1 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
3 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
4 around for store forwarding issue in the FPU on the UT699.
5 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
6 loads and operations if -mfix-ut699 is specified.
7 (divtf3_hq): Tweak attribute.
8 (sqrttf2_hq): Likewise.
10 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
12 * calls.c (store_one_arg): Remove incorrect const qualification on the
13 type of the temporary.
14 * cfgexpand.c (expand_return): Likewise.
15 * expr.c (expand_constructor): Likewise.
16 (expand_expr_real_1): Likewise.
18 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
20 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
23 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
26 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
28 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
30 * config/arm/aarch-common-protos.h
31 (alu_cost_table): Fix spelling of "extend".
32 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
34 2014-03-19 Richard Biener <rguenther@suse.de>
37 * tree-core.h (tree_type_common): Re-order pointer members
38 to reduce recursion depth during GC walks.
40 2014-03-19 Marek Polacek <polacek@redhat.com>
43 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
46 2014-03-19 Richard Biener <rguenther@suse.de>
49 * lto-streamer-in.c (input_function): In WPA stage do not drop
52 2014-03-19 Jakub Jelinek <jakub@redhat.com>
54 PR tree-optimization/60559
55 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
56 with build_zero_cst assignment.
58 2014-03-18 Kai Tietz <ktietz@redhat.com>
60 PR rtl-optimization/56356
61 * sdbout.c (sdbout_parms): Verify that parms'
62 incoming argument is valid.
63 (sdbout_reg_parms): Likewise.
65 2014-03-18 Richard Henderson <rth@redhat.com>
68 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
69 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
70 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
72 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
74 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
75 * doc/plugins.texi (Plugin callbacks): Mention
77 Italicize plugin event names in description. Explain that
78 PLUGIN_PRAGMAS has no sense for lto1. Explain
80 Remind that no GCC functions should be called after
82 Explain what pragmas with expansion are.
84 2014-03-18 Martin Liska <mliska@suse.cz>
86 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
87 gimple call statement is update.
88 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
89 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
91 2014-03-18 Jakub Jelinek <jakub@redhat.com>
94 * ubsan.c (ubsan_instrument_unreachable): Call
95 initialize_sanitizer_builtins.
96 (ubsan_pass): Likewise.
99 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
100 varpool_finalize_decl instead of rest_of_decl_compilation.
102 2014-03-18 Richard Biener <rguenther@suse.de>
104 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
105 by using bitmap_and_compl instead of bitmap_and_compl_into.
106 (df_rd_transfer_function): Likewise.
108 2014-03-18 Richard Biener <rguenther@suse.de>
110 * doc/lto.texi (fresolution): Fix typo.
112 2014-03-18 Richard Biener <rguenther@suse.de>
114 * doc/invoke.texi (flto): Update for changes in 4.9.
116 2014-03-18 Richard Biener <rguenther@suse.de>
118 * doc/loop.texi: Remove section on the removed lambda framework.
119 Update loop docs with recent changes in preserving loop structure.
121 2014-03-18 Richard Biener <rguenther@suse.de>
123 * doc/lto.texi (-fresolution): Document.
125 2014-03-18 Richard Biener <rguenther@suse.de>
127 * doc/contrib.texi: Adjust my name.
129 2014-03-18 Jakub Jelinek <jakub@redhat.com>
132 * internal-fn.c: Include diagnostic-core.h.
133 (expand_BUILTIN_EXPECT): New function.
134 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
135 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
136 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
137 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
139 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
140 Revert 3 argument __builtin_expect code.
141 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
142 * gimple-fold.c (gimple_fold_call): Likewise.
143 * tree.h (fold_builtin_expect): New prototype.
144 * builtins.c (build_builtin_expect_predicate): Add predictor
145 argument, if non-NULL, create 3 argument __builtin_expect.
146 (fold_builtin_expect): No longer static. Add ARG2 argument,
147 pass it through to build_builtin_expect_predicate.
148 (fold_builtin_2): Adjust caller.
149 (fold_builtin_3): Handle BUILT_IN_EXPECT.
150 * internal-fn.def (BUILTIN_EXPECT): New.
152 2014-03-18 Tobias Burnus <burnus@net-b.de>
155 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
156 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
157 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
159 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
162 * predict.c (combine_predictions_for_bb): Fix up formatting.
163 (expr_expected_value_1, expr_expected_value): Add predictor argument,
164 fill what it points to if non-NULL.
165 (tree_predict_by_opcode): Adjust caller, use the predictor.
166 * predict.def (PRED_COMPARE_AND_SWAP): Add.
168 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
170 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
171 proper constant for the store mode.
173 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
175 * symtab.c (change_decl_assembler_name): Fix transparent alias
178 2014-03-16 Renlin Li <Renlin.Li@arm.com>
180 * config/aarch64/aarch64.c: Correct the comments about the
181 aarch64 stack layout.
183 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
185 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
186 check for GF_OMP_FOR_KIND_FOR.
188 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
190 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
191 ymm and zmm register names.
193 2014-03-17 Jakub Jelinek <jakub@redhat.com>
196 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
197 note creation for the 2010-08-31 changes.
199 2014-03-17 Marek Polacek <polacek@redhat.com>
202 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
203 as -fno-tree-loop-vectorize.
204 (expand_omp_simd): Likewise.
206 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
208 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
209 (eligible_for_call_delay): New prototype.
210 * config/sparc/sparc.c (tls_call_delay): Rename into...
211 (eligible_for_call_delay): ...this. Return false if the instruction
212 cannot be put in the delay slot of a branch.
213 (eligible_for_restore_insn): Simplify.
214 (eligible_for_return_delay): Return false if the instruction cannot be
215 put in the delay slot of a branch and simplify.
216 (eligible_for_sibcall_delay): Return false if the instruction cannot be
217 put in the delay slot of a branch.
218 * config/sparc/sparc.md (fix_ut699): New attribute.
219 (tls_call_delay): Delete.
220 (in_call_delay): Reimplement.
221 (eligible_for_sibcall_delay): Rename into...
222 (in_sibcall_delay): ...this.
223 (eligible_for_return_delay): Rename into...
224 (in_return_delay): ...this.
225 (in_branch_delay): Reimplement.
226 (in_uncond_branch_delay): Delete.
227 (in_annul_branch_delay): Delete.
229 2014-03-14 Richard Henderson <rth@redhat.com>
232 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
233 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
234 (*floathi<X87MODEF>2_i387_with_temp): Remove.
235 (floathi splitters): Remove.
236 (float<SWI48x>xf2): New pattern.
237 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
238 code that tried to handle DImode for 32-bit, but which was excluded
239 by the pattern's condition. Drop allocation of stack temporary.
240 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
241 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
242 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
243 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
244 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
245 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
246 (*float<SWI48><MODEF>2_sse_interunit): Remove.
247 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
248 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
249 (*float<SWI48x><X87MODEF>2_i387): Remove.
250 (all float _with_temp splitters): Remove.
251 (*float<SWI48x><MODEF>2_i387): New pattern.
252 (*float<SWI48><MODEF>2_sse): New pattern.
253 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
254 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
256 2014-03-14 Jakub Jelinek <jakub@redhat.com>
257 Marek Polacek <polacek@redhat.com>
260 * common.opt (dump_base_name_prefixed): New Variable.
261 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
262 if x_dump_base_name_prefixed is already set, set it at the end.
264 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
266 PR rtl-optimization/60508
267 * lra-constraints.c (get_reload_reg): Add new parameter
269 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
270 Pass the new parameter values.
272 2014-03-14 Richard Biener <rguenther@suse.de>
274 * common.opt: Revert unintented changes from r205065.
277 2014-03-14 Richard Biener <rguenther@suse.de>
280 * cfghooks.c (split_block): Properly adjust all loops the
281 block was a latch of.
283 2014-03-14 Martin Jambor <mjambor@suse.cz>
286 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
289 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
292 * config/avr/avr.c (avr_set_current_function): Pass function name
293 through default_strip_name_encoding before sanity checking instead
294 of skipping the first char of the assembler name.
296 2014-03-13 Richard Henderson <rth@redhat.com>
299 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
300 (ix86_force_to_memory, ix86_free_from_memory): Remove.
301 * config/i386/i386-protos.h: Likewise.
302 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
303 in the expander instead of a splitter.
304 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
305 any possibility of requiring a memory.
306 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
307 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
308 (fp branch splitters): Update for ix86_split_fp_branch.
309 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
310 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
311 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
312 (*fop_<MODEF>_2_i387): Remove f/r alternative.
313 (*fop_<MODEF>_3_i387): Likewise.
314 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
315 (splitters for the fop_* register patterns): Remove.
316 (fscalexf4_i387): Rename from *fscalexf4_i387.
317 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
319 2014-03-13 Jakub Jelinek <jakub@redhat.com>
321 PR tree-optimization/59779
322 * tree-dfa.c (get_ref_base_and_extent): Use double_int
323 type for bitsize and maxsize instead of HOST_WIDE_INT.
325 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
327 PR rtl-optimization/57320
328 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
329 the CFG after thread_prologue_and_epilogue_insns.
331 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
333 PR rtl-optimization/57189
334 * lra-constraints.c (process_alt_operands): Disfavor spilling
337 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
339 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
342 2014-03-13 Jakub Jelinek <jakub@redhat.com>
344 PR tree-optimization/59025
346 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
347 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
349 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
352 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
353 calls of avr_out_plus_1.
355 2014-03-13 Bin Cheng <bin.cheng@arm.com>
357 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
358 BB's single pred and update the father loop's latch info later.
360 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
362 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
366 (VEC_base): Likewise.
367 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
368 registers, we need to swap double words in little endian mode.
370 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
371 to be a container mode for 128-bit integer operations added in ISA
372 2.07. Unlike TImode and PTImode, the preferred register set is
373 the Altivec/VMX registers for the 128-bit operations.
375 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
377 (rs6000_split_128bit_ok_p): Likewise.
379 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
380 macros for creating ISA 2.07 normal and overloaded builtin
381 functions with 3 arguments.
382 (BU_P8V_OVERLOAD_3): Likewise.
383 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
384 for use as overloaded functions.
385 (VPERM_1TI_UNS): Likewise.
386 (VSEL_1TI): Likewise.
387 (VSEL_1TI_UNS): Likewise.
388 (ST_INTERNAL_1ti): Likewise.
389 (LD_INTERNAL_1ti): Likewise.
390 (XXSEL_1TI): Likewise.
391 (XXSEL_1TI_UNS): Likewise.
392 (VPERM_1TI): Likewise.
393 (VPERM_1TI_UNS): Likewise.
394 (XXPERMDI_1TI): Likewise.
396 (LXVD2X_V1TI): Likewise.
397 (STXVD2X_V1TI): Likewise.
398 (VEC_INIT_V1TI): Likewise.
399 (VEC_SET_V1TI): Likewise.
400 (VEC_EXT_V1TI): Likewise.
401 (EQV_V1TI): Likewise.
402 (NAND_V1TI): Likewise.
403 (ORC_V1TI): Likewise.
404 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
405 added in ISA 2.07. Add both normal 'altivec' builtins, and the
409 (VADDEUQM): Likewise.
410 (VADDECUQ): Likewise.
411 (VSUBEUQM): Likewise.
412 (VSUBECUQ): Likewise.
414 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
415 __int128_t and __uint128_t types.
416 (__uint128_type): Likewise.
417 (altivec_categorize_keyword): Add support for vector __int128_t,
418 vector __uint128_t, vector __int128, and vector unsigned __int128
419 as a container type for TImode operations that need to be done in
420 VSX/Altivec registers.
421 (rs6000_macro_to_expand): Likewise.
422 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
423 to support 128-bit integer instructions vaddcuq, vadduqm,
424 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
425 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
427 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
428 for V1TImode, and set up preferences to use VSX/Altivec registers.
429 Setup VSX reload handlers.
430 (rs6000_debug_reg_global): Likewise.
431 (rs6000_init_hard_regno_mode_ok): Likewise.
432 (rs6000_preferred_simd_mode): Likewise.
433 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
434 (easy_altivec_constant): Likewise.
435 (output_vec_const_move): Likewise.
436 (rs6000_expand_vector_set): Convert V1TImode set and extract to
438 (rs6000_expand_vector_extract): Likewise.
439 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
441 (rs6000_const_vec): Add support for V1TImode.
442 (rs6000_emit_le_vsx_load): Swap double words when loading or
443 storing TImode/V1TImode.
444 (rs6000_emit_le_vsx_store): Likewise.
445 (rs6000_emit_le_vsx_move): Likewise.
446 (rs6000_emit_move): Add support for V1TImode.
447 (altivec_expand_ld_builtin): Likewise.
448 (altivec_expand_st_builtin): Likewise.
449 (altivec_expand_vec_init_builtin): Likewise.
450 (altivec_expand_builtin): Likewise.
451 (rs6000_init_builtins): Add support for V1TImode type. Add
452 support for ISA 2.07 128-bit integer builtins. Define type names
453 for the VSX/Altivec vector types.
454 (altivec_init_builtins): Add support for overloaded vector
455 functions with V1TImode type.
456 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
457 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
459 (rs6000_split_128bit_ok_p): Likewise.
460 (rs6000_handle_altivec_attribute): Create V1TImode from vector
461 __int128_t and vector __uint128_t.
463 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
471 (VS_scalar): Likewise.
472 (VS_double): Likewise.
473 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
475 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
476 we support the ISA 2.07 128-bit integer arithmetic instructions.
477 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
478 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
479 and TImode types for use with the builtin functions.
480 (V1TI_type_node): Likewise.
481 (unsigned_V1TI_type_node): Likewise.
482 (intTI_type_internal_node): Likewise.
483 (uintTI_type_internal_node): Likewise.
485 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
486 128-bit builtin functions.
487 (UNSPEC_VADDEUQM): Likewise.
488 (UNSPEC_VADDECUQ): Likewise.
489 (UNSPEC_VSUBCUQ): Likewise.
490 (UNSPEC_VSUBEUQM): Likewise.
491 (UNSPEC_VSUBECUQ): Likewise.
492 (VM): Add V1TImode to vector mode iterators.
495 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
496 (altivec_vaddcuq): Likewise.
497 (altivec_vsubuqm): Likewise.
498 (altivec_vsubcuq): Likewise.
499 (altivec_vaddeuqm): Likewise.
500 (altivec_vaddecuq): Likewise.
501 (altivec_vsubeuqm): Likewise.
502 (altivec_vsubecuq): Likewise.
504 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
506 (BOOL_128): Likewise.
507 (BOOL_REGS_OUTPUT): Likewise.
508 (BOOL_REGS_OP1): Likewise.
509 (BOOL_REGS_OP2): Likewise.
510 (BOOL_REGS_UNARY): Likewise.
511 (BOOL_REGS_AND_CR0): Likewise.
513 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
514 128-bit integer builtin support.
515 (vec_vadduqm): Likewise.
516 (vec_vaddecuq): Likewise.
517 (vec_vaddeuqm): Likewise.
518 (vec_vsubecuq): Likewise.
519 (vec_vsubeuqm): Likewise.
520 (vec_vsubcuq): Likewise.
521 (vec_vsubuqm): Likewise.
523 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
524 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
525 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
526 128-bit integer add/subtract to ISA 2.07.
528 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
530 * config/arc/arc.c (arc_predicate_delay_insns):
531 Fix third argument passed to conditionalize_nonjump.
533 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
535 * config/aarch64/aarch64-builtins.c
536 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
537 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
538 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
539 instead of __builtin_lfloor.
540 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
542 2014-03-12 Jakub Jelinek <jakub@redhat.com>
544 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
545 (tree_ssa_ifcombine_bb_1): New function.
546 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
547 is an empty forwarder block to then_bb or vice versa and then_bb
548 and else_bb are effectively swapped.
550 2014-03-12 Christian Bruel <christian.bruel@st.com>
553 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
554 REG_CFA_DEF_CFA note.
555 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
556 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
558 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
560 PR tree-optimization/60454
561 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
563 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
565 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
566 Do not define target_cpu_default2 to generic.
567 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
568 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
569 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
571 2014-03-12 Jakub Jelinek <jakub@redhat.com>
572 Marc Glisse <marc.glisse@inria.fr>
574 PR tree-optimization/60502
575 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
576 instead of build_low_bits_mask.
578 2014-03-12 Jakub Jelinek <jakub@redhat.com>
581 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
582 if there are multiple uses, but op doesn't live on E edge.
583 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
584 clobber stmts before __builtin_unreachable.
586 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
588 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
589 hard_frame_pointer_rtx.
590 * cse.c (cse_insn): Remove volatile check.
591 * cselib.c (cselib_process_insn): Likewise.
592 * dse.c (scan_insn): Likewise.
594 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
596 * config/arc/arc.c (conditionalize_nonjump): New function,
598 (arc_ifcvt): ... this.
599 (arc_predicate_delay_insns): Use it.
601 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
603 * config/arc/predicates.md (extend_operand): During/after reload,
604 allow const_int_operand.
605 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
606 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
607 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
609 (umulsi3_highpart_i): Likewise.
611 2014-03-11 Richard Biener <rguenther@suse.de>
613 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
614 Add asserts to guard possible wrong-code bugs.
616 2014-03-11 Richard Biener <rguenther@suse.de>
618 PR tree-optimization/60429
619 PR tree-optimization/60485
620 * tree-ssa-structalias.c (set_union_with_increment): Properly
621 take into account all fields that overlap the shifted vars.
622 (do_sd_constraint): Likewise.
623 (do_ds_constraint): Likewise.
624 (get_constraint_for_ptr_offset): Likewise.
626 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
628 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
629 (nios2_compute_frame_layout):
630 Add calculation of cfun->machine->fp_save_offset.
631 (nios2_expand_prologue): Correct setting of frame pointer register
633 (nios2_expand_epilogue): Update recovery of stack pointer from
634 frame pointer accordingly.
635 (nios2_initial_elimination_offset): Update calculation of offset
636 for eliminating to HARD_FRAME_POINTER_REGNUM.
638 2014-03-10 Jakub Jelinek <jakub@redhat.com>
641 * ipa.c (symtab_remove_unreachable_nodes): Don't call
642 cgraph_get_create_node on VAR_DECLs.
644 2014-03-10 Richard Biener <rguenther@suse.de>
647 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
649 2014-03-08 Douglas B Rupp <rupp@gnat.com>
651 * config/vms/vms.opt (vms_float_format): New variable.
653 2014-03-08 Tobias Burnus <burnus@net-b.de>
655 * doc/invoke.texi (-fcilkplus): Update implementation status.
657 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
658 Richard Biener <rguenther@suse.de>
660 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
661 consistently accross all TUs.
662 (run_gcc): Enable -fshort-double automatically at link at link-time
663 and disallow override.
665 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
668 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
669 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
670 if they can't be used.
672 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
674 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
675 for Solaris 11/x86 ld.
676 * configure: Regenerate.
678 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
680 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
681 (LIB_TLS_SPEC): Save as ld_tls_libs.
682 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
683 (HAVE_AS_IX86_TLSLDM): New test.
684 * configure, config.in: Regenerate.
685 * config/i386/i386.c (legitimize_tls_address): Fall back to
686 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
687 cannot support TLS_MODEL_LOCAL_DYNAMIC.
688 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
689 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
691 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
693 * common.opt (fira-loop-pressure): Mark as optimization.
695 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
697 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
698 an OpenMP mappable type.
700 2014-03-06 Matthias Klose <doko@ubuntu.com>
702 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
703 MULTILIB_OSDIRNAMES is not defined.
705 2014-03-06 Jakub Jelinek <jakub@redhat.com>
706 Meador Inge <meadori@codesourcery.com>
709 * config/arm/arm.c (arm_tls_symbol_p): Remove.
710 (arm_legitimize_address): Call legitimize_tls_address for any
711 arm_tls_referenced_p expression, handle constant addend. Call it
712 before testing for !TARGET_ARM.
713 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
715 2014-03-06 Richard Biener <rguenther@suse.de>
721 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
723 * tree-streamer.c (record_common_node): Assert we don't record
724 nodes with type double.
725 (preload_common_node): Skip type double, complex double and double
726 pointer since it is now frontend dependent due to fshort-double option.
728 2014-03-06 Richard Biener <rguenther@suse.de>
730 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
731 or -fno-lto is specified and the linker has full plugin support.
732 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
733 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
734 * lto-wrapper.c (merge_and_complain): Merge compile-time
736 (run_gcc): And pass it through to the link options.
738 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
742 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
744 * cselib.c (remove_useless_values): Skip to avoid quadratic
745 behavior if the condition moved from...
746 (cselib_process_insn): ... here holds.
748 2014-03-05 Jakub Jelinek <jakub@redhat.com>
751 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
752 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
755 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
756 (TM_H): Add x86-tune.def.
758 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
760 * config/aarch64/aarch64.c (generic_tunings):
761 Use cortexa57_extra_costs.
763 2014-03-05 Jakub Jelinek <jakub@redhat.com>
766 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
767 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
768 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
771 2014-03-04 Heiher <r@hev.cc>
773 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
774 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
776 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
778 * config/i386/predicates.md (const2356_operand): Change to ...
779 (const2367_operand): ... this.
780 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
782 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
783 (*avx512pf_scatterpf<mode>sf): Ditto.
784 (avx512pf_scatterpf<mode>df): Ditto.
785 (*avx512pf_scatterpf<mode>df_mask): Ditto.
786 (*avx512pf_scatterpf<mode>df): Ditto.
787 * config/i386/i386.c (ix86_expand_builtin): Update
788 incorrect hint operand error message.
790 2014-03-04 Richard Biener <rguenther@suse.de>
792 * lto-section-in.c (lto_get_section_data): Fix const cast.
794 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
796 * tree-streamer.c (record_common_node): Assert we don't record
797 nodes with type double.
798 (preload_common_node): Skip type double, complex double and double
799 pointer since it is now frontend dependent due to fshort-double option.
801 2014-03-04 Richard Biener <rguenther@suse.de>
804 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
805 (lto_input_toplevel_asms): Likewise.
806 * lto-section-in.c (lto_get_section_data): Instead do it here
809 2014-03-04 Richard Biener <rguenther@suse.de>
811 PR tree-optimization/60382
812 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
813 dead PHIs a reduction.
815 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
817 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
819 (_mm_prefetch): Move out of GCC target("sse") pragma.
820 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
821 GCC target("prfchw") pragma.
822 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
824 * config/i386/i386.c (ix86_option_override_internal): Enable
825 -mprfchw with -mprefetchwt1.
827 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
829 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
832 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
834 * opts.h (CL_PCH_IGNORE): Define.
835 * targhooks.c (option_affects_pch_p):
836 Return false for options that have CL_PCH_IGNORE set.
837 * opt-functions.awk: Process PchIgnore.
838 * doc/options.texi: Document PchIgnore.
840 * config/arc/arc.opt (misize): Add PchIgnore property.
842 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
844 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
845 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
846 constraint on constants to permit them being loaded into
847 GENERAL_REGS or BASE_REGS.
849 2014-03-03 Nick Clifton <nickc@redhat.com>
851 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
852 anti-cacnonical alternatives.
853 (negandhi3_real): New pattern.
854 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
856 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
858 * config/avr/avr-mcus.def: Remove atxmega16x1.
859 * config/avr/avr-tables.opt: Regenerate.
860 * config/avr/t-multilib: Regenerate.
861 * doc/avr-mmcu.texi: Regenerate.
863 2014-03-03 Tobias Grosser <tobias@grosser.es>
864 Mircea Namolaru <mircea.namolaru@inria.fr>
866 PR tree-optimization/58028
867 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
870 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
872 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
873 not handled by recognizers.
875 2014-03-03 Jakub Jelinek <jakub@redhat.com>
878 * function.c (expand_function_end): Don't emit
879 clobber_return_register sequence if clobber_after is a BARRIER.
880 * cfgexpand.c (construct_exit_block): Append instructions before
881 return_label to prev_bb.
883 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
885 * config/rs6000/constraints.md: Document reserved use of "wc".
887 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
890 * ipa.c (function_and_variable_visibility): When dissolving comdat
891 group, also set all symbols to local.
893 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
898 2013-12-14 Jan Hubicka <jh@suse.cz>
900 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
902 2014-03-02 Jon Beniston <jon@beniston.com>
908 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
909 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
910 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
911 (simple_return, *simple_return): New patterns
912 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
913 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
915 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
917 * dwarf2out.c (gen_subprogram_die): Tidy.
919 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
922 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
923 (*mov_t_msb_neg_negc): ... this new insn.
925 2014-02-28 Jason Merrill <jason@redhat.com>
928 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
931 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
934 * dwarf2out.c (decltype_auto_die): New static.
935 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
936 (gen_type_die_with_usage): Handle 'decltype(auto)'.
937 (is_cxx_auto): Likewise.
939 2014-02-28 Ian Bolton <ian.bolton@arm.com>
941 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
942 we are not using general regs only.
944 2014-02-28 Richard Biener <rguenther@suse.de>
947 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
948 previous fix and only allow to remove trivial pre-headers
949 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
950 (remove_forwarder_block): Properly update the latch of a loop.
952 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
955 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
956 (cselib_preserved_hash_table): New.
957 (preserve_constants_and_equivs): Move preserved vals to it.
958 (cselib_find_slot): Look it up first.
959 (cselib_init): Initialize it.
960 (cselib_finish): Release it.
961 (dump_cselib_table): Dump it.
963 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
966 * cselib.c (remove_useless_values): Skip to avoid quadratic
967 behavior if the condition moved from...
968 (cselib_process_insn): ... here holds.
970 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
973 * var-tracking.c (vt_initialize): Apply the same condition to
974 preserve the CFA base value.
976 2014-02-28 Joey Ye <joey.ye@arm.com>
979 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
980 if reload in progress or completed.
982 2014-02-28 Tobias Burnus <burnus@net-b.de>
985 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
988 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
990 * doc/tm.texi.in (Condition Code Status): Update documention for
991 relative locations of cc0-setter and cc0-user.
993 2014-02-27 Jeff Law <law@redhat.com>
995 PR rtl-optimization/52714
996 * combine.c (try_combine): When splitting an unrecognized PARALLEL
997 into two independent simple sets, if I3 is a jump, ensure the
998 pattern we place into I3 is a (set (pc) ...).
1000 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1001 Jeff Law <law@redhat.com>
1003 PR rtl-optimization/49847
1004 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1005 are in different blocks.
1006 * doc/tm.texi (Condition Code Status): Update documention for
1007 relative locations of cc0-setter and cc0-user.
1009 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1012 * lra.c (lra_emit_add): Check SUBREG too.
1014 2014-02-27 Andreas Schwab <schwab@suse.de>
1016 * config/m68k/m68k.c (m68k_option_override): Disable
1017 -flive-range-shrinkage for classic m68k.
1018 (m68k_override_options_after_change): Likewise.
1020 2014-02-27 Marek Polacek <polacek@redhat.com>
1023 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1024 -Wmaybe-uninitialized.
1026 2014-02-27 Alan Modra <amodra@gmail.com>
1029 * reload1.c (emit_input_reload_insns): When reload_override_in,
1030 set old to rl->in_reg when rl->in_reg is a subreg.
1032 2014-02-26 Richard Biener <rguenther@suse.de>
1035 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1037 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1039 * common/config/i386/predicates.md (const1256_operand): Remove.
1040 (const2356_operand): New.
1041 (const_1_to_2_operand): Remove.
1042 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1043 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1044 (*avx512pf_gatherpf<mode>sf): Ditto.
1045 (avx512pf_gatherpf<mode>df): Ditto.
1046 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1047 (*avx512pf_gatherpf<mode>df): Ditto.
1048 (avx512pf_scatterpf<mode>sf): Ditto.
1049 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1050 (*avx512pf_scatterpf<mode>sf): Ditto.
1051 (avx512pf_scatterpf<mode>df): Ditto.
1052 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1053 (*avx512pf_scatterpf<mode>df): Ditto.
1054 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1056 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1058 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1059 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1060 (_mm512_mask_testn_epi64_mask): Move to ...
1061 * config/i386/avx512cdintrin.h: Here.
1062 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1063 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1064 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1065 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1066 TARGET_AVX512F from TARGET_AVX512CD.
1068 2014-02-26 Richard Biener <rguenther@suse.de>
1071 * ipa.c (walk_polymorphic_call_targets): Properly guard
1072 call to inline_update_overall_summary.
1074 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1077 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1078 and latches only if requested. Fix latch if it is removed.
1079 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1080 LOOPS_HAVE_PREHEADERS.
1082 2014-02-25 Andrew Pinski <apinski@cavium.com>
1084 * builtins.c (expand_builtin_thread_pointer): Create a new target
1085 when the target is NULL.
1087 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1089 PR rtl-optimization/60317
1090 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1091 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1092 * lra-assigns.c: Include params.h.
1093 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1094 other reload pseudos considerations.
1096 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1098 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1099 to use canonical form for nor<mode>3.
1101 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1104 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1107 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1109 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1110 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1111 (ix86_handle_option): Handle OPT_mprefetchwt1.
1112 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1113 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1115 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1116 OPTION_MASK_ISA_PREFETCHWT1.
1117 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1118 (PTA_PREFETCHWT1): New.
1119 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1120 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1121 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1122 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1123 (*prefetch_avx512pf_<mode>_: Change into ...
1124 (*prefetch_prefetchwt1_<mode>: This.
1125 * config/i386/i386.opt (mprefetchwt1): New.
1126 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1127 (_mm_prefetch): Handle intent to write.
1128 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1130 2014-02-25 Richard Biener <rguenther@suse.de>
1133 * emit-rtl.c (mem_attrs_htab): Remove.
1134 (mem_attrs_htab_hash): Likewise.
1135 (mem_attrs_htab_eq): Likewise.
1136 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1137 (init_emit_once): Do not allocate mem_attrs_htab.
1139 2014-02-25 Richard Biener <rguenther@suse.de>
1142 * lto-opts.c (lto_write_options): Output non-explicit conservative
1143 -fwrapv, -fno-trapv and -fno-strict-overflow.
1144 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1145 (run_gcc): And pass them through.
1147 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1149 * sel-sched.c (calculate_new_fences): New parameter ptime.
1150 Calculate it as a maximum over all fence cycles.
1151 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1152 Print the final schedule timing when sched_verbose.
1154 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1156 PR rtl-optimization/60292
1157 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1158 bit fot the fence instruction.
1160 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1162 * calls.h: Fix typo in comment.
1164 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1166 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1167 adjusting offsetable addresses.
1169 2014-02-24 Guozhi Wei <carrot@google.com>
1171 * sparseset.h (sparseset_pop): Fix the wrong index.
1173 2014-02-24 Walter Lee <walt@tilera.com>
1175 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1176 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1178 * common/config/tilegx/tilegx-common.c
1179 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1180 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1182 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1183 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1184 (tilegx_gimplify_va_arg_expr): Handle big endian.
1185 (tilegx_expand_unaligned_load): Ditto.
1186 (tilegx_expand_unaligned_store): Ditto.
1187 (TARGET_RETURN_IN_MSB): New.
1188 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1189 (TARGET_ENDIAN_DEFAULT): New.
1190 (TARGET_BIG_ENDIAN): Handle big endian.
1191 (BYTES_BIG_ENDIAN): Ditto.
1192 (WORDS_BIG_ENDIAN): Ditto.
1193 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1196 * config/tilegx/tilegx.md (extv): Handle big endian.
1198 (insn_st<n>): Ditto.
1199 (insn_st<n>_add<bitsuffix>): Ditto.
1200 (insn_stnt<n>): Ditto.
1201 (insn_stnt<n>_add<bitsuffix>):Ditto.
1202 (vec_interleave_highv8qi): Handle big endian.
1203 (vec_interleave_highv8qi_be): New.
1204 (vec_interleave_highv8qi_le): New.
1205 (insn_v1int_h): Handle big endian.
1206 (vec_interleave_lowv8qi): Handle big endian.
1207 (vec_interleave_lowv8qi_be): New.
1208 (vec_interleave_lowv8qi_le): New.
1209 (insn_v1int_l): Handle big endian.
1210 (vec_interleave_highv4hi): Handle big endian.
1211 (vec_interleave_highv4hi_be): New.
1212 (vec_interleave_highv4hi_le): New.
1213 (insn_v2int_h): Handle big endian.
1214 (vec_interleave_lowv4hi): Handle big endian.
1215 (vec_interleave_lowv4hi_be): New.
1216 (vec_interleave_lowv4hi_le): New.
1217 (insn_v2int_l): Handle big endian.
1218 (vec_interleave_highv2si): Handle big endian.
1219 (vec_interleave_highv2si_be): New.
1220 (vec_interleave_highv2si_le): New.
1221 (insn_v4int_h): Handle big endian.
1222 (vec_interleave_lowv2si): Handle big endian.
1223 (vec_interleave_lowv2si_be): New.
1224 (vec_interleave_lowv2si_le): New.
1225 (insn_v4int_l): Handle big endian.
1226 * config/tilegx/tilegx.opt (mbig-endian): New option.
1227 (mlittle-endian): New option.
1228 * doc/install.texi: Document tilegxbe-linux.
1229 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1231 2014-02-24 Martin Jambor <mjambor@suse.cz>
1234 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1235 there are no parameter descriptors.
1237 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1239 PR rtl-optimization/60268
1240 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1241 initialization to ...
1242 (sched_rgn_init): ... here.
1243 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1245 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1247 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1250 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1252 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1255 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1257 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1258 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1260 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1262 * config/microblaze/predicates.md: Add cmp_op predicate.
1263 * config/microblaze/microblaze.md: Add branch_compare instruction
1264 which uses cmp_op predicate and emits cmp insn before branch.
1265 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1266 to microblaze_expand_conditional_branch and consolidate logic.
1267 (microblaze_expand_conditional_branch): emit branch_compare
1268 insn instead of handling cmp op separate from branch insn.
1270 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1272 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1275 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1277 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1278 define_insn with define_expand and new define_insn
1279 *altivec_lve<VI_char>x_internal.
1280 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1281 and new define_insn *altivec_stve<VI_char>x_internal.
1282 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1284 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1286 (altivec_expand_stvex_be): New function.
1288 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1290 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1291 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1292 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1293 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1295 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1298 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1299 instead of emit_move_insn.
1301 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1303 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1305 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1306 gen_altivec_vsumsws.
1308 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1310 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1311 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1312 (altivec_lvxl_<mode>): New define_expand incorporating
1313 -maltivec=be semantics where needed.
1314 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1315 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1316 semantics where needed.
1317 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1318 (altivec_stvx_<mode>): New define_expand incorporating
1319 -maltivec=be semantics where needed.
1320 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1321 VM2 iterator instead of V4SI.
1322 (altivec_stvxl_<mode>): New define_expand incorporating
1323 -maltivec=be semantics where needed.
1324 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1325 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1326 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1327 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1328 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1330 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1331 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1332 ALTIVEC_BUILTIN_STVXL.
1333 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1334 (altivec_expand_stvx_be): Likewise.
1335 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1336 (altivec_expand_lvx_be): Likewise.
1337 (altivec_expand_stvx_be): Likewise.
1338 (altivec_expand_builtin): Add cases for
1339 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1340 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1341 (altivec_init_builtins): Add definitions for
1342 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1343 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1345 2014-02-21 Catherine Moore <clm@codesourcery.com>
1347 * doc/invoke.texi (mvirt, mno-virt): Document.
1348 * config/mips/mips.opt (mvirt): New option.
1349 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1351 2014-02-21 Richard Biener <rguenther@suse.de>
1353 PR tree-optimization/60276
1354 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1355 (STMT_VINFO_MIN_NEG_DIST): New macro.
1356 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1357 STMT_VINFO_MIN_NEG_DIST.
1358 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1359 made for negative dependence distances still hold.
1361 2014-02-21 Richard Biener <rguenther@suse.de>
1364 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1365 DECL_INITIAL for globals not in the current function context.
1367 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1369 PR tree-optimization/56490
1370 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1371 * tree-ssa-uninit.c: Include params.h.
1372 (compute_control_dep_chain): Add num_calls argument, return false
1373 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1374 num_calls to recursive call.
1375 (find_predicates): Change dep_chain into normal array,
1376 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1377 variable and adjust compute_control_dep_chain caller.
1378 (find_def_preds): Likewise.
1380 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1382 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1383 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1385 2014-02-21 Nick Clifton <nickc@redhat.com>
1387 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1388 (pushhi1): Likewise.
1389 (popqi1): Add mode to pre_dec.
1392 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1394 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1395 mode for mask of V8SFmode permutation.
1397 2014-02-20 Richard Henderson <rth@redhat.com>
1400 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1401 a new pseudo for OLDVAL.
1403 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1406 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1407 gen_reg_rtx if d->testing_p.
1408 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1409 if d->testing_p and we will certainly return true.
1410 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1413 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1415 * emit-rtl.c (gen_reg_rtx): Assert that
1416 crtl->emit.regno_pointer_align_length is non-zero.
1418 2014-02-20 Richard Henderson <rth@redhat.com>
1421 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1422 on failure the store back into EXPECT.
1424 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1425 Sandra Loosemore <sandra@codesourcery.com>
1427 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1428 * config/nios2/nios2.c (nios2_function_profiler): Add
1429 -fPIC (flag_pic == 2) support.
1430 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1431 (nios2_large_offset_p): New function.
1432 (nios2_unspec_reloc_p): Move up position, update to use
1433 nios2_large_offset_p.
1434 (nios2_unspec_address): Remove function.
1435 (nios2_unspec_offset): New function.
1436 (nios2_large_got_address): New function.
1437 (nios2_got_address): Add large offset support.
1438 (nios2_legitimize_tls_address): Update usage of removed and new
1440 (nios2_symbol_binds_local_p): New function.
1441 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1442 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1443 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1444 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1445 processing for (const (unspec ...)).
1446 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1448 2014-02-20 Richard Biener <rguenther@suse.de>
1450 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1451 doing the substitution.
1452 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1454 2014-02-20 Martin Jambor <mjambor@suse.cz>
1457 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1458 info when checking whether lattices are bottom.
1460 2014-02-20 Richard Biener <rguenther@suse.de>
1463 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1466 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1469 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1470 parameter specifying the scaling.
1471 (inline_call): Update.
1472 (want_inline_recursively): Guard division by zero.
1473 (recursive_inlining): Update.
1474 * ipa-inline.h (clone_inlined_nodes): Update.
1476 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1479 * config/i386/i386.c (classify_argument): Pass structures of size
1480 64 bytes or less in register.
1482 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1483 Kirill Yukhin <kirill.yukhin@intel.com>
1485 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1486 (_mm_rcp28_round_ss): Ditto.
1487 (_mm_rsqrt28_round_sd): Ditto.
1488 (_mm_rsqrt28_round_ss): Ditto.
1489 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1490 (_mm_rcp14_round_ss): Ditto.
1491 (_mm_rsqrt14_round_sd): Ditto.
1492 (_mm_rsqrt14_round_ss): Ditto.
1493 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1494 the first input operand, get rid of match_dup.
1495 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1497 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1499 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1500 operand as the first input operand, set type attribute.
1501 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1503 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1504 operand as the first input operand, set type attribute.
1506 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1508 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1511 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1514 * config/i386/i386.c (construct_container): Remove TFmode check
1515 for X86_64_INTEGER_CLASS.
1517 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1520 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1521 only when -Wpsabi is enabled.
1523 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1526 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1527 passing arrays in registers are the same as for structs, so remove the
1528 special case for them.
1530 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1532 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1533 destination type, extract only the valid bits if the source type is not
1534 integral and has a different mode.
1536 2014-02-19 Richard Biener <rguenther@suse.de>
1539 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1542 2014-02-19 Richard Biener <rguenther@suse.de>
1545 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1546 (ipa_modify_call_arguments): Emit an argument load explicitely and
1547 preserve virtual SSA form there and for the replacement call.
1548 Do not update SSA form nor free dominance info.
1550 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1552 * ipa.c (function_and_variable_visibility): Also clear WEAK
1553 flag when disolving COMDAT_GROUP.
1555 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1557 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1558 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1560 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1561 do more sanity checks.
1562 (detect_type_change): Return true when giving up early.
1563 (compute_complex_assign_jump_func): Fix type parameter of
1564 ipa_set_ancestor_jf.
1565 (compute_complex_ancestor_jump_func): Likewise.
1566 (update_jump_functions_after_inlining): Fix updating of
1568 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1570 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1572 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1573 inline clones when edge disappears.
1575 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1578 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1579 Split 64-bit moves into 2 patterns. Do not allow the use of
1580 direct move for TDmode in little endian, since the decimal value
1581 has little endian bytes within a word, but the 64-bit pieces are
1582 ordered in a big endian fashion, and normal subreg's of TDmode are
1584 (mov<mode>_64bit_dm): Likewise.
1585 (movtd_64bit_nodm): Likewise.
1587 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1589 PR tree-optimization/60174
1590 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1591 statement of an SSA_NAME that occurs in an abnormal PHI node.
1593 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1596 * final.c (SEEN_BB): Remove.
1597 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1598 (final_scan_insn): Don't force_source_line on second
1599 NOTE_INSN_BASIC_BLOCK.
1601 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1604 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1605 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1606 (type_natural_mode): Warn ABI change when %zmm register is not
1607 available for AVX512F vector value passing.
1609 2014-02-18 Kai Tietz <ktietz@redhat.com>
1612 * config/i386/i386.c (ix86_expand_prologue): Use value in
1613 rax register as displacement when restoring %r10 or %rax.
1614 Fix wrong offset when restoring both registers.
1616 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1618 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1619 assertion with conditional return.
1621 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1622 Uros Bizjak <ubizjak@gmail.com>
1625 * config/i386/driver-i386.c (host_detect_local_cpu): If
1626 YMM state is not saved by the OS, also clear has_f16c. Move
1627 CPUID 0x80000001 handling before YMM state saving checking.
1629 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1631 PR rtl-optimization/58960
1632 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1633 factored out from ...
1634 (sched_init): ... here.
1635 (free_global_sched_pressure_data): New, factored out from ...
1636 (sched_finish): ... here.
1637 * sched-int.h (free_global_sched_pressure_data): Declare.
1638 * sched-rgn.c (nr_regions_initial): New static global.
1639 (haifa_find_rgns): Initialize it.
1640 (schedule_region): Disable sched-pressure for the newly
1643 2014-02-17 Richard Biener <rguenther@suse.de>
1645 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1646 release SSA defs of pattern stmts.
1648 2014-02-17 Richard Biener <rguenther@suse.de>
1650 * tree-inline.c (expand_call_inline): Release the virtual
1651 operand defined by the call we are about to inline.
1653 2014-02-17 Richard Biener <rguenther@suse.de>
1655 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1657 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1658 Ilya Tocar <ilya.tocar@intel.com>
1660 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1661 arguments order in builtin.
1662 (_mm512_permutexvar_epi64): Ditto.
1663 (_mm512_mask_permutexvar_epi64): Ditto
1664 (_mm512_maskz_permutexvar_epi32): Ditto
1665 (_mm512_permutexvar_epi32): Ditto
1666 (_mm512_mask_permutexvar_epi32): Ditto
1668 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1670 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1671 (p8_vmrgow): Likewise.
1673 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1675 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1678 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1681 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1682 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1683 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1684 using direct move instructions on ISA 2.07. Also adjust
1685 instruction length for 64-bit.
1686 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1687 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1689 2014-02-15 Alan Modra <amodra@gmail.com>
1693 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1694 find_replacement on parts of insn rtl that might be reloaded.
1696 2014-02-15 Richard Biener <rguenther@suse.de>
1698 PR tree-optimization/60183
1699 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1700 (tree_ssa_phiprop): Calculate and free post-dominators.
1702 2014-02-14 Jeff Law <law@redhat.com>
1704 PR rtl-optimization/60131
1705 * ree.c (get_extended_src_reg): New function.
1706 (combine_reaching_defs): Use it rather than assuming location of REG.
1707 (find_and_remove_re): Verify first operand of extension is
1708 a REG before adding the insns to the copy list.
1710 2014-02-14 Roland McGrath <mcgrathr@google.com>
1712 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1713 * configure: Regenerated.
1714 * config.in: Regenerated.
1715 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1716 instead of ASM_SHORT.
1718 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1719 Richard Earnshaw <rearnsha@arm.com>
1721 PR rtl-optimization/59535
1722 * lra-constraints.c (process_alt_operands): Encourage alternative
1723 when unassigned pseudo class is superset of the alternative class.
1724 (inherit_reload_reg): Don't inherit when optimizing for code size.
1725 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1726 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1727 modes not less than 4 for Thumb1.
1729 2014-02-14 Kyle McMartin <kyle@redhat.com>
1732 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1734 2014-02-14 Richard Biener <rguenther@suse.de>
1736 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1737 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1738 Do not strip INDIRECT_REFs.
1740 2014-02-14 Richard Biener <rguenther@suse.de>
1743 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1744 DECL_FUNCTION_SPECIFIC_TARGET.
1745 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1746 * tree-streamer-out.c (pack_ts_target_option): Remove.
1747 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1748 (write_ts_function_decl_tree_pointers): Do not stream
1749 DECL_FUNCTION_SPECIFIC_TARGET.
1750 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1751 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1752 (lto_input_ts_function_decl_tree_pointers): Do not stream
1753 DECL_FUNCTION_SPECIFIC_TARGET.
1755 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1757 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1758 (get_initial_def_for_induction, vectorizable_induction): Ignore
1759 debug stmts when looking for exit_phi.
1760 (vectorizable_live_operation): Fix up condition.
1762 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1764 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1765 nreverse() because it changes the content of original tree list.
1767 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1769 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1770 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1772 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1774 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1775 GNU coding standards.
1777 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1780 * dwarf2out.c (gen_subprogram_die): Don't call
1781 add_calling_convention_attribute if subr_die is old_die.
1783 2014-02-13 Sharad Singhai <singhai@google.com>
1785 * doc/optinfo.texi: Fix order of nodes.
1787 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1789 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1790 operands[2], not operands[3].
1792 2014-02-13 Richard Biener <rguenther@suse.de>
1795 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1796 mention the possibility of an in-tree build.
1797 (CLooG): Update recommended version to 0.18.1, mention the
1798 possibility of an in-tree build and clarify that the ISL
1799 bundled with CLooG does not work.
1801 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1804 * expr.c (compress_float_constant): If x is a hard register,
1805 extend into a pseudo and then move to x.
1807 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1809 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1810 caused by bad second argument to warning_at() with -mhotpatch and
1811 nested functions (e.g. with gfortran).
1813 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1815 * opts.c (option_name): Remove "enabled by default" rider.
1817 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1819 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1821 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1822 Uros Bizjak <ubizjak@gmail.com>
1825 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1826 * configure: Regenerated.
1828 2014-02-12 Richard Biener <rguenther@suse.de>
1830 * vec.c (vec_prefix::calculate_allocation): Move as
1831 inline variant to vec.h.
1832 (vec_prefix::calculate_allocation_1): New out-of-line version.
1833 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1834 (vec_prefix::m_has_auto_buf): Rename to ...
1835 (vec_prefix::m_using_auto_storage): ... this.
1836 (vec_prefix::calculate_allocation): Inline the easy cases
1837 and dispatch to calculate_allocation_1 which doesn't need the
1839 (va_heap::reserve): Use gcc_checking_assert.
1840 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1841 m_using_auto_storage.
1842 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1844 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1845 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1846 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1848 2014-02-12 Richard Biener <rguenther@suse.de>
1850 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1851 when we found a dependence.
1853 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1855 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1857 (maybe_fold_stmt): ... into this new function.
1858 * omp-low.c (lower_omp): Update comment.
1860 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1863 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1866 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1868 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1869 identifiers in comments.
1870 (cortexa53_extra_costs): Likewise.
1871 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1872 (cortexa7_extra_costs): Likewise.
1873 (cortexa12_extra_costs): Likewise.
1874 (cortexa15_extra_costs): Likewise.
1875 (v7m_extra_costs): Likewise.
1877 2014-02-12 Richard Biener <rguenther@suse.de>
1880 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1881 of posix_memalign being successful.
1882 (lower_stmt): Restrict lowering of posix_memalign to when
1883 -ftree-bit-ccp is enabled.
1885 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1887 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1889 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1891 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1893 PR rtl-optimization/60116
1894 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1895 other_insn once the combination has been validated.
1897 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1900 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1902 * ipa-devirt.c: Include demangle.h
1903 (odr_violation_reported): New static variable.
1904 (add_type_duplicate): Update odr_violations.
1905 (maybe_record_node): Add completep parameter; update it.
1906 (record_target_from_binfo): Add COMPLETEP parameter;
1907 update it as needed.
1908 (possible_polymorphic_call_targets_1): Likewise.
1909 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1910 rename FINAL to COMPLETE.
1911 (record_targets_from_bases): Sanity check we found the binfo;
1912 fix COMPLETEP updating.
1913 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1914 parameter, fix computing of COMPLETEP.
1915 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1916 at LTO time do demangling.
1917 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1918 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1920 (gimple_get_virt_method_for_binfo): Likewise.
1921 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1922 gimple_get_virt_method_for_vtable): Update prototypes.
1924 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1927 * genautomata.c (add_presence_absence): Fix typo with
1928 {final_}presence_list.
1930 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1933 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1934 for VSX/Altivec vectors that land in GPR registers.
1936 2014-02-11 Richard Henderson <rth@redhat.com>
1937 Jakub Jelinek <jakub@redhat.com>
1940 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1941 around drhs if type conversion to lacc->type is not useless.
1943 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1945 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1947 (cortex-a57.cortex-a53): Likewise.
1948 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1950 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1952 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1955 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1957 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1958 add_options_for_arm_vfp3.
1960 2014-02-11 Jeff Law <law@redhat.com>
1963 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1964 object with an undesirable mode.
1966 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1969 * config/i386/sol2-9.h: New file.
1970 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1971 *-*-solaris2.9*): Use it.
1973 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1975 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1976 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1978 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1980 * config/microblaze/microblaze.c: Extend mcpu version format
1982 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1984 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1986 2014-02-10 Richard Henderson <rth@redhat.com>
1989 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1990 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1991 ms-abi vs -mno-accumulate-outgoing-args.
1992 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1993 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1996 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1999 * cfgexpand.c (expand_asm_operands): Attach source location to
2000 ASM_INPUT rtx objects.
2001 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2003 2014-02-10 Nick Clifton <nickc@redhat.com>
2005 * config/mn10300/mn10300.c (popcount): New function.
2006 (mn10300_expand_prologue): Include saved registers in stack usage
2009 2014-02-10 Jeff Law <law@redhat.com>
2012 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2013 when changing the SET_DEST of a prior insn to avoid an input reload.
2015 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2017 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2018 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2019 -mcall-openbsd, or -mcall-linux.
2020 (CC1_ENDIAN_BIG_SPEC): Remove.
2021 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2022 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2023 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2024 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2025 and %cc1_endian_default.
2026 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2028 2014-02-10 Richard Biener <rguenther@suse.de>
2030 PR tree-optimization/60115
2031 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2032 MEM_REF handling. Properly verify that the accesses are not
2033 out of the objects bound.
2035 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2037 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2040 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2042 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2043 proper constants and fix formatting.
2044 (possible_polymorphic_call_targets): Fix formatting.
2046 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2047 Ilya Tocar <ilya.tocar@intel.com>
2049 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2050 (_mm512_loadu_epi32): Renamed into...
2051 (_mm512_loadu_si512): This.
2052 (_mm512_storeu_epi32): Renamed into...
2053 (_mm512_storeu_si512): This.
2054 (_mm512_maskz_ceil_ps): Removed.
2055 (_mm512_maskz_ceil_pd): Ditto.
2056 (_mm512_maskz_floor_ps): Ditto.
2057 (_mm512_maskz_floor_pd): Ditto.
2058 (_mm512_floor_round_ps): Ditto.
2059 (_mm512_floor_round_pd): Ditto.
2060 (_mm512_ceil_round_ps): Ditto.
2061 (_mm512_ceil_round_pd): Ditto.
2062 (_mm512_mask_floor_round_ps): Ditto.
2063 (_mm512_mask_floor_round_pd): Ditto.
2064 (_mm512_mask_ceil_round_ps): Ditto.
2065 (_mm512_mask_ceil_round_pd): Ditto.
2066 (_mm512_maskz_floor_round_ps): Ditto.
2067 (_mm512_maskz_floor_round_pd): Ditto.
2068 (_mm512_maskz_ceil_round_ps): Ditto.
2069 (_mm512_maskz_ceil_round_pd): Ditto.
2070 (_mm512_expand_pd): Ditto.
2071 (_mm512_expand_ps): Ditto.
2072 * config/i386/i386.c (ix86_builtins): Remove
2073 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2074 (bdesc_args): Ditto.
2075 * config/i386/predicates.md (const1256_operand): New.
2076 (const_1_to_2_operand): Ditto.
2077 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2078 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2079 (*avx512pf_gatherpf<mode>sf): Ditto.
2080 (avx512pf_gatherpf<mode>df): Ditto.
2081 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2082 (*avx512pf_gatherpf<mode>df): Ditto.
2083 (avx512pf_scatterpf<mode>sf): Ditto.
2084 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2085 (*avx512pf_scatterpf<mode>sf): Ditto.
2086 (avx512pf_scatterpf<mode>df): Ditto.
2087 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2088 (*avx512pf_scatterpf<mode>df): Ditto.
2089 (avx512f_expand<mode>): Removed.
2090 (<shift_insn><mode>3<mask_name>): Change predicate type.
2092 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2094 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2095 not at the end of datarefs vector use ordered_remove to avoid
2096 reordering datarefs vector.
2099 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2100 mark local addressable non-static vars as GOVD_PRIVATE
2101 instead of GOVD_LOCAL.
2102 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2103 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2107 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2108 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2109 assume_aligned or alloc_align attributes.
2110 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2111 arguments. Handle also assume_aligned and alloc_align attributes.
2112 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2113 calls to functions with assume_aligned or alloc_align attributes.
2114 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2116 2014-02-08 Terry Guo <terry.guo@arm.com>
2118 * doc/invoke.texi: Document ARM -march=armv7e-m.
2120 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2122 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2123 flag on __cilkrts_rethrow builtin.
2126 * ipa-cp.c (determine_versionability): Fail at -O0
2127 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2128 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2131 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2134 * tree-inline.c (copy_forbidden): Fail for
2135 __attribute__((optimize (0))) functions.
2137 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2139 * varpool.c: Include pointer-set.h.
2140 (varpool_remove_unreferenced_decls): Variables in other partitions
2141 will not be output; be however careful to not lose information
2144 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2146 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2147 lookup in the vtable constructor.
2149 2014-02-07 Jeff Law <law@redhat.com>
2152 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2153 define_insn_and_split.
2155 * ipa-inline.c (inline_small_functions): Fix typos.
2157 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2159 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2160 (s390_can_use_return_insn): Declare.
2161 * config/s390/s390.h (EPILOGUE_USES): Define.
2162 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2164 (s390_chunkify_start): Handle return JUMP_LABELs.
2165 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2166 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2167 (s390_can_use_return_insn): New functions.
2168 (s390_fix_long_loop_prediction): Handle conditional returns.
2169 (TARGET_SET_UP_BY_PROLOGUE): Define.
2170 * config/s390/s390.md (ANY_RETURN): New code iterator.
2171 (*creturn, *csimple_return, return, simple_return): New patterns.
2173 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2175 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2176 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2177 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2178 REG_CFA_RESTORE list when deciding not to restore a register.
2180 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2182 * config/s390/s390.c: Include tree-pass.h and context.h.
2183 (s390_early_mach): New function, split out from...
2184 (s390_emit_prologue): ...here.
2185 (pass_data_s390_early_mach): New pass structure.
2186 (pass_s390_early_mach): New class.
2187 (s390_option_override): Create and register early_mach pass.
2188 Move to end of file.
2190 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2192 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2193 to match for the exit block.
2195 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2197 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2198 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2199 Reject misaligned operands.
2201 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2203 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2205 2014-02-07 Richard Biener <rguenther@suse.de>
2208 * gimple-low.c (lower_builtin_posix_memalign): New function.
2209 (lower_stmt): Call it to lower posix_memalign in a way
2210 to make alignment info accessible.
2212 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2215 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2216 __builtin_setjmp_receiver.
2218 2014-02-07 Richard Biener <rguenther@suse.de>
2221 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2222 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2223 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2224 Handle BUILT_IN_POSIX_MEMALIGN.
2225 (find_func_clobbers): Likewise.
2226 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2227 (call_may_clobber_ref_p_1): Likewise.
2229 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2232 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2235 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2238 * lto-cgraph.c (lto_output_node): Use
2239 symtab_get_symbol_partitioning_class.
2240 (lto_output_varpool_node): likewise.
2241 (symtab_get_symbol_partitioning_class): Move here from
2243 * cgraph.h (symbol_partitioning_class): Likewise.
2244 (symtab_get_symbol_partitioning_class): Declare.
2246 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2248 * ggc.h (ggc_internal_cleared_alloc): New macro.
2249 * vec.h (vec_safe_copy): Handle memory stats.
2250 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2251 * target-globals.c (save_target_globals): Likewise.
2253 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2256 * expr.c (emit_move_resolve_push): Export; be bit more selective
2257 on when to clear alias set.
2258 * expr.h (emit_move_resolve_push): Declare.
2259 * function.h (struct function): Add tail_call_marked.
2260 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2261 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2262 * config/i386/i386.md (TImode move expander): De not call
2264 (FP push expanders): Preserve memory attributes.
2265 * config/i386/sse.md (push<mode>1): Remove.
2266 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2267 (ix86_expand_push): Remove.
2268 * config/i386/mmx.md (push<mode>1): Remove.
2270 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2272 PR rtl-optimization/60030
2273 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2274 lopart with paradoxical subreg before shifting it up by hprec.
2276 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2278 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2279 Remove extra newline at end of file.
2280 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2281 (arm_issue_rate): Handle cortexa57.
2282 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2283 (cortex-a57.cortex-a53): Likewise.
2285 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2288 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2289 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2291 (arm_expand_prologue): Adjust all callers.
2292 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2293 info, registers also at the lowest numbered registers side. Use
2294 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2298 * var-tracking.c (adjust_mems): Before adding a SET to
2299 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2301 2014-02-06 Alan Modra <amodra@gmail.com>
2304 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2305 change SDmode to DDmode when lra_in_progress.
2307 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2310 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2311 free_data_ref on the dr first, and before goto again also set dr
2312 to the next dr. For simd_lane_access, free old datarefs[i] before
2313 overwriting it. For get_vectype_for_scalar_type failure, don't
2314 free_data_ref if simd_lane_access.
2316 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2319 * tree.h (opts_for_fn): New inline function.
2320 (opt_for_fn): Define.
2321 * config/i386/i386.c (ix86_function_regparm): Use
2322 opt_for_fn (decl, optimize) instead of optimize.
2324 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2326 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2327 for SYMBOL_REF in large memory model.
2329 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2331 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2333 (cortex-a57): Likewise.
2334 (cortex-a57.cortex-a53): Likewise.
2336 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2337 Kugan Vivekanandarajah <kuganv@linaro.org>
2339 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2341 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2343 2014-02-06 Richard Biener <rguenther@suse.de>
2345 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2346 set_loop_copy and initialize_original_copy_tables.
2348 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2350 * config/aarch64/aarch64-simd.md
2351 (aarch64_ashr_simddi): Change QI to SI.
2353 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2354 Jakub Jelinek <jakub@redhat.com>
2357 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2360 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2362 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2363 CODE_FOR_altivec_vpku[hw]um to
2364 CODE_FOR_altivec_vpku[hw]um_direct.
2365 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2366 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2367 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2368 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2370 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2372 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2373 generation for -maltivec=be.
2374 (altivec_vsumsws): Simplify redundant test.
2376 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2378 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2379 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2380 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2381 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2382 gen_altivec_vpkuwum.
2383 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2385 (altivec_vpks<VI_char>ss): Likewise.
2386 (altivec_vpks<VI_char>us): Likewise.
2387 (altivec_vpku<VI_char>us): Likewise.
2388 (altivec_vpku<VI_char>um): Likewise.
2389 (altivec_vpku<VI_char>um_direct): New (copy of
2390 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2392 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2393 target is little endian and -maltivec=be is not specified.
2394 (*altivec_vupkhs<VU_char>_direct): New (copy of
2395 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2396 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2397 target is little endian and -maltivec=be is not specified.
2398 (*altivec_vupkls<VU_char>_direct): New (copy of
2399 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2400 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2401 little endian and -maltivec=be is not specified.
2402 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2403 little endian and -maltivec=be is not specified.
2405 2014-02-05 Richard Henderson <rth@redhat.com>
2408 * combine-stack-adj.c: Revert r206943.
2409 * sched-int.h (struct deps_desc): Add last_args_size.
2410 * sched-deps.c (init_deps): Initialize it.
2411 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2412 contain REG_ARGS_SIZE notes.
2414 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2416 * lto-cgraph.c (asm_nodes_output): Make global.
2417 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2418 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2419 (driver_handle_option): Handle OPT_fwpa.
2421 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2424 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2425 a comment typo and formatting issue. If odr_hash hasn't been
2426 created, return vNULL and set *completep to false.
2429 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2430 bb with no successors.
2432 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2435 * doc/invoke.texi (-march): Clarify documentation for ARM.
2439 2014-02-05 Richard Biener <rguenther@suse.de>
2441 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2442 when not vectorizing because of too many alias checks.
2443 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2444 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2446 2014-02-05 Nick Clifton <nickc@redhat.com>
2448 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2449 accept extended registers in any mode when compiling for the MN10300.
2451 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2453 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2454 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2455 sanitization attributes.
2456 (can_inline_edge_p): Likewise.
2457 (sanitize_attrs_match_for_inline_p): New function.
2459 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2461 * ipa-prop.c (detect_type_change): Shor circuit testing of
2462 type changes on THIS pointer.
2464 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2467 * config/pa/pa.c (legitimize_tls_address): Return original address
2468 if not passed a SYMBOL_REF rtx.
2469 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2471 (pa_emit_move_sequence): Simplify TLS source operands.
2472 (pa_legitimate_constant_p): Reject all TLS constants.
2473 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2474 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2476 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2478 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2479 groups when we know they are controlled by LTO.
2480 * varasm.c (default_binds_local_p_1): If object is in other partition,
2481 it will be resolved locally.
2483 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2485 * config/host-linux.c (linux_gt_pch_use_address): Don't
2486 use SSIZE_MAX because it is not always defined.
2488 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2491 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2492 threshold for pseudo splitting.
2493 (update_ebb_live_info): Process call argument hard registers and
2494 hard registers from insn definition too.
2495 (max_small_class_regs_num): New constant.
2496 (inherit_in_ebb): Update live hard regs through EBBs. Update
2497 reloads_num only for small register classes. Don't split for
2500 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2503 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2506 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2508 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2511 2014-02-04 Marek Polacek <polacek@redhat.com>
2513 * gdbinit.in (pel): Define.
2515 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2517 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2520 2014-02-04 Richard Biener <rguenther@suse.de>
2523 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2524 in function context local.
2525 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2526 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2527 similar to LTO_imported_decl_ref.
2529 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2531 PR tree-optimization/60002
2532 * cgraphclones.c (build_function_decl_skip_args): Clear
2535 PR tree-optimization/60023
2536 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2537 false to gsi_replace.
2538 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2539 has been in some EH region and vec_stmt could throw, add
2540 vec_stmt into the same EH region.
2541 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2542 has no lhs, ignore it.
2543 * internal-fn.c (expand_MASK_LOAD): Likewise.
2546 * tree-inline.c (copy_forbidden): Fail for
2547 __attribute__((optimize (0))) functions.
2550 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2551 is set, copy one less argument.
2552 (expand_simd_clones): Don't subtract clone_info->inbranch
2553 from simd_clone_struct_alloc argument.
2555 PR rtl-optimization/57915
2556 * recog.c (simplify_while_replacing): If all unary/binary/relational
2557 operation arguments are constant, attempt to simplify those.
2560 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2561 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2563 2014-02-04 Richard Biener <rguenther@suse.de>
2565 PR tree-optimization/60012
2566 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2567 TBAA disambiguation to all DDRs.
2569 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2572 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2573 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2575 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2578 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2580 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2582 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2583 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2585 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2588 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2589 to figure out targets of polymorphic calls with known decl.
2590 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2591 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2592 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2593 (get_polymorphic_call_info): ... here.
2594 (get_polymorphic_call_info_from_invariant): New function.
2596 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2598 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2599 lookup via vtable pointer; check for type consistency
2600 and turn inconsitent facts into UNREACHABLE.
2601 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2602 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2603 type inconsistent querries; return UNREACHABLE instead.
2605 2014-02-03 Richard Henderson <rth@twiddle.net>
2608 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2609 already processed this node.
2610 (normalize_one_pred_1): Pass along mark_set.
2611 (normalize_one_pred): Create and destroy a pointer_set_t.
2612 (normalize_one_pred_chain): Likewise.
2614 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2616 PR gcov-profile/58602
2617 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2619 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2622 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2623 -fno-devirtualize; try to devirtualize by the knowledge of
2624 virtual table pointer given by aggregate propagation.
2625 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2626 (ipa_print_node_jump_functions): Dump also offset that
2627 is relevant for polymorphic calls.
2628 (determine_known_aggregate_parts): Add arg_type parameter; use it
2629 instead of determining the type from pointer type.
2630 (ipa_compute_jump_functions_for_edge): Update call of
2631 determine_known_aggregate_parts.
2632 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2633 (gimple_get_virt_method_for_binfo): ... here; simplify using
2634 vtable_pointer_value_to_vtable.
2635 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2636 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2637 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2638 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2640 (vtable_pointer_value_to_binfo): ... here.
2641 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2643 2014-02-03 Teresa Johnson <tejohnson@google.com>
2645 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2646 redef of outer loop index variable.
2648 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2652 * doc/extend.texi (Function Attributes): Typo.
2654 2014-02-03 Cong Hou <congh@google.com>
2656 PR tree-optimization/60000
2657 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2658 if the vectorized statement is a store. A store statement can only
2659 appear at the end of pattern statements.
2661 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2663 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2664 (ix86_option_override_internal): Default long double to 64-bit for
2665 32-bit Bionic and to 128-bit for 64-bit Bionic.
2667 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2668 TARGET_LONG_DOUBLE_128 is true.
2669 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2671 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2672 (mlong-double-64): Negate -mlong-double-128.
2673 (mlong-double-128): New option.
2675 * config/i386/i386-c.c (ix86_target_macros): Define
2676 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2678 * doc/invoke.texi: Document -mlong-double-128.
2680 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2682 PR rtl-optimization/60024
2683 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2685 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2687 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2689 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2691 PR rtl-optimization/57662
2692 * sel-sched.c (code_motion_path_driver): Do not mark already not
2693 existing blocks in the visiting bitmap.
2695 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2697 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2698 on the insn being emitted.
2700 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2701 Will Deacon <will.deacon@arm.com>
2703 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2705 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2707 * config/arm/arm-tables.opt: Regenerate.
2709 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2711 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2712 for vector types other than V16QImode.
2713 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2714 define_expand, and call altivec_expand_vec_perm_le when producing
2715 code with little endian element order.
2716 (*altivec_vperm_<mode>_internal): New insn having previous
2717 behavior of altivec_vperm_<mode>.
2718 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2719 altivec_expand_vec_perm_le when producing code with little endian
2721 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2722 behavior of altivec_vperm_<mode>_uns.
2724 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2726 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2727 (altivec_vsumsws): Add handling for -maltivec=be with a little
2729 (altivec_vsumsws_direct): New.
2730 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2731 gen_altivec_vsumsws.
2733 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2735 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2736 vtable_pointer_value_to_binfo): New functions.
2737 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2738 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2740 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2742 * config/nios2/nios2.md (load_got_register): Initialize GOT
2743 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2744 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2746 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2748 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2749 preserverd by passthrough, do not propagate the type.
2751 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2753 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2754 (mips_atomic_assign_expand_fenv): New function.
2755 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2757 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2759 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2760 (__builtin_mips_set_fcsr): Likewise.
2761 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2762 MIPS_USI_FTYPE_VOID.
2763 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2764 (mips16_expand_set_fcsr): Likewise.
2765 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2766 (mips16_set_fcsr_stub): Likewise.
2767 (mips16_get_fcsr_one_only_stub): New class.
2768 (mips16_set_fcsr_one_only_stub): Likewise.
2769 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2770 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2771 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2772 (hard_float): New availability predicate.
2773 (mips_builtins): Add get_fcsr and set_fcsr.
2774 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2775 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2776 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2777 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2778 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2781 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2783 * config/mips/mips.c (mips_one_only_stub): New class.
2784 (mips_need_mips16_rdhwr_p): Replace with...
2785 (mips16_rdhwr_stub): ...this new variable.
2786 (mips16_stub_call_address): New function.
2787 (mips16_rdhwr_one_only_stub): New class.
2788 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2789 (mips_output_mips16_rdhwr): Delete.
2790 (mips_finish_stub): New function.
2791 (mips_code_end): Use it to handle rdhwr stubs.
2793 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2796 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2797 when calculating size of integer atomic types.
2799 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2801 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2803 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2805 PR tree-optimization/60003
2806 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2807 * profile.c (branch_prob): Use gimple_call_builtin_p
2808 to check for BUILT_IN_SETJMP_RECEIVER.
2809 * tree-inline.c (copy_bb): Call notice_special_calls.
2811 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2814 * lra-constraints.c (process_alt_operands): Update reload_sum only
2817 2014-01-31 Richard Henderson <rth@redhat.com>
2820 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2821 until after else_eh is processed.
2823 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2825 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2826 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2827 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2828 in smmintrin.h, remove them.
2829 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2830 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2831 * config/i386/i386.md (ROUND_SAE): Fix value.
2832 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2833 (const48_operand): New.
2834 * config/i386/subst.md (round), (round_expand): Use
2835 const_4_or_8_to_11_operand.
2836 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2838 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2840 * config/i386/constraints.md (Yk): Swap meaning with k.
2841 * config/i386/i386.md (movhi_internal): Change Yk to k.
2842 (movqi_internal): Ditto.
2843 (*k<logic><mode>): Ditto.
2846 (kandn<mode>): Ditto.
2847 (*<code>hi_1): Ditto.
2848 (*<code>qi_1): Ditto.
2849 (kxnor<mode>): Ditto.
2850 (kortestzhi): Ditto.
2851 (kortestchi): Ditto.
2853 (*one_cmplhi2_1): Ditto.
2854 (*one_cmplqi2_1): Ditto.
2855 * config/i386/sse.md (): Change k to Yk.
2856 (avx512f_load<mode>_mask): Ditto.
2857 (avx512f_blendm<mode>): Ditto.
2858 (avx512f_store<mode>_mask): Ditto.
2859 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2860 (avx512f_storedqu<mode>_mask): Ditto.
2861 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2863 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2864 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2865 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2866 (avx512f_maskcmp<mode>3): Ditto.
2867 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2868 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2869 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2870 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2871 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2872 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2873 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2874 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2875 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2876 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2877 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2878 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2879 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2880 (vec_extract_lo_<mode>_maskm): Ditto.
2881 (vec_extract_hi_<mode>_maskm): Ditto.
2882 (avx512f_vternlog<mode>_mask): Ditto.
2883 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2884 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2885 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2886 (avx512f_<code>v8div16qi2_mask): Ditto.
2887 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2888 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2889 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2890 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2891 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2892 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2893 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2894 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2895 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2896 (avx512cd_maskb_vec_dupv8di): Ditto.
2897 (avx512cd_maskw_vec_dupv16si): Ditto.
2898 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2899 (avx512f_vpermi2var<mode>3_mask): Ditto.
2900 (avx512f_vpermi2var<mode>3_mask): Ditto.
2901 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2902 (*avx512f_gathersi<mode>): Ditto.
2903 (*avx512f_gathersi<mode>_2): Ditto.
2904 (*avx512f_gatherdi<mode>): Ditto.
2905 (*avx512f_gatherdi<mode>_2): Ditto.
2906 (*avx512f_scattersi<mode>): Ditto.
2907 (*avx512f_scatterdi<mode>): Ditto.
2908 (avx512f_compress<mode>_mask): Ditto.
2909 (avx512f_compressstore<mode>_mask): Ditto.
2910 (avx512f_expand<mode>_mask): Ditto.
2911 * config/i386/subst.md (mask): Change k to Yk.
2912 (mask_scalar_merge): Ditto.
2915 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2917 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2919 2014-01-31 Richard Biener <rguenther@suse.de>
2922 * builtins.c (fold_builtin_memory_op): Make sure to not
2923 use a floating-point mode or a boolean or enumeral type for
2926 2014-01-30 DJ Delorie <dj@redhat.com>
2928 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2929 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2930 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2931 whenever main() has an epilogue.
2933 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2935 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2936 unused variable "field".
2937 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2938 (vsx_mergeh_<mode>): Likewise.
2939 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2940 (altivec_vmrghh): Likewise.
2941 (altivec_vmrghw): Likewise.
2942 (altivec_vmrglb): Likewise.
2943 (altivec_vmrglh): Likewise.
2944 (altivec_vmrglw): Likewise.
2945 (altivec_vspltb): Add missing uses.
2946 (altivec_vsplth): Likewise.
2947 (altivec_vspltw): Likewise.
2948 (altivec_vspltsf): Likewise.
2950 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2953 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2954 frame related instructions.
2956 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2958 PR rtl-optimization/59959
2959 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2960 any reload of register whose subreg is invalid.
2962 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2964 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2965 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2966 Add missing return type - void.
2968 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2970 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2971 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2972 remove element index adjustment for endian (now handled in vsx.md
2974 (altivec_expand_vec_perm_const): Use
2975 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2976 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2977 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2978 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2979 define_expand and a new define_insn *altivec_vspltb_internal;
2980 adjust for -maltivec=be on a little endian target.
2981 (altivec_vspltb_direct): New.
2982 (altivec_vsplth): Divide into a define_expand and a new
2983 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2984 little endian target.
2985 (altivec_vsplth_direct): New.
2986 (altivec_vspltw): Divide into a define_expand and a new
2987 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2988 little endian target.
2989 (altivec_vspltw_direct): New.
2990 (altivec_vspltsf): Divide into a define_expand and a new
2991 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2992 a little endian target.
2994 2014-01-30 Richard Biener <rguenther@suse.de>
2996 PR tree-optimization/59993
2997 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2998 can propagate form the earlier stmt and avoid the transform
2999 when the intermediate result is needed.
3001 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3003 * README.Portability: Fix typo.
3005 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3007 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3008 comparison_operator with ordered_comparison_operator.
3010 2014-01-30 Nick Clifton <nickc@redhat.com>
3012 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3013 Rename to mn10300_store_multiple_regs.
3014 * config/mn10300/mn10300.c: Likewise.
3015 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3016 store_multiple_regs.
3017 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3018 Call mn10300_store_multiple_regs.
3020 2014-01-30 Nick Clifton <nickc@redhat.com>
3021 DJ Delorie <dj@redhat.com>
3023 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3024 %fp 2 to keep registers after it properly word-aligned.
3025 (rl78_alloc_physical_registers_umul): Handle the case where both
3026 input operands are the same.
3028 2014-01-30 Richard Biener <rguenther@suse.de>
3030 PR tree-optimization/59903
3031 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3034 2014-01-30 Jason Merrill <jason@redhat.com>
3037 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3040 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3042 2014-01-30 Richard Biener <rguenther@suse.de>
3044 PR tree-optimization/59951
3045 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3047 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3050 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3051 SFmode to DFmode case.
3053 2014-01-29 DJ Delorie <dj@redhat.com>
3055 * config/msp430/msp430.opt (-minrt): New.
3056 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3058 (ENDFILE_SPEC): Likewise.
3060 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3062 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3063 (estimate_function_body_sizes): Use it.
3065 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3068 * dwarf2out.c (is_cxx_auto): New.
3069 (is_base_type): Use it.
3070 (gen_type_die_with_usage): Likewise.
3072 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3074 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3075 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3076 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3077 -maltivec=be with LE targets.
3078 (vsx_mergeh_<mode>): Likewise.
3079 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3080 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3081 (altivec_vmrghb): Replace with define_expand and new
3082 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3083 (altivec_vmrghb_direct): New define_insn.
3084 (altivec_vmrghh): Replace with define_expand and new
3085 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3086 (altivec_vmrghh_direct): New define_insn.
3087 (altivec_vmrghw): Replace with define_expand and new
3088 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3089 (altivec_vmrghw_direct): New define_insn.
3090 (*altivec_vmrghsf): Adjust for endianness.
3091 (altivec_vmrglb): Replace with define_expand and new
3092 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3093 (altivec_vmrglb_direct): New define_insn.
3094 (altivec_vmrglh): Replace with define_expand and new
3095 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3096 (altivec_vmrglh_direct): New define_insn.
3097 (altivec_vmrglw): Replace with define_expand and new
3098 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3099 (altivec_vmrglw_direct): New define_insn.
3100 (*altivec_vmrglsf): Adjust for endianness.
3101 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3102 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3103 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3104 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3105 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3106 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3107 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3108 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3110 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3112 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3113 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3116 2014-01-29 Richard Biener <rguenther@suse.de>
3118 PR tree-optimization/58742
3119 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3120 associate_pointerplus_align.
3121 (associate_pointerplus_diff): New function.
3122 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3123 and associate_pointerplus_diff.
3125 2014-01-29 Richard Biener <rguenther@suse.de>
3127 * lto-streamer.h (LTO_major_version): Bump to 3.
3128 (LTO_minor_version): Reset to 0.
3130 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3132 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3133 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3134 (arm_file_start): Generate correct asm header for armv7ve.
3135 * config/arm/bpabi.h: Add multilib support for armv7ve.
3136 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3137 and cortex-a15 to armv7ve.
3138 * config/arm/t-aprofile: Add multilib support for armv7ve.
3139 * doc/invoke.texi: Document -march=armv7ve.
3141 2014-01-29 Richard Biener <rguenther@suse.de>
3143 PR tree-optimization/58742
3144 * tree-ssa-forwprop.c (associate_plusminus): Return true
3145 if we changed sth, defer EH cleanup to ...
3146 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3147 (simplify_mult): New function.
3149 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3152 PR tree-optimization/59920
3153 * tree.c (build_common_builtin_nodes): Remove
3154 __builtin_setjmp_dispatcher initialization.
3155 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3156 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3157 instead of gsi_after_labels + manually skipping debug stmts.
3158 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3159 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3160 * tree-inline.c (copy_edges_for_bb): Remove
3161 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3162 argument. Ignore computed_goto_p stmts. Don't call
3163 make_abnormal_goto_edges. If a call might need abnormal edges
3164 for non-local gotos, see if it already has an edge to
3165 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3166 with true argument, don't do anything then, otherwise add
3167 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3168 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3170 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3171 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3172 (lower_stmt): Don't set data->calls_builtin_setjmp.
3173 (lower_builtin_setjmp): Adjust comment.
3174 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3175 * tree-cfg.c (found_computed_goto): Remove.
3176 (factor_computed_gotos): Remove.
3177 (make_goto_expr_edges): Return bool, true for computed gotos.
3178 Don't call make_abnormal_goto_edges.
3179 (build_gimple_cfg): Don't set found_computed_goto, don't call
3180 factor_computed_gotos.
3181 (computed_goto_p): No longer static.
3182 (make_blocks): Don't set found_computed_goto.
3183 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3184 (make_edges): If make_goto_expr_edges returns true, push bb
3185 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3186 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3187 vector. Record mapping between bbs and OpenMP regions if there
3188 are any, adjust make_gimple_omp_edges caller. Call
3189 handle_abnormal_edges.
3190 (make_abnormal_goto_edges): Remove.
3191 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3192 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3193 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3194 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3195 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3196 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3197 filling *region also set *region_idx to (*region)->entry->index.
3200 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3201 For REGs set ORIGINAL_REGNO.
3203 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3205 * doc/md.texi: Mention that a target shouldn't implement
3206 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3209 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3211 PR tree-optimization/59594
3212 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3213 a copy of the datarefs vector rather than the vector itself.
3215 2014-01-28 Jason Merrill <jason@redhat.com>
3218 * dwarf2out.c (auto_die): New static.
3219 (gen_type_die_with_usage): Handle C++1y 'auto'.
3220 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3223 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3226 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3227 (SPEC_X32): Likewise.
3228 (SPEC_64): Likewise.
3229 * config/i386/i386.c (ix86_option_override_internal): Turn off
3230 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3232 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3233 * config/i386/i386.h (TARGET_16BIT): New macro.
3234 (TARGET_16BIT_P): Likewise.
3235 * config/i386/i386.opt: Add m16.
3236 * doc/invoke.texi: Document -m16.
3238 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3240 PR preprocessor/59935
3241 * input.c (location_get_source_line): Bail out on when line number
3242 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3244 2014-01-28 Richard Biener <rguenther@suse.de>
3246 PR tree-optimization/58742
3247 * tree-ssa-forwprop.c (associate_plusminus): Handle
3248 pointer subtraction of the form (T)(P + A) - (T)P.
3250 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3252 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3255 2014-01-28 Richard Biener <rguenther@suse.de>
3258 2014-01-28 Richard Biener <rguenther@suse.de>
3260 PR rtl-optimization/45364
3261 PR rtl-optimization/59890
3262 * var-tracking.c (local_get_addr_clear_given_value): Handle
3263 already cleared slot.
3264 (val_reset): Handle not allocated local_get_addr_cache.
3265 (vt_find_locations): Use post-order on the inverted CFG.
3267 2014-01-28 Richard Biener <rguenther@suse.de>
3269 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3271 2014-01-28 Richard Biener <rguenther@suse.de>
3273 PR rtl-optimization/45364
3274 PR rtl-optimization/59890
3275 * var-tracking.c (local_get_addr_clear_given_value): Handle
3276 already cleared slot.
3277 (val_reset): Handle not allocated local_get_addr_cache.
3278 (vt_find_locations): Use post-order on the inverted CFG.
3280 2014-01-28 Alan Modra <amodra@gmail.com>
3282 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3283 * configure.ac <recursive call for build != host>: Define
3284 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3285 and LD_FOR_BUILD too.
3286 * configure: Regenerate.
3288 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3290 * config/i386/i386.c (get_builtin_code_for_version): Separate
3291 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3292 Broadwell from Haswell.
3294 2014-01-27 Steve Ellcey <sellcey@mips.com>
3296 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3297 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3298 * config/mips/mips.c (mips_option_override): Change setting
3300 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3301 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3302 Change from Mask to Var.
3304 2014-01-27 Jeff Law <law@redhat.com>
3306 * ipa-inline.c (inline_small_functions): Fix typo.
3308 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3310 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3311 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3312 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3313 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3314 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3315 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3316 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3317 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3318 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3319 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3320 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3321 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3322 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3323 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3324 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3325 (_mm512_storeu_epi64): Ditto.
3326 (_mm512_cmpge_epi32_mask): Ditto.
3327 (_mm512_cmpge_epu32_mask): Ditto.
3328 (_mm512_cmpge_epi64_mask): Ditto.
3329 (_mm512_cmpge_epu64_mask): Ditto.
3330 (_mm512_cmple_epi32_mask): Ditto.
3331 (_mm512_cmple_epu32_mask): Ditto.
3332 (_mm512_cmple_epi64_mask): Ditto.
3333 (_mm512_cmple_epu64_mask): Ditto.
3334 (_mm512_cmplt_epi32_mask): Ditto.
3335 (_mm512_cmplt_epu32_mask): Ditto.
3336 (_mm512_cmplt_epi64_mask): Ditto.
3337 (_mm512_cmplt_epu64_mask): Ditto.
3338 (_mm512_cmpneq_epi32_mask): Ditto.
3339 (_mm512_cmpneq_epu32_mask): Ditto.
3340 (_mm512_cmpneq_epi64_mask): Ditto.
3341 (_mm512_cmpneq_epu64_mask): Ditto.
3342 (_mm512_expand_pd): Ditto.
3343 (_mm512_expand_ps): Ditto.
3344 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3345 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3346 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3347 * config/i386/i386.c (ix86_builtins): Add
3348 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3349 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3350 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3351 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3352 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3353 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3354 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3355 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3356 IX86_BUILTIN_PMOVUSQW512_MEM.
3357 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3358 __builtin_ia32_pmovsqd512mem_mask,
3359 __builtin_ia32_pmovqd512mem_mask,
3360 __builtin_ia32_pmovusqw512mem_mask,
3361 __builtin_ia32_pmovsqw512mem_mask,
3362 __builtin_ia32_pmovqw512mem_mask,
3363 __builtin_ia32_pmovusdw512mem_mask,
3364 __builtin_ia32_pmovsdw512mem_mask,
3365 __builtin_ia32_pmovdw512mem_mask,
3366 __builtin_ia32_pmovqb512mem_mask,
3367 __builtin_ia32_pmovusqb512mem_mask,
3368 __builtin_ia32_pmovsqb512mem_mask,
3369 __builtin_ia32_pmovusdb512mem_mask,
3370 __builtin_ia32_pmovsdb512mem_mask,
3371 __builtin_ia32_pmovdb512mem_mask.
3372 (bdesc_args): Add __builtin_ia32_expanddf512,
3373 __builtin_ia32_expandsf512.
3374 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3375 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3376 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3377 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3378 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3379 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3380 (avx512f_<code>v8div16qi2_mask_store): This.
3381 (avx512f_expand<mode>): New.
3383 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3385 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3387 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3388 (_mm512_prefetch_i32scatter_pd): Ditto.
3389 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3390 (_mm512_prefetch_i64scatter_pd): Ditto.
3391 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3392 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3393 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3394 (_mm512_prefetch_i32scatter_ps): Ditto.
3395 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3396 (_mm512_prefetch_i64scatter_ps): Ditto.
3397 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3398 * config/i386/i386-builtin-types.def: Define
3399 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3400 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3401 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3402 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3403 IX86_BUILTIN_SCATTERPFQPD.
3404 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3405 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3406 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3407 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3408 __builtin_ia32_scatterpfqps.
3409 (ix86_expand_builtin): Expand new built-ins.
3410 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3411 fix memory access data type.
3412 (*avx512pf_gatherpf<mode>_mask): Ditto.
3413 (*avx512pf_gatherpf<mode>): Ditto.
3414 (avx512pf_scatterpf<mode>): Ditto.
3415 (*avx512pf_scatterpf<mode>_mask): Ditto.
3416 (*avx512pf_scatterpf<mode>): Ditto.
3417 (GATHER_SCATTER_SF_MEM_MODE): New.
3418 (avx512pf_gatherpf<mode>df): Ditto.
3419 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3420 (*avx512pf_scatterpf<mode>df): Ditto.
3422 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3425 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3426 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3429 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3431 * common/config/arm/arm-common.c
3432 (arm_rewrite_mcpu): Handle multiple names.
3434 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3436 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3438 * gimple-builder.h (create_gimple_tmp): Delete.
3440 2014-01-27 Christian Bruel <christian.bruel@st.com>
3442 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3445 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3447 * config/pa/pa.md (call): Generate indirect long calls to non-local
3448 functions when outputing 32-bit code.
3449 (call_value): Likewise except for special call to buggy powf function.
3451 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3452 portable runtime and PIC indirect calls.
3453 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3454 and PIC call sequences. Use ldo instead of blr to set return register
3455 in PIC call sequence.
3457 2014-01-25 Walter Lee <walt@tilera.com>
3459 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3460 avoid clobbering a live register.
3462 2014-01-25 Walter Lee <walt@tilera.com>
3464 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3465 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3466 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3467 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3469 2014-01-25 Walter Lee <walt@tilera.com>
3471 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3472 arguments on even registers.
3473 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3475 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3476 (BIGGEST_ALIGNMENT): Ditto.
3477 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3479 2014-01-25 Walter Lee <walt@tilera.com>
3481 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3482 insns before bundling.
3483 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3485 2014-01-25 Walter Lee <walt@tilera.com>
3487 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3488 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3489 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3491 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3493 * config/mips/constraints.md (kl): Delete.
3494 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3495 define expands, using...
3496 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3497 instructions for MIPS16.
3498 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3499 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3501 2014-01-25 Walter Lee <walt@tilera.com>
3503 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3507 2014-01-25 Walter Lee <walt@tilera.com>
3509 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3510 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3512 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3514 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3517 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3519 * print-rtl.c (in_call_function_usage): New var.
3520 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3521 EXPR_LIST mode as mode and not as reg note name.
3524 * cfgloopmanip.c (copy_loop_info): If
3525 loop->warned_aggressive_loop_optimizations, make sure
3526 the flag is set in target loop too.
3528 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3530 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3532 * builtins.def: Likewise.
3533 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3534 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3535 * ira.c (ira_setup_eliminable_regset): Likewise.
3536 * omp-low.c (gate_expand_omp): Likewise.
3537 (execute_lower_omp): Likewise.
3538 (diagnose_sb_0): Likewise.
3539 (gate_diagnose_omp_blocks): Likewise.
3540 (simd_clone_clauses_extract): Likewise.
3543 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3545 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3546 correction for little endian...
3547 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3550 2014-01-24 Jeff Law <law@redhat.com>
3552 PR tree-optimization/59919
3553 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3554 for non-returning calls.
3556 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3558 * common/config/aarch64/aarch64-common.c
3559 (aarch64_rewrite_mcpu): Handle multiple names.
3560 * config/aarch64/aarch64.h
3561 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3563 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3565 * input.c (add_file_to_cache_tab): Handle the case where fopen
3568 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3571 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3572 from push operand if code of push isn't PRE_DEC.
3574 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3577 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3578 -mquad-memory-atomic. Update -mquad-memory documentation to say
3579 it is only used for non-atomic loads/stores.
3581 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3582 -mquad-memory or -mquad-memory-atomic switches.
3584 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3585 -mquad-memory-atomic to ISA 2.07 support.
3587 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3588 to separate support of normal quad word memory operations (ldq, stq)
3589 from the atomic quad word memory operations.
3591 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3592 support to separate non-atomic quad word operations from atomic
3593 quad word operations. Disable non-atomic quad word operations in
3594 little endian mode so that we don't have to swap words after the
3595 load and before the store.
3596 (quad_load_store_p): Add comment about atomic quad word support.
3597 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3598 options printed with -mdebug=reg.
3600 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3601 -mquad-memory-atomic as the test for whether we have quad word
3602 atomic instructions.
3603 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3604 or -mp8-vector are used, allow byte/half-word atomic operations.
3606 * config/rs6000/sync.md (load_lockedti): Insure that the address
3607 is a proper indexed or indirect address for the lqarx instruction.
3608 On little endian systems, swap the hi/lo registers after the lqarx
3610 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3611 insure the address is valid for the lqarx instruction.
3612 (store_conditionalti): Insure that the address is a proper indexed
3613 or indirect address for the stqcrx. instruction. On little endian
3614 systems, swap the hi/lo registers before doing the stqcrx.
3616 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3617 insure the address is valid for the stqcrx. instruction.
3619 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3620 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3621 type of quad memory support is available.
3623 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3626 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3627 there is a danger of looping.
3629 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3632 force flag_ira_loop_pressure if set via command line.
3634 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3636 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3637 (ashr_simd): New builtin handling DI mode.
3638 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3639 (aarch64_sshr_simddi): New match pattern.
3640 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3641 (vshrd_n_s64): Likewise.
3642 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3644 2014-01-23 Nick Clifton <nickc@redhat.com>
3646 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3647 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3648 favour of mcu specific scripts.
3649 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3652 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3653 Alex Velenko <Alex.Velenko@arm.com>
3655 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3656 (vaddv_s16): Likewise.
3657 (vaddv_s32): Likewise.
3658 (vaddv_u8): Likewise.
3659 (vaddv_u16): Likewise.
3660 (vaddv_u32): Likewise.
3661 (vaddvq_s8): Likewise.
3662 (vaddvq_s16): Likewise.
3663 (vaddvq_s32): Likewise.
3664 (vaddvq_s64): Likewise.
3665 (vaddvq_u8): Likewise.
3666 (vaddvq_u16): Likewise.
3667 (vaddvq_u32): Likewise.
3668 (vaddvq_u64): Likewise.
3669 (vaddv_f32): Likewise.
3670 (vaddvq_f32): Likewise.
3671 (vaddvq_f64): Likewise.
3672 (vmaxv_f32): Likewise.
3673 (vmaxv_s8): Likewise.
3674 (vmaxv_s16): Likewise.
3675 (vmaxv_s32): Likewise.
3676 (vmaxv_u8): Likewise.
3677 (vmaxv_u16): Likewise.
3678 (vmaxv_u32): Likewise.
3679 (vmaxvq_f32): Likewise.
3680 (vmaxvq_f64): Likewise.
3681 (vmaxvq_s8): Likewise.
3682 (vmaxvq_s16): Likewise.
3683 (vmaxvq_s32): Likewise.
3684 (vmaxvq_u8): Likewise.
3685 (vmaxvq_u16): Likewise.
3686 (vmaxvq_u32): Likewise.
3687 (vmaxnmv_f32): Likewise.
3688 (vmaxnmvq_f32): Likewise.
3689 (vmaxnmvq_f64): Likewise.
3690 (vminv_f32): Likewise.
3691 (vminv_s8): Likewise.
3692 (vminv_s16): Likewise.
3693 (vminv_s32): Likewise.
3694 (vminv_u8): Likewise.
3695 (vminv_u16): Likewise.
3696 (vminv_u32): Likewise.
3697 (vminvq_f32): Likewise.
3698 (vminvq_f64): Likewise.
3699 (vminvq_s8): Likewise.
3700 (vminvq_s16): Likewise.
3701 (vminvq_s32): Likewise.
3702 (vminvq_u8): Likewise.
3703 (vminvq_u16): Likewise.
3704 (vminvq_u32): Likewise.
3705 (vminnmv_f32): Likewise.
3706 (vminnmvq_f32): Likewise.
3707 (vminnmvq_f64): Likewise.
3709 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3711 * config/aarch64/aarch64-simd.md
3712 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3713 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3714 (*aarch64_mul3_elt<mode>): Likewise.
3715 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3716 (*aarch64_mul3_elt_to_64v2df): Likewise.
3717 (*aarch64_mla_elt<mode>): Likewise.
3718 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3719 (*aarch64_mls_elt<mode>): Likewise.
3720 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3721 (*aarch64_fma4_elt<mode>): Likewise.
3722 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3723 (*aarch64_fma4_elt_to_64v2df): Likewise.
3724 (*aarch64_fnma4_elt<mode>): Likewise.
3725 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3726 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3727 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3728 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3729 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3730 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3731 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3732 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3733 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3735 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3737 * config/aarch64/aarch64-simd.md
3738 (aarch64_be_checked_get_lane<mode>): New define_expand.
3739 * config/aarch64/aarch64-simd-builtins.def
3740 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3741 New builtin definition.
3742 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3743 Use new safe be builtin.
3745 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3747 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3749 (aarch64_be_st1<mode>): Likewise.
3750 (aarch_ld1<VALL:mode>): Define_expand modified.
3751 (aarch_st1<VALL:mode>): Likewise.
3752 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3753 (UNSPEC_ST1): Likewise.
3755 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3757 * config/microblaze/microblaze.md: Add trap insn and attribute
3759 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3761 PR preprocessor/58580
3762 * input.h (location_get_source_line): Take an additional line_size
3764 (void diagnostics_file_cache_fini): Declare new function.
3765 * input.c (struct fcache): New type.
3766 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3767 New static constants.
3768 (diagnostic_file_cache_init, total_lines_num)
3769 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3770 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3771 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3772 (get_next_line, read_next_line, goto_next_line, read_line_num):
3773 New static function definitions.
3774 (diagnostic_file_cache_fini): New function.
3775 (location_get_source_line): Take an additional output line_len
3776 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3778 * diagnostic.c (diagnostic_finish): Call
3779 diagnostic_file_cache_fini.
3780 (adjust_line): Take an additional input parameter for the length
3781 of the line, rather than calculating it with strlen.
3782 (diagnostic_show_locus): Adjust the use of
3783 location_get_source_line and adjust_line with respect to their new
3784 signature. While displaying a line now, do not stop at the first
3785 null byte. Rather, display the zero byte as a space and keep
3786 going until we reach the size of the line.
3787 * Makefile.in: Add vec.o to OBJS-libcommon
3789 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3790 Ilya Tocar <ilya.tocar@intel.com>
3792 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3793 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3794 (__builtin_ia32_kmov16): Ditto.
3795 * config/i386/i386.md (UNSPEC_KMOV): New.
3798 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3800 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3801 (_mm512_storeu_si512): Ditto.
3803 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3806 * rtl.h (get_referenced_operands): Declare.
3807 * recog.c (get_referenced_operands): New function.
3808 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3809 operands have been referenced when recording LO_SUM references.
3811 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3813 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3815 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3817 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3818 Enable for generic and recent AMD targets.
3820 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3822 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3823 ARG_SIZE note when adjustment was eliminated.
3825 2014-01-22 Jeff Law <law@redhat.com>
3827 PR tree-optimization/59597
3828 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3829 in file. Accept new argument REGISTERING and use it to modify
3830 dump output appropriately.
3831 (register_jump_thread): Corresponding changes.
3832 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3833 thread paths involving joiner blocks. Add code to dump cancelled
3834 jump threading paths.
3836 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3838 PR rtl-optimization/59477
3839 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3840 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3842 2014-01-22 Tom Tromey <tromey@redhat.com>
3844 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3846 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3848 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3850 PR rtl-optimization/59896
3851 * lra-constraints.c (process_alt_operands): Check unused note for
3852 matched operands of insn with no output reloads.
3854 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3856 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3857 (mips_move_from_gpr_cost): Likewise.
3859 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3861 PR rtl-optimization/59858
3862 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3863 ira_class_hard_regs_num.
3864 (process_alt_operands): Increase reject for dying matched operand.
3866 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3869 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3870 smaller than size, perform several stores or loads and stores
3871 at dst + count - size to store or copy all of size bytes, rather
3872 than just last modesize bytes.
3874 2014-01-20 DJ Delorie <dj@redhat.com>
3876 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3877 that CLOBBERs are REGs before propogating their values.
3879 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3882 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3883 (cgraph_inline_failed_type): New function.
3884 * cgraph.h (DEFCIFCODE): Add type.
3885 (cgraph_inline_failed_type_t): New enum.
3886 (cgraph_inline_failed_type): New prototype.
3887 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3888 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3889 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3890 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3891 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3892 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3893 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3894 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3895 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3896 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3897 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3898 OPTIMIZATION_MISMATCH.
3899 * tree-inline.c (expand_call_inline): Emit errors during
3900 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3902 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3905 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3906 mode attribute in insn output.
3908 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3910 * output.h (output_constant): Delete.
3911 * varasm.c (output_constant): Make private.
3913 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3915 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3917 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3920 * tree.h (fold_builtin_strcat): New prototype.
3921 * builtins.c (fold_builtin_strcat): No longer static. Add len
3922 argument, if non-NULL, don't call c_strlen. Optimize
3923 directly into __builtin_memcpy instead of __builtin_strcpy.
3924 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3925 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3927 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3929 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3930 for SImode_address_operand operands, having only a REG argument.
3932 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3934 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3935 loader name using mbig-endian.
3936 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3938 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3940 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3944 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3946 * config/aarch64/aarch64-protos.h
3947 (aarch64_cannot_change_mode_class_ptr): Declare.
3948 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3949 aarch64_cannot_change_mode_class_ptr): New.
3950 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3951 backend hook aarch64_cannot_change_mode_class.
3953 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3955 * common/config/aarch64/aarch64-common.c
3956 (aarch64_handle_option): Don't handle any option order logic here.
3957 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3958 selected_cpu, warn on architecture version mismatch.
3959 (aarch64_override_options): Fix parsing order for option strings.
3961 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3962 Iain Sandoe <iain@codesourcery.com>
3965 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3966 warning. Amend comment to reflect current functionality.
3968 2014-01-20 Richard Biener <rguenther@suse.de>
3971 * builtins.c (fold_builtin_strcat): Remove case better handled
3972 by tree-ssa-strlen.c.
3974 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3976 * config/aarch64/aarch64.opt
3977 (mcpu, march, mtune): Make case-insensitive.
3979 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3982 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3983 if operands[1] is a REG or ZERO_EXTEND of a REG.
3985 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3987 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3989 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3991 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3992 long non-pic millicode calls.
3994 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3996 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3998 2014-01-19 Kito Cheng <kito@0xlab.org>
4000 * builtins.c (expand_movstr): Check movstr expand done or fail.
4002 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4003 H.J. Lu <hongjiu.lu@intel.com>
4006 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4007 to DImode for zero-extended addresses.
4009 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4011 PR rtl-optimization/57763
4012 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4013 on the new indirect jump_insn and increment LABEL_NUSES (label).
4015 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4019 * config.gcc (x86_archs): New variable.
4020 (x86_64_archs): Likewise.
4021 (x86_cpus): Likewise.
4022 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4023 --with-arch/--with-cpu= options.
4024 Support --with-arch=/--with-cpu={nehalem,westmere,
4025 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4027 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4029 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4030 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4032 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4034 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4036 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4039 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4040 clear cpp_get_options (parse_in)->warn_unused_macros for
4041 ix86_target_macros_internal with cpp_define.
4043 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4045 * jump.c (delete_related_insns): Keep (use (insn))s.
4046 * reorg.c (redundant_insn): Check for barriers too.
4048 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4050 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4052 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4054 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4055 call to $$dyncall when TARGET_LONG_CALLS is true.
4057 2014-01-17 Jeff Law <law@redhat.com>
4059 * ree.c (combine_set_extension): Temporarily disable test for
4060 changing number of hard registers.
4062 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4065 * ipa-inline-analysis.c (inline_free_summary):
4066 Do not free summary of aliases.
4068 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4071 * gimplify.c (gimplify_expr): Use create_tmp_var
4072 instead of create_tmp_var_raw. If cond doesn't have
4073 integral type, don't add the IFN_ANNOTATE builtin at all.
4075 2014-01-17 Martin Jambor <mjambor@suse.cz>
4078 * ipa-cp.c (prev_edge_clone): New variable.
4079 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4080 Also resize prev_edge_clone vector.
4081 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4082 (ipcp_edge_removal_hook): New function.
4083 (ipcp_driver): Register ipcp_edge_removal_hook.
4085 2014-01-17 Andrew Pinski <apinski@cavium.com>
4086 Steve Ellcey <sellcey@mips.com>
4089 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4092 2014-01-17 Jeff Law <law@redhat.com>
4095 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4096 so that pass_ccp runs first.
4098 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4100 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4101 (ix86_adjust_cost): Use !TARGET_XXX.
4102 (do_reorder_for_imul): Likewise.
4103 (swap_top_of_ready_list): Likewise.
4104 (ix86_sched_reorder): Likewise.
4106 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4108 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4109 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4110 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4111 (intel_memset): New. Duplicate slm_memset.
4112 (intel_cost): New. Duplicate slm_cost.
4113 (m_INTEL): New macro.
4114 (processor_target_table): Add "intel".
4115 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4116 with PROCESSOR_INTEL for "intel".
4117 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4118 PROCESSOR_SILVERMONT.
4119 (ix86_issue_rate): Likewise.
4120 (ix86_adjust_cost): Likewise.
4121 (ia32_multipass_dfa_lookahead): Likewise.
4122 (swap_top_of_ready_list): Likewise.
4123 (ix86_sched_reorder): Likewise.
4124 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4125 instead of TARGET_OPT_AGU.
4126 * config/i386/i386.h (TARGET_INTEL): New.
4127 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4128 (processor_type): Add PROCESSOR_INTEL.
4129 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4130 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4132 2014-01-17 Marek Polacek <polacek@redhat.com>
4135 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4138 2014-01-17 Richard Biener <rguenther@suse.de>
4140 PR tree-optimization/46590
4141 * opts.c (default_options_table): Add entries for
4142 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4143 all enabled at -O1 but not for -Og.
4144 * common.opt (fbranch-count-reg): Remove Init(1).
4145 (fmove-loop-invariants): Likewise.
4146 (ftree-pta): Likewise.
4148 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4150 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4151 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4152 decls to at least the GCC 4.8 used alignments.
4155 * tree-nested.c (convert_nonlocal_reference_stmt,
4156 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4157 of GIMPLE_BIND stmts, adjust associated decls.
4159 2014-01-17 Richard Biener <rguenther@suse.de>
4161 PR tree-optimization/46590
4162 * vec.h (vec<>::bseach): New member function implementing
4163 binary search according to C89 bsearch.
4164 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4165 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4166 bitmap pointer again. Make accesses_in_loop a flat array.
4167 (mem_ref_obstack): New global.
4168 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4169 (mark_ref_stored): Likewise.
4170 (ref_indep_loop_p_2): Likewise.
4171 (set_ref_stored_in_loop): New helper function.
4172 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4173 (memref_free): Adjust.
4174 (record_mem_ref_loc): Simplify.
4175 (gather_mem_refs_stmt): Adjust.
4176 (sort_locs_in_loop_postorder_cmp): New function.
4177 (analyze_memory_references): Sort accesses_in_loop after
4178 loop postorder number.
4179 (find_ref_loc_in_loop_cmp): New function.
4180 (for_all_locs_in_loop): Find relevant cluster of locs in
4181 accesses_in_loop and iterate without recursion.
4182 (execute_sm): Avoid uninit warning.
4183 (struct ref_always_accessed): Simplify.
4184 (ref_always_accessed::operator ()): Likewise.
4185 (ref_always_accessed_p): Likewise.
4186 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4187 loop postorder numbers here.
4188 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4191 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4194 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4195 on decls for which assemble_alias has been called.
4197 2014-01-17 Nick Clifton <nickc@redhat.com>
4199 * config/msp430/msp430.opt: (mcpu): New option.
4200 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4201 (msp430_option_override): Parse target_cpu. If the MCU name
4202 matches a generic string, clear target_mcu.
4203 (msp430_attr): Allow numeric interrupt values up to 63.
4204 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4205 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4207 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4209 * config/msp430/msp430.md (popm): Use %J rather than %I.
4210 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4211 (addhi_cy_i): Use immediate_operand for operand 2.
4212 * doc/invoke.texi: Document -mcpu option.
4214 2014-01-17 Richard Biener <rguenther@suse.de>
4216 PR rtl-optimization/38518
4217 * df.h (df_analyze_loop): Declare.
4218 * df-core.c: Include cfgloop.h.
4219 (df_analyze_1): Split out main part of df_analyze.
4220 (df_analyze): Adjust.
4221 (loop_inverted_post_order_compute): New function.
4222 (loop_post_order_compute): Likewise.
4223 (df_analyze_loop): New function avoiding whole-function
4225 * loop-invariant.c (find_defs): Use df_analyze_loop.
4226 (find_invariants): Adjust.
4227 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4229 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4231 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4232 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4234 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4236 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4237 traversal when removing references.
4239 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4242 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4244 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4247 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4248 pushing a reload for an autoinc when we had previously reloaded an
4249 inner part of the address.
4251 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4253 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4255 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4256 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4257 when not giving up or versioning for alias only because of
4259 (vect_analyze_data_ref_dependences): Set to true.
4260 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4262 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4263 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4267 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4270 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4271 operand 0 predicate for gathers, use a new pseudo as subtarget.
4273 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4276 * lra-constraints.c (process_alt_operands): Add printing debug info.
4277 Check absence of input/output reloads for matched operands too.
4279 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4281 PR rtl-optimization/59835
4282 * ira.c (ira_init_register_move_cost): Increase cost for
4285 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4287 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4289 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4292 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4293 non-register objects. Use gen_(high/low)part more consistently.
4296 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4299 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4300 endian support, remove tests for WORDS_BIG_ENDIAN.
4301 (p8_mfvsrd_3_<mode>): Likewise.
4302 (reload_gpr_from_vsx<mode>): Likewise.
4303 (reload_gpr_from_vsxsf): Likewise.
4304 (p8_mfvsrd_4_disf): Likewise.
4306 2014-01-16 Richard Biener <rguenther@suse.de>
4308 PR rtl-optimization/46590
4309 * lcm.c (compute_antinout_edge): Use postorder iteration.
4310 (compute_laterin): Use inverted postorder iteration.
4312 2014-01-16 Nick Clifton <nickc@redhat.com>
4315 * varasm.c (output_constant): Return the number of bytes actually
4317 (output_constructor_array_range): Update the field size with the
4318 number of bytes emitted by output_constant.
4319 (output_constructor_regular_field): Likewise. Also do not
4320 complain if the total number of bytes emitted is now greater
4321 than the expected fieldpos.
4322 * output.h (output_constant): Update prototype and descriptive comment.
4324 2014-01-16 Marek Polacek <polacek@redhat.com>
4327 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4328 it is error_mark_node.
4330 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4332 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4333 VALID_AVX256_REG_OR_OI_MODE.
4335 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4337 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4338 current procedure should be profiled.
4340 2014-01-15 Andrew Pinski <apinski@cavium.com>
4342 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4343 of moving from/to the STACK_REG register class.
4345 2014-01-15 Richard Henderson <rth@redhat.com>
4348 * reginfo.c (global_regs_decl): Globalize.
4349 * rtl.h (global_regs_decl): Declare.
4350 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4351 reserved via global_regs.
4353 2014-01-15 Teresa Johnson <tejohnson@google.com>
4355 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4357 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4359 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4360 and vmulosh rather than call gen_vec_widen_smult_*.
4361 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4362 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4363 (vec_widen_smult_even_v16qi): Likewise.
4364 (vec_widen_umult_even_v8hi): Likewise.
4365 (vec_widen_smult_even_v8hi): Likewise.
4366 (vec_widen_umult_odd_v16qi): Likewise.
4367 (vec_widen_smult_odd_v16qi): Likewise.
4368 (vec_widen_umult_odd_v8hi): Likewise.
4369 (vec_widen_smult_odd_v8hi): Likewise.
4370 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4371 vmuloub rather than call gen_vec_widen_umult_*.
4372 (vec_widen_umult_lo_v16qi): Likewise.
4373 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4374 vmulosb rather than call gen_vec_widen_smult_*.
4375 (vec_widen_smult_lo_v16qi): Likewise.
4376 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4377 rather than call gen_vec_widen_umult_*.
4378 (vec_widen_umult_lo_v8hi): Likewise.
4379 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4380 rather than call gen_vec_widen_smult_*.
4381 (vec_widen_smult_lo_v8hi): Likewise.
4383 2014-01-15 Jeff Law <law@redhat.com>
4385 PR tree-optimization/59747
4386 * ree.c (find_and_remove_re): Properly handle case where a second
4387 eliminated extension requires widening a copy created for elimination
4388 of a prior extension.
4389 (combine_set_extension): Ensure that the number of hard regs needed
4390 for a destination register does not change when we widen it.
4392 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4394 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4395 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4396 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4397 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4398 (avr-*-rtems*): Likewise.
4399 (bfin*-rtems*): Likewise.
4400 (moxie-*-rtems*): Likewise.
4401 (h8300-*-rtems*): Likewise.
4402 (i[34567]86-*-rtems*): Likewise.
4403 (lm32-*-rtems*): Likewise.
4404 (m32r-*-rtems*): Likewise.
4405 (m68k-*-rtems*): Likewise.
4406 (microblaze*-*-rtems*): Likewise.
4407 (mips*-*-rtems*): Likewise.
4408 (powerpc-*-rtems*): Likewise.
4409 (sh-*-rtems*): Likewise.
4410 (sparc-*-rtems*): Likewise.
4411 (sparc64-*-rtems*): Likewise.
4412 (v850-*-rtems*): Likewise.
4413 (m32c-*-rtems*): Likewise.
4415 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4417 PR rtl-optimization/59511
4418 * ira.c (ira_init_register_move_cost): Use memory costs for some
4419 cases of register move cost calculations.
4420 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4421 instead of BB frequency.
4422 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4423 * lra-assigns.c (find_hard_regno_for): Ditto.
4425 2014-01-15 Richard Biener <rguenther@suse.de>
4427 PR tree-optimization/59822
4428 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4429 (vectorizable_load): Use it to hoist defs of uses of invariant
4430 loads out of the loop.
4432 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4433 Kugan Vivekanandarajah <kuganv@linaro.org>
4436 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4439 2014-01-15 Richard Biener <rguenther@suse.de>
4441 PR rtl-optimization/59802
4442 * lcm.c (compute_available): Use inverted postorder to seed
4443 the initial worklist.
4445 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4448 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4449 ADDR_REGS for invalid symrefs in non-PIC code.
4451 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4454 * builtins.c (determine_block_size): Initialize *probable_max_size
4455 even if len_rtx is CONST_INT.
4457 2014-01-14 Andrew Pinski <apinski@cavium.com>
4459 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4460 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4461 (cortexa53_tunings): Likewise.
4462 (aarch64_sched_issue_rate): New function.
4463 (TARGET_SCHED_ISSUE_RATE): Define.
4465 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4467 * ira-costs.c (find_costs_and_classes): Add missed
4468 ira_init_register_move_cost_if_necessary.
4470 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4473 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4475 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4478 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4479 to indicate if type is used for function return value. Warn ABI
4480 change if the vector mode isn't available for function return value.
4481 (ix86_function_arg_advance): Pass false to type_natural_mode.
4482 (ix86_function_arg): Likewise.
4483 (ix86_gimplify_va_arg): Likewise.
4484 (function_arg_32): Don't warn ABI change.
4485 (ix86_function_value): Pass true to type_natural_mode.
4486 (ix86_return_in_memory): Likewise.
4487 (ix86_struct_value_rtx): Removed.
4488 (TARGET_STRUCT_VALUE_RTX): Likewise.
4490 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4492 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4493 converting a conditional jump into a conditional return.
4495 2014-01-14 Richard Biener <rguenther@suse.de>
4497 PR tree-optimization/58921
4498 PR tree-optimization/59006
4499 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4500 hoisting invariant stmts.
4501 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4502 invariant loads on the preheader edge if possible.
4504 2014-01-14 Joey Ye <joey.ye@arm.com>
4506 * doc/plugin.texi (Building GCC plugins): Update to C++.
4508 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4510 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4511 (_mm_rcp28_round_ss): Ditto.
4512 (_mm_rsqrt28_round_sd): Ditto.
4513 (_mm_rsqrt28_round_ss): Ditto.
4514 (_mm_rcp28_sd): Ditto.
4515 (_mm_rcp28_ss): Ditto.
4516 (_mm_rsqrt28_sd): Ditto.
4517 (_mm_rsqrt28_ss): Ditto.
4518 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4519 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4520 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4521 (IX86_BUILTIN_RCP28SD): Ditto.
4522 (IX86_BUILTIN_RCP28SS): Ditto.
4523 (IX86_BUILTIN_RSQRT28SD): Ditto.
4524 (IX86_BUILTIN_RSQRT28SS): Ditto.
4525 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4526 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4527 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4528 (ix86_expand_special_args_builtin): Expand new FTYPE.
4529 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4530 (srcp14<mode>): Make insn unary.
4531 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4532 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4533 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4534 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4535 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4536 Fix rounding: make it SAE only.
4537 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4539 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4541 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4542 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4543 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4544 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4545 (round_saeonly_mask_scalar_operand4): Ditto.
4546 (round_saeonly_mask_scalar_op3): Ditto.
4547 (round_saeonly_mask_scalar_op4): Ditto.
4549 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4551 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4552 Implement -maltivec=be for vec_insert and vec_extract.
4554 2014-01-10 DJ Delorie <dj@redhat.com>
4556 * config/msp430/msp430.md (call_internal): Don't allow memory
4557 references with SP as the base register.
4558 (call_value_internal): Likewise.
4559 * config/msp430/constraints.md (Yc): New. For memory references
4560 that don't use SP as a base register.
4562 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4563 "an integer without a # prefix"
4564 * config/msp430/msp430.md (epilogue_helper): Use it.
4566 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4569 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4570 AVX512F gather builtins.
4571 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4572 on gather decls with INTEGER_TYPE masktype.
4573 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4574 directly into the builtin rather than hoisting it before loop.
4576 PR tree-optimization/59387
4577 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4578 (scev_const_prop): If folded_casts and type has undefined overflow,
4579 use force_gimple_operand instead of force_gimple_operand_gsi and
4580 for each added stmt if it is assign with
4581 arith_code_with_undefined_signed_overflow, call
4582 rewrite_to_defined_overflow.
4583 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4584 gimple-fold.h instead.
4585 (arith_code_with_undefined_signed_overflow,
4586 rewrite_to_defined_overflow): Moved to ...
4587 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4588 rewrite_to_defined_overflow): ... here. No longer static.
4589 Include gimplify-me.h.
4590 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4591 rewrite_to_defined_overflow): New prototypes.
4593 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4595 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4597 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4599 * builtins.c (get_object_alignment_2): Minor tweak.
4600 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4602 2014-01-13 Christian Bruel <christian.bruel@st.com>
4604 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4605 optimized non constant lengths.
4607 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4610 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4611 load as __atomic_load_N if possible.
4613 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4615 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4617 (rs6000_expand_builtin): Adjust call.
4619 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4622 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4623 * config/rs6000/rs6000.c: Include target-globals.h.
4624 (rs6000_set_current_function): Instead of doing target_reinit
4625 unconditionally, use save_target_globals_default_opts and
4626 restore_target_globals.
4628 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4630 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4631 (rs6000_expand_builtin): Handle mffs and mtfsf.
4632 (rs6000_init_builtins): Define mffs and mtfsf.
4633 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4634 (rs6000_mffs): New pattern.
4635 (rs6000_mtfsf): New pattern.
4637 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4639 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4640 Start narrowing with START. Apply candidate-use pair
4641 and check overall cost in narrowing.
4642 (iv_ca_prune): Pass new argument.
4644 2014-01-10 Jeff Law <law@redhat.com>
4647 * ree.c (combine_reaching_defs): Ensure the defining statement
4648 occurs before the extension when optimizing extensions with
4649 different source and destination hard registers.
4651 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4654 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4655 vtables into the type inheritance graph.
4657 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4659 PR rtl-optimization/59754
4660 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4661 modes in the REGNO != REGNO case.
4663 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4665 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4667 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4669 PR tree-optimization/59745
4670 * tree-predcom.c (tree_predictive_commoning_loop): Call
4671 free_affine_expand_cache if giving up because components is NULL.
4673 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4674 GC in payload of target_globals struct instead of allocating them on
4675 the heap and the larger structs separately using GC.
4676 * target-globals.h (struct target_globals): Make regs, hard_regs,
4677 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4678 of GTY((skip)) and change type to void *.
4679 (reset_target_globals): Cast loads from those fields to corresponding
4682 2014-01-10 Steve Ellcey <sellcey@mips.com>
4685 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4686 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4687 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4689 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4692 * aarch64-modes.def (CC_Zmode): New flags mode.
4693 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4694 represents an equality.
4695 (aarch64_get_condition_code): Handle CC_Zmode.
4696 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4698 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4700 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4701 extraction in good case.
4703 2014-01-10 Richard Biener <rguenther@suse.de>
4705 PR tree-optimization/59374
4706 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4707 checking after SLP discovery. Mark stmts not participating
4708 in any SLP instance properly.
4710 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4712 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4713 when handling a SET rtx.
4715 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4717 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4718 (cortex-a57): Likewise.
4719 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4721 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4723 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4724 non-iwmmxt builtins.
4726 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4730 * ipa-devirt.c record_target_from_binfo): Take as argument
4731 stack of binfos and lookup matching one for virtual inheritance.
4732 (possible_polymorphic_call_targets_1): Update.
4734 2014-01-10 Huacai Chen <chenhc@lemote.com>
4736 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4737 kernel strings for Loongson-2E/2F/3A.
4739 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4742 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4743 is_gimple_call before calling gimple_call_internal_p.
4745 2014-01-09 Steve Ellcey <sellcey@mips.com>
4747 * Makefile.in (TREE_FLOW_H): Remove.
4748 (TREE_SSA_H): Add file names from tree-flow.h.
4749 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4750 * tree.h: Remove tree-flow.h reference.
4751 * hash-table.h: Remove tree-flow.h reference.
4752 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4753 reference with tree-ssa-loop.h.
4755 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4757 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4758 default element-order behavior for -maltivec.
4759 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4761 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4762 when targeting big endian, at least for now.
4763 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4765 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4768 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4769 var satisfies use_register_for_decl, just take into account type
4770 alignment, rather than decl alignment.
4772 PR tree-optimization/59622
4773 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4774 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4775 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4776 Don't devirtualize for inplace at all. For targets.length () == 1,
4777 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4779 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4781 * config/i386/i386.md (cpu): Remove the unused btver1.
4783 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4785 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4787 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4790 * tree-core.h (struct target_globals): New forward declaration.
4791 (struct tree_target_option): Add globals field.
4792 * tree.h (TREE_TARGET_GLOBALS): Define.
4793 (prepare_target_option_nodes_for_pch): New prototype.
4794 * target-globals.h (struct target_globals): Define even if
4796 * tree.c (prepare_target_option_node_for_pch,
4797 prepare_target_option_nodes_for_pch): New functions.
4798 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4799 * config/i386/i386.c: Include target-globals.h.
4800 (ix86_set_current_function): Instead of doing target_reinit
4801 unconditionally, use save_target_globals_default_opts and
4802 restore_target_globals.
4804 2014-01-09 Richard Biener <rguenther@suse.de>
4806 PR tree-optimization/59715
4807 * tree-cfg.h (split_critical_edges): Declare.
4808 * tree-cfg.c (split_critical_edges): Export.
4809 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4811 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4813 * cfgexpand.c (expand_stack_vars): Optionally disable
4814 asan stack protection.
4815 (expand_used_vars): Likewise.
4816 (partition_stack_vars): Likewise.
4817 * asan.c (asan_emit_stack_protection): Optionally disable
4818 after return stack usage.
4819 (instrument_derefs): Optionally disable memory access instrumentation.
4820 (instrument_builtin_call): Likewise.
4821 (instrument_strlen_call): Likewise.
4822 (asan_protect_global): Optionally disable global variables protection.
4823 * doc/invoke.texi: Added doc for new options.
4824 * params.def: Added new options.
4825 * params.h: Likewise.
4827 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4829 PR rtl-optimization/59724
4830 * ifcvt.c (cond_exec_process_if_block): Don't call
4831 flow_find_head_matching_sequence with 0 longest_match.
4832 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4833 non-active insns if !stop_after.
4834 (try_head_merge_bb): Revert 2014-01-07 changes.
4836 2014-01-08 Jeff Law <law@redhat.com>
4838 * ree.c (get_sub_rtx): New function, extracted from...
4839 (merge_def_and_ext): Here.
4840 (combine_reaching_defs): Use get_sub_rtx.
4842 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4844 * cgraph.h (varpool_variable_node): Do not choke on null node.
4846 2014-01-08 Catherine Moore <clm@codesourcery.com>
4848 * config/mips/mips.md (simple_return): Attempt to use JRC
4850 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4852 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4854 PR rtl-optimization/59137
4855 * reorg.c (steal_delay_list_from_target): Call update_block for
4857 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4859 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4861 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4862 two duplicate entries.
4864 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4867 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4869 * config/mips/mips.c (mips_truncated_op_cost): New function.
4870 (mips_rtx_costs): Adjust test for BADDU.
4871 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4873 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4875 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4876 (*baddu_si): ...this new pattern.
4878 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4881 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4883 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4886 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4888 (expand_expr, expand_normal): Adjust.
4889 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4890 inner_reference_p. Use inner_reference_p to expand inner references.
4891 (store_expr): Adjust.
4892 * cfgexpand.c (expand_call_stmt): Adjust.
4894 2014-01-08 Rong Xu <xur@google.com>
4896 * gcov-io.c (gcov_var): Move from gcov-io.h.
4897 (gcov_position): Ditto.
4898 (gcov_is_error): Ditto.
4899 (gcov_rewrite): Ditto.
4900 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4901 only part to libgcc/libgcov.h.
4903 2014-01-08 Marek Polacek <polacek@redhat.com>
4906 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4908 2014-01-08 Marek Polacek <polacek@redhat.com>
4911 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4913 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4915 PR rtl-optimization/59649
4916 * stor-layout.c (get_mode_bounds): For BImode return
4917 0 and STORE_FLAG_VALUE.
4919 2014-01-08 Richard Biener <rguenther@suse.de>
4922 * gimple.h (is_gimple_builtin_call): Remove.
4923 (gimple_builtin_call_types_compatible_p): New.
4924 (gimple_call_builtin_p): New overload.
4925 * gimple.c (is_gimple_builtin_call): Remove.
4926 (validate_call): Rename to ...
4927 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4929 (validate_type): New static function.
4930 (gimple_call_builtin_p): New overload and adjust.
4931 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4932 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4933 (gimple_fold_stmt_to_constant_1): Likewise.
4934 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4936 2014-01-08 Richard Biener <rguenther@suse.de>
4939 * gimplify.c (gimplify_expr): Gimplify register-register type
4940 VIEW_CONVERT_EXPRs to separate stmts.
4942 2014-01-07 Jeff Law <law@redhat.com>
4945 * ree.c (combine_set_extension): Handle case where source
4946 and destination registers in an extension insn are different.
4947 (combine_reaching_defs): Allow source and destination registers
4948 in extension to be different under limited circumstances.
4949 (add_removable_extension): Remove restriction that the
4950 source and destination registers in the extension are the same.
4951 (find_and_remove_re): Emit a copy from the extension's
4952 destination to its source after the defining insn if
4953 the source and destination registers are different.
4956 * ifcvt.c (merge_if_block): If we are merging a block with more than
4957 one successor with a block with no successors, remove any BARRIER
4958 after the second block.
4960 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4962 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4964 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4967 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4968 for 14-bit register offsets when INT14_OK_STRICT is false.
4970 2014-01-07 Roland Stigge <stigge@antcom.de>
4971 Michael Meissner <meissner@linux.vnet.ibm.com>
4974 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4975 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4977 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4979 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4982 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4984 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4985 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4986 rtx is const0_rtx or not.
4988 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4991 * target-globals.c (save_target_globals): Remove this_fn_optab
4993 * toplev.c: Include optabs.h.
4994 (target_reinit): Temporarily restore the global options if another
4995 set of options are in force.
4997 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4999 PR rtl-optimization/58668
5000 * cfgcleanup.c (flow_find_cross_jump): Don't count
5001 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5002 to determine what is counted.
5003 (flow_find_head_matching_sequence): Use active_insn_p to determine
5005 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5007 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5008 determine what is counted.
5010 PR tree-optimization/59643
5011 * tree-predcom.c (split_data_refs_to_components): If one dr is
5012 read and one write, determine_offset fails and the write isn't
5013 in the bad component, just put the read into the bad component.
5015 2014-01-07 Mike Stump <mikestump@comcast.net>
5016 Jakub Jelinek <jakub@redhat.com>
5019 * tree-core.h (struct tree_optimization_option): Change optabs
5020 type from unsigned char * to void *.
5021 * optabs.c (init_tree_optimization_optabs): Adjust
5022 TREE_OPTIMIZATION_OPTABS initialization.
5024 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5027 * config/i386/i386.h (struct machine_function): Add
5028 no_drap_save_restore field.
5029 * config/i386/i386.c (ix86_save_reg): Use
5030 !cfun->machine->no_drap_save_restore instead of
5031 crtl->stack_realign_needed.
5032 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5033 this function clears frame_pointer_needed. Set
5034 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5035 and DRAP reg is needed.
5037 2014-01-06 Marek Polacek <polacek@redhat.com>
5040 * doc/implement-c.texi: Mention that other integer types are
5041 permitted as bit-field types in strictly conforming mode.
5043 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5045 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5048 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5050 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5052 2014-01-06 Martin Jambor <mjambor@suse.cz>
5055 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5057 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5059 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5063 * var-tracking.c (add_stores): Preserve the value of the source even if
5064 we don't record the store.
5066 2014-01-06 Terry Guo <terry.guo@arm.com>
5068 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5070 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5073 * config/darwin.c (darwin_function_section): Adjust return values to
5074 correspond to optimisation changes made in r206070.
5076 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5078 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5079 from prefetch_block tune setting.
5080 (nocona_cost): Correct size of prefetch block to 64.
5082 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5084 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5085 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5086 used to save the static chain register in the computation of the offset
5087 from which the FP registers need to be restored.
5089 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5091 PR tree-optimization/59519
5092 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5093 ICE if get_current_def (current_new_name) is already non-NULL, as long
5094 as it is a phi result of some other phi in *new_exit_bb that has
5097 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5098 or vmovdqu* for misaligned_operand.
5099 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5100 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5101 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5102 aligned_mem for AVX512F masked aligned load and store builtins and for
5105 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5107 PR tree-optimization/59651
5108 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5109 Address range for negative step should be added by TYPE_SIZE_UNIT.
5111 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5113 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5114 overlapping registers also for registers other than the stack pointer.
5116 2014-01-03 Marek Polacek <polacek@redhat.com>
5119 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5122 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5125 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5128 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5129 (push splitter): Use <P:MODE_SIZE> instead of
5130 GET_MODE_SIZE (<P:MODE>mode).
5131 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5132 (mov -1, reg peephole2): Likewise.
5133 * config/i386/sse.md (*mov<mode>_internal,
5134 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5135 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5136 *<code><mode>3, *andnot<mode>3<mask_name>,
5137 <mask_codefor><code><mode>3<mask_name>): Likewise.
5138 * config/i386/subst.md (mask_mode512bit_condition,
5139 sd_mask_mode512bit_condition): Likewise.
5141 2014-01-02 Xinliang David Li <davidxl@google.com>
5143 PR tree-optimization/59303
5144 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5145 (dump_predicates): Better output format.
5146 (pred_equal_p): New function.
5147 (is_neq_relop_p): Ditto.
5148 (is_neq_zero_form_p): Ditto.
5149 (pred_expr_equal_p): Ditto.
5150 (pred_neg_p): Ditto.
5151 (simplify_pred): Ditto.
5152 (simplify_preds_2): Ditto.
5153 (simplify_preds_3): Ditto.
5154 (simplify_preds_4): Ditto.
5155 (simplify_preds): Ditto.
5157 (push_to_worklist): Ditto.
5158 (get_pred_info_from_cmp): Ditto.
5159 (is_degenerated_phi): Ditto.
5160 (normalize_one_pred_1): Ditto.
5161 (normalize_one_pred): Ditto.
5162 (normalize_one_pred_chain): Ditto.
5163 (normalize_preds): Ditto.
5164 (normalize_cond_1): Remove function.
5165 (normalize_cond): Ditto.
5166 (is_gcond_subset_of): Ditto.
5167 (is_subset_of_any): Ditto.
5168 (is_or_set_subset_of): Ditto.
5169 (is_and_set_subset_of): Ditto.
5170 (is_norm_cond_subset_of): Ditto.
5171 (pred_chain_length_cmp): Ditto.
5172 (convert_control_dep_chain_into_preds): Type change.
5173 (find_predicates): Ditto.
5174 (find_def_preds): Ditto.
5175 (destroy_predicates_vecs): Ditto.
5176 (find_matching_predicates_in_rest_chains): Ditto.
5177 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5178 (is_pred_expr_subset): Ditto.
5179 (is_pred_chain_subset_of): Ditto.
5180 (is_included_in): Ditto.
5181 (is_superset_of): Ditto.
5183 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5185 Update copyright years.
5187 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5189 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5190 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5191 config/arc/arc.md, config/arc/arc.opt,
5192 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5193 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5194 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5195 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5196 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5197 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5198 vtable-verify.c, vtable-verify.h: Use the standard form for the
5201 2014-01-02 Tobias Burnus <burnus@net-b.de>
5203 * gcc.c (process_command): Update copyright notice dates.
5204 * gcov-dump.c: Ditto.
5206 * doc/cpp.texi: Bump @copying's copyright year.
5207 * doc/cppinternals.texi: Ditto.
5208 * doc/gcc.texi: Ditto.
5209 * doc/gccint.texi: Ditto.
5210 * doc/gcov.texi: Ditto.
5211 * doc/install.texi: Ditto.
5212 * doc/invoke.texi: Ditto.
5214 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5216 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5218 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5220 * config/i386/sse.md (*mov<mode>_internal): Guard
5221 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5223 PR rtl-optimization/59647
5224 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5225 new_rtx into UNSIGNED_FLOAT rtxes.
5227 Copyright (C) 2014 Free Software Foundation, Inc.
5229 Copying and distribution of this file, with or without modification,
5230 are permitted in any medium without royalty provided the copyright
5231 notice and this notice are preserved.