1 2014-03-20 Jakub Jelinek <jakub@redhat.com>
4 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
5 into CONST, put pic register as first operand of PLUS. Use
6 gen_const_mem for both 32-bit and 64-bit PIC got loads.
8 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
12 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
14 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
15 around for store forwarding issue in the FPU on the UT699.
16 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
17 loads and operations if -mfix-ut699 is specified.
18 (divtf3_hq): Tweak attribute.
19 (sqrttf2_hq): Likewise.
21 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
23 * calls.c (store_one_arg): Remove incorrect const qualification on the
24 type of the temporary.
25 * cfgexpand.c (expand_return): Likewise.
26 * expr.c (expand_constructor): Likewise.
27 (expand_expr_real_1): Likewise.
29 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
31 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
34 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
37 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
39 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
41 * config/arm/aarch-common-protos.h
42 (alu_cost_table): Fix spelling of "extend".
43 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
45 2014-03-19 Richard Biener <rguenther@suse.de>
48 * tree-core.h (tree_type_common): Re-order pointer members
49 to reduce recursion depth during GC walks.
51 2014-03-19 Marek Polacek <polacek@redhat.com>
54 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
57 2014-03-19 Richard Biener <rguenther@suse.de>
60 * lto-streamer-in.c (input_function): In WPA stage do not drop
63 2014-03-19 Jakub Jelinek <jakub@redhat.com>
65 PR tree-optimization/60559
66 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
67 with build_zero_cst assignment.
69 2014-03-18 Kai Tietz <ktietz@redhat.com>
71 PR rtl-optimization/56356
72 * sdbout.c (sdbout_parms): Verify that parms'
73 incoming argument is valid.
74 (sdbout_reg_parms): Likewise.
76 2014-03-18 Richard Henderson <rth@redhat.com>
79 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
80 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
81 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
83 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
85 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
86 * doc/plugins.texi (Plugin callbacks): Mention
88 Italicize plugin event names in description. Explain that
89 PLUGIN_PRAGMAS has no sense for lto1. Explain
91 Remind that no GCC functions should be called after
93 Explain what pragmas with expansion are.
95 2014-03-18 Martin Liska <mliska@suse.cz>
97 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
98 gimple call statement is update.
99 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
100 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
102 2014-03-18 Jakub Jelinek <jakub@redhat.com>
105 * ubsan.c (ubsan_instrument_unreachable): Call
106 initialize_sanitizer_builtins.
107 (ubsan_pass): Likewise.
110 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
111 varpool_finalize_decl instead of rest_of_decl_compilation.
113 2014-03-18 Richard Biener <rguenther@suse.de>
115 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
116 by using bitmap_and_compl instead of bitmap_and_compl_into.
117 (df_rd_transfer_function): Likewise.
119 2014-03-18 Richard Biener <rguenther@suse.de>
121 * doc/lto.texi (fresolution): Fix typo.
123 2014-03-18 Richard Biener <rguenther@suse.de>
125 * doc/invoke.texi (flto): Update for changes in 4.9.
127 2014-03-18 Richard Biener <rguenther@suse.de>
129 * doc/loop.texi: Remove section on the removed lambda framework.
130 Update loop docs with recent changes in preserving loop structure.
132 2014-03-18 Richard Biener <rguenther@suse.de>
134 * doc/lto.texi (-fresolution): Document.
136 2014-03-18 Richard Biener <rguenther@suse.de>
138 * doc/contrib.texi: Adjust my name.
140 2014-03-18 Jakub Jelinek <jakub@redhat.com>
143 * internal-fn.c: Include diagnostic-core.h.
144 (expand_BUILTIN_EXPECT): New function.
145 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
146 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
147 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
148 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
150 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
151 Revert 3 argument __builtin_expect code.
152 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
153 * gimple-fold.c (gimple_fold_call): Likewise.
154 * tree.h (fold_builtin_expect): New prototype.
155 * builtins.c (build_builtin_expect_predicate): Add predictor
156 argument, if non-NULL, create 3 argument __builtin_expect.
157 (fold_builtin_expect): No longer static. Add ARG2 argument,
158 pass it through to build_builtin_expect_predicate.
159 (fold_builtin_2): Adjust caller.
160 (fold_builtin_3): Handle BUILT_IN_EXPECT.
161 * internal-fn.def (BUILTIN_EXPECT): New.
163 2014-03-18 Tobias Burnus <burnus@net-b.de>
166 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
167 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
168 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
170 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
173 * predict.c (combine_predictions_for_bb): Fix up formatting.
174 (expr_expected_value_1, expr_expected_value): Add predictor argument,
175 fill what it points to if non-NULL.
176 (tree_predict_by_opcode): Adjust caller, use the predictor.
177 * predict.def (PRED_COMPARE_AND_SWAP): Add.
179 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
181 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
182 proper constant for the store mode.
184 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
186 * symtab.c (change_decl_assembler_name): Fix transparent alias
189 2014-03-16 Renlin Li <Renlin.Li@arm.com>
191 * config/aarch64/aarch64.c: Correct the comments about the
192 aarch64 stack layout.
194 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
196 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
197 check for GF_OMP_FOR_KIND_FOR.
199 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
201 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
202 ymm and zmm register names.
204 2014-03-17 Jakub Jelinek <jakub@redhat.com>
207 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
208 note creation for the 2010-08-31 changes.
210 2014-03-17 Marek Polacek <polacek@redhat.com>
213 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
214 as -fno-tree-loop-vectorize.
215 (expand_omp_simd): Likewise.
217 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
219 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
220 (eligible_for_call_delay): New prototype.
221 * config/sparc/sparc.c (tls_call_delay): Rename into...
222 (eligible_for_call_delay): ...this. Return false if the instruction
223 cannot be put in the delay slot of a branch.
224 (eligible_for_restore_insn): Simplify.
225 (eligible_for_return_delay): Return false if the instruction cannot be
226 put in the delay slot of a branch and simplify.
227 (eligible_for_sibcall_delay): Return false if the instruction cannot be
228 put in the delay slot of a branch.
229 * config/sparc/sparc.md (fix_ut699): New attribute.
230 (tls_call_delay): Delete.
231 (in_call_delay): Reimplement.
232 (eligible_for_sibcall_delay): Rename into...
233 (in_sibcall_delay): ...this.
234 (eligible_for_return_delay): Rename into...
235 (in_return_delay): ...this.
236 (in_branch_delay): Reimplement.
237 (in_uncond_branch_delay): Delete.
238 (in_annul_branch_delay): Delete.
240 2014-03-14 Richard Henderson <rth@redhat.com>
243 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
244 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
245 (*floathi<X87MODEF>2_i387_with_temp): Remove.
246 (floathi splitters): Remove.
247 (float<SWI48x>xf2): New pattern.
248 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
249 code that tried to handle DImode for 32-bit, but which was excluded
250 by the pattern's condition. Drop allocation of stack temporary.
251 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
252 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
253 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
254 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
255 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
256 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
257 (*float<SWI48><MODEF>2_sse_interunit): Remove.
258 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
259 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
260 (*float<SWI48x><X87MODEF>2_i387): Remove.
261 (all float _with_temp splitters): Remove.
262 (*float<SWI48x><MODEF>2_i387): New pattern.
263 (*float<SWI48><MODEF>2_sse): New pattern.
264 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
265 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
267 2014-03-14 Jakub Jelinek <jakub@redhat.com>
268 Marek Polacek <polacek@redhat.com>
271 * common.opt (dump_base_name_prefixed): New Variable.
272 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
273 if x_dump_base_name_prefixed is already set, set it at the end.
275 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
277 PR rtl-optimization/60508
278 * lra-constraints.c (get_reload_reg): Add new parameter
280 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
281 Pass the new parameter values.
283 2014-03-14 Richard Biener <rguenther@suse.de>
285 * common.opt: Revert unintented changes from r205065.
288 2014-03-14 Richard Biener <rguenther@suse.de>
291 * cfghooks.c (split_block): Properly adjust all loops the
292 block was a latch of.
294 2014-03-14 Martin Jambor <mjambor@suse.cz>
297 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
300 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
303 * config/avr/avr.c (avr_set_current_function): Pass function name
304 through default_strip_name_encoding before sanity checking instead
305 of skipping the first char of the assembler name.
307 2014-03-13 Richard Henderson <rth@redhat.com>
310 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
311 (ix86_force_to_memory, ix86_free_from_memory): Remove.
312 * config/i386/i386-protos.h: Likewise.
313 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
314 in the expander instead of a splitter.
315 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
316 any possibility of requiring a memory.
317 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
318 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
319 (fp branch splitters): Update for ix86_split_fp_branch.
320 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
321 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
322 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
323 (*fop_<MODEF>_2_i387): Remove f/r alternative.
324 (*fop_<MODEF>_3_i387): Likewise.
325 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
326 (splitters for the fop_* register patterns): Remove.
327 (fscalexf4_i387): Rename from *fscalexf4_i387.
328 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
330 2014-03-13 Jakub Jelinek <jakub@redhat.com>
332 PR tree-optimization/59779
333 * tree-dfa.c (get_ref_base_and_extent): Use double_int
334 type for bitsize and maxsize instead of HOST_WIDE_INT.
336 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
338 PR rtl-optimization/57320
339 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
340 the CFG after thread_prologue_and_epilogue_insns.
342 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
344 PR rtl-optimization/57189
345 * lra-constraints.c (process_alt_operands): Disfavor spilling
348 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
350 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
353 2014-03-13 Jakub Jelinek <jakub@redhat.com>
355 PR tree-optimization/59025
357 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
358 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
360 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
363 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
364 calls of avr_out_plus_1.
366 2014-03-13 Bin Cheng <bin.cheng@arm.com>
368 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
369 BB's single pred and update the father loop's latch info later.
371 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
373 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
377 (VEC_base): Likewise.
378 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
379 registers, we need to swap double words in little endian mode.
381 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
382 to be a container mode for 128-bit integer operations added in ISA
383 2.07. Unlike TImode and PTImode, the preferred register set is
384 the Altivec/VMX registers for the 128-bit operations.
386 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
388 (rs6000_split_128bit_ok_p): Likewise.
390 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
391 macros for creating ISA 2.07 normal and overloaded builtin
392 functions with 3 arguments.
393 (BU_P8V_OVERLOAD_3): Likewise.
394 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
395 for use as overloaded functions.
396 (VPERM_1TI_UNS): Likewise.
397 (VSEL_1TI): Likewise.
398 (VSEL_1TI_UNS): Likewise.
399 (ST_INTERNAL_1ti): Likewise.
400 (LD_INTERNAL_1ti): Likewise.
401 (XXSEL_1TI): Likewise.
402 (XXSEL_1TI_UNS): Likewise.
403 (VPERM_1TI): Likewise.
404 (VPERM_1TI_UNS): Likewise.
405 (XXPERMDI_1TI): Likewise.
407 (LXVD2X_V1TI): Likewise.
408 (STXVD2X_V1TI): Likewise.
409 (VEC_INIT_V1TI): Likewise.
410 (VEC_SET_V1TI): Likewise.
411 (VEC_EXT_V1TI): Likewise.
412 (EQV_V1TI): Likewise.
413 (NAND_V1TI): Likewise.
414 (ORC_V1TI): Likewise.
415 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
416 added in ISA 2.07. Add both normal 'altivec' builtins, and the
420 (VADDEUQM): Likewise.
421 (VADDECUQ): Likewise.
422 (VSUBEUQM): Likewise.
423 (VSUBECUQ): Likewise.
425 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
426 __int128_t and __uint128_t types.
427 (__uint128_type): Likewise.
428 (altivec_categorize_keyword): Add support for vector __int128_t,
429 vector __uint128_t, vector __int128, and vector unsigned __int128
430 as a container type for TImode operations that need to be done in
431 VSX/Altivec registers.
432 (rs6000_macro_to_expand): Likewise.
433 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
434 to support 128-bit integer instructions vaddcuq, vadduqm,
435 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
436 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
438 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
439 for V1TImode, and set up preferences to use VSX/Altivec registers.
440 Setup VSX reload handlers.
441 (rs6000_debug_reg_global): Likewise.
442 (rs6000_init_hard_regno_mode_ok): Likewise.
443 (rs6000_preferred_simd_mode): Likewise.
444 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
445 (easy_altivec_constant): Likewise.
446 (output_vec_const_move): Likewise.
447 (rs6000_expand_vector_set): Convert V1TImode set and extract to
449 (rs6000_expand_vector_extract): Likewise.
450 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
452 (rs6000_const_vec): Add support for V1TImode.
453 (rs6000_emit_le_vsx_load): Swap double words when loading or
454 storing TImode/V1TImode.
455 (rs6000_emit_le_vsx_store): Likewise.
456 (rs6000_emit_le_vsx_move): Likewise.
457 (rs6000_emit_move): Add support for V1TImode.
458 (altivec_expand_ld_builtin): Likewise.
459 (altivec_expand_st_builtin): Likewise.
460 (altivec_expand_vec_init_builtin): Likewise.
461 (altivec_expand_builtin): Likewise.
462 (rs6000_init_builtins): Add support for V1TImode type. Add
463 support for ISA 2.07 128-bit integer builtins. Define type names
464 for the VSX/Altivec vector types.
465 (altivec_init_builtins): Add support for overloaded vector
466 functions with V1TImode type.
467 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
468 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
470 (rs6000_split_128bit_ok_p): Likewise.
471 (rs6000_handle_altivec_attribute): Create V1TImode from vector
472 __int128_t and vector __uint128_t.
474 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
482 (VS_scalar): Likewise.
483 (VS_double): Likewise.
484 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
486 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
487 we support the ISA 2.07 128-bit integer arithmetic instructions.
488 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
489 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
490 and TImode types for use with the builtin functions.
491 (V1TI_type_node): Likewise.
492 (unsigned_V1TI_type_node): Likewise.
493 (intTI_type_internal_node): Likewise.
494 (uintTI_type_internal_node): Likewise.
496 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
497 128-bit builtin functions.
498 (UNSPEC_VADDEUQM): Likewise.
499 (UNSPEC_VADDECUQ): Likewise.
500 (UNSPEC_VSUBCUQ): Likewise.
501 (UNSPEC_VSUBEUQM): Likewise.
502 (UNSPEC_VSUBECUQ): Likewise.
503 (VM): Add V1TImode to vector mode iterators.
506 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
507 (altivec_vaddcuq): Likewise.
508 (altivec_vsubuqm): Likewise.
509 (altivec_vsubcuq): Likewise.
510 (altivec_vaddeuqm): Likewise.
511 (altivec_vaddecuq): Likewise.
512 (altivec_vsubeuqm): Likewise.
513 (altivec_vsubecuq): Likewise.
515 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
517 (BOOL_128): Likewise.
518 (BOOL_REGS_OUTPUT): Likewise.
519 (BOOL_REGS_OP1): Likewise.
520 (BOOL_REGS_OP2): Likewise.
521 (BOOL_REGS_UNARY): Likewise.
522 (BOOL_REGS_AND_CR0): Likewise.
524 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
525 128-bit integer builtin support.
526 (vec_vadduqm): Likewise.
527 (vec_vaddecuq): Likewise.
528 (vec_vaddeuqm): Likewise.
529 (vec_vsubecuq): Likewise.
530 (vec_vsubeuqm): Likewise.
531 (vec_vsubcuq): Likewise.
532 (vec_vsubuqm): Likewise.
534 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
535 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
536 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
537 128-bit integer add/subtract to ISA 2.07.
539 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
541 * config/arc/arc.c (arc_predicate_delay_insns):
542 Fix third argument passed to conditionalize_nonjump.
544 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
546 * config/aarch64/aarch64-builtins.c
547 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
548 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
549 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
550 instead of __builtin_lfloor.
551 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
553 2014-03-12 Jakub Jelinek <jakub@redhat.com>
555 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
556 (tree_ssa_ifcombine_bb_1): New function.
557 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
558 is an empty forwarder block to then_bb or vice versa and then_bb
559 and else_bb are effectively swapped.
561 2014-03-12 Christian Bruel <christian.bruel@st.com>
564 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
565 REG_CFA_DEF_CFA note.
566 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
567 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
569 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
571 PR tree-optimization/60454
572 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
574 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
576 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
577 Do not define target_cpu_default2 to generic.
578 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
579 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
580 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
582 2014-03-12 Jakub Jelinek <jakub@redhat.com>
583 Marc Glisse <marc.glisse@inria.fr>
585 PR tree-optimization/60502
586 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
587 instead of build_low_bits_mask.
589 2014-03-12 Jakub Jelinek <jakub@redhat.com>
592 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
593 if there are multiple uses, but op doesn't live on E edge.
594 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
595 clobber stmts before __builtin_unreachable.
597 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
599 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
600 hard_frame_pointer_rtx.
601 * cse.c (cse_insn): Remove volatile check.
602 * cselib.c (cselib_process_insn): Likewise.
603 * dse.c (scan_insn): Likewise.
605 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
607 * config/arc/arc.c (conditionalize_nonjump): New function,
609 (arc_ifcvt): ... this.
610 (arc_predicate_delay_insns): Use it.
612 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
614 * config/arc/predicates.md (extend_operand): During/after reload,
615 allow const_int_operand.
616 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
617 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
618 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
620 (umulsi3_highpart_i): Likewise.
622 2014-03-11 Richard Biener <rguenther@suse.de>
624 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
625 Add asserts to guard possible wrong-code bugs.
627 2014-03-11 Richard Biener <rguenther@suse.de>
629 PR tree-optimization/60429
630 PR tree-optimization/60485
631 * tree-ssa-structalias.c (set_union_with_increment): Properly
632 take into account all fields that overlap the shifted vars.
633 (do_sd_constraint): Likewise.
634 (do_ds_constraint): Likewise.
635 (get_constraint_for_ptr_offset): Likewise.
637 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
639 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
640 (nios2_compute_frame_layout):
641 Add calculation of cfun->machine->fp_save_offset.
642 (nios2_expand_prologue): Correct setting of frame pointer register
644 (nios2_expand_epilogue): Update recovery of stack pointer from
645 frame pointer accordingly.
646 (nios2_initial_elimination_offset): Update calculation of offset
647 for eliminating to HARD_FRAME_POINTER_REGNUM.
649 2014-03-10 Jakub Jelinek <jakub@redhat.com>
652 * ipa.c (symtab_remove_unreachable_nodes): Don't call
653 cgraph_get_create_node on VAR_DECLs.
655 2014-03-10 Richard Biener <rguenther@suse.de>
658 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
660 2014-03-08 Douglas B Rupp <rupp@gnat.com>
662 * config/vms/vms.opt (vms_float_format): New variable.
664 2014-03-08 Tobias Burnus <burnus@net-b.de>
666 * doc/invoke.texi (-fcilkplus): Update implementation status.
668 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
669 Richard Biener <rguenther@suse.de>
671 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
672 consistently accross all TUs.
673 (run_gcc): Enable -fshort-double automatically at link at link-time
674 and disallow override.
676 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
679 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
680 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
681 if they can't be used.
683 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
685 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
686 for Solaris 11/x86 ld.
687 * configure: Regenerate.
689 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
691 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
692 (LIB_TLS_SPEC): Save as ld_tls_libs.
693 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
694 (HAVE_AS_IX86_TLSLDM): New test.
695 * configure, config.in: Regenerate.
696 * config/i386/i386.c (legitimize_tls_address): Fall back to
697 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
698 cannot support TLS_MODEL_LOCAL_DYNAMIC.
699 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
700 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
702 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
704 * common.opt (fira-loop-pressure): Mark as optimization.
706 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
708 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
709 an OpenMP mappable type.
711 2014-03-06 Matthias Klose <doko@ubuntu.com>
713 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
714 MULTILIB_OSDIRNAMES is not defined.
716 2014-03-06 Jakub Jelinek <jakub@redhat.com>
717 Meador Inge <meadori@codesourcery.com>
720 * config/arm/arm.c (arm_tls_symbol_p): Remove.
721 (arm_legitimize_address): Call legitimize_tls_address for any
722 arm_tls_referenced_p expression, handle constant addend. Call it
723 before testing for !TARGET_ARM.
724 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
726 2014-03-06 Richard Biener <rguenther@suse.de>
732 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
734 * tree-streamer.c (record_common_node): Assert we don't record
735 nodes with type double.
736 (preload_common_node): Skip type double, complex double and double
737 pointer since it is now frontend dependent due to fshort-double option.
739 2014-03-06 Richard Biener <rguenther@suse.de>
741 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
742 or -fno-lto is specified and the linker has full plugin support.
743 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
744 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
745 * lto-wrapper.c (merge_and_complain): Merge compile-time
747 (run_gcc): And pass it through to the link options.
749 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
753 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
755 * cselib.c (remove_useless_values): Skip to avoid quadratic
756 behavior if the condition moved from...
757 (cselib_process_insn): ... here holds.
759 2014-03-05 Jakub Jelinek <jakub@redhat.com>
762 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
763 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
766 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
767 (TM_H): Add x86-tune.def.
769 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
771 * config/aarch64/aarch64.c (generic_tunings):
772 Use cortexa57_extra_costs.
774 2014-03-05 Jakub Jelinek <jakub@redhat.com>
777 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
778 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
779 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
782 2014-03-04 Heiher <r@hev.cc>
784 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
785 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
787 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
789 * config/i386/predicates.md (const2356_operand): Change to ...
790 (const2367_operand): ... this.
791 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
793 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
794 (*avx512pf_scatterpf<mode>sf): Ditto.
795 (avx512pf_scatterpf<mode>df): Ditto.
796 (*avx512pf_scatterpf<mode>df_mask): Ditto.
797 (*avx512pf_scatterpf<mode>df): Ditto.
798 * config/i386/i386.c (ix86_expand_builtin): Update
799 incorrect hint operand error message.
801 2014-03-04 Richard Biener <rguenther@suse.de>
803 * lto-section-in.c (lto_get_section_data): Fix const cast.
805 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
807 * tree-streamer.c (record_common_node): Assert we don't record
808 nodes with type double.
809 (preload_common_node): Skip type double, complex double and double
810 pointer since it is now frontend dependent due to fshort-double option.
812 2014-03-04 Richard Biener <rguenther@suse.de>
815 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
816 (lto_input_toplevel_asms): Likewise.
817 * lto-section-in.c (lto_get_section_data): Instead do it here
820 2014-03-04 Richard Biener <rguenther@suse.de>
822 PR tree-optimization/60382
823 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
824 dead PHIs a reduction.
826 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
828 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
830 (_mm_prefetch): Move out of GCC target("sse") pragma.
831 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
832 GCC target("prfchw") pragma.
833 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
835 * config/i386/i386.c (ix86_option_override_internal): Enable
836 -mprfchw with -mprefetchwt1.
838 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
840 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
843 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
845 * opts.h (CL_PCH_IGNORE): Define.
846 * targhooks.c (option_affects_pch_p):
847 Return false for options that have CL_PCH_IGNORE set.
848 * opt-functions.awk: Process PchIgnore.
849 * doc/options.texi: Document PchIgnore.
851 * config/arc/arc.opt (misize): Add PchIgnore property.
853 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
855 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
856 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
857 constraint on constants to permit them being loaded into
858 GENERAL_REGS or BASE_REGS.
860 2014-03-03 Nick Clifton <nickc@redhat.com>
862 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
863 anti-cacnonical alternatives.
864 (negandhi3_real): New pattern.
865 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
867 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
869 * config/avr/avr-mcus.def: Remove atxmega16x1.
870 * config/avr/avr-tables.opt: Regenerate.
871 * config/avr/t-multilib: Regenerate.
872 * doc/avr-mmcu.texi: Regenerate.
874 2014-03-03 Tobias Grosser <tobias@grosser.es>
875 Mircea Namolaru <mircea.namolaru@inria.fr>
877 PR tree-optimization/58028
878 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
881 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
883 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
884 not handled by recognizers.
886 2014-03-03 Jakub Jelinek <jakub@redhat.com>
889 * function.c (expand_function_end): Don't emit
890 clobber_return_register sequence if clobber_after is a BARRIER.
891 * cfgexpand.c (construct_exit_block): Append instructions before
892 return_label to prev_bb.
894 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
896 * config/rs6000/constraints.md: Document reserved use of "wc".
898 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
901 * ipa.c (function_and_variable_visibility): When dissolving comdat
902 group, also set all symbols to local.
904 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
909 2013-12-14 Jan Hubicka <jh@suse.cz>
911 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
913 2014-03-02 Jon Beniston <jon@beniston.com>
919 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
920 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
921 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
922 (simple_return, *simple_return): New patterns
923 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
924 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
926 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
928 * dwarf2out.c (gen_subprogram_die): Tidy.
930 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
933 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
934 (*mov_t_msb_neg_negc): ... this new insn.
936 2014-02-28 Jason Merrill <jason@redhat.com>
939 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
942 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
945 * dwarf2out.c (decltype_auto_die): New static.
946 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
947 (gen_type_die_with_usage): Handle 'decltype(auto)'.
948 (is_cxx_auto): Likewise.
950 2014-02-28 Ian Bolton <ian.bolton@arm.com>
952 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
953 we are not using general regs only.
955 2014-02-28 Richard Biener <rguenther@suse.de>
958 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
959 previous fix and only allow to remove trivial pre-headers
960 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
961 (remove_forwarder_block): Properly update the latch of a loop.
963 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
966 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
967 (cselib_preserved_hash_table): New.
968 (preserve_constants_and_equivs): Move preserved vals to it.
969 (cselib_find_slot): Look it up first.
970 (cselib_init): Initialize it.
971 (cselib_finish): Release it.
972 (dump_cselib_table): Dump it.
974 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
977 * cselib.c (remove_useless_values): Skip to avoid quadratic
978 behavior if the condition moved from...
979 (cselib_process_insn): ... here holds.
981 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
984 * var-tracking.c (vt_initialize): Apply the same condition to
985 preserve the CFA base value.
987 2014-02-28 Joey Ye <joey.ye@arm.com>
990 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
991 if reload in progress or completed.
993 2014-02-28 Tobias Burnus <burnus@net-b.de>
996 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
999 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1001 * doc/tm.texi.in (Condition Code Status): Update documention for
1002 relative locations of cc0-setter and cc0-user.
1004 2014-02-27 Jeff Law <law@redhat.com>
1006 PR rtl-optimization/52714
1007 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1008 into two independent simple sets, if I3 is a jump, ensure the
1009 pattern we place into I3 is a (set (pc) ...).
1011 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1012 Jeff Law <law@redhat.com>
1014 PR rtl-optimization/49847
1015 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1016 are in different blocks.
1017 * doc/tm.texi (Condition Code Status): Update documention for
1018 relative locations of cc0-setter and cc0-user.
1020 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1023 * lra.c (lra_emit_add): Check SUBREG too.
1025 2014-02-27 Andreas Schwab <schwab@suse.de>
1027 * config/m68k/m68k.c (m68k_option_override): Disable
1028 -flive-range-shrinkage for classic m68k.
1029 (m68k_override_options_after_change): Likewise.
1031 2014-02-27 Marek Polacek <polacek@redhat.com>
1034 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1035 -Wmaybe-uninitialized.
1037 2014-02-27 Alan Modra <amodra@gmail.com>
1040 * reload1.c (emit_input_reload_insns): When reload_override_in,
1041 set old to rl->in_reg when rl->in_reg is a subreg.
1043 2014-02-26 Richard Biener <rguenther@suse.de>
1046 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1048 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1050 * common/config/i386/predicates.md (const1256_operand): Remove.
1051 (const2356_operand): New.
1052 (const_1_to_2_operand): Remove.
1053 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1054 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1055 (*avx512pf_gatherpf<mode>sf): Ditto.
1056 (avx512pf_gatherpf<mode>df): Ditto.
1057 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1058 (*avx512pf_gatherpf<mode>df): Ditto.
1059 (avx512pf_scatterpf<mode>sf): Ditto.
1060 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1061 (*avx512pf_scatterpf<mode>sf): Ditto.
1062 (avx512pf_scatterpf<mode>df): Ditto.
1063 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1064 (*avx512pf_scatterpf<mode>df): Ditto.
1065 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1067 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1069 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1070 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1071 (_mm512_mask_testn_epi64_mask): Move to ...
1072 * config/i386/avx512cdintrin.h: Here.
1073 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1074 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1075 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1076 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1077 TARGET_AVX512F from TARGET_AVX512CD.
1079 2014-02-26 Richard Biener <rguenther@suse.de>
1082 * ipa.c (walk_polymorphic_call_targets): Properly guard
1083 call to inline_update_overall_summary.
1085 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1088 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1089 and latches only if requested. Fix latch if it is removed.
1090 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1091 LOOPS_HAVE_PREHEADERS.
1093 2014-02-25 Andrew Pinski <apinski@cavium.com>
1095 * builtins.c (expand_builtin_thread_pointer): Create a new target
1096 when the target is NULL.
1098 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1100 PR rtl-optimization/60317
1101 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1102 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1103 * lra-assigns.c: Include params.h.
1104 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1105 other reload pseudos considerations.
1107 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1109 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1110 to use canonical form for nor<mode>3.
1112 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1115 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1118 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1120 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1121 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1122 (ix86_handle_option): Handle OPT_mprefetchwt1.
1123 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1124 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1126 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1127 OPTION_MASK_ISA_PREFETCHWT1.
1128 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1129 (PTA_PREFETCHWT1): New.
1130 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1131 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1132 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1133 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1134 (*prefetch_avx512pf_<mode>_: Change into ...
1135 (*prefetch_prefetchwt1_<mode>: This.
1136 * config/i386/i386.opt (mprefetchwt1): New.
1137 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1138 (_mm_prefetch): Handle intent to write.
1139 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1141 2014-02-25 Richard Biener <rguenther@suse.de>
1144 * emit-rtl.c (mem_attrs_htab): Remove.
1145 (mem_attrs_htab_hash): Likewise.
1146 (mem_attrs_htab_eq): Likewise.
1147 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1148 (init_emit_once): Do not allocate mem_attrs_htab.
1150 2014-02-25 Richard Biener <rguenther@suse.de>
1153 * lto-opts.c (lto_write_options): Output non-explicit conservative
1154 -fwrapv, -fno-trapv and -fno-strict-overflow.
1155 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1156 (run_gcc): And pass them through.
1158 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1160 * sel-sched.c (calculate_new_fences): New parameter ptime.
1161 Calculate it as a maximum over all fence cycles.
1162 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1163 Print the final schedule timing when sched_verbose.
1165 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1167 PR rtl-optimization/60292
1168 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1169 bit fot the fence instruction.
1171 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1173 * calls.h: Fix typo in comment.
1175 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1177 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1178 adjusting offsetable addresses.
1180 2014-02-24 Guozhi Wei <carrot@google.com>
1182 * sparseset.h (sparseset_pop): Fix the wrong index.
1184 2014-02-24 Walter Lee <walt@tilera.com>
1186 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1187 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1189 * common/config/tilegx/tilegx-common.c
1190 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1191 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1193 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1194 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1195 (tilegx_gimplify_va_arg_expr): Handle big endian.
1196 (tilegx_expand_unaligned_load): Ditto.
1197 (tilegx_expand_unaligned_store): Ditto.
1198 (TARGET_RETURN_IN_MSB): New.
1199 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1200 (TARGET_ENDIAN_DEFAULT): New.
1201 (TARGET_BIG_ENDIAN): Handle big endian.
1202 (BYTES_BIG_ENDIAN): Ditto.
1203 (WORDS_BIG_ENDIAN): Ditto.
1204 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1207 * config/tilegx/tilegx.md (extv): Handle big endian.
1209 (insn_st<n>): Ditto.
1210 (insn_st<n>_add<bitsuffix>): Ditto.
1211 (insn_stnt<n>): Ditto.
1212 (insn_stnt<n>_add<bitsuffix>):Ditto.
1213 (vec_interleave_highv8qi): Handle big endian.
1214 (vec_interleave_highv8qi_be): New.
1215 (vec_interleave_highv8qi_le): New.
1216 (insn_v1int_h): Handle big endian.
1217 (vec_interleave_lowv8qi): Handle big endian.
1218 (vec_interleave_lowv8qi_be): New.
1219 (vec_interleave_lowv8qi_le): New.
1220 (insn_v1int_l): Handle big endian.
1221 (vec_interleave_highv4hi): Handle big endian.
1222 (vec_interleave_highv4hi_be): New.
1223 (vec_interleave_highv4hi_le): New.
1224 (insn_v2int_h): Handle big endian.
1225 (vec_interleave_lowv4hi): Handle big endian.
1226 (vec_interleave_lowv4hi_be): New.
1227 (vec_interleave_lowv4hi_le): New.
1228 (insn_v2int_l): Handle big endian.
1229 (vec_interleave_highv2si): Handle big endian.
1230 (vec_interleave_highv2si_be): New.
1231 (vec_interleave_highv2si_le): New.
1232 (insn_v4int_h): Handle big endian.
1233 (vec_interleave_lowv2si): Handle big endian.
1234 (vec_interleave_lowv2si_be): New.
1235 (vec_interleave_lowv2si_le): New.
1236 (insn_v4int_l): Handle big endian.
1237 * config/tilegx/tilegx.opt (mbig-endian): New option.
1238 (mlittle-endian): New option.
1239 * doc/install.texi: Document tilegxbe-linux.
1240 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1242 2014-02-24 Martin Jambor <mjambor@suse.cz>
1245 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1246 there are no parameter descriptors.
1248 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1250 PR rtl-optimization/60268
1251 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1252 initialization to ...
1253 (sched_rgn_init): ... here.
1254 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1256 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1258 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1261 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1263 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1266 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1268 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1269 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1271 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1273 * config/microblaze/predicates.md: Add cmp_op predicate.
1274 * config/microblaze/microblaze.md: Add branch_compare instruction
1275 which uses cmp_op predicate and emits cmp insn before branch.
1276 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1277 to microblaze_expand_conditional_branch and consolidate logic.
1278 (microblaze_expand_conditional_branch): emit branch_compare
1279 insn instead of handling cmp op separate from branch insn.
1281 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1283 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1286 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1288 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1289 define_insn with define_expand and new define_insn
1290 *altivec_lve<VI_char>x_internal.
1291 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1292 and new define_insn *altivec_stve<VI_char>x_internal.
1293 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1295 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1297 (altivec_expand_stvex_be): New function.
1299 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1301 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1302 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1303 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1304 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1306 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1309 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1310 instead of emit_move_insn.
1312 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1314 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1316 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1317 gen_altivec_vsumsws.
1319 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1321 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1322 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1323 (altivec_lvxl_<mode>): New define_expand incorporating
1324 -maltivec=be semantics where needed.
1325 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1326 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1327 semantics where needed.
1328 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1329 (altivec_stvx_<mode>): New define_expand incorporating
1330 -maltivec=be semantics where needed.
1331 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1332 VM2 iterator instead of V4SI.
1333 (altivec_stvxl_<mode>): New define_expand incorporating
1334 -maltivec=be semantics where needed.
1335 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1336 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1337 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1338 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1339 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1340 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1341 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1342 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1343 ALTIVEC_BUILTIN_STVXL.
1344 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1345 (altivec_expand_stvx_be): Likewise.
1346 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1347 (altivec_expand_lvx_be): Likewise.
1348 (altivec_expand_stvx_be): Likewise.
1349 (altivec_expand_builtin): Add cases for
1350 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1351 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1352 (altivec_init_builtins): Add definitions for
1353 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1354 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1356 2014-02-21 Catherine Moore <clm@codesourcery.com>
1358 * doc/invoke.texi (mvirt, mno-virt): Document.
1359 * config/mips/mips.opt (mvirt): New option.
1360 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1362 2014-02-21 Richard Biener <rguenther@suse.de>
1364 PR tree-optimization/60276
1365 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1366 (STMT_VINFO_MIN_NEG_DIST): New macro.
1367 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1368 STMT_VINFO_MIN_NEG_DIST.
1369 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1370 made for negative dependence distances still hold.
1372 2014-02-21 Richard Biener <rguenther@suse.de>
1375 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1376 DECL_INITIAL for globals not in the current function context.
1378 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1380 PR tree-optimization/56490
1381 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1382 * tree-ssa-uninit.c: Include params.h.
1383 (compute_control_dep_chain): Add num_calls argument, return false
1384 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1385 num_calls to recursive call.
1386 (find_predicates): Change dep_chain into normal array,
1387 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1388 variable and adjust compute_control_dep_chain caller.
1389 (find_def_preds): Likewise.
1391 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1393 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1394 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1396 2014-02-21 Nick Clifton <nickc@redhat.com>
1398 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1399 (pushhi1): Likewise.
1400 (popqi1): Add mode to pre_dec.
1403 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1405 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1406 mode for mask of V8SFmode permutation.
1408 2014-02-20 Richard Henderson <rth@redhat.com>
1411 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1412 a new pseudo for OLDVAL.
1414 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1417 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1418 gen_reg_rtx if d->testing_p.
1419 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1420 if d->testing_p and we will certainly return true.
1421 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1424 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1426 * emit-rtl.c (gen_reg_rtx): Assert that
1427 crtl->emit.regno_pointer_align_length is non-zero.
1429 2014-02-20 Richard Henderson <rth@redhat.com>
1432 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1433 on failure the store back into EXPECT.
1435 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1436 Sandra Loosemore <sandra@codesourcery.com>
1438 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1439 * config/nios2/nios2.c (nios2_function_profiler): Add
1440 -fPIC (flag_pic == 2) support.
1441 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1442 (nios2_large_offset_p): New function.
1443 (nios2_unspec_reloc_p): Move up position, update to use
1444 nios2_large_offset_p.
1445 (nios2_unspec_address): Remove function.
1446 (nios2_unspec_offset): New function.
1447 (nios2_large_got_address): New function.
1448 (nios2_got_address): Add large offset support.
1449 (nios2_legitimize_tls_address): Update usage of removed and new
1451 (nios2_symbol_binds_local_p): New function.
1452 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1453 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1454 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1455 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1456 processing for (const (unspec ...)).
1457 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1459 2014-02-20 Richard Biener <rguenther@suse.de>
1461 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1462 doing the substitution.
1463 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1465 2014-02-20 Martin Jambor <mjambor@suse.cz>
1468 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1469 info when checking whether lattices are bottom.
1471 2014-02-20 Richard Biener <rguenther@suse.de>
1474 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1477 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1480 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1481 parameter specifying the scaling.
1482 (inline_call): Update.
1483 (want_inline_recursively): Guard division by zero.
1484 (recursive_inlining): Update.
1485 * ipa-inline.h (clone_inlined_nodes): Update.
1487 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1490 * config/i386/i386.c (classify_argument): Pass structures of size
1491 64 bytes or less in register.
1493 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1494 Kirill Yukhin <kirill.yukhin@intel.com>
1496 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1497 (_mm_rcp28_round_ss): Ditto.
1498 (_mm_rsqrt28_round_sd): Ditto.
1499 (_mm_rsqrt28_round_ss): Ditto.
1500 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1501 (_mm_rcp14_round_ss): Ditto.
1502 (_mm_rsqrt14_round_sd): Ditto.
1503 (_mm_rsqrt14_round_ss): Ditto.
1504 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1505 the first input operand, get rid of match_dup.
1506 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1508 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1510 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1511 operand as the first input operand, set type attribute.
1512 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1514 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1515 operand as the first input operand, set type attribute.
1517 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1519 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1522 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1525 * config/i386/i386.c (construct_container): Remove TFmode check
1526 for X86_64_INTEGER_CLASS.
1528 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1531 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1532 only when -Wpsabi is enabled.
1534 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1537 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1538 passing arrays in registers are the same as for structs, so remove the
1539 special case for them.
1541 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1543 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1544 destination type, extract only the valid bits if the source type is not
1545 integral and has a different mode.
1547 2014-02-19 Richard Biener <rguenther@suse.de>
1550 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1553 2014-02-19 Richard Biener <rguenther@suse.de>
1556 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1557 (ipa_modify_call_arguments): Emit an argument load explicitely and
1558 preserve virtual SSA form there and for the replacement call.
1559 Do not update SSA form nor free dominance info.
1561 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1563 * ipa.c (function_and_variable_visibility): Also clear WEAK
1564 flag when disolving COMDAT_GROUP.
1566 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1568 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1569 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1571 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1572 do more sanity checks.
1573 (detect_type_change): Return true when giving up early.
1574 (compute_complex_assign_jump_func): Fix type parameter of
1575 ipa_set_ancestor_jf.
1576 (compute_complex_ancestor_jump_func): Likewise.
1577 (update_jump_functions_after_inlining): Fix updating of
1579 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1581 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1583 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1584 inline clones when edge disappears.
1586 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1589 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1590 Split 64-bit moves into 2 patterns. Do not allow the use of
1591 direct move for TDmode in little endian, since the decimal value
1592 has little endian bytes within a word, but the 64-bit pieces are
1593 ordered in a big endian fashion, and normal subreg's of TDmode are
1595 (mov<mode>_64bit_dm): Likewise.
1596 (movtd_64bit_nodm): Likewise.
1598 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1600 PR tree-optimization/60174
1601 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1602 statement of an SSA_NAME that occurs in an abnormal PHI node.
1604 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1607 * final.c (SEEN_BB): Remove.
1608 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1609 (final_scan_insn): Don't force_source_line on second
1610 NOTE_INSN_BASIC_BLOCK.
1612 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1615 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1616 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1617 (type_natural_mode): Warn ABI change when %zmm register is not
1618 available for AVX512F vector value passing.
1620 2014-02-18 Kai Tietz <ktietz@redhat.com>
1623 * config/i386/i386.c (ix86_expand_prologue): Use value in
1624 rax register as displacement when restoring %r10 or %rax.
1625 Fix wrong offset when restoring both registers.
1627 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1629 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1630 assertion with conditional return.
1632 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1633 Uros Bizjak <ubizjak@gmail.com>
1636 * config/i386/driver-i386.c (host_detect_local_cpu): If
1637 YMM state is not saved by the OS, also clear has_f16c. Move
1638 CPUID 0x80000001 handling before YMM state saving checking.
1640 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1642 PR rtl-optimization/58960
1643 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1644 factored out from ...
1645 (sched_init): ... here.
1646 (free_global_sched_pressure_data): New, factored out from ...
1647 (sched_finish): ... here.
1648 * sched-int.h (free_global_sched_pressure_data): Declare.
1649 * sched-rgn.c (nr_regions_initial): New static global.
1650 (haifa_find_rgns): Initialize it.
1651 (schedule_region): Disable sched-pressure for the newly
1654 2014-02-17 Richard Biener <rguenther@suse.de>
1656 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1657 release SSA defs of pattern stmts.
1659 2014-02-17 Richard Biener <rguenther@suse.de>
1661 * tree-inline.c (expand_call_inline): Release the virtual
1662 operand defined by the call we are about to inline.
1664 2014-02-17 Richard Biener <rguenther@suse.de>
1666 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1668 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1669 Ilya Tocar <ilya.tocar@intel.com>
1671 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1672 arguments order in builtin.
1673 (_mm512_permutexvar_epi64): Ditto.
1674 (_mm512_mask_permutexvar_epi64): Ditto
1675 (_mm512_maskz_permutexvar_epi32): Ditto
1676 (_mm512_permutexvar_epi32): Ditto
1677 (_mm512_mask_permutexvar_epi32): Ditto
1679 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1681 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1682 (p8_vmrgow): Likewise.
1684 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1686 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1689 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1692 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1693 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1694 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1695 using direct move instructions on ISA 2.07. Also adjust
1696 instruction length for 64-bit.
1697 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1698 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1700 2014-02-15 Alan Modra <amodra@gmail.com>
1704 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1705 find_replacement on parts of insn rtl that might be reloaded.
1707 2014-02-15 Richard Biener <rguenther@suse.de>
1709 PR tree-optimization/60183
1710 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1711 (tree_ssa_phiprop): Calculate and free post-dominators.
1713 2014-02-14 Jeff Law <law@redhat.com>
1715 PR rtl-optimization/60131
1716 * ree.c (get_extended_src_reg): New function.
1717 (combine_reaching_defs): Use it rather than assuming location of REG.
1718 (find_and_remove_re): Verify first operand of extension is
1719 a REG before adding the insns to the copy list.
1721 2014-02-14 Roland McGrath <mcgrathr@google.com>
1723 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1724 * configure: Regenerated.
1725 * config.in: Regenerated.
1726 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1727 instead of ASM_SHORT.
1729 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1730 Richard Earnshaw <rearnsha@arm.com>
1732 PR rtl-optimization/59535
1733 * lra-constraints.c (process_alt_operands): Encourage alternative
1734 when unassigned pseudo class is superset of the alternative class.
1735 (inherit_reload_reg): Don't inherit when optimizing for code size.
1736 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1737 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1738 modes not less than 4 for Thumb1.
1740 2014-02-14 Kyle McMartin <kyle@redhat.com>
1743 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1745 2014-02-14 Richard Biener <rguenther@suse.de>
1747 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1748 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1749 Do not strip INDIRECT_REFs.
1751 2014-02-14 Richard Biener <rguenther@suse.de>
1754 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1755 DECL_FUNCTION_SPECIFIC_TARGET.
1756 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1757 * tree-streamer-out.c (pack_ts_target_option): Remove.
1758 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1759 (write_ts_function_decl_tree_pointers): Do not stream
1760 DECL_FUNCTION_SPECIFIC_TARGET.
1761 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1762 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1763 (lto_input_ts_function_decl_tree_pointers): Do not stream
1764 DECL_FUNCTION_SPECIFIC_TARGET.
1766 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1768 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1769 (get_initial_def_for_induction, vectorizable_induction): Ignore
1770 debug stmts when looking for exit_phi.
1771 (vectorizable_live_operation): Fix up condition.
1773 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1775 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1776 nreverse() because it changes the content of original tree list.
1778 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1780 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1781 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1783 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1785 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1786 GNU coding standards.
1788 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1791 * dwarf2out.c (gen_subprogram_die): Don't call
1792 add_calling_convention_attribute if subr_die is old_die.
1794 2014-02-13 Sharad Singhai <singhai@google.com>
1796 * doc/optinfo.texi: Fix order of nodes.
1798 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1800 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1801 operands[2], not operands[3].
1803 2014-02-13 Richard Biener <rguenther@suse.de>
1806 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1807 mention the possibility of an in-tree build.
1808 (CLooG): Update recommended version to 0.18.1, mention the
1809 possibility of an in-tree build and clarify that the ISL
1810 bundled with CLooG does not work.
1812 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1815 * expr.c (compress_float_constant): If x is a hard register,
1816 extend into a pseudo and then move to x.
1818 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1820 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1821 caused by bad second argument to warning_at() with -mhotpatch and
1822 nested functions (e.g. with gfortran).
1824 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1826 * opts.c (option_name): Remove "enabled by default" rider.
1828 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1830 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1832 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1833 Uros Bizjak <ubizjak@gmail.com>
1836 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1837 * configure: Regenerated.
1839 2014-02-12 Richard Biener <rguenther@suse.de>
1841 * vec.c (vec_prefix::calculate_allocation): Move as
1842 inline variant to vec.h.
1843 (vec_prefix::calculate_allocation_1): New out-of-line version.
1844 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1845 (vec_prefix::m_has_auto_buf): Rename to ...
1846 (vec_prefix::m_using_auto_storage): ... this.
1847 (vec_prefix::calculate_allocation): Inline the easy cases
1848 and dispatch to calculate_allocation_1 which doesn't need the
1850 (va_heap::reserve): Use gcc_checking_assert.
1851 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1852 m_using_auto_storage.
1853 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1855 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1856 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1857 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1859 2014-02-12 Richard Biener <rguenther@suse.de>
1861 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1862 when we found a dependence.
1864 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1866 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1868 (maybe_fold_stmt): ... into this new function.
1869 * omp-low.c (lower_omp): Update comment.
1871 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1874 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1877 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1879 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1880 identifiers in comments.
1881 (cortexa53_extra_costs): Likewise.
1882 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1883 (cortexa7_extra_costs): Likewise.
1884 (cortexa12_extra_costs): Likewise.
1885 (cortexa15_extra_costs): Likewise.
1886 (v7m_extra_costs): Likewise.
1888 2014-02-12 Richard Biener <rguenther@suse.de>
1891 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1892 of posix_memalign being successful.
1893 (lower_stmt): Restrict lowering of posix_memalign to when
1894 -ftree-bit-ccp is enabled.
1896 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1898 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1900 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1902 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1904 PR rtl-optimization/60116
1905 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1906 other_insn once the combination has been validated.
1908 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1911 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1913 * ipa-devirt.c: Include demangle.h
1914 (odr_violation_reported): New static variable.
1915 (add_type_duplicate): Update odr_violations.
1916 (maybe_record_node): Add completep parameter; update it.
1917 (record_target_from_binfo): Add COMPLETEP parameter;
1918 update it as needed.
1919 (possible_polymorphic_call_targets_1): Likewise.
1920 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1921 rename FINAL to COMPLETE.
1922 (record_targets_from_bases): Sanity check we found the binfo;
1923 fix COMPLETEP updating.
1924 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1925 parameter, fix computing of COMPLETEP.
1926 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1927 at LTO time do demangling.
1928 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1929 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1931 (gimple_get_virt_method_for_binfo): Likewise.
1932 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1933 gimple_get_virt_method_for_vtable): Update prototypes.
1935 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1938 * genautomata.c (add_presence_absence): Fix typo with
1939 {final_}presence_list.
1941 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1944 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1945 for VSX/Altivec vectors that land in GPR registers.
1947 2014-02-11 Richard Henderson <rth@redhat.com>
1948 Jakub Jelinek <jakub@redhat.com>
1951 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1952 around drhs if type conversion to lacc->type is not useless.
1954 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1956 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1958 (cortex-a57.cortex-a53): Likewise.
1959 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1961 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1963 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1966 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1968 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1969 add_options_for_arm_vfp3.
1971 2014-02-11 Jeff Law <law@redhat.com>
1974 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1975 object with an undesirable mode.
1977 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1980 * config/i386/sol2-9.h: New file.
1981 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1982 *-*-solaris2.9*): Use it.
1984 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1986 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1987 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1989 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1991 * config/microblaze/microblaze.c: Extend mcpu version format
1993 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1995 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1997 2014-02-10 Richard Henderson <rth@redhat.com>
2000 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2001 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2002 ms-abi vs -mno-accumulate-outgoing-args.
2003 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2004 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2007 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2010 * cfgexpand.c (expand_asm_operands): Attach source location to
2011 ASM_INPUT rtx objects.
2012 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2014 2014-02-10 Nick Clifton <nickc@redhat.com>
2016 * config/mn10300/mn10300.c (popcount): New function.
2017 (mn10300_expand_prologue): Include saved registers in stack usage
2020 2014-02-10 Jeff Law <law@redhat.com>
2023 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2024 when changing the SET_DEST of a prior insn to avoid an input reload.
2026 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2028 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2029 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2030 -mcall-openbsd, or -mcall-linux.
2031 (CC1_ENDIAN_BIG_SPEC): Remove.
2032 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2033 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2034 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2035 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2036 and %cc1_endian_default.
2037 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2039 2014-02-10 Richard Biener <rguenther@suse.de>
2041 PR tree-optimization/60115
2042 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2043 MEM_REF handling. Properly verify that the accesses are not
2044 out of the objects bound.
2046 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2048 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2051 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2053 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2054 proper constants and fix formatting.
2055 (possible_polymorphic_call_targets): Fix formatting.
2057 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2058 Ilya Tocar <ilya.tocar@intel.com>
2060 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2061 (_mm512_loadu_epi32): Renamed into...
2062 (_mm512_loadu_si512): This.
2063 (_mm512_storeu_epi32): Renamed into...
2064 (_mm512_storeu_si512): This.
2065 (_mm512_maskz_ceil_ps): Removed.
2066 (_mm512_maskz_ceil_pd): Ditto.
2067 (_mm512_maskz_floor_ps): Ditto.
2068 (_mm512_maskz_floor_pd): Ditto.
2069 (_mm512_floor_round_ps): Ditto.
2070 (_mm512_floor_round_pd): Ditto.
2071 (_mm512_ceil_round_ps): Ditto.
2072 (_mm512_ceil_round_pd): Ditto.
2073 (_mm512_mask_floor_round_ps): Ditto.
2074 (_mm512_mask_floor_round_pd): Ditto.
2075 (_mm512_mask_ceil_round_ps): Ditto.
2076 (_mm512_mask_ceil_round_pd): Ditto.
2077 (_mm512_maskz_floor_round_ps): Ditto.
2078 (_mm512_maskz_floor_round_pd): Ditto.
2079 (_mm512_maskz_ceil_round_ps): Ditto.
2080 (_mm512_maskz_ceil_round_pd): Ditto.
2081 (_mm512_expand_pd): Ditto.
2082 (_mm512_expand_ps): Ditto.
2083 * config/i386/i386.c (ix86_builtins): Remove
2084 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2085 (bdesc_args): Ditto.
2086 * config/i386/predicates.md (const1256_operand): New.
2087 (const_1_to_2_operand): Ditto.
2088 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2089 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2090 (*avx512pf_gatherpf<mode>sf): Ditto.
2091 (avx512pf_gatherpf<mode>df): Ditto.
2092 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2093 (*avx512pf_gatherpf<mode>df): Ditto.
2094 (avx512pf_scatterpf<mode>sf): Ditto.
2095 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2096 (*avx512pf_scatterpf<mode>sf): Ditto.
2097 (avx512pf_scatterpf<mode>df): Ditto.
2098 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2099 (*avx512pf_scatterpf<mode>df): Ditto.
2100 (avx512f_expand<mode>): Removed.
2101 (<shift_insn><mode>3<mask_name>): Change predicate type.
2103 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2105 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2106 not at the end of datarefs vector use ordered_remove to avoid
2107 reordering datarefs vector.
2110 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2111 mark local addressable non-static vars as GOVD_PRIVATE
2112 instead of GOVD_LOCAL.
2113 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2114 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2118 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2119 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2120 assume_aligned or alloc_align attributes.
2121 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2122 arguments. Handle also assume_aligned and alloc_align attributes.
2123 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2124 calls to functions with assume_aligned or alloc_align attributes.
2125 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2127 2014-02-08 Terry Guo <terry.guo@arm.com>
2129 * doc/invoke.texi: Document ARM -march=armv7e-m.
2131 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2133 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2134 flag on __cilkrts_rethrow builtin.
2137 * ipa-cp.c (determine_versionability): Fail at -O0
2138 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2139 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2142 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2145 * tree-inline.c (copy_forbidden): Fail for
2146 __attribute__((optimize (0))) functions.
2148 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2150 * varpool.c: Include pointer-set.h.
2151 (varpool_remove_unreferenced_decls): Variables in other partitions
2152 will not be output; be however careful to not lose information
2155 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2157 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2158 lookup in the vtable constructor.
2160 2014-02-07 Jeff Law <law@redhat.com>
2163 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2164 define_insn_and_split.
2166 * ipa-inline.c (inline_small_functions): Fix typos.
2168 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2170 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2171 (s390_can_use_return_insn): Declare.
2172 * config/s390/s390.h (EPILOGUE_USES): Define.
2173 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2175 (s390_chunkify_start): Handle return JUMP_LABELs.
2176 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2177 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2178 (s390_can_use_return_insn): New functions.
2179 (s390_fix_long_loop_prediction): Handle conditional returns.
2180 (TARGET_SET_UP_BY_PROLOGUE): Define.
2181 * config/s390/s390.md (ANY_RETURN): New code iterator.
2182 (*creturn, *csimple_return, return, simple_return): New patterns.
2184 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2186 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2187 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2188 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2189 REG_CFA_RESTORE list when deciding not to restore a register.
2191 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2193 * config/s390/s390.c: Include tree-pass.h and context.h.
2194 (s390_early_mach): New function, split out from...
2195 (s390_emit_prologue): ...here.
2196 (pass_data_s390_early_mach): New pass structure.
2197 (pass_s390_early_mach): New class.
2198 (s390_option_override): Create and register early_mach pass.
2199 Move to end of file.
2201 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2203 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2204 to match for the exit block.
2206 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2208 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2209 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2210 Reject misaligned operands.
2212 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2214 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2216 2014-02-07 Richard Biener <rguenther@suse.de>
2219 * gimple-low.c (lower_builtin_posix_memalign): New function.
2220 (lower_stmt): Call it to lower posix_memalign in a way
2221 to make alignment info accessible.
2223 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2226 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2227 __builtin_setjmp_receiver.
2229 2014-02-07 Richard Biener <rguenther@suse.de>
2232 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2233 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2234 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2235 Handle BUILT_IN_POSIX_MEMALIGN.
2236 (find_func_clobbers): Likewise.
2237 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2238 (call_may_clobber_ref_p_1): Likewise.
2240 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2243 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2246 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2249 * lto-cgraph.c (lto_output_node): Use
2250 symtab_get_symbol_partitioning_class.
2251 (lto_output_varpool_node): likewise.
2252 (symtab_get_symbol_partitioning_class): Move here from
2254 * cgraph.h (symbol_partitioning_class): Likewise.
2255 (symtab_get_symbol_partitioning_class): Declare.
2257 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2259 * ggc.h (ggc_internal_cleared_alloc): New macro.
2260 * vec.h (vec_safe_copy): Handle memory stats.
2261 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2262 * target-globals.c (save_target_globals): Likewise.
2264 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2267 * expr.c (emit_move_resolve_push): Export; be bit more selective
2268 on when to clear alias set.
2269 * expr.h (emit_move_resolve_push): Declare.
2270 * function.h (struct function): Add tail_call_marked.
2271 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2272 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2273 * config/i386/i386.md (TImode move expander): De not call
2275 (FP push expanders): Preserve memory attributes.
2276 * config/i386/sse.md (push<mode>1): Remove.
2277 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2278 (ix86_expand_push): Remove.
2279 * config/i386/mmx.md (push<mode>1): Remove.
2281 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2283 PR rtl-optimization/60030
2284 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2285 lopart with paradoxical subreg before shifting it up by hprec.
2287 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2289 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2290 Remove extra newline at end of file.
2291 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2292 (arm_issue_rate): Handle cortexa57.
2293 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2294 (cortex-a57.cortex-a53): Likewise.
2296 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2299 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2300 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2302 (arm_expand_prologue): Adjust all callers.
2303 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2304 info, registers also at the lowest numbered registers side. Use
2305 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2309 * var-tracking.c (adjust_mems): Before adding a SET to
2310 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2312 2014-02-06 Alan Modra <amodra@gmail.com>
2315 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2316 change SDmode to DDmode when lra_in_progress.
2318 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2321 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2322 free_data_ref on the dr first, and before goto again also set dr
2323 to the next dr. For simd_lane_access, free old datarefs[i] before
2324 overwriting it. For get_vectype_for_scalar_type failure, don't
2325 free_data_ref if simd_lane_access.
2327 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2330 * tree.h (opts_for_fn): New inline function.
2331 (opt_for_fn): Define.
2332 * config/i386/i386.c (ix86_function_regparm): Use
2333 opt_for_fn (decl, optimize) instead of optimize.
2335 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2337 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2338 for SYMBOL_REF in large memory model.
2340 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2342 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2344 (cortex-a57): Likewise.
2345 (cortex-a57.cortex-a53): Likewise.
2347 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2348 Kugan Vivekanandarajah <kuganv@linaro.org>
2350 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2352 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2354 2014-02-06 Richard Biener <rguenther@suse.de>
2356 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2357 set_loop_copy and initialize_original_copy_tables.
2359 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2361 * config/aarch64/aarch64-simd.md
2362 (aarch64_ashr_simddi): Change QI to SI.
2364 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2365 Jakub Jelinek <jakub@redhat.com>
2368 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2371 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2373 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2374 CODE_FOR_altivec_vpku[hw]um to
2375 CODE_FOR_altivec_vpku[hw]um_direct.
2376 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2377 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2378 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2379 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2381 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2383 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2384 generation for -maltivec=be.
2385 (altivec_vsumsws): Simplify redundant test.
2387 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2389 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2390 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2391 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2392 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2393 gen_altivec_vpkuwum.
2394 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2396 (altivec_vpks<VI_char>ss): Likewise.
2397 (altivec_vpks<VI_char>us): Likewise.
2398 (altivec_vpku<VI_char>us): Likewise.
2399 (altivec_vpku<VI_char>um): Likewise.
2400 (altivec_vpku<VI_char>um_direct): New (copy of
2401 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2403 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2404 target is little endian and -maltivec=be is not specified.
2405 (*altivec_vupkhs<VU_char>_direct): New (copy of
2406 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2407 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2408 target is little endian and -maltivec=be is not specified.
2409 (*altivec_vupkls<VU_char>_direct): New (copy of
2410 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2411 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2412 little endian and -maltivec=be is not specified.
2413 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2414 little endian and -maltivec=be is not specified.
2416 2014-02-05 Richard Henderson <rth@redhat.com>
2419 * combine-stack-adj.c: Revert r206943.
2420 * sched-int.h (struct deps_desc): Add last_args_size.
2421 * sched-deps.c (init_deps): Initialize it.
2422 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2423 contain REG_ARGS_SIZE notes.
2425 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2427 * lto-cgraph.c (asm_nodes_output): Make global.
2428 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2429 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2430 (driver_handle_option): Handle OPT_fwpa.
2432 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2435 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2436 a comment typo and formatting issue. If odr_hash hasn't been
2437 created, return vNULL and set *completep to false.
2440 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2441 bb with no successors.
2443 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2446 * doc/invoke.texi (-march): Clarify documentation for ARM.
2450 2014-02-05 Richard Biener <rguenther@suse.de>
2452 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2453 when not vectorizing because of too many alias checks.
2454 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2455 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2457 2014-02-05 Nick Clifton <nickc@redhat.com>
2459 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2460 accept extended registers in any mode when compiling for the MN10300.
2462 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2464 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2465 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2466 sanitization attributes.
2467 (can_inline_edge_p): Likewise.
2468 (sanitize_attrs_match_for_inline_p): New function.
2470 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2472 * ipa-prop.c (detect_type_change): Shor circuit testing of
2473 type changes on THIS pointer.
2475 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2478 * config/pa/pa.c (legitimize_tls_address): Return original address
2479 if not passed a SYMBOL_REF rtx.
2480 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2482 (pa_emit_move_sequence): Simplify TLS source operands.
2483 (pa_legitimate_constant_p): Reject all TLS constants.
2484 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2485 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2487 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2489 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2490 groups when we know they are controlled by LTO.
2491 * varasm.c (default_binds_local_p_1): If object is in other partition,
2492 it will be resolved locally.
2494 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2496 * config/host-linux.c (linux_gt_pch_use_address): Don't
2497 use SSIZE_MAX because it is not always defined.
2499 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2502 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2503 threshold for pseudo splitting.
2504 (update_ebb_live_info): Process call argument hard registers and
2505 hard registers from insn definition too.
2506 (max_small_class_regs_num): New constant.
2507 (inherit_in_ebb): Update live hard regs through EBBs. Update
2508 reloads_num only for small register classes. Don't split for
2511 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2514 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2517 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2519 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2522 2014-02-04 Marek Polacek <polacek@redhat.com>
2524 * gdbinit.in (pel): Define.
2526 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2528 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2531 2014-02-04 Richard Biener <rguenther@suse.de>
2534 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2535 in function context local.
2536 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2537 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2538 similar to LTO_imported_decl_ref.
2540 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2542 PR tree-optimization/60002
2543 * cgraphclones.c (build_function_decl_skip_args): Clear
2546 PR tree-optimization/60023
2547 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2548 false to gsi_replace.
2549 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2550 has been in some EH region and vec_stmt could throw, add
2551 vec_stmt into the same EH region.
2552 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2553 has no lhs, ignore it.
2554 * internal-fn.c (expand_MASK_LOAD): Likewise.
2557 * tree-inline.c (copy_forbidden): Fail for
2558 __attribute__((optimize (0))) functions.
2561 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2562 is set, copy one less argument.
2563 (expand_simd_clones): Don't subtract clone_info->inbranch
2564 from simd_clone_struct_alloc argument.
2566 PR rtl-optimization/57915
2567 * recog.c (simplify_while_replacing): If all unary/binary/relational
2568 operation arguments are constant, attempt to simplify those.
2571 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2572 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2574 2014-02-04 Richard Biener <rguenther@suse.de>
2576 PR tree-optimization/60012
2577 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2578 TBAA disambiguation to all DDRs.
2580 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2583 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2584 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2586 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2589 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2591 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2593 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2594 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2596 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2599 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2600 to figure out targets of polymorphic calls with known decl.
2601 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2602 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2603 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2604 (get_polymorphic_call_info): ... here.
2605 (get_polymorphic_call_info_from_invariant): New function.
2607 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2609 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2610 lookup via vtable pointer; check for type consistency
2611 and turn inconsitent facts into UNREACHABLE.
2612 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2613 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2614 type inconsistent querries; return UNREACHABLE instead.
2616 2014-02-03 Richard Henderson <rth@twiddle.net>
2619 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2620 already processed this node.
2621 (normalize_one_pred_1): Pass along mark_set.
2622 (normalize_one_pred): Create and destroy a pointer_set_t.
2623 (normalize_one_pred_chain): Likewise.
2625 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2627 PR gcov-profile/58602
2628 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2630 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2633 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2634 -fno-devirtualize; try to devirtualize by the knowledge of
2635 virtual table pointer given by aggregate propagation.
2636 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2637 (ipa_print_node_jump_functions): Dump also offset that
2638 is relevant for polymorphic calls.
2639 (determine_known_aggregate_parts): Add arg_type parameter; use it
2640 instead of determining the type from pointer type.
2641 (ipa_compute_jump_functions_for_edge): Update call of
2642 determine_known_aggregate_parts.
2643 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2644 (gimple_get_virt_method_for_binfo): ... here; simplify using
2645 vtable_pointer_value_to_vtable.
2646 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2647 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2648 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2649 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2651 (vtable_pointer_value_to_binfo): ... here.
2652 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2654 2014-02-03 Teresa Johnson <tejohnson@google.com>
2656 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2657 redef of outer loop index variable.
2659 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2663 * doc/extend.texi (Function Attributes): Typo.
2665 2014-02-03 Cong Hou <congh@google.com>
2667 PR tree-optimization/60000
2668 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2669 if the vectorized statement is a store. A store statement can only
2670 appear at the end of pattern statements.
2672 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2674 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2675 (ix86_option_override_internal): Default long double to 64-bit for
2676 32-bit Bionic and to 128-bit for 64-bit Bionic.
2678 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2679 TARGET_LONG_DOUBLE_128 is true.
2680 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2682 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2683 (mlong-double-64): Negate -mlong-double-128.
2684 (mlong-double-128): New option.
2686 * config/i386/i386-c.c (ix86_target_macros): Define
2687 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2689 * doc/invoke.texi: Document -mlong-double-128.
2691 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2693 PR rtl-optimization/60024
2694 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2696 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2698 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2700 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2702 PR rtl-optimization/57662
2703 * sel-sched.c (code_motion_path_driver): Do not mark already not
2704 existing blocks in the visiting bitmap.
2706 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2708 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2709 on the insn being emitted.
2711 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2712 Will Deacon <will.deacon@arm.com>
2714 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2716 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2718 * config/arm/arm-tables.opt: Regenerate.
2720 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2722 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2723 for vector types other than V16QImode.
2724 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2725 define_expand, and call altivec_expand_vec_perm_le when producing
2726 code with little endian element order.
2727 (*altivec_vperm_<mode>_internal): New insn having previous
2728 behavior of altivec_vperm_<mode>.
2729 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2730 altivec_expand_vec_perm_le when producing code with little endian
2732 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2733 behavior of altivec_vperm_<mode>_uns.
2735 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2737 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2738 (altivec_vsumsws): Add handling for -maltivec=be with a little
2740 (altivec_vsumsws_direct): New.
2741 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2742 gen_altivec_vsumsws.
2744 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2746 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2747 vtable_pointer_value_to_binfo): New functions.
2748 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2749 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2751 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2753 * config/nios2/nios2.md (load_got_register): Initialize GOT
2754 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2755 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2757 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2759 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2760 preserverd by passthrough, do not propagate the type.
2762 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2764 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2765 (mips_atomic_assign_expand_fenv): New function.
2766 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2768 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2770 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2771 (__builtin_mips_set_fcsr): Likewise.
2772 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2773 MIPS_USI_FTYPE_VOID.
2774 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2775 (mips16_expand_set_fcsr): Likewise.
2776 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2777 (mips16_set_fcsr_stub): Likewise.
2778 (mips16_get_fcsr_one_only_stub): New class.
2779 (mips16_set_fcsr_one_only_stub): Likewise.
2780 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2781 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2782 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2783 (hard_float): New availability predicate.
2784 (mips_builtins): Add get_fcsr and set_fcsr.
2785 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2786 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2787 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2788 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2789 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2792 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2794 * config/mips/mips.c (mips_one_only_stub): New class.
2795 (mips_need_mips16_rdhwr_p): Replace with...
2796 (mips16_rdhwr_stub): ...this new variable.
2797 (mips16_stub_call_address): New function.
2798 (mips16_rdhwr_one_only_stub): New class.
2799 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2800 (mips_output_mips16_rdhwr): Delete.
2801 (mips_finish_stub): New function.
2802 (mips_code_end): Use it to handle rdhwr stubs.
2804 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2807 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2808 when calculating size of integer atomic types.
2810 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2812 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2814 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2816 PR tree-optimization/60003
2817 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2818 * profile.c (branch_prob): Use gimple_call_builtin_p
2819 to check for BUILT_IN_SETJMP_RECEIVER.
2820 * tree-inline.c (copy_bb): Call notice_special_calls.
2822 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2825 * lra-constraints.c (process_alt_operands): Update reload_sum only
2828 2014-01-31 Richard Henderson <rth@redhat.com>
2831 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2832 until after else_eh is processed.
2834 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2836 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2837 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2838 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2839 in smmintrin.h, remove them.
2840 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2841 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2842 * config/i386/i386.md (ROUND_SAE): Fix value.
2843 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2844 (const48_operand): New.
2845 * config/i386/subst.md (round), (round_expand): Use
2846 const_4_or_8_to_11_operand.
2847 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2849 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2851 * config/i386/constraints.md (Yk): Swap meaning with k.
2852 * config/i386/i386.md (movhi_internal): Change Yk to k.
2853 (movqi_internal): Ditto.
2854 (*k<logic><mode>): Ditto.
2857 (kandn<mode>): Ditto.
2858 (*<code>hi_1): Ditto.
2859 (*<code>qi_1): Ditto.
2860 (kxnor<mode>): Ditto.
2861 (kortestzhi): Ditto.
2862 (kortestchi): Ditto.
2864 (*one_cmplhi2_1): Ditto.
2865 (*one_cmplqi2_1): Ditto.
2866 * config/i386/sse.md (): Change k to Yk.
2867 (avx512f_load<mode>_mask): Ditto.
2868 (avx512f_blendm<mode>): Ditto.
2869 (avx512f_store<mode>_mask): Ditto.
2870 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2871 (avx512f_storedqu<mode>_mask): Ditto.
2872 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2874 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2875 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2876 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2877 (avx512f_maskcmp<mode>3): Ditto.
2878 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2879 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2880 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2881 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2882 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2883 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2884 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2885 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2886 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2887 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2888 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2889 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2890 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2891 (vec_extract_lo_<mode>_maskm): Ditto.
2892 (vec_extract_hi_<mode>_maskm): Ditto.
2893 (avx512f_vternlog<mode>_mask): Ditto.
2894 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2895 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2896 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2897 (avx512f_<code>v8div16qi2_mask): Ditto.
2898 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2899 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2900 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2901 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2902 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2903 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2904 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2905 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2906 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2907 (avx512cd_maskb_vec_dupv8di): Ditto.
2908 (avx512cd_maskw_vec_dupv16si): Ditto.
2909 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2910 (avx512f_vpermi2var<mode>3_mask): Ditto.
2911 (avx512f_vpermi2var<mode>3_mask): Ditto.
2912 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2913 (*avx512f_gathersi<mode>): Ditto.
2914 (*avx512f_gathersi<mode>_2): Ditto.
2915 (*avx512f_gatherdi<mode>): Ditto.
2916 (*avx512f_gatherdi<mode>_2): Ditto.
2917 (*avx512f_scattersi<mode>): Ditto.
2918 (*avx512f_scatterdi<mode>): Ditto.
2919 (avx512f_compress<mode>_mask): Ditto.
2920 (avx512f_compressstore<mode>_mask): Ditto.
2921 (avx512f_expand<mode>_mask): Ditto.
2922 * config/i386/subst.md (mask): Change k to Yk.
2923 (mask_scalar_merge): Ditto.
2926 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2928 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2930 2014-01-31 Richard Biener <rguenther@suse.de>
2933 * builtins.c (fold_builtin_memory_op): Make sure to not
2934 use a floating-point mode or a boolean or enumeral type for
2937 2014-01-30 DJ Delorie <dj@redhat.com>
2939 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2940 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2941 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2942 whenever main() has an epilogue.
2944 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2946 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2947 unused variable "field".
2948 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2949 (vsx_mergeh_<mode>): Likewise.
2950 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2951 (altivec_vmrghh): Likewise.
2952 (altivec_vmrghw): Likewise.
2953 (altivec_vmrglb): Likewise.
2954 (altivec_vmrglh): Likewise.
2955 (altivec_vmrglw): Likewise.
2956 (altivec_vspltb): Add missing uses.
2957 (altivec_vsplth): Likewise.
2958 (altivec_vspltw): Likewise.
2959 (altivec_vspltsf): Likewise.
2961 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2964 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2965 frame related instructions.
2967 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2969 PR rtl-optimization/59959
2970 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2971 any reload of register whose subreg is invalid.
2973 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2975 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2976 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2977 Add missing return type - void.
2979 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2981 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2982 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2983 remove element index adjustment for endian (now handled in vsx.md
2985 (altivec_expand_vec_perm_const): Use
2986 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2987 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2988 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2989 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2990 define_expand and a new define_insn *altivec_vspltb_internal;
2991 adjust for -maltivec=be on a little endian target.
2992 (altivec_vspltb_direct): New.
2993 (altivec_vsplth): Divide into a define_expand and a new
2994 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2995 little endian target.
2996 (altivec_vsplth_direct): New.
2997 (altivec_vspltw): Divide into a define_expand and a new
2998 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2999 little endian target.
3000 (altivec_vspltw_direct): New.
3001 (altivec_vspltsf): Divide into a define_expand and a new
3002 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3003 a little endian target.
3005 2014-01-30 Richard Biener <rguenther@suse.de>
3007 PR tree-optimization/59993
3008 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3009 can propagate form the earlier stmt and avoid the transform
3010 when the intermediate result is needed.
3012 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3014 * README.Portability: Fix typo.
3016 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3018 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3019 comparison_operator with ordered_comparison_operator.
3021 2014-01-30 Nick Clifton <nickc@redhat.com>
3023 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3024 Rename to mn10300_store_multiple_regs.
3025 * config/mn10300/mn10300.c: Likewise.
3026 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3027 store_multiple_regs.
3028 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3029 Call mn10300_store_multiple_regs.
3031 2014-01-30 Nick Clifton <nickc@redhat.com>
3032 DJ Delorie <dj@redhat.com>
3034 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3035 %fp 2 to keep registers after it properly word-aligned.
3036 (rl78_alloc_physical_registers_umul): Handle the case where both
3037 input operands are the same.
3039 2014-01-30 Richard Biener <rguenther@suse.de>
3041 PR tree-optimization/59903
3042 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3045 2014-01-30 Jason Merrill <jason@redhat.com>
3048 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3051 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3053 2014-01-30 Richard Biener <rguenther@suse.de>
3055 PR tree-optimization/59951
3056 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3058 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3061 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3062 SFmode to DFmode case.
3064 2014-01-29 DJ Delorie <dj@redhat.com>
3066 * config/msp430/msp430.opt (-minrt): New.
3067 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3069 (ENDFILE_SPEC): Likewise.
3071 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3073 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3074 (estimate_function_body_sizes): Use it.
3076 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3079 * dwarf2out.c (is_cxx_auto): New.
3080 (is_base_type): Use it.
3081 (gen_type_die_with_usage): Likewise.
3083 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3085 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3086 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3087 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3088 -maltivec=be with LE targets.
3089 (vsx_mergeh_<mode>): Likewise.
3090 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3091 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3092 (altivec_vmrghb): Replace with define_expand and new
3093 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3094 (altivec_vmrghb_direct): New define_insn.
3095 (altivec_vmrghh): Replace with define_expand and new
3096 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3097 (altivec_vmrghh_direct): New define_insn.
3098 (altivec_vmrghw): Replace with define_expand and new
3099 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3100 (altivec_vmrghw_direct): New define_insn.
3101 (*altivec_vmrghsf): Adjust for endianness.
3102 (altivec_vmrglb): Replace with define_expand and new
3103 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3104 (altivec_vmrglb_direct): New define_insn.
3105 (altivec_vmrglh): Replace with define_expand and new
3106 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3107 (altivec_vmrglh_direct): New define_insn.
3108 (altivec_vmrglw): Replace with define_expand and new
3109 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3110 (altivec_vmrglw_direct): New define_insn.
3111 (*altivec_vmrglsf): Adjust for endianness.
3112 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3113 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3114 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3115 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3116 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3117 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3118 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3119 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3121 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3123 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3124 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3127 2014-01-29 Richard Biener <rguenther@suse.de>
3129 PR tree-optimization/58742
3130 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3131 associate_pointerplus_align.
3132 (associate_pointerplus_diff): New function.
3133 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3134 and associate_pointerplus_diff.
3136 2014-01-29 Richard Biener <rguenther@suse.de>
3138 * lto-streamer.h (LTO_major_version): Bump to 3.
3139 (LTO_minor_version): Reset to 0.
3141 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3143 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3144 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3145 (arm_file_start): Generate correct asm header for armv7ve.
3146 * config/arm/bpabi.h: Add multilib support for armv7ve.
3147 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3148 and cortex-a15 to armv7ve.
3149 * config/arm/t-aprofile: Add multilib support for armv7ve.
3150 * doc/invoke.texi: Document -march=armv7ve.
3152 2014-01-29 Richard Biener <rguenther@suse.de>
3154 PR tree-optimization/58742
3155 * tree-ssa-forwprop.c (associate_plusminus): Return true
3156 if we changed sth, defer EH cleanup to ...
3157 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3158 (simplify_mult): New function.
3160 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3163 PR tree-optimization/59920
3164 * tree.c (build_common_builtin_nodes): Remove
3165 __builtin_setjmp_dispatcher initialization.
3166 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3167 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3168 instead of gsi_after_labels + manually skipping debug stmts.
3169 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3170 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3171 * tree-inline.c (copy_edges_for_bb): Remove
3172 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3173 argument. Ignore computed_goto_p stmts. Don't call
3174 make_abnormal_goto_edges. If a call might need abnormal edges
3175 for non-local gotos, see if it already has an edge to
3176 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3177 with true argument, don't do anything then, otherwise add
3178 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3179 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3181 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3182 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3183 (lower_stmt): Don't set data->calls_builtin_setjmp.
3184 (lower_builtin_setjmp): Adjust comment.
3185 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3186 * tree-cfg.c (found_computed_goto): Remove.
3187 (factor_computed_gotos): Remove.
3188 (make_goto_expr_edges): Return bool, true for computed gotos.
3189 Don't call make_abnormal_goto_edges.
3190 (build_gimple_cfg): Don't set found_computed_goto, don't call
3191 factor_computed_gotos.
3192 (computed_goto_p): No longer static.
3193 (make_blocks): Don't set found_computed_goto.
3194 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3195 (make_edges): If make_goto_expr_edges returns true, push bb
3196 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3197 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3198 vector. Record mapping between bbs and OpenMP regions if there
3199 are any, adjust make_gimple_omp_edges caller. Call
3200 handle_abnormal_edges.
3201 (make_abnormal_goto_edges): Remove.
3202 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3203 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3204 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3205 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3206 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3207 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3208 filling *region also set *region_idx to (*region)->entry->index.
3211 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3212 For REGs set ORIGINAL_REGNO.
3214 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3216 * doc/md.texi: Mention that a target shouldn't implement
3217 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3220 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3222 PR tree-optimization/59594
3223 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3224 a copy of the datarefs vector rather than the vector itself.
3226 2014-01-28 Jason Merrill <jason@redhat.com>
3229 * dwarf2out.c (auto_die): New static.
3230 (gen_type_die_with_usage): Handle C++1y 'auto'.
3231 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3234 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3237 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3238 (SPEC_X32): Likewise.
3239 (SPEC_64): Likewise.
3240 * config/i386/i386.c (ix86_option_override_internal): Turn off
3241 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3243 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3244 * config/i386/i386.h (TARGET_16BIT): New macro.
3245 (TARGET_16BIT_P): Likewise.
3246 * config/i386/i386.opt: Add m16.
3247 * doc/invoke.texi: Document -m16.
3249 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3251 PR preprocessor/59935
3252 * input.c (location_get_source_line): Bail out on when line number
3253 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3255 2014-01-28 Richard Biener <rguenther@suse.de>
3257 PR tree-optimization/58742
3258 * tree-ssa-forwprop.c (associate_plusminus): Handle
3259 pointer subtraction of the form (T)(P + A) - (T)P.
3261 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3263 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3266 2014-01-28 Richard Biener <rguenther@suse.de>
3269 2014-01-28 Richard Biener <rguenther@suse.de>
3271 PR rtl-optimization/45364
3272 PR rtl-optimization/59890
3273 * var-tracking.c (local_get_addr_clear_given_value): Handle
3274 already cleared slot.
3275 (val_reset): Handle not allocated local_get_addr_cache.
3276 (vt_find_locations): Use post-order on the inverted CFG.
3278 2014-01-28 Richard Biener <rguenther@suse.de>
3280 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3282 2014-01-28 Richard Biener <rguenther@suse.de>
3284 PR rtl-optimization/45364
3285 PR rtl-optimization/59890
3286 * var-tracking.c (local_get_addr_clear_given_value): Handle
3287 already cleared slot.
3288 (val_reset): Handle not allocated local_get_addr_cache.
3289 (vt_find_locations): Use post-order on the inverted CFG.
3291 2014-01-28 Alan Modra <amodra@gmail.com>
3293 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3294 * configure.ac <recursive call for build != host>: Define
3295 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3296 and LD_FOR_BUILD too.
3297 * configure: Regenerate.
3299 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3301 * config/i386/i386.c (get_builtin_code_for_version): Separate
3302 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3303 Broadwell from Haswell.
3305 2014-01-27 Steve Ellcey <sellcey@mips.com>
3307 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3308 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3309 * config/mips/mips.c (mips_option_override): Change setting
3311 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3312 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3313 Change from Mask to Var.
3315 2014-01-27 Jeff Law <law@redhat.com>
3317 * ipa-inline.c (inline_small_functions): Fix typo.
3319 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3321 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3322 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3323 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3324 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3325 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3326 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3327 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3328 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3329 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3330 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3331 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3332 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3333 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3334 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3335 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3336 (_mm512_storeu_epi64): Ditto.
3337 (_mm512_cmpge_epi32_mask): Ditto.
3338 (_mm512_cmpge_epu32_mask): Ditto.
3339 (_mm512_cmpge_epi64_mask): Ditto.
3340 (_mm512_cmpge_epu64_mask): Ditto.
3341 (_mm512_cmple_epi32_mask): Ditto.
3342 (_mm512_cmple_epu32_mask): Ditto.
3343 (_mm512_cmple_epi64_mask): Ditto.
3344 (_mm512_cmple_epu64_mask): Ditto.
3345 (_mm512_cmplt_epi32_mask): Ditto.
3346 (_mm512_cmplt_epu32_mask): Ditto.
3347 (_mm512_cmplt_epi64_mask): Ditto.
3348 (_mm512_cmplt_epu64_mask): Ditto.
3349 (_mm512_cmpneq_epi32_mask): Ditto.
3350 (_mm512_cmpneq_epu32_mask): Ditto.
3351 (_mm512_cmpneq_epi64_mask): Ditto.
3352 (_mm512_cmpneq_epu64_mask): Ditto.
3353 (_mm512_expand_pd): Ditto.
3354 (_mm512_expand_ps): Ditto.
3355 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3356 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3357 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3358 * config/i386/i386.c (ix86_builtins): Add
3359 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3360 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3361 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3362 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3363 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3364 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3365 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3366 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3367 IX86_BUILTIN_PMOVUSQW512_MEM.
3368 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3369 __builtin_ia32_pmovsqd512mem_mask,
3370 __builtin_ia32_pmovqd512mem_mask,
3371 __builtin_ia32_pmovusqw512mem_mask,
3372 __builtin_ia32_pmovsqw512mem_mask,
3373 __builtin_ia32_pmovqw512mem_mask,
3374 __builtin_ia32_pmovusdw512mem_mask,
3375 __builtin_ia32_pmovsdw512mem_mask,
3376 __builtin_ia32_pmovdw512mem_mask,
3377 __builtin_ia32_pmovqb512mem_mask,
3378 __builtin_ia32_pmovusqb512mem_mask,
3379 __builtin_ia32_pmovsqb512mem_mask,
3380 __builtin_ia32_pmovusdb512mem_mask,
3381 __builtin_ia32_pmovsdb512mem_mask,
3382 __builtin_ia32_pmovdb512mem_mask.
3383 (bdesc_args): Add __builtin_ia32_expanddf512,
3384 __builtin_ia32_expandsf512.
3385 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3386 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3387 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3388 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3389 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3390 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3391 (avx512f_<code>v8div16qi2_mask_store): This.
3392 (avx512f_expand<mode>): New.
3394 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3396 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3398 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3399 (_mm512_prefetch_i32scatter_pd): Ditto.
3400 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3401 (_mm512_prefetch_i64scatter_pd): Ditto.
3402 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3403 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3404 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3405 (_mm512_prefetch_i32scatter_ps): Ditto.
3406 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3407 (_mm512_prefetch_i64scatter_ps): Ditto.
3408 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3409 * config/i386/i386-builtin-types.def: Define
3410 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3411 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3412 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3413 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3414 IX86_BUILTIN_SCATTERPFQPD.
3415 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3416 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3417 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3418 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3419 __builtin_ia32_scatterpfqps.
3420 (ix86_expand_builtin): Expand new built-ins.
3421 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3422 fix memory access data type.
3423 (*avx512pf_gatherpf<mode>_mask): Ditto.
3424 (*avx512pf_gatherpf<mode>): Ditto.
3425 (avx512pf_scatterpf<mode>): Ditto.
3426 (*avx512pf_scatterpf<mode>_mask): Ditto.
3427 (*avx512pf_scatterpf<mode>): Ditto.
3428 (GATHER_SCATTER_SF_MEM_MODE): New.
3429 (avx512pf_gatherpf<mode>df): Ditto.
3430 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3431 (*avx512pf_scatterpf<mode>df): Ditto.
3433 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3436 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3437 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3440 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3442 * common/config/arm/arm-common.c
3443 (arm_rewrite_mcpu): Handle multiple names.
3445 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3447 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3449 * gimple-builder.h (create_gimple_tmp): Delete.
3451 2014-01-27 Christian Bruel <christian.bruel@st.com>
3453 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3456 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3458 * config/pa/pa.md (call): Generate indirect long calls to non-local
3459 functions when outputing 32-bit code.
3460 (call_value): Likewise except for special call to buggy powf function.
3462 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3463 portable runtime and PIC indirect calls.
3464 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3465 and PIC call sequences. Use ldo instead of blr to set return register
3466 in PIC call sequence.
3468 2014-01-25 Walter Lee <walt@tilera.com>
3470 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3471 avoid clobbering a live register.
3473 2014-01-25 Walter Lee <walt@tilera.com>
3475 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3476 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3477 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3478 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3480 2014-01-25 Walter Lee <walt@tilera.com>
3482 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3483 arguments on even registers.
3484 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3486 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3487 (BIGGEST_ALIGNMENT): Ditto.
3488 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3490 2014-01-25 Walter Lee <walt@tilera.com>
3492 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3493 insns before bundling.
3494 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3496 2014-01-25 Walter Lee <walt@tilera.com>
3498 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3499 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3500 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3502 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3504 * config/mips/constraints.md (kl): Delete.
3505 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3506 define expands, using...
3507 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3508 instructions for MIPS16.
3509 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3510 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3512 2014-01-25 Walter Lee <walt@tilera.com>
3514 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3518 2014-01-25 Walter Lee <walt@tilera.com>
3520 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3521 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3523 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3525 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3528 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3530 * print-rtl.c (in_call_function_usage): New var.
3531 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3532 EXPR_LIST mode as mode and not as reg note name.
3535 * cfgloopmanip.c (copy_loop_info): If
3536 loop->warned_aggressive_loop_optimizations, make sure
3537 the flag is set in target loop too.
3539 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3541 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3543 * builtins.def: Likewise.
3544 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3545 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3546 * ira.c (ira_setup_eliminable_regset): Likewise.
3547 * omp-low.c (gate_expand_omp): Likewise.
3548 (execute_lower_omp): Likewise.
3549 (diagnose_sb_0): Likewise.
3550 (gate_diagnose_omp_blocks): Likewise.
3551 (simd_clone_clauses_extract): Likewise.
3554 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3556 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3557 correction for little endian...
3558 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3561 2014-01-24 Jeff Law <law@redhat.com>
3563 PR tree-optimization/59919
3564 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3565 for non-returning calls.
3567 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3569 * common/config/aarch64/aarch64-common.c
3570 (aarch64_rewrite_mcpu): Handle multiple names.
3571 * config/aarch64/aarch64.h
3572 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3574 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3576 * input.c (add_file_to_cache_tab): Handle the case where fopen
3579 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3582 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3583 from push operand if code of push isn't PRE_DEC.
3585 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3588 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3589 -mquad-memory-atomic. Update -mquad-memory documentation to say
3590 it is only used for non-atomic loads/stores.
3592 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3593 -mquad-memory or -mquad-memory-atomic switches.
3595 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3596 -mquad-memory-atomic to ISA 2.07 support.
3598 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3599 to separate support of normal quad word memory operations (ldq, stq)
3600 from the atomic quad word memory operations.
3602 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3603 support to separate non-atomic quad word operations from atomic
3604 quad word operations. Disable non-atomic quad word operations in
3605 little endian mode so that we don't have to swap words after the
3606 load and before the store.
3607 (quad_load_store_p): Add comment about atomic quad word support.
3608 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3609 options printed with -mdebug=reg.
3611 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3612 -mquad-memory-atomic as the test for whether we have quad word
3613 atomic instructions.
3614 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3615 or -mp8-vector are used, allow byte/half-word atomic operations.
3617 * config/rs6000/sync.md (load_lockedti): Insure that the address
3618 is a proper indexed or indirect address for the lqarx instruction.
3619 On little endian systems, swap the hi/lo registers after the lqarx
3621 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3622 insure the address is valid for the lqarx instruction.
3623 (store_conditionalti): Insure that the address is a proper indexed
3624 or indirect address for the stqcrx. instruction. On little endian
3625 systems, swap the hi/lo registers before doing the stqcrx.
3627 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3628 insure the address is valid for the stqcrx. instruction.
3630 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3631 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3632 type of quad memory support is available.
3634 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3637 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3638 there is a danger of looping.
3640 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3642 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3643 force flag_ira_loop_pressure if set via command line.
3645 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3647 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3648 (ashr_simd): New builtin handling DI mode.
3649 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3650 (aarch64_sshr_simddi): New match pattern.
3651 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3652 (vshrd_n_s64): Likewise.
3653 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3655 2014-01-23 Nick Clifton <nickc@redhat.com>
3657 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3658 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3659 favour of mcu specific scripts.
3660 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3663 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3664 Alex Velenko <Alex.Velenko@arm.com>
3666 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3667 (vaddv_s16): Likewise.
3668 (vaddv_s32): Likewise.
3669 (vaddv_u8): Likewise.
3670 (vaddv_u16): Likewise.
3671 (vaddv_u32): Likewise.
3672 (vaddvq_s8): Likewise.
3673 (vaddvq_s16): Likewise.
3674 (vaddvq_s32): Likewise.
3675 (vaddvq_s64): Likewise.
3676 (vaddvq_u8): Likewise.
3677 (vaddvq_u16): Likewise.
3678 (vaddvq_u32): Likewise.
3679 (vaddvq_u64): Likewise.
3680 (vaddv_f32): Likewise.
3681 (vaddvq_f32): Likewise.
3682 (vaddvq_f64): Likewise.
3683 (vmaxv_f32): Likewise.
3684 (vmaxv_s8): Likewise.
3685 (vmaxv_s16): Likewise.
3686 (vmaxv_s32): Likewise.
3687 (vmaxv_u8): Likewise.
3688 (vmaxv_u16): Likewise.
3689 (vmaxv_u32): Likewise.
3690 (vmaxvq_f32): Likewise.
3691 (vmaxvq_f64): Likewise.
3692 (vmaxvq_s8): Likewise.
3693 (vmaxvq_s16): Likewise.
3694 (vmaxvq_s32): Likewise.
3695 (vmaxvq_u8): Likewise.
3696 (vmaxvq_u16): Likewise.
3697 (vmaxvq_u32): Likewise.
3698 (vmaxnmv_f32): Likewise.
3699 (vmaxnmvq_f32): Likewise.
3700 (vmaxnmvq_f64): Likewise.
3701 (vminv_f32): Likewise.
3702 (vminv_s8): Likewise.
3703 (vminv_s16): Likewise.
3704 (vminv_s32): Likewise.
3705 (vminv_u8): Likewise.
3706 (vminv_u16): Likewise.
3707 (vminv_u32): Likewise.
3708 (vminvq_f32): Likewise.
3709 (vminvq_f64): Likewise.
3710 (vminvq_s8): Likewise.
3711 (vminvq_s16): Likewise.
3712 (vminvq_s32): Likewise.
3713 (vminvq_u8): Likewise.
3714 (vminvq_u16): Likewise.
3715 (vminvq_u32): Likewise.
3716 (vminnmv_f32): Likewise.
3717 (vminnmvq_f32): Likewise.
3718 (vminnmvq_f64): Likewise.
3720 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3722 * config/aarch64/aarch64-simd.md
3723 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3724 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3725 (*aarch64_mul3_elt<mode>): Likewise.
3726 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3727 (*aarch64_mul3_elt_to_64v2df): Likewise.
3728 (*aarch64_mla_elt<mode>): Likewise.
3729 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3730 (*aarch64_mls_elt<mode>): Likewise.
3731 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3732 (*aarch64_fma4_elt<mode>): Likewise.
3733 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3734 (*aarch64_fma4_elt_to_64v2df): Likewise.
3735 (*aarch64_fnma4_elt<mode>): Likewise.
3736 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3737 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3738 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3739 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3740 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3741 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3742 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3743 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3744 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3746 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3748 * config/aarch64/aarch64-simd.md
3749 (aarch64_be_checked_get_lane<mode>): New define_expand.
3750 * config/aarch64/aarch64-simd-builtins.def
3751 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3752 New builtin definition.
3753 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3754 Use new safe be builtin.
3756 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3758 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3760 (aarch64_be_st1<mode>): Likewise.
3761 (aarch_ld1<VALL:mode>): Define_expand modified.
3762 (aarch_st1<VALL:mode>): Likewise.
3763 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3764 (UNSPEC_ST1): Likewise.
3766 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3768 * config/microblaze/microblaze.md: Add trap insn and attribute
3770 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3772 PR preprocessor/58580
3773 * input.h (location_get_source_line): Take an additional line_size
3775 (void diagnostics_file_cache_fini): Declare new function.
3776 * input.c (struct fcache): New type.
3777 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3778 New static constants.
3779 (diagnostic_file_cache_init, total_lines_num)
3780 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3781 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3782 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3783 (get_next_line, read_next_line, goto_next_line, read_line_num):
3784 New static function definitions.
3785 (diagnostic_file_cache_fini): New function.
3786 (location_get_source_line): Take an additional output line_len
3787 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3789 * diagnostic.c (diagnostic_finish): Call
3790 diagnostic_file_cache_fini.
3791 (adjust_line): Take an additional input parameter for the length
3792 of the line, rather than calculating it with strlen.
3793 (diagnostic_show_locus): Adjust the use of
3794 location_get_source_line and adjust_line with respect to their new
3795 signature. While displaying a line now, do not stop at the first
3796 null byte. Rather, display the zero byte as a space and keep
3797 going until we reach the size of the line.
3798 * Makefile.in: Add vec.o to OBJS-libcommon
3800 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3801 Ilya Tocar <ilya.tocar@intel.com>
3803 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3804 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3805 (__builtin_ia32_kmov16): Ditto.
3806 * config/i386/i386.md (UNSPEC_KMOV): New.
3809 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3811 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3812 (_mm512_storeu_si512): Ditto.
3814 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3817 * rtl.h (get_referenced_operands): Declare.
3818 * recog.c (get_referenced_operands): New function.
3819 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3820 operands have been referenced when recording LO_SUM references.
3822 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3824 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3826 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3828 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3829 Enable for generic and recent AMD targets.
3831 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3833 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3834 ARG_SIZE note when adjustment was eliminated.
3836 2014-01-22 Jeff Law <law@redhat.com>
3838 PR tree-optimization/59597
3839 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3840 in file. Accept new argument REGISTERING and use it to modify
3841 dump output appropriately.
3842 (register_jump_thread): Corresponding changes.
3843 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3844 thread paths involving joiner blocks. Add code to dump cancelled
3845 jump threading paths.
3847 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3849 PR rtl-optimization/59477
3850 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3851 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3853 2014-01-22 Tom Tromey <tromey@redhat.com>
3855 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3857 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3859 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3861 PR rtl-optimization/59896
3862 * lra-constraints.c (process_alt_operands): Check unused note for
3863 matched operands of insn with no output reloads.
3865 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3867 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3868 (mips_move_from_gpr_cost): Likewise.
3870 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3872 PR rtl-optimization/59858
3873 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3874 ira_class_hard_regs_num.
3875 (process_alt_operands): Increase reject for dying matched operand.
3877 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3880 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3881 smaller than size, perform several stores or loads and stores
3882 at dst + count - size to store or copy all of size bytes, rather
3883 than just last modesize bytes.
3885 2014-01-20 DJ Delorie <dj@redhat.com>
3887 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3888 that CLOBBERs are REGs before propogating their values.
3890 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3893 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3894 (cgraph_inline_failed_type): New function.
3895 * cgraph.h (DEFCIFCODE): Add type.
3896 (cgraph_inline_failed_type_t): New enum.
3897 (cgraph_inline_failed_type): New prototype.
3898 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3899 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3900 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3901 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3902 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3903 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3904 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3905 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3906 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3907 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3908 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3909 OPTIMIZATION_MISMATCH.
3910 * tree-inline.c (expand_call_inline): Emit errors during
3911 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3913 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3916 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3917 mode attribute in insn output.
3919 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3921 * output.h (output_constant): Delete.
3922 * varasm.c (output_constant): Make private.
3924 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3926 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3928 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3931 * tree.h (fold_builtin_strcat): New prototype.
3932 * builtins.c (fold_builtin_strcat): No longer static. Add len
3933 argument, if non-NULL, don't call c_strlen. Optimize
3934 directly into __builtin_memcpy instead of __builtin_strcpy.
3935 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3936 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3938 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3940 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3941 for SImode_address_operand operands, having only a REG argument.
3943 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3945 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3946 loader name using mbig-endian.
3947 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3949 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3951 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3955 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3957 * config/aarch64/aarch64-protos.h
3958 (aarch64_cannot_change_mode_class_ptr): Declare.
3959 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3960 aarch64_cannot_change_mode_class_ptr): New.
3961 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3962 backend hook aarch64_cannot_change_mode_class.
3964 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3966 * common/config/aarch64/aarch64-common.c
3967 (aarch64_handle_option): Don't handle any option order logic here.
3968 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3969 selected_cpu, warn on architecture version mismatch.
3970 (aarch64_override_options): Fix parsing order for option strings.
3972 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3973 Iain Sandoe <iain@codesourcery.com>
3976 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3977 warning. Amend comment to reflect current functionality.
3979 2014-01-20 Richard Biener <rguenther@suse.de>
3982 * builtins.c (fold_builtin_strcat): Remove case better handled
3983 by tree-ssa-strlen.c.
3985 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3987 * config/aarch64/aarch64.opt
3988 (mcpu, march, mtune): Make case-insensitive.
3990 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3993 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3994 if operands[1] is a REG or ZERO_EXTEND of a REG.
3996 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3998 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4000 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4002 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4003 long non-pic millicode calls.
4005 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4007 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4009 2014-01-19 Kito Cheng <kito@0xlab.org>
4011 * builtins.c (expand_movstr): Check movstr expand done or fail.
4013 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4014 H.J. Lu <hongjiu.lu@intel.com>
4017 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4018 to DImode for zero-extended addresses.
4020 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4022 PR rtl-optimization/57763
4023 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4024 on the new indirect jump_insn and increment LABEL_NUSES (label).
4026 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4030 * config.gcc (x86_archs): New variable.
4031 (x86_64_archs): Likewise.
4032 (x86_cpus): Likewise.
4033 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4034 --with-arch/--with-cpu= options.
4035 Support --with-arch=/--with-cpu={nehalem,westmere,
4036 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4038 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4040 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4041 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4043 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4045 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4047 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4050 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4051 clear cpp_get_options (parse_in)->warn_unused_macros for
4052 ix86_target_macros_internal with cpp_define.
4054 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4056 * jump.c (delete_related_insns): Keep (use (insn))s.
4057 * reorg.c (redundant_insn): Check for barriers too.
4059 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4061 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4063 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4065 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4066 call to $$dyncall when TARGET_LONG_CALLS is true.
4068 2014-01-17 Jeff Law <law@redhat.com>
4070 * ree.c (combine_set_extension): Temporarily disable test for
4071 changing number of hard registers.
4073 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4076 * ipa-inline-analysis.c (inline_free_summary):
4077 Do not free summary of aliases.
4079 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4082 * gimplify.c (gimplify_expr): Use create_tmp_var
4083 instead of create_tmp_var_raw. If cond doesn't have
4084 integral type, don't add the IFN_ANNOTATE builtin at all.
4086 2014-01-17 Martin Jambor <mjambor@suse.cz>
4089 * ipa-cp.c (prev_edge_clone): New variable.
4090 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4091 Also resize prev_edge_clone vector.
4092 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4093 (ipcp_edge_removal_hook): New function.
4094 (ipcp_driver): Register ipcp_edge_removal_hook.
4096 2014-01-17 Andrew Pinski <apinski@cavium.com>
4097 Steve Ellcey <sellcey@mips.com>
4100 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4103 2014-01-17 Jeff Law <law@redhat.com>
4106 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4107 so that pass_ccp runs first.
4109 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4111 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4112 (ix86_adjust_cost): Use !TARGET_XXX.
4113 (do_reorder_for_imul): Likewise.
4114 (swap_top_of_ready_list): Likewise.
4115 (ix86_sched_reorder): Likewise.
4117 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4119 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4120 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4121 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4122 (intel_memset): New. Duplicate slm_memset.
4123 (intel_cost): New. Duplicate slm_cost.
4124 (m_INTEL): New macro.
4125 (processor_target_table): Add "intel".
4126 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4127 with PROCESSOR_INTEL for "intel".
4128 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4129 PROCESSOR_SILVERMONT.
4130 (ix86_issue_rate): Likewise.
4131 (ix86_adjust_cost): Likewise.
4132 (ia32_multipass_dfa_lookahead): Likewise.
4133 (swap_top_of_ready_list): Likewise.
4134 (ix86_sched_reorder): Likewise.
4135 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4136 instead of TARGET_OPT_AGU.
4137 * config/i386/i386.h (TARGET_INTEL): New.
4138 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4139 (processor_type): Add PROCESSOR_INTEL.
4140 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4141 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4143 2014-01-17 Marek Polacek <polacek@redhat.com>
4146 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4149 2014-01-17 Richard Biener <rguenther@suse.de>
4151 PR tree-optimization/46590
4152 * opts.c (default_options_table): Add entries for
4153 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4154 all enabled at -O1 but not for -Og.
4155 * common.opt (fbranch-count-reg): Remove Init(1).
4156 (fmove-loop-invariants): Likewise.
4157 (ftree-pta): Likewise.
4159 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4161 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4162 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4163 decls to at least the GCC 4.8 used alignments.
4166 * tree-nested.c (convert_nonlocal_reference_stmt,
4167 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4168 of GIMPLE_BIND stmts, adjust associated decls.
4170 2014-01-17 Richard Biener <rguenther@suse.de>
4172 PR tree-optimization/46590
4173 * vec.h (vec<>::bseach): New member function implementing
4174 binary search according to C89 bsearch.
4175 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4176 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4177 bitmap pointer again. Make accesses_in_loop a flat array.
4178 (mem_ref_obstack): New global.
4179 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4180 (mark_ref_stored): Likewise.
4181 (ref_indep_loop_p_2): Likewise.
4182 (set_ref_stored_in_loop): New helper function.
4183 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4184 (memref_free): Adjust.
4185 (record_mem_ref_loc): Simplify.
4186 (gather_mem_refs_stmt): Adjust.
4187 (sort_locs_in_loop_postorder_cmp): New function.
4188 (analyze_memory_references): Sort accesses_in_loop after
4189 loop postorder number.
4190 (find_ref_loc_in_loop_cmp): New function.
4191 (for_all_locs_in_loop): Find relevant cluster of locs in
4192 accesses_in_loop and iterate without recursion.
4193 (execute_sm): Avoid uninit warning.
4194 (struct ref_always_accessed): Simplify.
4195 (ref_always_accessed::operator ()): Likewise.
4196 (ref_always_accessed_p): Likewise.
4197 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4198 loop postorder numbers here.
4199 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4202 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4205 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4206 on decls for which assemble_alias has been called.
4208 2014-01-17 Nick Clifton <nickc@redhat.com>
4210 * config/msp430/msp430.opt: (mcpu): New option.
4211 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4212 (msp430_option_override): Parse target_cpu. If the MCU name
4213 matches a generic string, clear target_mcu.
4214 (msp430_attr): Allow numeric interrupt values up to 63.
4215 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4216 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4218 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4220 * config/msp430/msp430.md (popm): Use %J rather than %I.
4221 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4222 (addhi_cy_i): Use immediate_operand for operand 2.
4223 * doc/invoke.texi: Document -mcpu option.
4225 2014-01-17 Richard Biener <rguenther@suse.de>
4227 PR rtl-optimization/38518
4228 * df.h (df_analyze_loop): Declare.
4229 * df-core.c: Include cfgloop.h.
4230 (df_analyze_1): Split out main part of df_analyze.
4231 (df_analyze): Adjust.
4232 (loop_inverted_post_order_compute): New function.
4233 (loop_post_order_compute): Likewise.
4234 (df_analyze_loop): New function avoiding whole-function
4236 * loop-invariant.c (find_defs): Use df_analyze_loop.
4237 (find_invariants): Adjust.
4238 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4240 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4242 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4243 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4245 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4247 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4248 traversal when removing references.
4250 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4253 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4255 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4258 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4259 pushing a reload for an autoinc when we had previously reloaded an
4260 inner part of the address.
4262 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4264 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4266 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4267 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4268 when not giving up or versioning for alias only because of
4270 (vect_analyze_data_ref_dependences): Set to true.
4271 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4273 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4274 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4278 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4281 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4282 operand 0 predicate for gathers, use a new pseudo as subtarget.
4284 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4287 * lra-constraints.c (process_alt_operands): Add printing debug info.
4288 Check absence of input/output reloads for matched operands too.
4290 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4292 PR rtl-optimization/59835
4293 * ira.c (ira_init_register_move_cost): Increase cost for
4296 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4298 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4300 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4303 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4304 non-register objects. Use gen_(high/low)part more consistently.
4307 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4310 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4311 endian support, remove tests for WORDS_BIG_ENDIAN.
4312 (p8_mfvsrd_3_<mode>): Likewise.
4313 (reload_gpr_from_vsx<mode>): Likewise.
4314 (reload_gpr_from_vsxsf): Likewise.
4315 (p8_mfvsrd_4_disf): Likewise.
4317 2014-01-16 Richard Biener <rguenther@suse.de>
4319 PR rtl-optimization/46590
4320 * lcm.c (compute_antinout_edge): Use postorder iteration.
4321 (compute_laterin): Use inverted postorder iteration.
4323 2014-01-16 Nick Clifton <nickc@redhat.com>
4326 * varasm.c (output_constant): Return the number of bytes actually
4328 (output_constructor_array_range): Update the field size with the
4329 number of bytes emitted by output_constant.
4330 (output_constructor_regular_field): Likewise. Also do not
4331 complain if the total number of bytes emitted is now greater
4332 than the expected fieldpos.
4333 * output.h (output_constant): Update prototype and descriptive comment.
4335 2014-01-16 Marek Polacek <polacek@redhat.com>
4338 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4339 it is error_mark_node.
4341 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4343 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4344 VALID_AVX256_REG_OR_OI_MODE.
4346 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4348 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4349 current procedure should be profiled.
4351 2014-01-15 Andrew Pinski <apinski@cavium.com>
4353 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4354 of moving from/to the STACK_REG register class.
4356 2014-01-15 Richard Henderson <rth@redhat.com>
4359 * reginfo.c (global_regs_decl): Globalize.
4360 * rtl.h (global_regs_decl): Declare.
4361 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4362 reserved via global_regs.
4364 2014-01-15 Teresa Johnson <tejohnson@google.com>
4366 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4368 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4370 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4371 and vmulosh rather than call gen_vec_widen_smult_*.
4372 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4373 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4374 (vec_widen_smult_even_v16qi): Likewise.
4375 (vec_widen_umult_even_v8hi): Likewise.
4376 (vec_widen_smult_even_v8hi): Likewise.
4377 (vec_widen_umult_odd_v16qi): Likewise.
4378 (vec_widen_smult_odd_v16qi): Likewise.
4379 (vec_widen_umult_odd_v8hi): Likewise.
4380 (vec_widen_smult_odd_v8hi): Likewise.
4381 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4382 vmuloub rather than call gen_vec_widen_umult_*.
4383 (vec_widen_umult_lo_v16qi): Likewise.
4384 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4385 vmulosb rather than call gen_vec_widen_smult_*.
4386 (vec_widen_smult_lo_v16qi): Likewise.
4387 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4388 rather than call gen_vec_widen_umult_*.
4389 (vec_widen_umult_lo_v8hi): Likewise.
4390 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4391 rather than call gen_vec_widen_smult_*.
4392 (vec_widen_smult_lo_v8hi): Likewise.
4394 2014-01-15 Jeff Law <law@redhat.com>
4396 PR tree-optimization/59747
4397 * ree.c (find_and_remove_re): Properly handle case where a second
4398 eliminated extension requires widening a copy created for elimination
4399 of a prior extension.
4400 (combine_set_extension): Ensure that the number of hard regs needed
4401 for a destination register does not change when we widen it.
4403 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4405 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4406 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4407 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4408 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4409 (avr-*-rtems*): Likewise.
4410 (bfin*-rtems*): Likewise.
4411 (moxie-*-rtems*): Likewise.
4412 (h8300-*-rtems*): Likewise.
4413 (i[34567]86-*-rtems*): Likewise.
4414 (lm32-*-rtems*): Likewise.
4415 (m32r-*-rtems*): Likewise.
4416 (m68k-*-rtems*): Likewise.
4417 (microblaze*-*-rtems*): Likewise.
4418 (mips*-*-rtems*): Likewise.
4419 (powerpc-*-rtems*): Likewise.
4420 (sh-*-rtems*): Likewise.
4421 (sparc-*-rtems*): Likewise.
4422 (sparc64-*-rtems*): Likewise.
4423 (v850-*-rtems*): Likewise.
4424 (m32c-*-rtems*): Likewise.
4426 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4428 PR rtl-optimization/59511
4429 * ira.c (ira_init_register_move_cost): Use memory costs for some
4430 cases of register move cost calculations.
4431 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4432 instead of BB frequency.
4433 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4434 * lra-assigns.c (find_hard_regno_for): Ditto.
4436 2014-01-15 Richard Biener <rguenther@suse.de>
4438 PR tree-optimization/59822
4439 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4440 (vectorizable_load): Use it to hoist defs of uses of invariant
4441 loads out of the loop.
4443 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4444 Kugan Vivekanandarajah <kuganv@linaro.org>
4447 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4450 2014-01-15 Richard Biener <rguenther@suse.de>
4452 PR rtl-optimization/59802
4453 * lcm.c (compute_available): Use inverted postorder to seed
4454 the initial worklist.
4456 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4459 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4460 ADDR_REGS for invalid symrefs in non-PIC code.
4462 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4465 * builtins.c (determine_block_size): Initialize *probable_max_size
4466 even if len_rtx is CONST_INT.
4468 2014-01-14 Andrew Pinski <apinski@cavium.com>
4470 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4471 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4472 (cortexa53_tunings): Likewise.
4473 (aarch64_sched_issue_rate): New function.
4474 (TARGET_SCHED_ISSUE_RATE): Define.
4476 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4478 * ira-costs.c (find_costs_and_classes): Add missed
4479 ira_init_register_move_cost_if_necessary.
4481 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4484 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4486 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4489 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4490 to indicate if type is used for function return value. Warn ABI
4491 change if the vector mode isn't available for function return value.
4492 (ix86_function_arg_advance): Pass false to type_natural_mode.
4493 (ix86_function_arg): Likewise.
4494 (ix86_gimplify_va_arg): Likewise.
4495 (function_arg_32): Don't warn ABI change.
4496 (ix86_function_value): Pass true to type_natural_mode.
4497 (ix86_return_in_memory): Likewise.
4498 (ix86_struct_value_rtx): Removed.
4499 (TARGET_STRUCT_VALUE_RTX): Likewise.
4501 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4503 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4504 converting a conditional jump into a conditional return.
4506 2014-01-14 Richard Biener <rguenther@suse.de>
4508 PR tree-optimization/58921
4509 PR tree-optimization/59006
4510 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4511 hoisting invariant stmts.
4512 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4513 invariant loads on the preheader edge if possible.
4515 2014-01-14 Joey Ye <joey.ye@arm.com>
4517 * doc/plugin.texi (Building GCC plugins): Update to C++.
4519 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4521 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4522 (_mm_rcp28_round_ss): Ditto.
4523 (_mm_rsqrt28_round_sd): Ditto.
4524 (_mm_rsqrt28_round_ss): Ditto.
4525 (_mm_rcp28_sd): Ditto.
4526 (_mm_rcp28_ss): Ditto.
4527 (_mm_rsqrt28_sd): Ditto.
4528 (_mm_rsqrt28_ss): Ditto.
4529 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4530 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4531 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4532 (IX86_BUILTIN_RCP28SD): Ditto.
4533 (IX86_BUILTIN_RCP28SS): Ditto.
4534 (IX86_BUILTIN_RSQRT28SD): Ditto.
4535 (IX86_BUILTIN_RSQRT28SS): Ditto.
4536 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4537 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4538 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4539 (ix86_expand_special_args_builtin): Expand new FTYPE.
4540 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4541 (srcp14<mode>): Make insn unary.
4542 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4543 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4544 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4545 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4546 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4547 Fix rounding: make it SAE only.
4548 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4550 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4552 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4553 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4554 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4555 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4556 (round_saeonly_mask_scalar_operand4): Ditto.
4557 (round_saeonly_mask_scalar_op3): Ditto.
4558 (round_saeonly_mask_scalar_op4): Ditto.
4560 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4562 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4563 Implement -maltivec=be for vec_insert and vec_extract.
4565 2014-01-10 DJ Delorie <dj@redhat.com>
4567 * config/msp430/msp430.md (call_internal): Don't allow memory
4568 references with SP as the base register.
4569 (call_value_internal): Likewise.
4570 * config/msp430/constraints.md (Yc): New. For memory references
4571 that don't use SP as a base register.
4573 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4574 "an integer without a # prefix"
4575 * config/msp430/msp430.md (epilogue_helper): Use it.
4577 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4580 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4581 AVX512F gather builtins.
4582 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4583 on gather decls with INTEGER_TYPE masktype.
4584 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4585 directly into the builtin rather than hoisting it before loop.
4587 PR tree-optimization/59387
4588 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4589 (scev_const_prop): If folded_casts and type has undefined overflow,
4590 use force_gimple_operand instead of force_gimple_operand_gsi and
4591 for each added stmt if it is assign with
4592 arith_code_with_undefined_signed_overflow, call
4593 rewrite_to_defined_overflow.
4594 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4595 gimple-fold.h instead.
4596 (arith_code_with_undefined_signed_overflow,
4597 rewrite_to_defined_overflow): Moved to ...
4598 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4599 rewrite_to_defined_overflow): ... here. No longer static.
4600 Include gimplify-me.h.
4601 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4602 rewrite_to_defined_overflow): New prototypes.
4604 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4606 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4608 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4610 * builtins.c (get_object_alignment_2): Minor tweak.
4611 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4613 2014-01-13 Christian Bruel <christian.bruel@st.com>
4615 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4616 optimized non constant lengths.
4618 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4621 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4622 load as __atomic_load_N if possible.
4624 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4626 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4628 (rs6000_expand_builtin): Adjust call.
4630 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4633 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4634 * config/rs6000/rs6000.c: Include target-globals.h.
4635 (rs6000_set_current_function): Instead of doing target_reinit
4636 unconditionally, use save_target_globals_default_opts and
4637 restore_target_globals.
4639 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4641 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4642 (rs6000_expand_builtin): Handle mffs and mtfsf.
4643 (rs6000_init_builtins): Define mffs and mtfsf.
4644 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4645 (rs6000_mffs): New pattern.
4646 (rs6000_mtfsf): New pattern.
4648 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4650 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4651 Start narrowing with START. Apply candidate-use pair
4652 and check overall cost in narrowing.
4653 (iv_ca_prune): Pass new argument.
4655 2014-01-10 Jeff Law <law@redhat.com>
4658 * ree.c (combine_reaching_defs): Ensure the defining statement
4659 occurs before the extension when optimizing extensions with
4660 different source and destination hard registers.
4662 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4665 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4666 vtables into the type inheritance graph.
4668 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4670 PR rtl-optimization/59754
4671 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4672 modes in the REGNO != REGNO case.
4674 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4676 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4678 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4680 PR tree-optimization/59745
4681 * tree-predcom.c (tree_predictive_commoning_loop): Call
4682 free_affine_expand_cache if giving up because components is NULL.
4684 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4685 GC in payload of target_globals struct instead of allocating them on
4686 the heap and the larger structs separately using GC.
4687 * target-globals.h (struct target_globals): Make regs, hard_regs,
4688 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4689 of GTY((skip)) and change type to void *.
4690 (reset_target_globals): Cast loads from those fields to corresponding
4693 2014-01-10 Steve Ellcey <sellcey@mips.com>
4696 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4697 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4698 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4700 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4703 * aarch64-modes.def (CC_Zmode): New flags mode.
4704 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4705 represents an equality.
4706 (aarch64_get_condition_code): Handle CC_Zmode.
4707 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4709 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4711 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4712 extraction in good case.
4714 2014-01-10 Richard Biener <rguenther@suse.de>
4716 PR tree-optimization/59374
4717 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4718 checking after SLP discovery. Mark stmts not participating
4719 in any SLP instance properly.
4721 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4723 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4724 when handling a SET rtx.
4726 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4728 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4729 (cortex-a57): Likewise.
4730 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4732 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4734 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4735 non-iwmmxt builtins.
4737 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4741 * ipa-devirt.c record_target_from_binfo): Take as argument
4742 stack of binfos and lookup matching one for virtual inheritance.
4743 (possible_polymorphic_call_targets_1): Update.
4745 2014-01-10 Huacai Chen <chenhc@lemote.com>
4747 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4748 kernel strings for Loongson-2E/2F/3A.
4750 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4753 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4754 is_gimple_call before calling gimple_call_internal_p.
4756 2014-01-09 Steve Ellcey <sellcey@mips.com>
4758 * Makefile.in (TREE_FLOW_H): Remove.
4759 (TREE_SSA_H): Add file names from tree-flow.h.
4760 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4761 * tree.h: Remove tree-flow.h reference.
4762 * hash-table.h: Remove tree-flow.h reference.
4763 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4764 reference with tree-ssa-loop.h.
4766 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4768 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4769 default element-order behavior for -maltivec.
4770 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4771 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4772 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4773 when targeting big endian, at least for now.
4774 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4776 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4779 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4780 var satisfies use_register_for_decl, just take into account type
4781 alignment, rather than decl alignment.
4783 PR tree-optimization/59622
4784 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4785 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4786 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4787 Don't devirtualize for inplace at all. For targets.length () == 1,
4788 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4790 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4792 * config/i386/i386.md (cpu): Remove the unused btver1.
4794 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4796 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4798 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4801 * tree-core.h (struct target_globals): New forward declaration.
4802 (struct tree_target_option): Add globals field.
4803 * tree.h (TREE_TARGET_GLOBALS): Define.
4804 (prepare_target_option_nodes_for_pch): New prototype.
4805 * target-globals.h (struct target_globals): Define even if
4807 * tree.c (prepare_target_option_node_for_pch,
4808 prepare_target_option_nodes_for_pch): New functions.
4809 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4810 * config/i386/i386.c: Include target-globals.h.
4811 (ix86_set_current_function): Instead of doing target_reinit
4812 unconditionally, use save_target_globals_default_opts and
4813 restore_target_globals.
4815 2014-01-09 Richard Biener <rguenther@suse.de>
4817 PR tree-optimization/59715
4818 * tree-cfg.h (split_critical_edges): Declare.
4819 * tree-cfg.c (split_critical_edges): Export.
4820 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4822 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4824 * cfgexpand.c (expand_stack_vars): Optionally disable
4825 asan stack protection.
4826 (expand_used_vars): Likewise.
4827 (partition_stack_vars): Likewise.
4828 * asan.c (asan_emit_stack_protection): Optionally disable
4829 after return stack usage.
4830 (instrument_derefs): Optionally disable memory access instrumentation.
4831 (instrument_builtin_call): Likewise.
4832 (instrument_strlen_call): Likewise.
4833 (asan_protect_global): Optionally disable global variables protection.
4834 * doc/invoke.texi: Added doc for new options.
4835 * params.def: Added new options.
4836 * params.h: Likewise.
4838 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4840 PR rtl-optimization/59724
4841 * ifcvt.c (cond_exec_process_if_block): Don't call
4842 flow_find_head_matching_sequence with 0 longest_match.
4843 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4844 non-active insns if !stop_after.
4845 (try_head_merge_bb): Revert 2014-01-07 changes.
4847 2014-01-08 Jeff Law <law@redhat.com>
4849 * ree.c (get_sub_rtx): New function, extracted from...
4850 (merge_def_and_ext): Here.
4851 (combine_reaching_defs): Use get_sub_rtx.
4853 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4855 * cgraph.h (varpool_variable_node): Do not choke on null node.
4857 2014-01-08 Catherine Moore <clm@codesourcery.com>
4859 * config/mips/mips.md (simple_return): Attempt to use JRC
4861 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4863 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4865 PR rtl-optimization/59137
4866 * reorg.c (steal_delay_list_from_target): Call update_block for
4868 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4870 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4872 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4873 two duplicate entries.
4875 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4878 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4880 * config/mips/mips.c (mips_truncated_op_cost): New function.
4881 (mips_rtx_costs): Adjust test for BADDU.
4882 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4884 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4886 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4887 (*baddu_si): ...this new pattern.
4889 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4892 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4894 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4897 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4899 (expand_expr, expand_normal): Adjust.
4900 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4901 inner_reference_p. Use inner_reference_p to expand inner references.
4902 (store_expr): Adjust.
4903 * cfgexpand.c (expand_call_stmt): Adjust.
4905 2014-01-08 Rong Xu <xur@google.com>
4907 * gcov-io.c (gcov_var): Move from gcov-io.h.
4908 (gcov_position): Ditto.
4909 (gcov_is_error): Ditto.
4910 (gcov_rewrite): Ditto.
4911 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4912 only part to libgcc/libgcov.h.
4914 2014-01-08 Marek Polacek <polacek@redhat.com>
4917 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4919 2014-01-08 Marek Polacek <polacek@redhat.com>
4922 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4924 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4926 PR rtl-optimization/59649
4927 * stor-layout.c (get_mode_bounds): For BImode return
4928 0 and STORE_FLAG_VALUE.
4930 2014-01-08 Richard Biener <rguenther@suse.de>
4933 * gimple.h (is_gimple_builtin_call): Remove.
4934 (gimple_builtin_call_types_compatible_p): New.
4935 (gimple_call_builtin_p): New overload.
4936 * gimple.c (is_gimple_builtin_call): Remove.
4937 (validate_call): Rename to ...
4938 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4940 (validate_type): New static function.
4941 (gimple_call_builtin_p): New overload and adjust.
4942 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4943 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4944 (gimple_fold_stmt_to_constant_1): Likewise.
4945 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4947 2014-01-08 Richard Biener <rguenther@suse.de>
4950 * gimplify.c (gimplify_expr): Gimplify register-register type
4951 VIEW_CONVERT_EXPRs to separate stmts.
4953 2014-01-07 Jeff Law <law@redhat.com>
4956 * ree.c (combine_set_extension): Handle case where source
4957 and destination registers in an extension insn are different.
4958 (combine_reaching_defs): Allow source and destination registers
4959 in extension to be different under limited circumstances.
4960 (add_removable_extension): Remove restriction that the
4961 source and destination registers in the extension are the same.
4962 (find_and_remove_re): Emit a copy from the extension's
4963 destination to its source after the defining insn if
4964 the source and destination registers are different.
4967 * ifcvt.c (merge_if_block): If we are merging a block with more than
4968 one successor with a block with no successors, remove any BARRIER
4969 after the second block.
4971 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4973 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4975 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4978 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4979 for 14-bit register offsets when INT14_OK_STRICT is false.
4981 2014-01-07 Roland Stigge <stigge@antcom.de>
4982 Michael Meissner <meissner@linux.vnet.ibm.com>
4985 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4986 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4988 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4990 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4993 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4995 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4996 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4997 rtx is const0_rtx or not.
4999 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5002 * target-globals.c (save_target_globals): Remove this_fn_optab
5004 * toplev.c: Include optabs.h.
5005 (target_reinit): Temporarily restore the global options if another
5006 set of options are in force.
5008 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5010 PR rtl-optimization/58668
5011 * cfgcleanup.c (flow_find_cross_jump): Don't count
5012 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5013 to determine what is counted.
5014 (flow_find_head_matching_sequence): Use active_insn_p to determine
5016 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5018 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5019 determine what is counted.
5021 PR tree-optimization/59643
5022 * tree-predcom.c (split_data_refs_to_components): If one dr is
5023 read and one write, determine_offset fails and the write isn't
5024 in the bad component, just put the read into the bad component.
5026 2014-01-07 Mike Stump <mikestump@comcast.net>
5027 Jakub Jelinek <jakub@redhat.com>
5030 * tree-core.h (struct tree_optimization_option): Change optabs
5031 type from unsigned char * to void *.
5032 * optabs.c (init_tree_optimization_optabs): Adjust
5033 TREE_OPTIMIZATION_OPTABS initialization.
5035 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5038 * config/i386/i386.h (struct machine_function): Add
5039 no_drap_save_restore field.
5040 * config/i386/i386.c (ix86_save_reg): Use
5041 !cfun->machine->no_drap_save_restore instead of
5042 crtl->stack_realign_needed.
5043 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5044 this function clears frame_pointer_needed. Set
5045 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5046 and DRAP reg is needed.
5048 2014-01-06 Marek Polacek <polacek@redhat.com>
5051 * doc/implement-c.texi: Mention that other integer types are
5052 permitted as bit-field types in strictly conforming mode.
5054 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5056 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5059 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5061 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5063 2014-01-06 Martin Jambor <mjambor@suse.cz>
5066 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5068 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5070 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5074 * var-tracking.c (add_stores): Preserve the value of the source even if
5075 we don't record the store.
5077 2014-01-06 Terry Guo <terry.guo@arm.com>
5079 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5081 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5084 * config/darwin.c (darwin_function_section): Adjust return values to
5085 correspond to optimisation changes made in r206070.
5087 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5089 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5090 from prefetch_block tune setting.
5091 (nocona_cost): Correct size of prefetch block to 64.
5093 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5095 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5096 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5097 used to save the static chain register in the computation of the offset
5098 from which the FP registers need to be restored.
5100 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5102 PR tree-optimization/59519
5103 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5104 ICE if get_current_def (current_new_name) is already non-NULL, as long
5105 as it is a phi result of some other phi in *new_exit_bb that has
5108 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5109 or vmovdqu* for misaligned_operand.
5110 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5111 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5112 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5113 aligned_mem for AVX512F masked aligned load and store builtins and for
5116 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5118 PR tree-optimization/59651
5119 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5120 Address range for negative step should be added by TYPE_SIZE_UNIT.
5122 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5124 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5125 overlapping registers also for registers other than the stack pointer.
5127 2014-01-03 Marek Polacek <polacek@redhat.com>
5130 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5133 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5136 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5139 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5140 (push splitter): Use <P:MODE_SIZE> instead of
5141 GET_MODE_SIZE (<P:MODE>mode).
5142 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5143 (mov -1, reg peephole2): Likewise.
5144 * config/i386/sse.md (*mov<mode>_internal,
5145 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5146 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5147 *<code><mode>3, *andnot<mode>3<mask_name>,
5148 <mask_codefor><code><mode>3<mask_name>): Likewise.
5149 * config/i386/subst.md (mask_mode512bit_condition,
5150 sd_mask_mode512bit_condition): Likewise.
5152 2014-01-02 Xinliang David Li <davidxl@google.com>
5154 PR tree-optimization/59303
5155 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5156 (dump_predicates): Better output format.
5157 (pred_equal_p): New function.
5158 (is_neq_relop_p): Ditto.
5159 (is_neq_zero_form_p): Ditto.
5160 (pred_expr_equal_p): Ditto.
5161 (pred_neg_p): Ditto.
5162 (simplify_pred): Ditto.
5163 (simplify_preds_2): Ditto.
5164 (simplify_preds_3): Ditto.
5165 (simplify_preds_4): Ditto.
5166 (simplify_preds): Ditto.
5168 (push_to_worklist): Ditto.
5169 (get_pred_info_from_cmp): Ditto.
5170 (is_degenerated_phi): Ditto.
5171 (normalize_one_pred_1): Ditto.
5172 (normalize_one_pred): Ditto.
5173 (normalize_one_pred_chain): Ditto.
5174 (normalize_preds): Ditto.
5175 (normalize_cond_1): Remove function.
5176 (normalize_cond): Ditto.
5177 (is_gcond_subset_of): Ditto.
5178 (is_subset_of_any): Ditto.
5179 (is_or_set_subset_of): Ditto.
5180 (is_and_set_subset_of): Ditto.
5181 (is_norm_cond_subset_of): Ditto.
5182 (pred_chain_length_cmp): Ditto.
5183 (convert_control_dep_chain_into_preds): Type change.
5184 (find_predicates): Ditto.
5185 (find_def_preds): Ditto.
5186 (destroy_predicates_vecs): Ditto.
5187 (find_matching_predicates_in_rest_chains): Ditto.
5188 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5189 (is_pred_expr_subset): Ditto.
5190 (is_pred_chain_subset_of): Ditto.
5191 (is_included_in): Ditto.
5192 (is_superset_of): Ditto.
5194 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5196 Update copyright years.
5198 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5200 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5201 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5202 config/arc/arc.md, config/arc/arc.opt,
5203 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5204 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5205 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5206 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5207 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5208 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5209 vtable-verify.c, vtable-verify.h: Use the standard form for the
5212 2014-01-02 Tobias Burnus <burnus@net-b.de>
5214 * gcc.c (process_command): Update copyright notice dates.
5215 * gcov-dump.c: Ditto.
5217 * doc/cpp.texi: Bump @copying's copyright year.
5218 * doc/cppinternals.texi: Ditto.
5219 * doc/gcc.texi: Ditto.
5220 * doc/gccint.texi: Ditto.
5221 * doc/gcov.texi: Ditto.
5222 * doc/install.texi: Ditto.
5223 * doc/invoke.texi: Ditto.
5225 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5227 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5229 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5231 * config/i386/sse.md (*mov<mode>_internal): Guard
5232 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5234 PR rtl-optimization/59647
5235 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5236 new_rtx into UNSIGNED_FLOAT rtxes.
5238 Copyright (C) 2014 Free Software Foundation, Inc.
5240 Copying and distribution of this file, with or without modification,
5241 are permitted in any medium without royalty provided the copyright
5242 notice and this notice are preserved.