUMP_LABELs.
3204 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3205 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3206 (s390_can_use_return_insn): New functions.
3207 (s390_fix_long_loop_prediction): Handle conditional returns.
3208 (TARGET_SET_UP_BY_PROLOGUE): Define.
3209 * config/s390/s390.md (ANY_RETURN): New code iterator.
3210 (*creturn, *csimple_return, return, simple_return): New patterns.
3212 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3214 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3215 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3216 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3217 REG_CFA_RESTORE list when deciding not to restore a register.
3219 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3221 * config/s390/s390.c: Include tree-pass.h and context.h.
3222 (s390_early_mach): New function, split out from...
3223 (s390_emit_prologue): ...here.
3224 (pass_data_s390_early_mach): New pass structure.
3225 (pass_s390_early_mach): New class.
3226 (s390_option_override): Create and register early_mach pass.
3227 Move to end of file.
3229 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3231 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3232 to match for the exit block.
3234 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3236 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3237 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3238 Reject misaligned operands.
3240 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3242 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3244 2014-02-07 Richard Biener <rguenther@suse.de>
3247 * gimple-low.c (lower_builtin_posix_memalign): New function.
3248 (lower_stmt): Call it to lower posix_memalign in a way
3249 to make alignment info accessible.
3251 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3254 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3255 __builtin_setjmp_receiver.
3257 2014-02-07 Richard Biener <rguenther@suse.de>
3260 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3261 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3262 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3263 Handle BUILT_IN_POSIX_MEMALIGN.
3264 (find_func_clobbers): Likewise.
3265 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3266 (call_may_clobber_ref_p_1): Likewise.
3268 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3271 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3274 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3277 * lto-cgraph.c (lto_output_node): Use
3278 symtab_get_symbol_partitioning_class.
3279 (lto_output_varpool_node): likewise.
3280 (symtab_get_symbol_partitioning_class): Move here from
3282 * cgraph.h (symbol_partitioning_class): Likewise.
3283 (symtab_get_symbol_partitioning_class): Declare.
3285 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3287 * ggc.h (ggc_internal_cleared_alloc): New macro.
3288 * vec.h (vec_safe_copy): Handle memory stats.
3289 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3290 * target-globals.c (save_target_globals): Likewise.
3292 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3295 * expr.c (emit_move_resolve_push): Export; be bit more selective
3296 on when to clear alias set.
3297 * expr.h (emit_move_resolve_push): Declare.
3298 * function.h (struct function): Add tail_call_marked.
3299 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3300 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3301 * config/i386/i386.md (TImode move expander): De not call
3303 (FP push expanders): Preserve memory attributes.
3304 * config/i386/sse.md (push<mode>1): Remove.
3305 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3306 (ix86_expand_push): Remove.
3307 * config/i386/mmx.md (push<mode>1): Remove.
3309 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3311 PR rtl-optimization/60030
3312 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3313 lopart with paradoxical subreg before shifting it up by hprec.
3315 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3317 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3318 Remove extra newline at end of file.
3319 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3320 (arm_issue_rate): Handle cortexa57.
3321 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3322 (cortex-a57.cortex-a53): Likewise.
3324 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3327 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3328 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3330 (arm_expand_prologue): Adjust all callers.
3331 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3332 info, registers also at the lowest numbered registers side. Use
3333 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3337 * var-tracking.c (adjust_mems): Before adding a SET to
3338 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3340 2014-02-06 Alan Modra <amodra@gmail.com>
3343 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3344 change SDmode to DDmode when lra_in_progress.
3346 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3349 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3350 free_data_ref on the dr first, and before goto again also set dr
3351 to the next dr. For simd_lane_access, free old datarefs[i] before
3352 overwriting it. For get_vectype_for_scalar_type failure, don't
3353 free_data_ref if simd_lane_access.
3355 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3358 * tree.h (opts_for_fn): New inline function.
3359 (opt_for_fn): Define.
3360 * config/i386/i386.c (ix86_function_regparm): Use
3361 opt_for_fn (decl, optimize) instead of optimize.
3363 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3365 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3366 for SYMBOL_REF in large memory model.
3368 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3370 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3372 (cortex-a57): Likewise.
3373 (cortex-a57.cortex-a53): Likewise.
3375 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3376 Kugan Vivekanandarajah <kuganv@linaro.org>
3378 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3380 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3382 2014-02-06 Richard Biener <rguenther@suse.de>
3384 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3385 set_loop_copy and initialize_original_copy_tables.
3387 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3389 * config/aarch64/aarch64-simd.md
3390 (aarch64_ashr_simddi): Change QI to SI.
3392 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3393 Jakub Jelinek <jakub@redhat.com>
3396 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3399 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3401 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3402 CODE_FOR_altivec_vpku[hw]um to
3403 CODE_FOR_altivec_vpku[hw]um_direct.
3404 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3405 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3406 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3407 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3409 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3411 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3412 generation for -maltivec=be.
3413 (altivec_vsumsws): Simplify redundant test.
3415 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3417 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3418 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3419 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3420 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3421 gen_altivec_vpkuwum.
3422 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3424 (altivec_vpks<VI_char>ss): Likewise.
3425 (altivec_vpks<VI_char>us): Likewise.
3426 (altivec_vpku<VI_char>us): Likewise.
3427 (altivec_vpku<VI_char>um): Likewise.
3428 (altivec_vpku<VI_char>um_direct): New (copy of
3429 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3431 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3432 target is little endian and -maltivec=be is not specified.
3433 (*altivec_vupkhs<VU_char>_direct): New (copy of
3434 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3435 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3436 target is little endian and -maltivec=be is not specified.
3437 (*altivec_vupkls<VU_char>_direct): New (copy of
3438 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3439 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3440 little endian and -maltivec=be is not specified.
3441 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3442 little endian and -maltivec=be is not specified.
3444 2014-02-05 Richard Henderson <rth@redhat.com>
3447 * combine-stack-adj.c: Revert r206943.
3448 * sched-int.h (struct deps_desc): Add last_args_size.
3449 * sched-deps.c (init_deps): Initialize it.
3450 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3451 contain REG_ARGS_SIZE notes.
3453 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3455 * lto-cgraph.c (asm_nodes_output): Make global.
3456 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3457 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3458 (driver_handle_option): Handle OPT_fwpa.
3460 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3463 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3464 a comment typo and formatting issue. If odr_hash hasn't been
3465 created, return vNULL and set *completep to false.
3468 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3469 bb with no successors.
3471 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3474 * doc/invoke.texi (-march): Clarify documentation for ARM.
3478 2014-02-05 Richard Biener <rguenther@suse.de>
3480 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3481 when not vectorizing because of too many alias checks.
3482 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3483 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3485 2014-02-05 Nick Clifton <nickc@redhat.com>
3487 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3488 accept extended registers in any mode when compiling for the MN10300.
3490 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3492 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3493 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3494 sanitization attributes.
3495 (can_inline_edge_p): Likewise.
3496 (sanitize_attrs_match_for_inline_p): New function.
3498 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3500 * ipa-prop.c (detect_type_change): Shor circuit testing of
3501 type changes on THIS pointer.
3503 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3506 * config/pa/pa.c (legitimize_tls_address): Return original address
3507 if not passed a SYMBOL_REF rtx.
3508 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3510 (pa_emit_move_sequence): Simplify TLS source operands.
3511 (pa_legitimate_constant_p): Reject all TLS constants.
3512 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3513 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3515 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3517 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3518 groups when we know they are controlled by LTO.
3519 * varasm.c (default_binds_local_p_1): If object is in other partition,
3520 it will be resolved locally.
3522 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3524 * config/host-linux.c (linux_gt_pch_use_address): Don't
3525 use SSIZE_MAX because it is not always defined.
3527 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3530 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3531 threshold for pseudo splitting.
3532 (update_ebb_live_info): Process call argument hard registers and
3533 hard registers from insn definition too.
3534 (max_small_class_regs_num): New constant.
3535 (inherit_in_ebb): Update live hard regs through EBBs. Update
3536 reloads_num only for small register classes. Don't split for
3539 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3542 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3545 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3547 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3550 2014-02-04 Marek Polacek <polacek@redhat.com>
3552 * gdbinit.in (pel): Define.
3554 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3556 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3559 2014-02-04 Richard Biener <rguenther@suse.de>
3562 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3563 in function context local.
3564 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3565 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3566 similar to LTO_imported_decl_ref.
3568 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3570 PR tree-optimization/60002
3571 * cgraphclones.c (build_function_decl_skip_args): Clear
3574 PR tree-optimization/60023
3575 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3576 false to gsi_replace.
3577 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3578 has been in some EH region and vec_stmt could throw, add
3579 vec_stmt into the same EH region.
3580 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3581 has no lhs, ignore it.
3582 * internal-fn.c (expand_MASK_LOAD): Likewise.
3585 * tree-inline.c (copy_forbidden): Fail for
3586 __attribute__((optimize (0))) functions.
3589 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3590 is set, copy one less argument.
3591 (expand_simd_clones): Don't subtract clone_info->inbranch
3592 from simd_clone_struct_alloc argument.
3594 PR rtl-optimization/57915
3595 * recog.c (simplify_while_replacing): If all unary/binary/relational
3596 operation arguments are constant, attempt to simplify those.
3599 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3600 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3602 2014-02-04 Richard Biener <rguenther@suse.de>
3604 PR tree-optimization/60012
3605 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3606 TBAA disambiguation to all DDRs.
3608 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3611 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3612 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3614 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3617 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3619 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3621 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3622 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3624 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3627 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3628 to figure out targets of polymorphic calls with known decl.
3629 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3630 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3631 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3632 (get_polymorphic_call_info): ... here.
3633 (get_polymorphic_call_info_from_invariant): New function.
3635 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3637 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3638 lookup via vtable pointer; check for type consistency
3639 and turn inconsitent facts into UNREACHABLE.
3640 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3641 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3642 type inconsistent querries; return UNREACHABLE instead.
3644 2014-02-03 Richard Henderson <rth@twiddle.net>
3647 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3648 already processed this node.
3649 (normalize_one_pred_1): Pass along mark_set.
3650 (normalize_one_pred): Create and destroy a pointer_set_t.
3651 (normalize_one_pred_chain): Likewise.
3653 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3655 PR gcov-profile/58602
3656 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3658 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3661 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3662 -fno-devirtualize; try to devirtualize by the knowledge of
3663 virtual table pointer given by aggregate propagation.
3664 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3665 (ipa_print_node_jump_functions): Dump also offset that
3666 is relevant for polymorphic calls.
3667 (determine_known_aggregate_parts): Add arg_type parameter; use it
3668 instead of determining the type from pointer type.
3669 (ipa_compute_jump_functions_for_edge): Update call of
3670 determine_known_aggregate_parts.
3671 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3672 (gimple_get_virt_method_for_binfo): ... here; simplify using
3673 vtable_pointer_value_to_vtable.
3674 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3675 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3676 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3677 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3679 (vtable_pointer_value_to_binfo): ... here.
3680 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3682 2014-02-03 Teresa Johnson <tejohnson@google.com>
3684 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3685 redef of outer loop index variable.
3687 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3691 * doc/extend.texi (Function Attributes): Typo.
3693 2014-02-03 Cong Hou <congh@google.com>
3695 PR tree-optimization/60000
3696 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3697 if the vectorized statement is a store. A store statement can only
3698 appear at the end of pattern statements.
3700 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3702 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3703 (ix86_option_override_internal): Default long double to 64-bit for
3704 32-bit Bionic and to 128-bit for 64-bit Bionic.
3706 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3707 TARGET_LONG_DOUBLE_128 is true.
3708 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3710 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3711 (mlong-double-64): Negate -mlong-double-128.
3712 (mlong-double-128): New option.
3714 * config/i386/i386-c.c (ix86_target_macros): Define
3715 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3717 * doc/invoke.texi: Document -mlong-double-128.
3719 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3721 PR rtl-optimization/60024
3722 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3724 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3726 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3728 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3730 PR rtl-optimization/57662
3731 * sel-sched.c (code_motion_path_driver): Do not mark already not
3732 existing blocks in the visiting bitmap.
3734 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3736 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3737 on the insn being emitted.
3739 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3740 Will Deacon <will.deacon@arm.com>
3742 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3744 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3746 * config/arm/arm-tables.opt: Regenerate.
3748 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3750 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3751 for vector types other than V16QImode.
3752 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3753 define_expand, and call altivec_expand_vec_perm_le when producing
3754 code with little endian element order.
3755 (*altivec_vperm_<mode>_internal): New insn having previous
3756 behavior of altivec_vperm_<mode>.
3757 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3758 altivec_expand_vec_perm_le when producing code with little endian
3760 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3761 behavior of altivec_vperm_<mode>_uns.
3763 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3765 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3766 (altivec_vsumsws): Add handling for -maltivec=be with a little
3768 (altivec_vsumsws_direct): New.
3769 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3770 gen_altivec_vsumsws.
3772 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3774 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3775 vtable_pointer_value_to_binfo): New functions.
3776 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3777 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3779 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3781 * config/nios2/nios2.md (load_got_register): Initialize GOT
3782 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3783 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3785 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3787 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3788 preserverd by passthrough, do not propagate the type.
3790 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3792 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3793 (mips_atomic_assign_expand_fenv): New function.
3794 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3796 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3798 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3799 (__builtin_mips_set_fcsr): Likewise.
3800 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3801 MIPS_USI_FTYPE_VOID.
3802 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3803 (mips16_expand_set_fcsr): Likewise.
3804 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3805 (mips16_set_fcsr_stub): Likewise.
3806 (mips16_get_fcsr_one_only_stub): New class.
3807 (mips16_set_fcsr_one_only_stub): Likewise.
3808 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3809 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3810 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3811 (hard_float): New availability predicate.
3812 (mips_builtins): Add get_fcsr and set_fcsr.
3813 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3814 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3815 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3816 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3817 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3820 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3822 * config/mips/mips.c (mips_one_only_stub): New class.
3823 (mips_need_mips16_rdhwr_p): Replace with...
3824 (mips16_rdhwr_stub): ...this new variable.
3825 (mips16_stub_call_address): New function.
3826 (mips16_rdhwr_one_only_stub): New class.
3827 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3828 (mips_output_mips16_rdhwr): Delete.
3829 (mips_finish_stub): New function.
3830 (mips_code_end): Use it to handle rdhwr stubs.
3832 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3835 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3836 when calculating size of integer atomic types.
3838 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3840 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3842 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3844 PR tree-optimization/60003
3845 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3846 * profile.c (branch_prob): Use gimple_call_builtin_p
3847 to check for BUILT_IN_SETJMP_RECEIVER.
3848 * tree-inline.c (copy_bb): Call notice_special_calls.
3850 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3853 * lra-constraints.c (process_alt_operands): Update reload_sum only
3856 2014-01-31 Richard Henderson <rth@redhat.com>
3859 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3860 until after else_eh is processed.
3862 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3864 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3865 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3866 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3867 in smmintrin.h, remove them.
3868 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3869 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3870 * config/i386/i386.md (ROUND_SAE): Fix value.
3871 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3872 (const48_operand): New.
3873 * config/i386/subst.md (round), (round_expand): Use
3874 const_4_or_8_to_11_operand.
3875 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3877 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3879 * config/i386/constraints.md (Yk): Swap meaning with k.
3880 * config/i386/i386.md (movhi_internal): Change Yk to k.
3881 (movqi_internal): Ditto.
3882 (*k<logic><mode>): Ditto.
3885 (kandn<mode>): Ditto.
3886 (*<code>hi_1): Ditto.
3887 (*<code>qi_1): Ditto.
3888 (kxnor<mode>): Ditto.
3889 (kortestzhi): Ditto.
3890 (kortestchi): Ditto.
3892 (*one_cmplhi2_1): Ditto.
3893 (*one_cmplqi2_1): Ditto.
3894 * config/i386/sse.md (): Change k to Yk.
3895 (avx512f_load<mode>_mask): Ditto.
3896 (avx512f_blendm<mode>): Ditto.
3897 (avx512f_store<mode>_mask): Ditto.
3898 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3899 (avx512f_storedqu<mode>_mask): Ditto.
3900 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3902 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3903 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3904 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3905 (avx512f_maskcmp<mode>3): Ditto.
3906 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3907 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3908 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3909 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3910 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3911 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3912 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3913 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3914 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3915 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3916 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3917 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3918 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3919 (vec_extract_lo_<mode>_maskm): Ditto.
3920 (vec_extract_hi_<mode>_maskm): Ditto.
3921 (avx512f_vternlog<mode>_mask): Ditto.
3922 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3923 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3924 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3925 (avx512f_<code>v8div16qi2_mask): Ditto.
3926 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3927 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3928 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3929 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3930 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3931 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3932 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3933 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3934 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3935 (avx512cd_maskb_vec_dupv8di): Ditto.
3936 (avx512cd_maskw_vec_dupv16si): Ditto.
3937 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3938 (avx512f_vpermi2var<mode>3_mask): Ditto.
3939 (avx512f_vpermi2var<mode>3_mask): Ditto.
3940 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3941 (*avx512f_gathersi<mode>): Ditto.
3942 (*avx512f_gathersi<mode>_2): Ditto.
3943 (*avx512f_gatherdi<mode>): Ditto.
3944 (*avx512f_gatherdi<mode>_2): Ditto.
3945 (*avx512f_scattersi<mode>): Ditto.
3946 (*avx512f_scatterdi<mode>): Ditto.
3947 (avx512f_compress<mode>_mask): Ditto.
3948 (avx512f_compressstore<mode>_mask): Ditto.
3949 (avx512f_expand<mode>_mask): Ditto.
3950 * config/i386/subst.md (mask): Change k to Yk.
3951 (mask_scalar_merge): Ditto.
3954 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3956 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3958 2014-01-31 Richard Biener <rguenther@suse.de>
3961 * builtins.c (fold_builtin_memory_op): Make sure to not
3962 use a floating-point mode or a boolean or enumeral type for
3965 2014-01-30 DJ Delorie <dj@redhat.com>
3967 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3968 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3969 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3970 whenever main() has an epilogue.
3972 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3974 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3975 unused variable "field".
3976 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3977 (vsx_mergeh_<mode>): Likewise.
3978 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3979 (altivec_vmrghh): Likewise.
3980 (altivec_vmrghw): Likewise.
3981 (altivec_vmrglb): Likewise.
3982 (altivec_vmrglh): Likewise.
3983 (altivec_vmrglw): Likewise.
3984 (altivec_vspltb): Add missing uses.
3985 (altivec_vsplth): Likewise.
3986 (altivec_vspltw): Likewise.
3987 (altivec_vspltsf): Likewise.
3989 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3992 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3993 frame related instructions.
3995 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3997 PR rtl-optimization/59959
3998 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3999 any reload of register whose subreg is invalid.
4001 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4003 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4004 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4005 Add missing return type - void.
4007 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4009 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4010 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4011 remove element index adjustment for endian (now handled in vsx.md
4013 (altivec_expand_vec_perm_const): Use
4014 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4015 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4016 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4017 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4018 define_expand and a new define_insn *altivec_vspltb_internal;
4019 adjust for -maltivec=be on a little endian target.
4020 (altivec_vspltb_direct): New.
4021 (altivec_vsplth): Divide into a define_expand and a new
4022 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4023 little endian target.
4024 (altivec_vsplth_direct): New.
4025 (altivec_vspltw): Divide into a define_expand and a new
4026 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4027 little endian target.
4028 (altivec_vspltw_direct): New.
4029 (altivec_vspltsf): Divide into a define_expand and a new
4030 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4031 a little endian target.
4033 2014-01-30 Richard Biener <rguenther@suse.de>
4035 PR tree-optimization/59993
4036 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4037 can propagate form the earlier stmt and avoid the transform
4038 when the intermediate result is needed.
4040 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4042 * README.Portability: Fix typo.
4044 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4046 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4047 comparison_operator with ordered_comparison_operator.
4049 2014-01-30 Nick Clifton <nickc@redhat.com>
4051 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4052 Rename to mn10300_store_multiple_regs.
4053 * config/mn10300/mn10300.c: Likewise.
4054 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4055 store_multiple_regs.
4056 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4057 Call mn10300_store_multiple_regs.
4059 2014-01-30 Nick Clifton <nickc@redhat.com>
4060 DJ Delorie <dj@redhat.com>
4062 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4063 %fp 2 to keep registers after it properly word-aligned.
4064 (rl78_alloc_physical_registers_umul): Handle the case where both
4065 input operands are the same.
4067 2014-01-30 Richard Biener <rguenther@suse.de>
4069 PR tree-optimization/59903
4070 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4073 2014-01-30 Jason Merrill <jason@redhat.com>
4076 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4079 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4081 2014-01-30 Richard Biener <rguenther@suse.de>
4083 PR tree-optimization/59951
4084 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4086 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4089 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4090 SFmode to DFmode case.
4092 2014-01-29 DJ Delorie <dj@redhat.com>
4094 * config/msp430/msp430.opt (-minrt): New.
4095 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4097 (ENDFILE_SPEC): Likewise.
4099 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4101 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4102 (estimate_function_body_sizes): Use it.
4104 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4107 * dwarf2out.c (is_cxx_auto): New.
4108 (is_base_type): Use it.
4109 (gen_type_die_with_usage): Likewise.
4111 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4113 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4114 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4115 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4116 -maltivec=be with LE targets.
4117 (vsx_mergeh_<mode>): Likewise.
4118 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4119 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4120 (altivec_vmrghb): Replace with define_expand and new
4121 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4122 (altivec_vmrghb_direct): New define_insn.
4123 (altivec_vmrghh): Replace with define_expand and new
4124 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4125 (altivec_vmrghh_direct): New define_insn.
4126 (altivec_vmrghw): Replace with define_expand and new
4127 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4128 (altivec_vmrghw_direct): New define_insn.
4129 (*altivec_vmrghsf): Adjust for endianness.
4130 (altivec_vmrglb): Replace with define_expand and new
4131 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4132 (altivec_vmrglb_direct): New define_insn.
4133 (altivec_vmrglh): Replace with define_expand and new
4134 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4135 (altivec_vmrglh_direct): New define_insn.
4136 (altivec_vmrglw): Replace with define_expand and new
4137 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4138 (altivec_vmrglw_direct): New define_insn.
4139 (*altivec_vmrglsf): Adjust for endianness.
4140 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4141 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4142 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4143 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4144 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4145 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4146 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4147 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4149 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4151 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4152 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4155 2014-01-29 Richard Biener <rguenther@suse.de>
4157 PR tree-optimization/58742
4158 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4159 associate_pointerplus_align.
4160 (associate_pointerplus_diff): New function.
4161 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4162 and associate_pointerplus_diff.
4164 2014-01-29 Richard Biener <rguenther@suse.de>
4166 * lto-streamer.h (LTO_major_version): Bump to 3.
4167 (LTO_minor_version): Reset to 0.
4169 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4171 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4172 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4173 (arm_file_start): Generate correct asm header for armv7ve.
4174 * config/arm/bpabi.h: Add multilib support for armv7ve.
4175 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4176 and cortex-a15 to armv7ve.
4177 * config/arm/t-aprofile: Add multilib support for armv7ve.
4178 * doc/invoke.texi: Document -march=armv7ve.
4180 2014-01-29 Richard Biener <rguenther@suse.de>
4182 PR tree-optimization/58742
4183 * tree-ssa-forwprop.c (associate_plusminus): Return true
4184 if we changed sth, defer EH cleanup to ...
4185 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4186 (simplify_mult): New function.
4188 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4191 PR tree-optimization/59920
4192 * tree.c (build_common_builtin_nodes): Remove
4193 __builtin_setjmp_dispatcher initialization.
4194 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4195 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4196 instead of gsi_after_labels + manually skipping debug stmts.
4197 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4198 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4199 * tree-inline.c (copy_edges_for_bb): Remove
4200 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4201 argument. Ignore computed_goto_p stmts. Don't call
4202 make_abnormal_goto_edges. If a call might need abnormal edges
4203 for non-local gotos, see if it already has an edge to
4204 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4205 with true argument, don't do anything then, otherwise add
4206 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4207 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4209 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4210 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4211 (lower_stmt): Don't set data->calls_builtin_setjmp.
4212 (lower_builtin_setjmp): Adjust comment.
4213 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4214 * tree-cfg.c (found_computed_goto): Remove.
4215 (factor_computed_gotos): Remove.
4216 (make_goto_expr_edges): Return bool, true for computed gotos.
4217 Don't call make_abnormal_goto_edges.
4218 (build_gimple_cfg): Don't set found_computed_goto, don't call
4219 factor_computed_gotos.
4220 (computed_goto_p): No longer static.
4221 (make_blocks): Don't set found_computed_goto.
4222 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4223 (make_edges): If make_goto_expr_edges returns true, push bb
4224 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4225 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4226 vector. Record mapping between bbs and OpenMP regions if there
4227 are any, adjust make_gimple_omp_edges caller. Call
4228 handle_abnormal_edges.
4229 (make_abnormal_goto_edges): Remove.
4230 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4231 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4232 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4233 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4234 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4235 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4236 filling *region also set *region_idx to (*region)->entry->index.
4239 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4240 For REGs set ORIGINAL_REGNO.
4242 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4244 * doc/md.texi: Mention that a target shouldn't implement
4245 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4248 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4250 PR tree-optimization/59594
4251 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4252 a copy of the datarefs vector rather than the vector itself.
4254 2014-01-28 Jason Merrill <jason@redhat.com>
4257 * dwarf2out.c (auto_die): New static.
4258 (gen_type_die_with_usage): Handle C++1y 'auto'.
4259 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4262 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4265 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4266 (SPEC_X32): Likewise.
4267 (SPEC_64): Likewise.
4268 * config/i386/i386.c (ix86_option_override_internal): Turn off
4269 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4271 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4272 * config/i386/i386.h (TARGET_16BIT): New macro.
4273 (TARGET_16BIT_P): Likewise.
4274 * config/i386/i386.opt: Add m16.
4275 * doc/invoke.texi: Document -m16.
4277 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4279 PR preprocessor/59935
4280 * input.c (location_get_source_line): Bail out on when line number
4281 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4283 2014-01-28 Richard Biener <rguenther@suse.de>
4285 PR tree-optimization/58742
4286 * tree-ssa-forwprop.c (associate_plusminus): Handle
4287 pointer subtraction of the form (T)(P + A) - (T)P.
4289 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4291 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4294 2014-01-28 Richard Biener <rguenther@suse.de>
4297 2014-01-28 Richard Biener <rguenther@suse.de>
4299 PR rtl-optimization/45364
4300 PR rtl-optimization/59890
4301 * var-tracking.c (local_get_addr_clear_given_value): Handle
4302 already cleared slot.
4303 (val_reset): Handle not allocated local_get_addr_cache.
4304 (vt_find_locations): Use post-order on the inverted CFG.
4306 2014-01-28 Richard Biener <rguenther@suse.de>
4308 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4310 2014-01-28 Richard Biener <rguenther@suse.de>
4312 PR rtl-optimization/45364
4313 PR rtl-optimization/59890
4314 * var-tracking.c (local_get_addr_clear_given_value): Handle
4315 already cleared slot.
4316 (val_reset): Handle not allocated local_get_addr_cache.
4317 (vt_find_locations): Use post-order on the inverted CFG.
4319 2014-01-28 Alan Modra <amodra@gmail.com>
4321 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4322 * configure.ac <recursive call for build != host>: Define
4323 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4324 and LD_FOR_BUILD too.
4325 * configure: Regenerate.
4327 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4329 * config/i386/i386.c (get_builtin_code_for_version): Separate
4330 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4331 Broadwell from Haswell.
4333 2014-01-27 Steve Ellcey <sellcey@mips.com>
4335 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4336 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4337 * config/mips/mips.c (mips_option_override): Change setting
4339 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4340 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4341 Change from Mask to Var.
4343 2014-01-27 Jeff Law <law@redhat.com>
4345 * ipa-inline.c (inline_small_functions): Fix typo.
4347 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4349 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4350 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4351 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4352 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4353 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4354 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4355 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4356 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4357 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4358 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4359 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4360 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4361 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4362 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4363 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4364 (_mm512_storeu_epi64): Ditto.
4365 (_mm512_cmpge_epi32_mask): Ditto.
4366 (_mm512_cmpge_epu32_mask): Ditto.
4367 (_mm512_cmpge_epi64_mask): Ditto.
4368 (_mm512_cmpge_epu64_mask): Ditto.
4369 (_mm512_cmple_epi32_mask): Ditto.
4370 (_mm512_cmple_epu32_mask): Ditto.
4371 (_mm512_cmple_epi64_mask): Ditto.
4372 (_mm512_cmple_epu64_mask): Ditto.
4373 (_mm512_cmplt_epi32_mask): Ditto.
4374 (_mm512_cmplt_epu32_mask): Ditto.
4375 (_mm512_cmplt_epi64_mask): Ditto.
4376 (_mm512_cmplt_epu64_mask): Ditto.
4377 (_mm512_cmpneq_epi32_mask): Ditto.
4378 (_mm512_cmpneq_epu32_mask): Ditto.
4379 (_mm512_cmpneq_epi64_mask): Ditto.
4380 (_mm512_cmpneq_epu64_mask): Ditto.
4381 (_mm512_expand_pd): Ditto.
4382 (_mm512_expand_ps): Ditto.
4383 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4384 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4385 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4386 * config/i386/i386.c (ix86_builtins): Add
4387 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4388 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4389 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4390 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4391 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4392 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4393 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4394 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4395 IX86_BUILTIN_PMOVUSQW512_MEM.
4396 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4397 __builtin_ia32_pmovsqd512mem_mask,
4398 __builtin_ia32_pmovqd512mem_mask,
4399 __builtin_ia32_pmovusqw512mem_mask,
4400 __builtin_ia32_pmovsqw512mem_mask,
4401 __builtin_ia32_pmovqw512mem_mask,
4402 __builtin_ia32_pmovusdw512mem_mask,
4403 __builtin_ia32_pmovsdw512mem_mask,
4404 __builtin_ia32_pmovdw512mem_mask,
4405 __builtin_ia32_pmovqb512mem_mask,
4406 __builtin_ia32_pmovusqb512mem_mask,
4407 __builtin_ia32_pmovsqb512mem_mask,
4408 __builtin_ia32_pmovusdb512mem_mask,
4409 __builtin_ia32_pmovsdb512mem_mask,
4410 __builtin_ia32_pmovdb512mem_mask.
4411 (bdesc_args): Add __builtin_ia32_expanddf512,
4412 __builtin_ia32_expandsf512.
4413 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4414 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4415 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4416 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4417 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4418 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4419 (avx512f_<code>v8div16qi2_mask_store): This.
4420 (avx512f_expand<mode>): New.
4422 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4424 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4426 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4427 (_mm512_prefetch_i32scatter_pd): Ditto.
4428 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4429 (_mm512_prefetch_i64scatter_pd): Ditto.
4430 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4431 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4432 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4433 (_mm512_prefetch_i32scatter_ps): Ditto.
4434 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4435 (_mm512_prefetch_i64scatter_ps): Ditto.
4436 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4437 * config/i386/i386-builtin-types.def: Define
4438 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4439 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4440 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4441 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4442 IX86_BUILTIN_SCATTERPFQPD.
4443 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4444 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4445 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4446 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4447 __builtin_ia32_scatterpfqps.
4448 (ix86_expand_builtin): Expand new built-ins.
4449 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4450 fix memory access data type.
4451 (*avx512pf_gatherpf<mode>_mask): Ditto.
4452 (*avx512pf_gatherpf<mode>): Ditto.
4453 (avx512pf_scatterpf<mode>): Ditto.
4454 (*avx512pf_scatterpf<mode>_mask): Ditto.
4455 (*avx512pf_scatterpf<mode>): Ditto.
4456 (GATHER_SCATTER_SF_MEM_MODE): New.
4457 (avx512pf_gatherpf<mode>df): Ditto.
4458 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4459 (*avx512pf_scatterpf<mode>df): Ditto.
4461 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4464 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4465 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4468 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4470 * common/config/arm/arm-common.c
4471 (arm_rewrite_mcpu): Handle multiple names.
4473 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4475 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4477 * gimple-builder.h (create_gimple_tmp): Delete.
4479 2014-01-27 Christian Bruel <christian.bruel@st.com>
4481 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4484 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4486 * config/pa/pa.md (call): Generate indirect long calls to non-local
4487 functions when outputing 32-bit code.
4488 (call_value): Likewise except for special call to buggy powf function.
4490 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4491 portable runtime and PIC indirect calls.
4492 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4493 and PIC call sequences. Use ldo instead of blr to set return register
4494 in PIC call sequence.
4496 2014-01-25 Walter Lee <walt@tilera.com>
4498 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4499 avoid clobbering a live register.
4501 2014-01-25 Walter Lee <walt@tilera.com>
4503 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4504 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4505 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4506 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4508 2014-01-25 Walter Lee <walt@tilera.com>
4510 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4511 arguments on even registers.
4512 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4514 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4515 (BIGGEST_ALIGNMENT): Ditto.
4516 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4518 2014-01-25 Walter Lee <walt@tilera.com>
4520 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4521 insns before bundling.
4522 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4524 2014-01-25 Walter Lee <walt@tilera.com>
4526 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4527 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4528 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4530 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4532 * config/mips/constraints.md (kl): Delete.
4533 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4534 define expands, using...
4535 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4536 instructions for MIPS16.
4537 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4538 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4540 2014-01-25 Walter Lee <walt@tilera.com>
4542 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4546 2014-01-25 Walter Lee <walt@tilera.com>
4548 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4549 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4551 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4553 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4556 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4558 * print-rtl.c (in_call_function_usage): New var.
4559 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4560 EXPR_LIST mode as mode and not as reg note name.
4563 * cfgloopmanip.c (copy_loop_info): If
4564 loop->warned_aggressive_loop_optimizations, make sure
4565 the flag is set in target loop too.
4567 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4569 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4571 * builtins.def: Likewise.
4572 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4573 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4574 * ira.c (ira_setup_eliminable_regset): Likewise.
4575 * omp-low.c (gate_expand_omp): Likewise.
4576 (execute_lower_omp): Likewise.
4577 (diagnose_sb_0): Likewise.
4578 (gate_diagnose_omp_blocks): Likewise.
4579 (simd_clone_clauses_extract): Likewise.
4582 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4584 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4585 correction for little endian...
4586 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4589 2014-01-24 Jeff Law <law@redhat.com>
4591 PR tree-optimization/59919
4592 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4593 for non-returning calls.
4595 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4597 * common/config/aarch64/aarch64-common.c
4598 (aarch64_rewrite_mcpu): Handle multiple names.
4599 * config/aarch64/aarch64.h
4600 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4602 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4604 * input.c (add_file_to_cache_tab): Handle the case where fopen
4607 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4610 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4611 from push operand if code of push isn't PRE_DEC.
4613 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4616 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4617 -mquad-memory-atomic. Update -mquad-memory documentation to say
4618 it is only used for non-atomic loads/stores.
4620 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4621 -mquad-memory or -mquad-memory-atomic switches.
4623 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4624 -mquad-memory-atomic to ISA 2.07 support.
4626 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4627 to separate support of normal quad word memory operations (ldq, stq)
4628 from the atomic quad word memory operations.
4630 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4631 support to separate non-atomic quad word operations from atomic
4632 quad word operations. Disable non-atomic quad word operations in
4633 little endian mode so that we don't have to swap words after the
4634 load and before the store.
4635 (quad_load_store_p): Add comment about atomic quad word support.
4636 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4637 options printed with -mdebug=reg.
4639 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4640 -mquad-memory-atomic as the test for whether we have quad word
4641 atomic instructions.
4642 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4643 or -mp8-vector are used, allow byte/half-word atomic operations.
4645 * config/rs6000/sync.md (load_lockedti): Insure that the address
4646 is a proper indexed or indirect address for the lqarx instruction.
4647 On little endian systems, swap the hi/lo registers after the lqarx
4649 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4650 insure the address is valid for the lqarx instruction.
4651 (store_conditionalti): Insure that the address is a proper indexed
4652 or indirect address for the stqcrx. instruction. On little endian
4653 systems, swap the hi/lo registers before doing the stqcrx.
4655 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4656 insure the address is valid for the stqcrx. instruction.
4658 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4659 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4660 type of quad memory support is available.
4662 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4665 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4666 there is a danger of looping.
4668 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4670 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4671 force flag_ira_loop_pressure if set via command line.
4673 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4675 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4676 (ashr_simd): New builtin handling DI mode.
4677 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4678 (aarch64_sshr_simddi): New match pattern.
4679 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4680 (vshrd_n_s64): Likewise.
4681 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4683 2014-01-23 Nick Clifton <nickc@redhat.com>
4685 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4686 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4687 favour of mcu specific scripts.
4688 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4691 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4692 Alex Velenko <Alex.Velenko@arm.com>
4694 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4695 (vaddv_s16): Likewise.
4696 (vaddv_s32): Likewise.
4697 (vaddv_u8): Likewise.
4698 (vaddv_u16): Likewise.
4699 (vaddv_u32): Likewise.
4700 (vaddvq_s8): Likewise.
4701 (vaddvq_s16): Likewise.
4702 (vaddvq_s32): Likewise.
4703 (vaddvq_s64): Likewise.
4704 (vaddvq_u8): Likewise.
4705 (vaddvq_u16): Likewise.
4706 (vaddvq_u32): Likewise.
4707 (vaddvq_u64): Likewise.
4708 (vaddv_f32): Likewise.
4709 (vaddvq_f32): Likewise.
4710 (vaddvq_f64): Likewise.
4711 (vmaxv_f32): Likewise.
4712 (vmaxv_s8): Likewise.
4713 (vmaxv_s16): Likewise.
4714 (vmaxv_s32): Likewise.
4715 (vmaxv_u8): Likewise.
4716 (vmaxv_u16): Likewise.
4717 (vmaxv_u32): Likewise.
4718 (vmaxvq_f32): Likewise.
4719 (vmaxvq_f64): Likewise.
4720 (vmaxvq_s8): Likewise.
4721 (vmaxvq_s16): Likewise.
4722 (vmaxvq_s32): Likewise.
4723 (vmaxvq_u8): Likewise.
4724 (vmaxvq_u16): Likewise.
4725 (vmaxvq_u32): Likewise.
4726 (vmaxnmv_f32): Likewise.
4727 (vmaxnmvq_f32): Likewise.
4728 (vmaxnmvq_f64): Likewise.
4729 (vminv_f32): Likewise.
4730 (vminv_s8): Likewise.
4731 (vminv_s16): Likewise.
4732 (vminv_s32): Likewise.
4733 (vminv_u8): Likewise.
4734 (vminv_u16): Likewise.
4735 (vminv_u32): Likewise.
4736 (vminvq_f32): Likewise.
4737 (vminvq_f64): Likewise.
4738 (vminvq_s8): Likewise.
4739 (vminvq_s16): Likewise.
4740 (vminvq_s32): Likewise.
4741 (vminvq_u8): Likewise.
4742 (vminvq_u16): Likewise.
4743 (vminvq_u32): Likewise.
4744 (vminnmv_f32): Likewise.
4745 (vminnmvq_f32): Likewise.
4746 (vminnmvq_f64): Likewise.
4748 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4750 * config/aarch64/aarch64-simd.md
4751 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4752 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4753 (*aarch64_mul3_elt<mode>): Likewise.
4754 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4755 (*aarch64_mul3_elt_to_64v2df): Likewise.
4756 (*aarch64_mla_elt<mode>): Likewise.
4757 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4758 (*aarch64_mls_elt<mode>): Likewise.
4759 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4760 (*aarch64_fma4_elt<mode>): Likewise.
4761 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4762 (*aarch64_fma4_elt_to_64v2df): Likewise.
4763 (*aarch64_fnma4_elt<mode>): Likewise.
4764 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4765 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4766 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4767 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4768 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4769 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4770 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4771 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4772 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4774 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4776 * config/aarch64/aarch64-simd.md
4777 (aarch64_be_checked_get_lane<mode>): New define_expand.
4778 * config/aarch64/aarch64-simd-builtins.def
4779 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4780 New builtin definition.
4781 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4782 Use new safe be builtin.
4784 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4786 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4788 (aarch64_be_st1<mode>): Likewise.
4789 (aarch_ld1<VALL:mode>): Define_expand modified.
4790 (aarch_st1<VALL:mode>): Likewise.
4791 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4792 (UNSPEC_ST1): Likewise.
4794 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4796 * config/microblaze/microblaze.md: Add trap insn and attribute
4798 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4800 PR preprocessor/58580
4801 * input.h (location_get_source_line): Take an additional line_size
4803 (void diagnostics_file_cache_fini): Declare new function.
4804 * input.c (struct fcache): New type.
4805 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4806 New static constants.
4807 (diagnostic_file_cache_init, total_lines_num)
4808 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4809 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4810 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4811 (get_next_line, read_next_line, goto_next_line, read_line_num):
4812 New static function definitions.
4813 (diagnostic_file_cache_fini): New function.
4814 (location_get_source_line): Take an additional output line_len
4815 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4817 * diagnostic.c (diagnostic_finish): Call
4818 diagnostic_file_cache_fini.
4819 (adjust_line): Take an additional input parameter for the length
4820 of the line, rather than calculating it with strlen.
4821 (diagnostic_show_locus): Adjust the use of
4822 location_get_source_line and adjust_line with respect to their new
4823 signature. While displaying a line now, do not stop at the first
4824 null byte. Rather, display the zero byte as a space and keep
4825 going until we reach the size of the line.
4826 * Makefile.in: Add vec.o to OBJS-libcommon
4828 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4829 Ilya Tocar <ilya.tocar@intel.com>
4831 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4832 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4833 (__builtin_ia32_kmov16): Ditto.
4834 * config/i386/i386.md (UNSPEC_KMOV): New.
4837 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4839 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4840 (_mm512_storeu_si512): Ditto.
4842 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4845 * rtl.h (get_referenced_operands): Declare.
4846 * recog.c (get_referenced_operands): New function.
4847 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4848 operands have been referenced when recording LO_SUM references.
4850 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4852 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4854 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4856 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4857 Enable for generic and recent AMD targets.
4859 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4861 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4862 ARG_SIZE note when adjustment was eliminated.
4864 2014-01-22 Jeff Law <law@redhat.com>
4866 PR tree-optimization/59597
4867 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4868 in file. Accept new argument REGISTERING and use it to modify
4869 dump output appropriately.
4870 (register_jump_thread): Corresponding changes.
4871 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4872 thread paths involving joiner blocks. Add code to dump cancelled
4873 jump threading paths.
4875 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4877 PR rtl-optimization/59477
4878 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4879 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4881 2014-01-22 Tom Tromey <tromey@redhat.com>
4883 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4885 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4887 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4889 PR rtl-optimization/59896
4890 * lra-constraints.c (process_alt_operands): Check unused note for
4891 matched operands of insn with no output reloads.
4893 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4895 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4896 (mips_move_from_gpr_cost): Likewise.
4898 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4900 PR rtl-optimization/59858
4901 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4902 ira_class_hard_regs_num.
4903 (process_alt_operands): Increase reject for dying matched operand.
4905 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4908 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4909 smaller than size, perform several stores or loads and stores
4910 at dst + count - size to store or copy all of size bytes, rather
4911 than just last modesize bytes.
4913 2014-01-20 DJ Delorie <dj@redhat.com>
4915 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4916 that CLOBBERs are REGs before propogating their values.
4918 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4921 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4922 (cgraph_inline_failed_type): New function.
4923 * cgraph.h (DEFCIFCODE): Add type.
4924 (cgraph_inline_failed_type_t): New enum.
4925 (cgraph_inline_failed_type): New prototype.
4926 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4927 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4928 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4929 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4930 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4931 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4932 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4933 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4934 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4935 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4936 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4937 OPTIMIZATION_MISMATCH.
4938 * tree-inline.c (expand_call_inline): Emit errors during
4939 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4941 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4944 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4945 mode attribute in insn output.
4947 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4949 * output.h (output_constant): Delete.
4950 * varasm.c (output_constant): Make private.
4952 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4954 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4956 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4959 * tree.h (fold_builtin_strcat): New prototype.
4960 * builtins.c (fold_builtin_strcat): No longer static. Add len
4961 argument, if non-NULL, don't call c_strlen. Optimize
4962 directly into __builtin_memcpy instead of __builtin_strcpy.
4963 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4964 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4966 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4968 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4969 for SImode_address_operand operands, having only a REG argument.
4971 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4973 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4974 loader name using mbig-endian.
4975 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4977 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4979 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4983 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4985 * config/aarch64/aarch64-protos.h
4986 (aarch64_cannot_change_mode_class_ptr): Declare.
4987 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4988 aarch64_cannot_change_mode_class_ptr): New.
4989 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4990 backend hook aarch64_cannot_change_mode_class.
4992 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4994 * common/config/aarch64/aarch64-common.c
4995 (aarch64_handle_option): Don't handle any option order logic here.
4996 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4997 selected_cpu, warn on architecture version mismatch.
4998 (aarch64_override_options): Fix parsing order for option strings.
5000 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5001 Iain Sandoe <iain@codesourcery.com>
5004 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5005 warning. Amend comment to reflect current functionality.
5007 2014-01-20 Richard Biener <rguenther@suse.de>
5010 * builtins.c (fold_builtin_strcat): Remove case better handled
5011 by tree-ssa-strlen.c.
5013 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5015 * config/aarch64/aarch64.opt
5016 (mcpu, march, mtune): Make case-insensitive.
5018 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5021 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5022 if operands[1] is a REG or ZERO_EXTEND of a REG.
5024 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5026 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5028 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5030 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5031 long non-pic millicode calls.
5033 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5035 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5037 2014-01-19 Kito Cheng <kito@0xlab.org>
5039 * builtins.c (expand_movstr): Check movstr expand done or fail.
5041 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5042 H.J. Lu <hongjiu.lu@intel.com>
5045 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5046 to DImode for zero-extended addresses.
5048 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5050 PR rtl-optimization/57763
5051 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5052 on the new indirect jump_insn and increment LABEL_NUSES (label).
5054 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5058 * config.gcc (x86_archs): New variable.
5059 (x86_64_archs): Likewise.
5060 (x86_cpus): Likewise.
5061 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5062 --with-arch/--with-cpu= options.
5063 Support --with-arch=/--with-cpu={nehalem,westmere,
5064 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5066 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5068 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5069 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5071 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5073 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5075 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5078 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5079 clear cpp_get_options (parse_in)->warn_unused_macros for
5080 ix86_target_macros_internal with cpp_define.
5082 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5084 * jump.c (delete_related_insns): Keep (use (insn))s.
5085 * reorg.c (redundant_insn): Check for barriers too.
5087 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5089 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5091 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5093 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5094 call to $$dyncall when TARGET_LONG_CALLS is true.
5096 2014-01-17 Jeff Law <law@redhat.com>
5098 * ree.c (combine_set_extension): Temporarily disable test for
5099 changing number of hard registers.
5101 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5104 * ipa-inline-analysis.c (inline_free_summary):
5105 Do not free summary of aliases.
5107 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5110 * gimplify.c (gimplify_expr): Use create_tmp_var
5111 instead of create_tmp_var_raw. If cond doesn't have
5112 integral type, don't add the IFN_ANNOTATE builtin at all.
5114 2014-01-17 Martin Jambor <mjambor@suse.cz>
5117 * ipa-cp.c (prev_edge_clone): New variable.
5118 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5119 Also resize prev_edge_clone vector.
5120 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5121 (ipcp_edge_removal_hook): New function.
5122 (ipcp_driver): Register ipcp_edge_removal_hook.
5124 2014-01-17 Andrew Pinski <apinski@cavium.com>
5125 Steve Ellcey <sellcey@mips.com>
5128 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5131 2014-01-17 Jeff Law <law@redhat.com>
5134 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5135 so that pass_ccp runs first.
5137 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5139 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5140 (ix86_adjust_cost): Use !TARGET_XXX.
5141 (do_reorder_for_imul): Likewise.
5142 (swap_top_of_ready_list): Likewise.
5143 (ix86_sched_reorder): Likewise.
5145 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5147 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5148 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5149 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5150 (intel_memset): New. Duplicate slm_memset.
5151 (intel_cost): New. Duplicate slm_cost.
5152 (m_INTEL): New macro.
5153 (processor_target_table): Add "intel".
5154 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5155 with PROCESSOR_INTEL for "intel".
5156 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5157 PROCESSOR_SILVERMONT.
5158 (ix86_issue_rate): Likewise.
5159 (ix86_adjust_cost): Likewise.
5160 (ia32_multipass_dfa_lookahead): Likewise.
5161 (swap_top_of_ready_list): Likewise.
5162 (ix86_sched_reorder): Likewise.
5163 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5164 instead of TARGET_OPT_AGU.
5165 * config/i386/i386.h (TARGET_INTEL): New.
5166 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5167 (processor_type): Add PROCESSOR_INTEL.
5168 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5169 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5171 2014-01-17 Marek Polacek <polacek@redhat.com>
5174 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5177 2014-01-17 Richard Biener <rguenther@suse.de>
5179 PR tree-optimization/46590
5180 * opts.c (default_options_table): Add entries for
5181 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5182 all enabled at -O1 but not for -Og.
5183 * common.opt (fbranch-count-reg): Remove Init(1).
5184 (fmove-loop-invariants): Likewise.
5185 (ftree-pta): Likewise.
5187 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5189 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5190 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5191 decls to at least the GCC 4.8 used alignments.
5194 * tree-nested.c (convert_nonlocal_reference_stmt,
5195 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5196 of GIMPLE_BIND stmts, adjust associated decls.
5198 2014-01-17 Richard Biener <rguenther@suse.de>
5200 PR tree-optimization/46590
5201 * vec.h (vec<>::bseach): New member function implementing
5202 binary search according to C89 bsearch.
5203 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5204 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5205 bitmap pointer again. Make accesses_in_loop a flat array.
5206 (mem_ref_obstack): New global.
5207 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5208 (mark_ref_stored): Likewise.
5209 (ref_indep_loop_p_2): Likewise.
5210 (set_ref_stored_in_loop): New helper function.
5211 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5212 (memref_free): Adjust.
5213 (record_mem_ref_loc): Simplify.
5214 (gather_mem_refs_stmt): Adjust.
5215 (sort_locs_in_loop_postorder_cmp): New function.
5216 (analyze_memory_references): Sort accesses_in_loop after
5217 loop postorder number.
5218 (find_ref_loc_in_loop_cmp): New function.
5219 (for_all_locs_in_loop): Find relevant cluster of locs in
5220 accesses_in_loop and iterate without recursion.
5221 (execute_sm): Avoid uninit warning.
5222 (struct ref_always_accessed): Simplify.
5223 (ref_always_accessed::operator ()): Likewise.
5224 (ref_always_accessed_p): Likewise.
5225 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5226 loop postorder numbers here.
5227 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5230 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5233 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5234 on decls for which assemble_alias has been called.
5236 2014-01-17 Nick Clifton <nickc@redhat.com>
5238 * config/msp430/msp430.opt: (mcpu): New option.
5239 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5240 (msp430_option_override): Parse target_cpu. If the MCU name
5241 matches a generic string, clear target_mcu.
5242 (msp430_attr): Allow numeric interrupt values up to 63.
5243 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5244 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5246 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5248 * config/msp430/msp430.md (popm): Use %J rather than %I.
5249 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5250 (addhi_cy_i): Use immediate_operand for operand 2.
5251 * doc/invoke.texi: Document -mcpu option.
5253 2014-01-17 Richard Biener <rguenther@suse.de>
5255 PR rtl-optimization/38518
5256 * df.h (df_analyze_loop): Declare.
5257 * df-core.c: Include cfgloop.h.
5258 (df_analyze_1): Split out main part of df_analyze.
5259 (df_analyze): Adjust.
5260 (loop_inverted_post_order_compute): New function.
5261 (loop_post_order_compute): Likewise.
5262 (df_analyze_loop): New function avoiding whole-function
5264 * loop-invariant.c (find_defs): Use df_analyze_loop.
5265 (find_invariants): Adjust.
5266 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5268 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5270 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5271 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5273 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5275 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5276 traversal when removing references.
5278 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5281 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5283 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5286 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5287 pushing a reload for an autoinc when we had previously reloaded an
5288 inner part of the address.
5290 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5292 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5294 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5295 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5296 when not giving up or versioning for alias only because of
5298 (vect_analyze_data_ref_dependences): Set to true.
5299 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5301 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5302 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5306 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5309 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5310 operand 0 predicate for gathers, use a new pseudo as subtarget.
5312 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5315 * lra-constraints.c (process_alt_operands): Add printing debug info.
5316 Check absence of input/output reloads for matched operands too.
5318 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5320 PR rtl-optimization/59835
5321 * ira.c (ira_init_register_move_cost): Increase cost for
5324 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5326 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5328 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5331 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5332 non-register objects. Use gen_(high/low)part more consistently.
5335 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5338 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5339 endian support, remove tests for WORDS_BIG_ENDIAN.
5340 (p8_mfvsrd_3_<mode>): Likewise.
5341 (reload_gpr_from_vsx<mode>): Likewise.
5342 (reload_gpr_from_vsxsf): Likewise.
5343 (p8_mfvsrd_4_disf): Likewise.
5345 2014-01-16 Richard Biener <rguenther@suse.de>
5347 PR rtl-optimization/46590
5348 * lcm.c (compute_antinout_edge): Use postorder iteration.
5349 (compute_laterin): Use inverted postorder iteration.
5351 2014-01-16 Nick Clifton <nickc@redhat.com>
5354 * varasm.c (output_constant): Return the number of bytes actually
5356 (output_constructor_array_range): Update the field size with the
5357 number of bytes emitted by output_constant.
5358 (output_constructor_regular_field): Likewise. Also do not
5359 complain if the total number of bytes emitted is now greater
5360 than the expected fieldpos.
5361 * output.h (output_constant): Update prototype and descriptive comment.
5363 2014-01-16 Marek Polacek <polacek@redhat.com>
5366 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5367 it is error_mark_node.
5369 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5371 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5372 VALID_AVX256_REG_OR_OI_MODE.
5374 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5376 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5377 current procedure should be profiled.
5379 2014-01-15 Andrew Pinski <apinski@cavium.com>
5381 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5382 of moving from/to the STACK_REG register class.
5384 2014-01-15 Richard Henderson <rth@redhat.com>
5387 * reginfo.c (global_regs_decl): Globalize.
5388 * rtl.h (global_regs_decl): Declare.
5389 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5390 reserved via global_regs.
5392 2014-01-15 Teresa Johnson <tejohnson@google.com>
5394 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5396 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5398 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5399 and vmulosh rather than call gen_vec_widen_smult_*.
5400 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5401 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5402 (vec_widen_smult_even_v16qi): Likewise.
5403 (vec_widen_umult_even_v8hi): Likewise.
5404 (vec_widen_smult_even_v8hi): Likewise.
5405 (vec_widen_umult_odd_v16qi): Likewise.
5406 (vec_widen_smult_odd_v16qi): Likewise.
5407 (vec_widen_umult_odd_v8hi): Likewise.
5408 (vec_widen_smult_odd_v8hi): Likewise.
5409 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5410 vmuloub rather than call gen_vec_widen_umult_*.
5411 (vec_widen_umult_lo_v16qi): Likewise.
5412 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5413 vmulosb rather than call gen_vec_widen_smult_*.
5414 (vec_widen_smult_lo_v16qi): Likewise.
5415 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5416 rather than call gen_vec_widen_umult_*.
5417 (vec_widen_umult_lo_v8hi): Likewise.
5418 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5419 rather than call gen_vec_widen_smult_*.
5420 (vec_widen_smult_lo_v8hi): Likewise.
5422 2014-01-15 Jeff Law <law@redhat.com>
5424 PR tree-optimization/59747
5425 * ree.c (find_and_remove_re): Properly handle case where a second
5426 eliminated extension requires widening a copy created for elimination
5427 of a prior extension.
5428 (combine_set_extension): Ensure that the number of hard regs needed
5429 for a destination register does not change when we widen it.
5431 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5433 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5434 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5435 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5436 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5437 (avr-*-rtems*): Likewise.
5438 (bfin*-rtems*): Likewise.
5439 (moxie-*-rtems*): Likewise.
5440 (h8300-*-rtems*): Likewise.
5441 (i[34567]86-*-rtems*): Likewise.
5442 (lm32-*-rtems*): Likewise.
5443 (m32r-*-rtems*): Likewise.
5444 (m68k-*-rtems*): Likewise.
5445 (microblaze*-*-rtems*): Likewise.
5446 (mips*-*-rtems*): Likewise.
5447 (powerpc-*-rtems*): Likewise.
5448 (sh-*-rtems*): Likewise.
5449 (sparc-*-rtems*): Likewise.
5450 (sparc64-*-rtems*): Likewise.
5451 (v850-*-rtems*): Likewise.
5452 (m32c-*-rtems*): Likewise.
5454 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5456 PR rtl-optimization/59511
5457 * ira.c (ira_init_register_move_cost): Use memory costs for some
5458 cases of register move cost calculations.
5459 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5460 instead of BB frequency.
5461 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5462 * lra-assigns.c (find_hard_regno_for): Ditto.
5464 2014-01-15 Richard Biener <rguenther@suse.de>
5466 PR tree-optimization/59822
5467 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5468 (vectorizable_load): Use it to hoist defs of uses of invariant
5469 loads out of the loop.
5471 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5472 Kugan Vivekanandarajah <kuganv@linaro.org>
5475 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5478 2014-01-15 Richard Biener <rguenther@suse.de>
5480 PR rtl-optimization/59802
5481 * lcm.c (compute_available): Use inverted postorder to seed
5482 the initial worklist.
5484 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5487 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5488 ADDR_REGS for invalid symrefs in non-PIC code.
5490 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5493 * builtins.c (determine_block_size): Initialize *probable_max_size
5494 even if len_rtx is CONST_INT.
5496 2014-01-14 Andrew Pinski <apinski@cavium.com>
5498 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5499 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5500 (cortexa53_tunings): Likewise.
5501 (aarch64_sched_issue_rate): New function.
5502 (TARGET_SCHED_ISSUE_RATE): Define.
5504 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5506 * ira-costs.c (find_costs_and_classes): Add missed
5507 ira_init_register_move_cost_if_necessary.
5509 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5512 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5514 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5517 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5518 to indicate if type is used for function return value. Warn ABI
5519 change if the vector mode isn't available for function return value.
5520 (ix86_function_arg_advance): Pass false to type_natural_mode.
5521 (ix86_function_arg): Likewise.
5522 (ix86_gimplify_va_arg): Likewise.
5523 (function_arg_32): Don't warn ABI change.
5524 (ix86_function_value): Pass true to type_natural_mode.
5525 (ix86_return_in_memory): Likewise.
5526 (ix86_struct_value_rtx): Removed.
5527 (TARGET_STRUCT_VALUE_RTX): Likewise.
5529 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5531 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5532 converting a conditional jump into a conditional return.
5534 2014-01-14 Richard Biener <rguenther@suse.de>
5536 PR tree-optimization/58921
5537 PR tree-optimization/59006
5538 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5539 hoisting invariant stmts.
5540 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5541 invariant loads on the preheader edge if possible.
5543 2014-01-14 Joey Ye <joey.ye@arm.com>
5545 * doc/plugin.texi (Building GCC plugins): Update to C++.
5547 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5549 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5550 (_mm_rcp28_round_ss): Ditto.
5551 (_mm_rsqrt28_round_sd): Ditto.
5552 (_mm_rsqrt28_round_ss): Ditto.
5553 (_mm_rcp28_sd): Ditto.
5554 (_mm_rcp28_ss): Ditto.
5555 (_mm_rsqrt28_sd): Ditto.
5556 (_mm_rsqrt28_ss): Ditto.
5557 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5558 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5559 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5560 (IX86_BUILTIN_RCP28SD): Ditto.
5561 (IX86_BUILTIN_RCP28SS): Ditto.
5562 (IX86_BUILTIN_RSQRT28SD): Ditto.
5563 (IX86_BUILTIN_RSQRT28SS): Ditto.
5564 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5565 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5566 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5567 (ix86_expand_special_args_builtin): Expand new FTYPE.
5568 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5569 (srcp14<mode>): Make insn unary.
5570 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5571 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5572 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5573 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5574 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5575 Fix rounding: make it SAE only.
5576 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5578 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5580 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5581 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5582 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5583 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5584 (round_saeonly_mask_scalar_operand4): Ditto.
5585 (round_saeonly_mask_scalar_op3): Ditto.
5586 (round_saeonly_mask_scalar_op4): Ditto.
5588 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5590 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5591 Implement -maltivec=be for vec_insert and vec_extract.
5593 2014-01-10 DJ Delorie <dj@redhat.com>
5595 * config/msp430/msp430.md (call_internal): Don't allow memory
5596 references with SP as the base register.
5597 (call_value_internal): Likewise.
5598 * config/msp430/constraints.md (Yc): New. For memory references
5599 that don't use SP as a base register.
5601 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5602 "an integer without a # prefix"
5603 * config/msp430/msp430.md (epilogue_helper): Use it.
5605 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5608 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5609 AVX512F gather builtins.
5610 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5611 on gather decls with INTEGER_TYPE masktype.
5612 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5613 directly into the builtin rather than hoisting it before loop.
5615 PR tree-optimization/59387
5616 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5617 (scev_const_prop): If folded_casts and type has undefined overflow,
5618 use force_gimple_operand instead of force_gimple_operand_gsi and
5619 for each added stmt if it is assign with
5620 arith_code_with_undefined_signed_overflow, call
5621 rewrite_to_defined_overflow.
5622 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5623 gimple-fold.h instead.
5624 (arith_code_with_undefined_signed_overflow,
5625 rewrite_to_defined_overflow): Moved to ...
5626 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5627 rewrite_to_defined_overflow): ... here. No longer static.
5628 Include gimplify-me.h.
5629 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5630 rewrite_to_defined_overflow): New prototypes.
5632 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5634 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5636 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5638 * builtins.c (get_object_alignment_2): Minor tweak.
5639 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5641 2014-01-13 Christian Bruel <christian.bruel@st.com>
5643 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5644 optimized non constant lengths.
5646 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5649 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5650 load as __atomic_load_N if possible.
5652 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5654 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5656 (rs6000_expand_builtin): Adjust call.
5658 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5661 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5662 * config/rs6000/rs6000.c: Include target-globals.h.
5663 (rs6000_set_current_function): Instead of doing target_reinit
5664 unconditionally, use save_target_globals_default_opts and
5665 restore_target_globals.
5667 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5669 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5670 (rs6000_expand_builtin): Handle mffs and mtfsf.
5671 (rs6000_init_builtins): Define mffs and mtfsf.
5672 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5673 (rs6000_mffs): New pattern.
5674 (rs6000_mtfsf): New pattern.
5676 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5678 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5679 Start narrowing with START. Apply candidate-use pair
5680 and check overall cost in narrowing.
5681 (iv_ca_prune): Pass new argument.
5683 2014-01-10 Jeff Law <law@redhat.com>
5686 * ree.c (combine_reaching_defs): Ensure the defining statement
5687 occurs before the extension when optimizing extensions with
5688 different source and destination hard registers.
5690 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5693 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5694 vtables into the type inheritance graph.
5696 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5698 PR rtl-optimization/59754
5699 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5700 modes in the REGNO != REGNO case.
5702 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5704 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5706 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5708 PR tree-optimization/59745
5709 * tree-predcom.c (tree_predictive_commoning_loop): Call
5710 free_affine_expand_cache if giving up because components is NULL.
5712 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5713 GC in payload of target_globals struct instead of allocating them on
5714 the heap and the larger structs separately using GC.
5715 * target-globals.h (struct target_globals): Make regs, hard_regs,
5716 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5717 of GTY((skip)) and change type to void *.
5718 (reset_target_globals): Cast loads from those fields to corresponding
5721 2014-01-10 Steve Ellcey <sellcey@mips.com>
5724 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5725 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5726 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5728 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5731 * aarch64-modes.def (CC_Zmode): New flags mode.
5732 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5733 represents an equality.
5734 (aarch64_get_condition_code): Handle CC_Zmode.
5735 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5737 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5739 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5740 extraction in good case.
5742 2014-01-10 Richard Biener <rguenther@suse.de>
5744 PR tree-optimization/59374
5745 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5746 checking after SLP discovery. Mark stmts not participating
5747 in any SLP instance properly.
5749 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5751 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5752 when handling a SET rtx.
5754 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5756 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5757 (cortex-a57): Likewise.
5758 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5760 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5762 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5763 non-iwmmxt builtins.
5765 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5769 * ipa-devirt.c record_target_from_binfo): Take as argument
5770 stack of binfos and lookup matching one for virtual inheritance.
5771 (possible_polymorphic_call_targets_1): Update.
5773 2014-01-10 Huacai Chen <chenhc@lemote.com>
5775 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5776 kernel strings for Loongson-2E/2F/3A.
5778 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5781 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5782 is_gimple_call before calling gimple_call_internal_p.
5784 2014-01-09 Steve Ellcey <sellcey@mips.com>
5786 * Makefile.in (TREE_FLOW_H): Remove.
5787 (TREE_SSA_H): Add file names from tree-flow.h.
5788 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5789 * tree.h: Remove tree-flow.h reference.
5790 * hash-table.h: Remove tree-flow.h reference.
5791 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5792 reference with tree-ssa-loop.h.
5794 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5796 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5797 default element-order behavior for -maltivec.
5798 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5799 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5800 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5801 when targeting big endian, at least for now.
5802 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5804 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5807 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5808 var satisfies use_register_for_decl, just take into account type
5809 alignment, rather than decl alignment.
5811 PR tree-optimization/59622
5812 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5813 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5814 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5815 Don't devirtualize for inplace at all. For targets.length () == 1,
5816 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5818 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5820 * config/i386/i386.md (cpu): Remove the unused btver1.
5822 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5824 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5826 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5829 * tree-core.h (struct target_globals): New forward declaration.
5830 (struct tree_target_option): Add globals field.
5831 * tree.h (TREE_TARGET_GLOBALS): Define.
5832 (prepare_target_option_nodes_for_pch): New prototype.
5833 * target-globals.h (struct target_globals): Define even if
5835 * tree.c (prepare_target_option_node_for_pch,
5836 prepare_target_option_nodes_for_pch): New functions.
5837 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5838 * config/i386/i386.c: Include target-globals.h.
5839 (ix86_set_current_function): Instead of doing target_reinit
5840 unconditionally, use save_target_globals_default_opts and
5841 restore_target_globals.
5843 2014-01-09 Richard Biener <rguenther@suse.de>
5845 PR tree-optimization/59715
5846 * tree-cfg.h (split_critical_edges): Declare.
5847 * tree-cfg.c (split_critical_edges): Export.
5848 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5850 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5852 * cfgexpand.c (expand_stack_vars): Optionally disable
5853 asan stack protection.
5854 (expand_used_vars): Likewise.
5855 (partition_stack_vars): Likewise.
5856 * asan.c (asan_emit_stack_protection): Optionally disable
5857 after return stack usage.
5858 (instrument_derefs): Optionally disable memory access instrumentation.
5859 (instrument_builtin_call): Likewise.
5860 (instrument_strlen_call): Likewise.
5861 (asan_protect_global): Optionally disable global variables protection.
5862 * doc/invoke.texi: Added doc for new options.
5863 * params.def: Added new options.
5864 * params.h: Likewise.
5866 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5868 PR rtl-optimization/59724
5869 * ifcvt.c (cond_exec_process_if_block): Don't call
5870 flow_find_head_matching_sequence with 0 longest_match.
5871 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5872 non-active insns if !stop_after.
5873 (try_head_merge_bb): Revert 2014-01-07 changes.
5875 2014-01-08 Jeff Law <law@redhat.com>
5877 * ree.c (get_sub_rtx): New function, extracted from...
5878 (merge_def_and_ext): Here.
5879 (combine_reaching_defs): Use get_sub_rtx.
5881 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5883 * cgraph.h (varpool_variable_node): Do not choke on null node.
5885 2014-01-08 Catherine Moore <clm@codesourcery.com>
5887 * config/mips/mips.md (simple_return): Attempt to use JRC
5889 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5891 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5893 PR rtl-optimization/59137
5894 * reorg.c (steal_delay_list_from_target): Call update_block for
5896 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5898 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5900 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5901 two duplicate entries.
5903 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5906 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5908 * config/mips/mips.c (mips_truncated_op_cost): New function.
5909 (mips_rtx_costs): Adjust test for BADDU.
5910 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5912 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5914 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5915 (*baddu_si): ...this new pattern.
5917 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5920 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5922 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5925 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5927 (expand_expr, expand_normal): Adjust.
5928 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5929 inner_reference_p. Use inner_reference_p to expand inner references.
5930 (store_expr): Adjust.
5931 * cfgexpand.c (expand_call_stmt): Adjust.
5933 2014-01-08 Rong Xu <xur@google.com>
5935 * gcov-io.c (gcov_var): Move from gcov-io.h.
5936 (gcov_position): Ditto.
5937 (gcov_is_error): Ditto.
5938 (gcov_rewrite): Ditto.
5939 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5940 only part to libgcc/libgcov.h.
5942 2014-01-08 Marek Polacek <polacek@redhat.com>
5945 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5947 2014-01-08 Marek Polacek <polacek@redhat.com>
5950 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5952 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5954 PR rtl-optimization/59649
5955 * stor-layout.c (get_mode_bounds): For BImode return
5956 0 and STORE_FLAG_VALUE.
5958 2014-01-08 Richard Biener <rguenther@suse.de>
5961 * gimple.h (is_gimple_builtin_call): Remove.
5962 (gimple_builtin_call_types_compatible_p): New.
5963 (gimple_call_builtin_p): New overload.
5964 * gimple.c (is_gimple_builtin_call): Remove.
5965 (validate_call): Rename to ...
5966 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5968 (validate_type): New static function.
5969 (gimple_call_builtin_p): New overload and adjust.
5970 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5971 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5972 (gimple_fold_stmt_to_constant_1): Likewise.
5973 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5975 2014-01-08 Richard Biener <rguenther@suse.de>
5978 * gimplify.c (gimplify_expr): Gimplify register-register type
5979 VIEW_CONVERT_EXPRs to separate stmts.
5981 2014-01-07 Jeff Law <law@redhat.com>
5984 * ree.c (combine_set_extension): Handle case where source
5985 and destination registers in an extension insn are different.
5986 (combine_reaching_defs): Allow source and destination registers
5987 in extension to be different under limited circumstances.
5988 (add_removable_extension): Remove restriction that the
5989 source and destination registers in the extension are the same.
5990 (find_and_remove_re): Emit a copy from the extension's
5991 destination to its source after the defining insn if
5992 the source and destination registers are different.
5995 * ifcvt.c (merge_if_block): If we are merging a block with more than
5996 one successor with a block with no successors, remove any BARRIER
5997 after the second block.
5999 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6001 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6003 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6006 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6007 for 14-bit register offsets when INT14_OK_STRICT is false.
6009 2014-01-07 Roland Stigge <stigge@antcom.de>
6010 Michael Meissner <meissner@linux.vnet.ibm.com>
6013 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6014 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6016 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6018 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6021 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6023 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6024 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6025 rtx is const0_rtx or not.
6027 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6030 * target-globals.c (save_target_globals): Remove this_fn_optab
6032 * toplev.c: Include optabs.h.
6033 (target_reinit): Temporarily restore the global options if another
6034 set of options are in force.
6036 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6038 PR rtl-optimization/58668
6039 * cfgcleanup.c (flow_find_cross_jump): Don't count
6040 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6041 to determine what is counted.
6042 (flow_find_head_matching_sequence): Use active_insn_p to determine
6044 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6046 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6047 determine what is counted.
6049 PR tree-optimization/59643
6050 * tree-predcom.c (split_data_refs_to_components): If one dr is
6051 read and one write, determine_offset fails and the write isn't
6052 in the bad component, just put the read into the bad component.
6054 2014-01-07 Mike Stump <mikestump@comcast.net>
6055 Jakub Jelinek <jakub@redhat.com>
6058 * tree-core.h (struct tree_optimization_option): Change optabs
6059 type from unsigned char * to void *.
6060 * optabs.c (init_tree_optimization_optabs): Adjust
6061 TREE_OPTIMIZATION_OPTABS initialization.
6063 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6066 * config/i386/i386.h (struct machine_function): Add
6067 no_drap_save_restore field.
6068 * config/i386/i386.c (ix86_save_reg): Use
6069 !cfun->machine->no_drap_save_restore instead of
6070 crtl->stack_realign_needed.
6071 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6072 this function clears frame_pointer_needed. Set
6073 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6074 and DRAP reg is needed.
6076 2014-01-06 Marek Polacek <polacek@redhat.com>
6079 * doc/implement-c.texi: Mention that other integer types are
6080 permitted as bit-field types in strictly conforming mode.
6082 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6084 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6087 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6089 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6091 2014-01-06 Martin Jambor <mjambor@suse.cz>
6094 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6096 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6098 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6102 * var-tracking.c (add_stores): Preserve the value of the source even if
6103 we don't record the store.
6105 2014-01-06 Terry Guo <terry.guo@arm.com>
6107 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6109 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6112 * config/darwin.c (darwin_function_section): Adjust return values to
6113 correspond to optimisation changes made in r206070.
6115 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6117 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6118 from prefetch_block tune setting.
6119 (nocona_cost): Correct size of prefetch block to 64.
6121 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6123 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6124 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6125 used to save the static chain register in the computation of the offset
6126 from which the FP registers need to be restored.
6128 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6130 PR tree-optimization/59519
6131 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6132 ICE if get_current_def (current_new_name) is already non-NULL, as long
6133 as it is a phi result of some other phi in *new_exit_bb that has
6136 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6137 or vmovdqu* for misaligned_operand.
6138 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6139 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6140 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6141 aligned_mem for AVX512F masked aligned load and store builtins and for
6144 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6146 PR tree-optimization/59651
6147 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6148 Address range for negative step should be added by TYPE_SIZE_UNIT.
6150 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6152 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6153 overlapping registers also for registers other than the stack pointer.
6155 2014-01-03 Marek Polacek <polacek@redhat.com>
6158 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6161 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6164 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6167 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6168 (push splitter): Use <P:MODE_SIZE> instead of
6169 GET_MODE_SIZE (<P:MODE>mode).
6170 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6171 (mov -1, reg peephole2): Likewise.
6172 * config/i386/sse.md (*mov<mode>_internal,
6173 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6174 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6175 *<code><mode>3, *andnot<mode>3<mask_name>,
6176 <mask_codefor><code><mode>3<mask_name>): Likewise.
6177 * config/i386/subst.md (mask_mode512bit_condition,
6178 sd_mask_mode512bit_condition): Likewise.
6180 2014-01-02 Xinliang David Li <davidxl@google.com>
6182 PR tree-optimization/59303
6183 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6184 (dump_predicates): Better output format.
6185 (pred_equal_p): New function.
6186 (is_neq_relop_p): Ditto.
6187 (is_neq_zero_form_p): Ditto.
6188 (pred_expr_equal_p): Ditto.
6189 (pred_neg_p): Ditto.
6190 (simplify_pred): Ditto.
6191 (simplify_preds_2): Ditto.
6192 (simplify_preds_3): Ditto.
6193 (simplify_preds_4): Ditto.
6194 (simplify_preds): Ditto.
6196 (push_to_worklist): Ditto.
6197 (get_pred_info_from_cmp): Ditto.
6198 (is_degenerated_phi): Ditto.
6199 (normalize_one_pred_1): Ditto.
6200 (normalize_one_pred): Ditto.
6201 (normalize_one_pred_chain): Ditto.
6202 (normalize_preds): Ditto.
6203 (normalize_cond_1): Remove function.
6204 (normalize_cond): Ditto.
6205 (is_gcond_subset_of): Ditto.
6206 (is_subset_of_any): Ditto.
6207 (is_or_set_subset_of): Ditto.
6208 (is_and_set_subset_of): Ditto.
6209 (is_norm_cond_subset_of): Ditto.
6210 (pred_chain_length_cmp): Ditto.
6211 (convert_control_dep_chain_into_preds): Type change.
6212 (find_predicates): Ditto.
6213 (find_def_preds): Ditto.
6214 (destroy_predicates_vecs): Ditto.
6215 (find_matching_predicates_in_rest_chains): Ditto.
6216 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6217 (is_pred_expr_subset): Ditto.
6218 (is_pred_chain_subset_of): Ditto.
6219 (is_included_in): Ditto.
6220 (is_superset_of): Ditto.
6222 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6224 Update copyright years.
6226 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6228 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6229 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6230 config/arc/arc.md, config/arc/arc.opt,
6231 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6232 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6233 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6234 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6235 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6236 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6237 vtable-verify.c, vtable-verify.h: Use the standard form for the
6240 2014-01-02 Tobias Burnus <burnus@net-b.de>
6242 * gcc.c (process_command): Update copyright notice dates.
6243 * gcov-dump.c: Ditto.
6245 * doc/cpp.texi: Bump @copying's copyright year.
6246 * doc/cppinternals.texi: Ditto.
6247 * doc/gcc.texi: Ditto.
6248 * doc/gccint.texi: Ditto.
6249 * doc/gcov.texi: Ditto.
6250 * doc/install.texi: Ditto.
6251 * doc/invoke.texi: Ditto.
6253 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6255 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6257 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6259 * config/i386/sse.md (*mov<mode>_internal): Guard
6260 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6262 PR rtl-optimization/59647
6263 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6264 new_rtx into UNSIGNED_FLOAT rtxes.
6266 Copyright (C) 2014 Free Software Foundation, Inc.
6268 Copying and distribution of this file, with or without modification,
6269 are permitted in any medium without royalty provided the copyright
6270 notice and this notice are preserved.