1 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
3 * calls.c (store_one_arg): Remove incorrect const qualification on the
5 * cfgexpand.c (expand_return): Likewise.
6 * expr.c (expand_constructor): Likewise.
7 (expand_expr_real_1): Likewise.
9 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
11 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
14 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
17 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
19 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
21 * config/arm/aarch-common-protos.h
22 (alu_cost_table): Fix spelling of "extend".
23 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
25 2014-03-19 Richard Biener <rguenther@suse.de>
28 * tree-core.h (tree_type_common): Re-order pointer members
29 to reduce recursion depth during GC walks.
31 2014-03-19 Marek Polacek <polacek@redhat.com>
34 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
37 2014-03-19 Richard Biener <rguenther@suse.de>
40 * lto-streamer-in.c (input_function): In WPA stage do not drop
43 2014-03-19 Jakub Jelinek <jakub@redhat.com>
45 PR tree-optimization/60559
46 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
47 with build_zero_cst assignment.
49 2014-03-18 Kai Tietz <ktietz@redhat.com>
51 PR rtl-optimization/56356
52 * sdbout.c (sdbout_parms): Verify that parms'
53 incoming argument is valid.
54 (sdbout_reg_parms): Likewise.
56 2014-03-18 Richard Henderson <rth@redhat.com>
59 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
60 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
61 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
63 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
65 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
66 * doc/plugins.texi (Plugin callbacks): Mention
68 Italicize plugin event names in description. Explain that
69 PLUGIN_PRAGMAS has no sense for lto1. Explain
71 Remind that no GCC functions should be called after
73 Explain what pragmas with expansion are.
75 2014-03-18 Martin Liska <mliska@suse.cz>
77 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
78 gimple call statement is update.
79 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
80 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
82 2014-03-18 Jakub Jelinek <jakub@redhat.com>
85 * ubsan.c (ubsan_instrument_unreachable): Call
86 initialize_sanitizer_builtins.
87 (ubsan_pass): Likewise.
90 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
91 varpool_finalize_decl instead of rest_of_decl_compilation.
93 2014-03-18 Richard Biener <rguenther@suse.de>
95 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
96 by using bitmap_and_compl instead of bitmap_and_compl_into.
97 (df_rd_transfer_function): Likewise.
99 2014-03-18 Richard Biener <rguenther@suse.de>
101 * doc/lto.texi (fresolution): Fix typo.
103 2014-03-18 Richard Biener <rguenther@suse.de>
105 * doc/invoke.texi (flto): Update for changes in 4.9.
107 2014-03-18 Richard Biener <rguenther@suse.de>
109 * doc/loop.texi: Remove section on the removed lambda framework.
110 Update loop docs with recent changes in preserving loop structure.
112 2014-03-18 Richard Biener <rguenther@suse.de>
114 * doc/lto.texi (-fresolution): Document.
116 2014-03-18 Richard Biener <rguenther@suse.de>
118 * doc/contrib.texi: Adjust my name.
120 2014-03-18 Jakub Jelinek <jakub@redhat.com>
123 * internal-fn.c: Include diagnostic-core.h.
124 (expand_BUILTIN_EXPECT): New function.
125 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
126 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
127 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
128 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
130 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
131 Revert 3 argument __builtin_expect code.
132 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
133 * gimple-fold.c (gimple_fold_call): Likewise.
134 * tree.h (fold_builtin_expect): New prototype.
135 * builtins.c (build_builtin_expect_predicate): Add predictor
136 argument, if non-NULL, create 3 argument __builtin_expect.
137 (fold_builtin_expect): No longer static. Add ARG2 argument,
138 pass it through to build_builtin_expect_predicate.
139 (fold_builtin_2): Adjust caller.
140 (fold_builtin_3): Handle BUILT_IN_EXPECT.
141 * internal-fn.def (BUILTIN_EXPECT): New.
143 2014-03-18 Tobias Burnus <burnus@net-b.de>
146 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
147 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
148 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
150 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
153 * predict.c (combine_predictions_for_bb): Fix up formatting.
154 (expr_expected_value_1, expr_expected_value): Add predictor argument,
155 fill what it points to if non-NULL.
156 (tree_predict_by_opcode): Adjust caller, use the predictor.
157 * predict.def (PRED_COMPARE_AND_SWAP): Add.
159 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
161 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
162 proper constant for the store mode.
164 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
166 * symtab.c (change_decl_assembler_name): Fix transparent alias
169 2014-03-16 Renlin Li <Renlin.Li@arm.com>
171 * config/aarch64/aarch64.c: Correct the comments about the
172 aarch64 stack layout.
174 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
176 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
177 check for GF_OMP_FOR_KIND_FOR.
179 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
181 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
182 ymm and zmm register names.
184 2014-03-17 Jakub Jelinek <jakub@redhat.com>
187 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
188 note creation for the 2010-08-31 changes.
190 2014-03-17 Marek Polacek <polacek@redhat.com>
193 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
194 as -fno-tree-loop-vectorize.
195 (expand_omp_simd): Likewise.
197 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
199 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
200 (eligible_for_call_delay): New prototype.
201 * config/sparc/sparc.c (tls_call_delay): Rename into...
202 (eligible_for_call_delay): ...this. Return false if the instruction
203 cannot be put in the delay slot of a branch.
204 (eligible_for_restore_insn): Simplify.
205 (eligible_for_return_delay): Return false if the instruction cannot be
206 put in the delay slot of a branch and simplify.
207 (eligible_for_sibcall_delay): Return false if the instruction cannot be
208 put in the delay slot of a branch.
209 * config/sparc/sparc.md (fix_ut699): New attribute.
210 (tls_call_delay): Delete.
211 (in_call_delay): Reimplement.
212 (eligible_for_sibcall_delay): Rename into...
213 (in_sibcall_delay): ...this.
214 (eligible_for_return_delay): Rename into...
215 (in_return_delay): ...this.
216 (in_branch_delay): Reimplement.
217 (in_uncond_branch_delay): Delete.
218 (in_annul_branch_delay): Delete.
220 2014-03-14 Richard Henderson <rth@redhat.com>
223 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
224 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
225 (*floathi<X87MODEF>2_i387_with_temp): Remove.
226 (floathi splitters): Remove.
227 (float<SWI48x>xf2): New pattern.
228 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
229 code that tried to handle DImode for 32-bit, but which was excluded
230 by the pattern's condition. Drop allocation of stack temporary.
231 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
232 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
233 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
234 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
235 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
236 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
237 (*float<SWI48><MODEF>2_sse_interunit): Remove.
238 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
239 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
240 (*float<SWI48x><X87MODEF>2_i387): Remove.
241 (all float _with_temp splitters): Remove.
242 (*float<SWI48x><MODEF>2_i387): New pattern.
243 (*float<SWI48><MODEF>2_sse): New pattern.
244 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
245 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
247 2014-03-14 Jakub Jelinek <jakub@redhat.com>
248 Marek Polacek <polacek@redhat.com>
251 * common.opt (dump_base_name_prefixed): New Variable.
252 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
253 if x_dump_base_name_prefixed is already set, set it at the end.
255 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
257 PR rtl-optimization/60508
258 * lra-constraints.c (get_reload_reg): Add new parameter
260 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
261 Pass the new parameter values.
263 2014-03-14 Richard Biener <rguenther@suse.de>
265 * common.opt: Revert unintented changes from r205065.
268 2014-03-14 Richard Biener <rguenther@suse.de>
271 * cfghooks.c (split_block): Properly adjust all loops the
272 block was a latch of.
274 2014-03-14 Martin Jambor <mjambor@suse.cz>
277 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
280 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
283 * config/avr/avr.c (avr_set_current_function): Pass function name
284 through default_strip_name_encoding before sanity checking instead
285 of skipping the first char of the assembler name.
287 2014-03-13 Richard Henderson <rth@redhat.com>
290 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
291 (ix86_force_to_memory, ix86_free_from_memory): Remove.
292 * config/i386/i386-protos.h: Likewise.
293 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
294 in the expander instead of a splitter.
295 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
296 any possibility of requiring a memory.
297 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
298 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
299 (fp branch splitters): Update for ix86_split_fp_branch.
300 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
301 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
302 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
303 (*fop_<MODEF>_2_i387): Remove f/r alternative.
304 (*fop_<MODEF>_3_i387): Likewise.
305 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
306 (splitters for the fop_* register patterns): Remove.
307 (fscalexf4_i387): Rename from *fscalexf4_i387.
308 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
310 2014-03-13 Jakub Jelinek <jakub@redhat.com>
312 PR tree-optimization/59779
313 * tree-dfa.c (get_ref_base_and_extent): Use double_int
314 type for bitsize and maxsize instead of HOST_WIDE_INT.
316 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
318 PR rtl-optimization/57320
319 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
320 the CFG after thread_prologue_and_epilogue_insns.
322 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
324 PR rtl-optimization/57189
325 * lra-constraints.c (process_alt_operands): Disfavor spilling
328 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
330 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
333 2014-03-13 Jakub Jelinek <jakub@redhat.com>
335 PR tree-optimization/59025
337 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
338 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
340 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
343 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
344 calls of avr_out_plus_1.
346 2014-03-13 Bin Cheng <bin.cheng@arm.com>
348 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
349 BB's single pred and update the father loop's latch info later.
351 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
353 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
357 (VEC_base): Likewise.
358 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
359 registers, we need to swap double words in little endian mode.
361 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
362 to be a container mode for 128-bit integer operations added in ISA
363 2.07. Unlike TImode and PTImode, the preferred register set is
364 the Altivec/VMX registers for the 128-bit operations.
366 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
368 (rs6000_split_128bit_ok_p): Likewise.
370 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
371 macros for creating ISA 2.07 normal and overloaded builtin
372 functions with 3 arguments.
373 (BU_P8V_OVERLOAD_3): Likewise.
374 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
375 for use as overloaded functions.
376 (VPERM_1TI_UNS): Likewise.
377 (VSEL_1TI): Likewise.
378 (VSEL_1TI_UNS): Likewise.
379 (ST_INTERNAL_1ti): Likewise.
380 (LD_INTERNAL_1ti): Likewise.
381 (XXSEL_1TI): Likewise.
382 (XXSEL_1TI_UNS): Likewise.
383 (VPERM_1TI): Likewise.
384 (VPERM_1TI_UNS): Likewise.
385 (XXPERMDI_1TI): Likewise.
387 (LXVD2X_V1TI): Likewise.
388 (STXVD2X_V1TI): Likewise.
389 (VEC_INIT_V1TI): Likewise.
390 (VEC_SET_V1TI): Likewise.
391 (VEC_EXT_V1TI): Likewise.
392 (EQV_V1TI): Likewise.
393 (NAND_V1TI): Likewise.
394 (ORC_V1TI): Likewise.
395 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
396 added in ISA 2.07. Add both normal 'altivec' builtins, and the
400 (VADDEUQM): Likewise.
401 (VADDECUQ): Likewise.
402 (VSUBEUQM): Likewise.
403 (VSUBECUQ): Likewise.
405 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
406 __int128_t and __uint128_t types.
407 (__uint128_type): Likewise.
408 (altivec_categorize_keyword): Add support for vector __int128_t,
409 vector __uint128_t, vector __int128, and vector unsigned __int128
410 as a container type for TImode operations that need to be done in
411 VSX/Altivec registers.
412 (rs6000_macro_to_expand): Likewise.
413 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
414 to support 128-bit integer instructions vaddcuq, vadduqm,
415 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
416 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
418 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
419 for V1TImode, and set up preferences to use VSX/Altivec registers.
420 Setup VSX reload handlers.
421 (rs6000_debug_reg_global): Likewise.
422 (rs6000_init_hard_regno_mode_ok): Likewise.
423 (rs6000_preferred_simd_mode): Likewise.
424 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
425 (easy_altivec_constant): Likewise.
426 (output_vec_const_move): Likewise.
427 (rs6000_expand_vector_set): Convert V1TImode set and extract to
429 (rs6000_expand_vector_extract): Likewise.
430 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
432 (rs6000_const_vec): Add support for V1TImode.
433 (rs6000_emit_le_vsx_load): Swap double words when loading or
434 storing TImode/V1TImode.
435 (rs6000_emit_le_vsx_store): Likewise.
436 (rs6000_emit_le_vsx_move): Likewise.
437 (rs6000_emit_move): Add support for V1TImode.
438 (altivec_expand_ld_builtin): Likewise.
439 (altivec_expand_st_builtin): Likewise.
440 (altivec_expand_vec_init_builtin): Likewise.
441 (altivec_expand_builtin): Likewise.
442 (rs6000_init_builtins): Add support for V1TImode type. Add
443 support for ISA 2.07 128-bit integer builtins. Define type names
444 for the VSX/Altivec vector types.
445 (altivec_init_builtins): Add support for overloaded vector
446 functions with V1TImode type.
447 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
448 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
450 (rs6000_split_128bit_ok_p): Likewise.
451 (rs6000_handle_altivec_attribute): Create V1TImode from vector
452 __int128_t and vector __uint128_t.
454 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
462 (VS_scalar): Likewise.
463 (VS_double): Likewise.
464 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
466 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
467 we support the ISA 2.07 128-bit integer arithmetic instructions.
468 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
469 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
470 and TImode types for use with the builtin functions.
471 (V1TI_type_node): Likewise.
472 (unsigned_V1TI_type_node): Likewise.
473 (intTI_type_internal_node): Likewise.
474 (uintTI_type_internal_node): Likewise.
476 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
477 128-bit builtin functions.
478 (UNSPEC_VADDEUQM): Likewise.
479 (UNSPEC_VADDECUQ): Likewise.
480 (UNSPEC_VSUBCUQ): Likewise.
481 (UNSPEC_VSUBEUQM): Likewise.
482 (UNSPEC_VSUBECUQ): Likewise.
483 (VM): Add V1TImode to vector mode iterators.
486 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
487 (altivec_vaddcuq): Likewise.
488 (altivec_vsubuqm): Likewise.
489 (altivec_vsubcuq): Likewise.
490 (altivec_vaddeuqm): Likewise.
491 (altivec_vaddecuq): Likewise.
492 (altivec_vsubeuqm): Likewise.
493 (altivec_vsubecuq): Likewise.
495 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
497 (BOOL_128): Likewise.
498 (BOOL_REGS_OUTPUT): Likewise.
499 (BOOL_REGS_OP1): Likewise.
500 (BOOL_REGS_OP2): Likewise.
501 (BOOL_REGS_UNARY): Likewise.
502 (BOOL_REGS_AND_CR0): Likewise.
504 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
505 128-bit integer builtin support.
506 (vec_vadduqm): Likewise.
507 (vec_vaddecuq): Likewise.
508 (vec_vaddeuqm): Likewise.
509 (vec_vsubecuq): Likewise.
510 (vec_vsubeuqm): Likewise.
511 (vec_vsubcuq): Likewise.
512 (vec_vsubuqm): Likewise.
514 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
515 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
516 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
517 128-bit integer add/subtract to ISA 2.07.
519 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
521 * config/arc/arc.c (arc_predicate_delay_insns):
522 Fix third argument passed to conditionalize_nonjump.
524 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
526 * config/aarch64/aarch64-builtins.c
527 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
528 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
529 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
530 instead of __builtin_lfloor.
531 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
533 2014-03-12 Jakub Jelinek <jakub@redhat.com>
535 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
536 (tree_ssa_ifcombine_bb_1): New function.
537 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
538 is an empty forwarder block to then_bb or vice versa and then_bb
539 and else_bb are effectively swapped.
541 2014-03-12 Christian Bruel <christian.bruel@st.com>
544 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
545 REG_CFA_DEF_CFA note.
546 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
547 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
549 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
551 PR tree-optimization/60454
552 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
554 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
556 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
557 Do not define target_cpu_default2 to generic.
558 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
559 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
560 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
562 2014-03-12 Jakub Jelinek <jakub@redhat.com>
563 Marc Glisse <marc.glisse@inria.fr>
565 PR tree-optimization/60502
566 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
567 instead of build_low_bits_mask.
569 2014-03-12 Jakub Jelinek <jakub@redhat.com>
572 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
573 if there are multiple uses, but op doesn't live on E edge.
574 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
575 clobber stmts before __builtin_unreachable.
577 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
579 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
580 hard_frame_pointer_rtx.
581 * cse.c (cse_insn): Remove volatile check.
582 * cselib.c (cselib_process_insn): Likewise.
583 * dse.c (scan_insn): Likewise.
585 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
587 * config/arc/arc.c (conditionalize_nonjump): New function,
589 (arc_ifcvt): ... this.
590 (arc_predicate_delay_insns): Use it.
592 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
594 * config/arc/predicates.md (extend_operand): During/after reload,
595 allow const_int_operand.
596 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
597 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
598 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
600 (umulsi3_highpart_i): Likewise.
602 2014-03-11 Richard Biener <rguenther@suse.de>
604 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
605 Add asserts to guard possible wrong-code bugs.
607 2014-03-11 Richard Biener <rguenther@suse.de>
609 PR tree-optimization/60429
610 PR tree-optimization/60485
611 * tree-ssa-structalias.c (set_union_with_increment): Properly
612 take into account all fields that overlap the shifted vars.
613 (do_sd_constraint): Likewise.
614 (do_ds_constraint): Likewise.
615 (get_constraint_for_ptr_offset): Likewise.
617 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
619 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
620 (nios2_compute_frame_layout):
621 Add calculation of cfun->machine->fp_save_offset.
622 (nios2_expand_prologue): Correct setting of frame pointer register
624 (nios2_expand_epilogue): Update recovery of stack pointer from
625 frame pointer accordingly.
626 (nios2_initial_elimination_offset): Update calculation of offset
627 for eliminating to HARD_FRAME_POINTER_REGNUM.
629 2014-03-10 Jakub Jelinek <jakub@redhat.com>
632 * ipa.c (symtab_remove_unreachable_nodes): Don't call
633 cgraph_get_create_node on VAR_DECLs.
635 2014-03-10 Richard Biener <rguenther@suse.de>
638 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
640 2014-03-08 Douglas B Rupp <rupp@gnat.com>
642 * config/vms/vms.opt (vms_float_format): New variable.
644 2014-03-08 Tobias Burnus <burnus@net-b.de>
646 * doc/invoke.texi (-fcilkplus): Update implementation status.
648 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
649 Richard Biener <rguenther@suse.de>
651 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
652 consistently accross all TUs.
653 (run_gcc): Enable -fshort-double automatically at link at link-time
654 and disallow override.
656 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
659 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
660 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
661 if they can't be used.
663 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
665 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
666 for Solaris 11/x86 ld.
667 * configure: Regenerate.
669 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
671 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
672 (LIB_TLS_SPEC): Save as ld_tls_libs.
673 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
674 (HAVE_AS_IX86_TLSLDM): New test.
675 * configure, config.in: Regenerate.
676 * config/i386/i386.c (legitimize_tls_address): Fall back to
677 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
678 cannot support TLS_MODEL_LOCAL_DYNAMIC.
679 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
680 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
682 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
684 * common.opt (fira-loop-pressure): Mark as optimization.
686 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
688 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
689 an OpenMP mappable type.
691 2014-03-06 Matthias Klose <doko@ubuntu.com>
693 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
694 MULTILIB_OSDIRNAMES is not defined.
696 2014-03-06 Jakub Jelinek <jakub@redhat.com>
697 Meador Inge <meadori@codesourcery.com>
700 * config/arm/arm.c (arm_tls_symbol_p): Remove.
701 (arm_legitimize_address): Call legitimize_tls_address for any
702 arm_tls_referenced_p expression, handle constant addend. Call it
703 before testing for !TARGET_ARM.
704 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
706 2014-03-06 Richard Biener <rguenther@suse.de>
712 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
714 * tree-streamer.c (record_common_node): Assert we don't record
715 nodes with type double.
716 (preload_common_node): Skip type double, complex double and double
717 pointer since it is now frontend dependent due to fshort-double option.
719 2014-03-06 Richard Biener <rguenther@suse.de>
721 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
722 or -fno-lto is specified and the linker has full plugin support.
723 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
724 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
725 * lto-wrapper.c (merge_and_complain): Merge compile-time
727 (run_gcc): And pass it through to the link options.
729 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
733 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
735 * cselib.c (remove_useless_values): Skip to avoid quadratic
736 behavior if the condition moved from...
737 (cselib_process_insn): ... here holds.
739 2014-03-05 Jakub Jelinek <jakub@redhat.com>
742 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
743 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
746 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
747 (TM_H): Add x86-tune.def.
749 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
751 * config/aarch64/aarch64.c (generic_tunings):
752 Use cortexa57_extra_costs.
754 2014-03-05 Jakub Jelinek <jakub@redhat.com>
757 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
758 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
759 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
762 2014-03-04 Heiher <r@hev.cc>
764 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
765 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
767 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
769 * config/i386/predicates.md (const2356_operand): Change to ...
770 (const2367_operand): ... this.
771 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
773 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
774 (*avx512pf_scatterpf<mode>sf): Ditto.
775 (avx512pf_scatterpf<mode>df): Ditto.
776 (*avx512pf_scatterpf<mode>df_mask): Ditto.
777 (*avx512pf_scatterpf<mode>df): Ditto.
778 * config/i386/i386.c (ix86_expand_builtin): Update
779 incorrect hint operand error message.
781 2014-03-04 Richard Biener <rguenther@suse.de>
783 * lto-section-in.c (lto_get_section_data): Fix const cast.
785 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
787 * tree-streamer.c (record_common_node): Assert we don't record
788 nodes with type double.
789 (preload_common_node): Skip type double, complex double and double
790 pointer since it is now frontend dependent due to fshort-double option.
792 2014-03-04 Richard Biener <rguenther@suse.de>
795 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
796 (lto_input_toplevel_asms): Likewise.
797 * lto-section-in.c (lto_get_section_data): Instead do it here
800 2014-03-04 Richard Biener <rguenther@suse.de>
802 PR tree-optimization/60382
803 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
804 dead PHIs a reduction.
806 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
808 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
810 (_mm_prefetch): Move out of GCC target("sse") pragma.
811 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
812 GCC target("prfchw") pragma.
813 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
815 * config/i386/i386.c (ix86_option_override_internal): Enable
816 -mprfchw with -mprefetchwt1.
818 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
820 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
823 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
825 * opts.h (CL_PCH_IGNORE): Define.
826 * targhooks.c (option_affects_pch_p):
827 Return false for options that have CL_PCH_IGNORE set.
828 * opt-functions.awk: Process PchIgnore.
829 * doc/options.texi: Document PchIgnore.
831 * config/arc/arc.opt (misize): Add PchIgnore property.
833 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
835 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
836 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
837 constraint on constants to permit them being loaded into
838 GENERAL_REGS or BASE_REGS.
840 2014-03-03 Nick Clifton <nickc@redhat.com>
842 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
843 anti-cacnonical alternatives.
844 (negandhi3_real): New pattern.
845 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
847 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
849 * config/avr/avr-mcus.def: Remove atxmega16x1.
850 * config/avr/avr-tables.opt: Regenerate.
851 * config/avr/t-multilib: Regenerate.
852 * doc/avr-mmcu.texi: Regenerate.
854 2014-03-03 Tobias Grosser <tobias@grosser.es>
855 Mircea Namolaru <mircea.namolaru@inria.fr>
857 PR tree-optimization/58028
858 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
861 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
863 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
864 not handled by recognizers.
866 2014-03-03 Jakub Jelinek <jakub@redhat.com>
869 * function.c (expand_function_end): Don't emit
870 clobber_return_register sequence if clobber_after is a BARRIER.
871 * cfgexpand.c (construct_exit_block): Append instructions before
872 return_label to prev_bb.
874 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
876 * config/rs6000/constraints.md: Document reserved use of "wc".
878 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
881 * ipa.c (function_and_variable_visibility): When dissolving comdat
882 group, also set all symbols to local.
884 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
889 2013-12-14 Jan Hubicka <jh@suse.cz>
891 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
893 2014-03-02 Jon Beniston <jon@beniston.com>
899 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
900 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
901 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
902 (simple_return, *simple_return): New patterns
903 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
904 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
906 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
908 * dwarf2out.c (gen_subprogram_die): Tidy.
910 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
913 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
914 (*mov_t_msb_neg_negc): ... this new insn.
916 2014-02-28 Jason Merrill <jason@redhat.com>
919 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
922 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
925 * dwarf2out.c (decltype_auto_die): New static.
926 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
927 (gen_type_die_with_usage): Handle 'decltype(auto)'.
928 (is_cxx_auto): Likewise.
930 2014-02-28 Ian Bolton <ian.bolton@arm.com>
932 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
933 we are not using general regs only.
935 2014-02-28 Richard Biener <rguenther@suse.de>
938 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
939 previous fix and only allow to remove trivial pre-headers
940 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
941 (remove_forwarder_block): Properly update the latch of a loop.
943 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
946 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
947 (cselib_preserved_hash_table): New.
948 (preserve_constants_and_equivs): Move preserved vals to it.
949 (cselib_find_slot): Look it up first.
950 (cselib_init): Initialize it.
951 (cselib_finish): Release it.
952 (dump_cselib_table): Dump it.
954 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
957 * cselib.c (remove_useless_values): Skip to avoid quadratic
958 behavior if the condition moved from...
959 (cselib_process_insn): ... here holds.
961 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
964 * var-tracking.c (vt_initialize): Apply the same condition to
965 preserve the CFA base value.
967 2014-02-28 Joey Ye <joey.ye@arm.com>
970 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
971 if reload in progress or completed.
973 2014-02-28 Tobias Burnus <burnus@net-b.de>
976 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
979 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
981 * doc/tm.texi.in (Condition Code Status): Update documention for
982 relative locations of cc0-setter and cc0-user.
984 2014-02-27 Jeff Law <law@redhat.com>
986 PR rtl-optimization/52714
987 * combine.c (try_combine): When splitting an unrecognized PARALLEL
988 into two independent simple sets, if I3 is a jump, ensure the
989 pattern we place into I3 is a (set (pc) ...).
991 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
992 Jeff Law <law@redhat.com>
994 PR rtl-optimization/49847
995 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
996 are in different blocks.
997 * doc/tm.texi (Condition Code Status): Update documention for
998 relative locations of cc0-setter and cc0-user.
1000 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1003 * lra.c (lra_emit_add): Check SUBREG too.
1005 2014-02-27 Andreas Schwab <schwab@suse.de>
1007 * config/m68k/m68k.c (m68k_option_override): Disable
1008 -flive-range-shrinkage for classic m68k.
1009 (m68k_override_options_after_change): Likewise.
1011 2014-02-27 Marek Polacek <polacek@redhat.com>
1014 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1015 -Wmaybe-uninitialized.
1017 2014-02-27 Alan Modra <amodra@gmail.com>
1020 * reload1.c (emit_input_reload_insns): When reload_override_in,
1021 set old to rl->in_reg when rl->in_reg is a subreg.
1023 2014-02-26 Richard Biener <rguenther@suse.de>
1026 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1028 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1030 * common/config/i386/predicates.md (const1256_operand): Remove.
1031 (const2356_operand): New.
1032 (const_1_to_2_operand): Remove.
1033 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1034 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1035 (*avx512pf_gatherpf<mode>sf): Ditto.
1036 (avx512pf_gatherpf<mode>df): Ditto.
1037 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1038 (*avx512pf_gatherpf<mode>df): Ditto.
1039 (avx512pf_scatterpf<mode>sf): Ditto.
1040 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1041 (*avx512pf_scatterpf<mode>sf): Ditto.
1042 (avx512pf_scatterpf<mode>df): Ditto.
1043 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1044 (*avx512pf_scatterpf<mode>df): Ditto.
1045 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1047 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1049 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1050 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1051 (_mm512_mask_testn_epi64_mask): Move to ...
1052 * config/i386/avx512cdintrin.h: Here.
1053 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1054 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1055 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1056 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1057 TARGET_AVX512F from TARGET_AVX512CD.
1059 2014-02-26 Richard Biener <rguenther@suse.de>
1062 * ipa.c (walk_polymorphic_call_targets): Properly guard
1063 call to inline_update_overall_summary.
1065 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1068 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1069 and latches only if requested. Fix latch if it is removed.
1070 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1071 LOOPS_HAVE_PREHEADERS.
1073 2014-02-25 Andrew Pinski <apinski@cavium.com>
1075 * builtins.c (expand_builtin_thread_pointer): Create a new target
1076 when the target is NULL.
1078 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1080 PR rtl-optimization/60317
1081 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1082 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1083 * lra-assigns.c: Include params.h.
1084 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1085 other reload pseudos considerations.
1087 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1089 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1090 to use canonical form for nor<mode>3.
1092 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1095 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1098 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1100 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1101 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1102 (ix86_handle_option): Handle OPT_mprefetchwt1.
1103 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1104 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1106 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1107 OPTION_MASK_ISA_PREFETCHWT1.
1108 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1109 (PTA_PREFETCHWT1): New.
1110 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1111 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1112 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1113 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1114 (*prefetch_avx512pf_<mode>_: Change into ...
1115 (*prefetch_prefetchwt1_<mode>: This.
1116 * config/i386/i386.opt (mprefetchwt1): New.
1117 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1118 (_mm_prefetch): Handle intent to write.
1119 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1121 2014-02-25 Richard Biener <rguenther@suse.de>
1124 * emit-rtl.c (mem_attrs_htab): Remove.
1125 (mem_attrs_htab_hash): Likewise.
1126 (mem_attrs_htab_eq): Likewise.
1127 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1128 (init_emit_once): Do not allocate mem_attrs_htab.
1130 2014-02-25 Richard Biener <rguenther@suse.de>
1133 * lto-opts.c (lto_write_options): Output non-explicit conservative
1134 -fwrapv, -fno-trapv and -fno-strict-overflow.
1135 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1136 (run_gcc): And pass them through.
1138 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1140 * sel-sched.c (calculate_new_fences): New parameter ptime.
1141 Calculate it as a maximum over all fence cycles.
1142 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1143 Print the final schedule timing when sched_verbose.
1145 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1147 PR rtl-optimization/60292
1148 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1149 bit fot the fence instruction.
1151 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1153 * calls.h: Fix typo in comment.
1155 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1157 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1158 adjusting offsetable addresses.
1160 2014-02-24 Guozhi Wei <carrot@google.com>
1162 * sparseset.h (sparseset_pop): Fix the wrong index.
1164 2014-02-24 Walter Lee <walt@tilera.com>
1166 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1167 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1169 * common/config/tilegx/tilegx-common.c
1170 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1171 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1173 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1174 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1175 (tilegx_gimplify_va_arg_expr): Handle big endian.
1176 (tilegx_expand_unaligned_load): Ditto.
1177 (tilegx_expand_unaligned_store): Ditto.
1178 (TARGET_RETURN_IN_MSB): New.
1179 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1180 (TARGET_ENDIAN_DEFAULT): New.
1181 (TARGET_BIG_ENDIAN): Handle big endian.
1182 (BYTES_BIG_ENDIAN): Ditto.
1183 (WORDS_BIG_ENDIAN): Ditto.
1184 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1187 * config/tilegx/tilegx.md (extv): Handle big endian.
1189 (insn_st<n>): Ditto.
1190 (insn_st<n>_add<bitsuffix>): Ditto.
1191 (insn_stnt<n>): Ditto.
1192 (insn_stnt<n>_add<bitsuffix>):Ditto.
1193 (vec_interleave_highv8qi): Handle big endian.
1194 (vec_interleave_highv8qi_be): New.
1195 (vec_interleave_highv8qi_le): New.
1196 (insn_v1int_h): Handle big endian.
1197 (vec_interleave_lowv8qi): Handle big endian.
1198 (vec_interleave_lowv8qi_be): New.
1199 (vec_interleave_lowv8qi_le): New.
1200 (insn_v1int_l): Handle big endian.
1201 (vec_interleave_highv4hi): Handle big endian.
1202 (vec_interleave_highv4hi_be): New.
1203 (vec_interleave_highv4hi_le): New.
1204 (insn_v2int_h): Handle big endian.
1205 (vec_interleave_lowv4hi): Handle big endian.
1206 (vec_interleave_lowv4hi_be): New.
1207 (vec_interleave_lowv4hi_le): New.
1208 (insn_v2int_l): Handle big endian.
1209 (vec_interleave_highv2si): Handle big endian.
1210 (vec_interleave_highv2si_be): New.
1211 (vec_interleave_highv2si_le): New.
1212 (insn_v4int_h): Handle big endian.
1213 (vec_interleave_lowv2si): Handle big endian.
1214 (vec_interleave_lowv2si_be): New.
1215 (vec_interleave_lowv2si_le): New.
1216 (insn_v4int_l): Handle big endian.
1217 * config/tilegx/tilegx.opt (mbig-endian): New option.
1218 (mlittle-endian): New option.
1219 * doc/install.texi: Document tilegxbe-linux.
1220 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1222 2014-02-24 Martin Jambor <mjambor@suse.cz>
1225 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1226 there are no parameter descriptors.
1228 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1230 PR rtl-optimization/60268
1231 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1232 initialization to ...
1233 (sched_rgn_init): ... here.
1234 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1236 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1238 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1241 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1243 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1246 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1248 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1249 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1251 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1253 * config/microblaze/predicates.md: Add cmp_op predicate.
1254 * config/microblaze/microblaze.md: Add branch_compare instruction
1255 which uses cmp_op predicate and emits cmp insn before branch.
1256 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1257 to microblaze_expand_conditional_branch and consolidate logic.
1258 (microblaze_expand_conditional_branch): emit branch_compare
1259 insn instead of handling cmp op separate from branch insn.
1261 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1263 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1266 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1268 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1269 define_insn with define_expand and new define_insn
1270 *altivec_lve<VI_char>x_internal.
1271 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1272 and new define_insn *altivec_stve<VI_char>x_internal.
1273 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1275 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1277 (altivec_expand_stvex_be): New function.
1279 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1281 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1282 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1283 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1284 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1286 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1289 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1290 instead of emit_move_insn.
1292 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1294 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1296 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1297 gen_altivec_vsumsws.
1299 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1301 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1302 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1303 (altivec_lvxl_<mode>): New define_expand incorporating
1304 -maltivec=be semantics where needed.
1305 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1306 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1307 semantics where needed.
1308 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1309 (altivec_stvx_<mode>): New define_expand incorporating
1310 -maltivec=be semantics where needed.
1311 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1312 VM2 iterator instead of V4SI.
1313 (altivec_stvxl_<mode>): New define_expand incorporating
1314 -maltivec=be semantics where needed.
1315 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1316 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1317 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1318 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1319 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1320 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1321 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1322 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1323 ALTIVEC_BUILTIN_STVXL.
1324 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1325 (altivec_expand_stvx_be): Likewise.
1326 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1327 (altivec_expand_lvx_be): Likewise.
1328 (altivec_expand_stvx_be): Likewise.
1329 (altivec_expand_builtin): Add cases for
1330 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1331 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1332 (altivec_init_builtins): Add definitions for
1333 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1334 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1336 2014-02-21 Catherine Moore <clm@codesourcery.com>
1338 * doc/invoke.texi (mvirt, mno-virt): Document.
1339 * config/mips/mips.opt (mvirt): New option.
1340 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1342 2014-02-21 Richard Biener <rguenther@suse.de>
1344 PR tree-optimization/60276
1345 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1346 (STMT_VINFO_MIN_NEG_DIST): New macro.
1347 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1348 STMT_VINFO_MIN_NEG_DIST.
1349 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1350 made for negative dependence distances still hold.
1352 2014-02-21 Richard Biener <rguenther@suse.de>
1355 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1356 DECL_INITIAL for globals not in the current function context.
1358 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1360 PR tree-optimization/56490
1361 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1362 * tree-ssa-uninit.c: Include params.h.
1363 (compute_control_dep_chain): Add num_calls argument, return false
1364 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1365 num_calls to recursive call.
1366 (find_predicates): Change dep_chain into normal array,
1367 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1368 variable and adjust compute_control_dep_chain caller.
1369 (find_def_preds): Likewise.
1371 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1373 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1374 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1376 2014-02-21 Nick Clifton <nickc@redhat.com>
1378 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1379 (pushhi1): Likewise.
1380 (popqi1): Add mode to pre_dec.
1383 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1385 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1386 mode for mask of V8SFmode permutation.
1388 2014-02-20 Richard Henderson <rth@redhat.com>
1391 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1392 a new pseudo for OLDVAL.
1394 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1397 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1398 gen_reg_rtx if d->testing_p.
1399 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1400 if d->testing_p and we will certainly return true.
1401 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1404 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1406 * emit-rtl.c (gen_reg_rtx): Assert that
1407 crtl->emit.regno_pointer_align_length is non-zero.
1409 2014-02-20 Richard Henderson <rth@redhat.com>
1412 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1413 on failure the store back into EXPECT.
1415 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1416 Sandra Loosemore <sandra@codesourcery.com>
1418 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1419 * config/nios2/nios2.c (nios2_function_profiler): Add
1420 -fPIC (flag_pic == 2) support.
1421 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1422 (nios2_large_offset_p): New function.
1423 (nios2_unspec_reloc_p): Move up position, update to use
1424 nios2_large_offset_p.
1425 (nios2_unspec_address): Remove function.
1426 (nios2_unspec_offset): New function.
1427 (nios2_large_got_address): New function.
1428 (nios2_got_address): Add large offset support.
1429 (nios2_legitimize_tls_address): Update usage of removed and new
1431 (nios2_symbol_binds_local_p): New function.
1432 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1433 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1434 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1435 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1436 processing for (const (unspec ...)).
1437 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1439 2014-02-20 Richard Biener <rguenther@suse.de>
1441 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1442 doing the substitution.
1443 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1445 2014-02-20 Martin Jambor <mjambor@suse.cz>
1448 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1449 info when checking whether lattices are bottom.
1451 2014-02-20 Richard Biener <rguenther@suse.de>
1454 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1457 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1460 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1461 parameter specifying the scaling.
1462 (inline_call): Update.
1463 (want_inline_recursively): Guard division by zero.
1464 (recursive_inlining): Update.
1465 * ipa-inline.h (clone_inlined_nodes): Update.
1467 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1470 * config/i386/i386.c (classify_argument): Pass structures of size
1471 64 bytes or less in register.
1473 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1474 Kirill Yukhin <kirill.yukhin@intel.com>
1476 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1477 (_mm_rcp28_round_ss): Ditto.
1478 (_mm_rsqrt28_round_sd): Ditto.
1479 (_mm_rsqrt28_round_ss): Ditto.
1480 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1481 (_mm_rcp14_round_ss): Ditto.
1482 (_mm_rsqrt14_round_sd): Ditto.
1483 (_mm_rsqrt14_round_ss): Ditto.
1484 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1485 the first input operand, get rid of match_dup.
1486 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1488 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1490 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1491 operand as the first input operand, set type attribute.
1492 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1494 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1495 operand as the first input operand, set type attribute.
1497 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1499 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1502 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1505 * config/i386/i386.c (construct_container): Remove TFmode check
1506 for X86_64_INTEGER_CLASS.
1508 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1511 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1512 only when -Wpsabi is enabled.
1514 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1517 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1518 passing arrays in registers are the same as for structs, so remove the
1519 special case for them.
1521 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1523 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1524 destination type, extract only the valid bits if the source type is not
1525 integral and has a different mode.
1527 2014-02-19 Richard Biener <rguenther@suse.de>
1530 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1533 2014-02-19 Richard Biener <rguenther@suse.de>
1536 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1537 (ipa_modify_call_arguments): Emit an argument load explicitely and
1538 preserve virtual SSA form there and for the replacement call.
1539 Do not update SSA form nor free dominance info.
1541 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1543 * ipa.c (function_and_variable_visibility): Also clear WEAK
1544 flag when disolving COMDAT_GROUP.
1546 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1548 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1549 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1551 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1552 do more sanity checks.
1553 (detect_type_change): Return true when giving up early.
1554 (compute_complex_assign_jump_func): Fix type parameter of
1555 ipa_set_ancestor_jf.
1556 (compute_complex_ancestor_jump_func): Likewise.
1557 (update_jump_functions_after_inlining): Fix updating of
1559 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1561 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1563 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1564 inline clones when edge disappears.
1566 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1569 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1570 Split 64-bit moves into 2 patterns. Do not allow the use of
1571 direct move for TDmode in little endian, since the decimal value
1572 has little endian bytes within a word, but the 64-bit pieces are
1573 ordered in a big endian fashion, and normal subreg's of TDmode are
1575 (mov<mode>_64bit_dm): Likewise.
1576 (movtd_64bit_nodm): Likewise.
1578 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1580 PR tree-optimization/60174
1581 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1582 statement of an SSA_NAME that occurs in an abnormal PHI node.
1584 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1587 * final.c (SEEN_BB): Remove.
1588 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1589 (final_scan_insn): Don't force_source_line on second
1590 NOTE_INSN_BASIC_BLOCK.
1592 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1595 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1596 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1597 (type_natural_mode): Warn ABI change when %zmm register is not
1598 available for AVX512F vector value passing.
1600 2014-02-18 Kai Tietz <ktietz@redhat.com>
1603 * config/i386/i386.c (ix86_expand_prologue): Use value in
1604 rax register as displacement when restoring %r10 or %rax.
1605 Fix wrong offset when restoring both registers.
1607 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1609 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1610 assertion with conditional return.
1612 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1613 Uros Bizjak <ubizjak@gmail.com>
1616 * config/i386/driver-i386.c (host_detect_local_cpu): If
1617 YMM state is not saved by the OS, also clear has_f16c. Move
1618 CPUID 0x80000001 handling before YMM state saving checking.
1620 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1622 PR rtl-optimization/58960
1623 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1624 factored out from ...
1625 (sched_init): ... here.
1626 (free_global_sched_pressure_data): New, factored out from ...
1627 (sched_finish): ... here.
1628 * sched-int.h (free_global_sched_pressure_data): Declare.
1629 * sched-rgn.c (nr_regions_initial): New static global.
1630 (haifa_find_rgns): Initialize it.
1631 (schedule_region): Disable sched-pressure for the newly
1634 2014-02-17 Richard Biener <rguenther@suse.de>
1636 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1637 release SSA defs of pattern stmts.
1639 2014-02-17 Richard Biener <rguenther@suse.de>
1641 * tree-inline.c (expand_call_inline): Release the virtual
1642 operand defined by the call we are about to inline.
1644 2014-02-17 Richard Biener <rguenther@suse.de>
1646 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1648 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1649 Ilya Tocar <ilya.tocar@intel.com>
1651 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1652 arguments order in builtin.
1653 (_mm512_permutexvar_epi64): Ditto.
1654 (_mm512_mask_permutexvar_epi64): Ditto
1655 (_mm512_maskz_permutexvar_epi32): Ditto
1656 (_mm512_permutexvar_epi32): Ditto
1657 (_mm512_mask_permutexvar_epi32): Ditto
1659 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1661 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1662 (p8_vmrgow): Likewise.
1664 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1666 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1669 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1672 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1673 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1674 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1675 using direct move instructions on ISA 2.07. Also adjust
1676 instruction length for 64-bit.
1677 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1678 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1680 2014-02-15 Alan Modra <amodra@gmail.com>
1684 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1685 find_replacement on parts of insn rtl that might be reloaded.
1687 2014-02-15 Richard Biener <rguenther@suse.de>
1689 PR tree-optimization/60183
1690 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1691 (tree_ssa_phiprop): Calculate and free post-dominators.
1693 2014-02-14 Jeff Law <law@redhat.com>
1695 PR rtl-optimization/60131
1696 * ree.c (get_extended_src_reg): New function.
1697 (combine_reaching_defs): Use it rather than assuming location of REG.
1698 (find_and_remove_re): Verify first operand of extension is
1699 a REG before adding the insns to the copy list.
1701 2014-02-14 Roland McGrath <mcgrathr@google.com>
1703 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1704 * configure: Regenerated.
1705 * config.in: Regenerated.
1706 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1707 instead of ASM_SHORT.
1709 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1710 Richard Earnshaw <rearnsha@arm.com>
1712 PR rtl-optimization/59535
1713 * lra-constraints.c (process_alt_operands): Encourage alternative
1714 when unassigned pseudo class is superset of the alternative class.
1715 (inherit_reload_reg): Don't inherit when optimizing for code size.
1716 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1717 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1718 modes not less than 4 for Thumb1.
1720 2014-02-14 Kyle McMartin <kyle@redhat.com>
1723 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1725 2014-02-14 Richard Biener <rguenther@suse.de>
1727 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1728 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1729 Do not strip INDIRECT_REFs.
1731 2014-02-14 Richard Biener <rguenther@suse.de>
1734 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1735 DECL_FUNCTION_SPECIFIC_TARGET.
1736 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1737 * tree-streamer-out.c (pack_ts_target_option): Remove.
1738 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1739 (write_ts_function_decl_tree_pointers): Do not stream
1740 DECL_FUNCTION_SPECIFIC_TARGET.
1741 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1742 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1743 (lto_input_ts_function_decl_tree_pointers): Do not stream
1744 DECL_FUNCTION_SPECIFIC_TARGET.
1746 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1748 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1749 (get_initial_def_for_induction, vectorizable_induction): Ignore
1750 debug stmts when looking for exit_phi.
1751 (vectorizable_live_operation): Fix up condition.
1753 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1755 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1756 nreverse() because it changes the content of original tree list.
1758 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1760 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1761 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1763 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1765 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1766 GNU coding standards.
1768 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1771 * dwarf2out.c (gen_subprogram_die): Don't call
1772 add_calling_convention_attribute if subr_die is old_die.
1774 2014-02-13 Sharad Singhai <singhai@google.com>
1776 * doc/optinfo.texi: Fix order of nodes.
1778 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1780 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1781 operands[2], not operands[3].
1783 2014-02-13 Richard Biener <rguenther@suse.de>
1786 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1787 mention the possibility of an in-tree build.
1788 (CLooG): Update recommended version to 0.18.1, mention the
1789 possibility of an in-tree build and clarify that the ISL
1790 bundled with CLooG does not work.
1792 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1795 * expr.c (compress_float_constant): If x is a hard register,
1796 extend into a pseudo and then move to x.
1798 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1800 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1801 caused by bad second argument to warning_at() with -mhotpatch and
1802 nested functions (e.g. with gfortran).
1804 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1806 * opts.c (option_name): Remove "enabled by default" rider.
1808 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1810 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1812 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1813 Uros Bizjak <ubizjak@gmail.com>
1816 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1817 * configure: Regenerated.
1819 2014-02-12 Richard Biener <rguenther@suse.de>
1821 * vec.c (vec_prefix::calculate_allocation): Move as
1822 inline variant to vec.h.
1823 (vec_prefix::calculate_allocation_1): New out-of-line version.
1824 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1825 (vec_prefix::m_has_auto_buf): Rename to ...
1826 (vec_prefix::m_using_auto_storage): ... this.
1827 (vec_prefix::calculate_allocation): Inline the easy cases
1828 and dispatch to calculate_allocation_1 which doesn't need the
1830 (va_heap::reserve): Use gcc_checking_assert.
1831 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1832 m_using_auto_storage.
1833 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1835 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1836 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1837 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1839 2014-02-12 Richard Biener <rguenther@suse.de>
1841 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1842 when we found a dependence.
1844 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1846 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1848 (maybe_fold_stmt): ... into this new function.
1849 * omp-low.c (lower_omp): Update comment.
1851 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1854 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1857 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1859 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1860 identifiers in comments.
1861 (cortexa53_extra_costs): Likewise.
1862 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1863 (cortexa7_extra_costs): Likewise.
1864 (cortexa12_extra_costs): Likewise.
1865 (cortexa15_extra_costs): Likewise.
1866 (v7m_extra_costs): Likewise.
1868 2014-02-12 Richard Biener <rguenther@suse.de>
1871 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1872 of posix_memalign being successful.
1873 (lower_stmt): Restrict lowering of posix_memalign to when
1874 -ftree-bit-ccp is enabled.
1876 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1878 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1880 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1882 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1884 PR rtl-optimization/60116
1885 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1886 other_insn once the combination has been validated.
1888 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1891 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1893 * ipa-devirt.c: Include demangle.h
1894 (odr_violation_reported): New static variable.
1895 (add_type_duplicate): Update odr_violations.
1896 (maybe_record_node): Add completep parameter; update it.
1897 (record_target_from_binfo): Add COMPLETEP parameter;
1898 update it as needed.
1899 (possible_polymorphic_call_targets_1): Likewise.
1900 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1901 rename FINAL to COMPLETE.
1902 (record_targets_from_bases): Sanity check we found the binfo;
1903 fix COMPLETEP updating.
1904 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1905 parameter, fix computing of COMPLETEP.
1906 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1907 at LTO time do demangling.
1908 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1909 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1911 (gimple_get_virt_method_for_binfo): Likewise.
1912 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1913 gimple_get_virt_method_for_vtable): Update prototypes.
1915 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1918 * genautomata.c (add_presence_absence): Fix typo with
1919 {final_}presence_list.
1921 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1924 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1925 for VSX/Altivec vectors that land in GPR registers.
1927 2014-02-11 Richard Henderson <rth@redhat.com>
1928 Jakub Jelinek <jakub@redhat.com>
1931 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1932 around drhs if type conversion to lacc->type is not useless.
1934 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1936 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1938 (cortex-a57.cortex-a53): Likewise.
1939 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1941 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1943 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1946 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1948 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1949 add_options_for_arm_vfp3.
1951 2014-02-11 Jeff Law <law@redhat.com>
1954 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1955 object with an undesirable mode.
1957 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1960 * config/i386/sol2-9.h: New file.
1961 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1962 *-*-solaris2.9*): Use it.
1964 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1966 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1967 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1969 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1971 * config/microblaze/microblaze.c: Extend mcpu version format
1973 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1975 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1977 2014-02-10 Richard Henderson <rth@redhat.com>
1980 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1981 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1982 ms-abi vs -mno-accumulate-outgoing-args.
1983 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1984 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1987 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1990 * cfgexpand.c (expand_asm_operands): Attach source location to
1991 ASM_INPUT rtx objects.
1992 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1994 2014-02-10 Nick Clifton <nickc@redhat.com>
1996 * config/mn10300/mn10300.c (popcount): New function.
1997 (mn10300_expand_prologue): Include saved registers in stack usage
2000 2014-02-10 Jeff Law <law@redhat.com>
2003 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2004 when changing the SET_DEST of a prior insn to avoid an input reload.
2006 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2008 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2009 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2010 -mcall-openbsd, or -mcall-linux.
2011 (CC1_ENDIAN_BIG_SPEC): Remove.
2012 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2013 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2014 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2015 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2016 and %cc1_endian_default.
2017 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2019 2014-02-10 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/60115
2022 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2023 MEM_REF handling. Properly verify that the accesses are not
2024 out of the objects bound.
2026 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2028 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2031 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2033 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2034 proper constants and fix formatting.
2035 (possible_polymorphic_call_targets): Fix formatting.
2037 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2038 Ilya Tocar <ilya.tocar@intel.com>
2040 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2041 (_mm512_loadu_epi32): Renamed into...
2042 (_mm512_loadu_si512): This.
2043 (_mm512_storeu_epi32): Renamed into...
2044 (_mm512_storeu_si512): This.
2045 (_mm512_maskz_ceil_ps): Removed.
2046 (_mm512_maskz_ceil_pd): Ditto.
2047 (_mm512_maskz_floor_ps): Ditto.
2048 (_mm512_maskz_floor_pd): Ditto.
2049 (_mm512_floor_round_ps): Ditto.
2050 (_mm512_floor_round_pd): Ditto.
2051 (_mm512_ceil_round_ps): Ditto.
2052 (_mm512_ceil_round_pd): Ditto.
2053 (_mm512_mask_floor_round_ps): Ditto.
2054 (_mm512_mask_floor_round_pd): Ditto.
2055 (_mm512_mask_ceil_round_ps): Ditto.
2056 (_mm512_mask_ceil_round_pd): Ditto.
2057 (_mm512_maskz_floor_round_ps): Ditto.
2058 (_mm512_maskz_floor_round_pd): Ditto.
2059 (_mm512_maskz_ceil_round_ps): Ditto.
2060 (_mm512_maskz_ceil_round_pd): Ditto.
2061 (_mm512_expand_pd): Ditto.
2062 (_mm512_expand_ps): Ditto.
2063 * config/i386/i386.c (ix86_builtins): Remove
2064 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2065 (bdesc_args): Ditto.
2066 * config/i386/predicates.md (const1256_operand): New.
2067 (const_1_to_2_operand): Ditto.
2068 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2069 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2070 (*avx512pf_gatherpf<mode>sf): Ditto.
2071 (avx512pf_gatherpf<mode>df): Ditto.
2072 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2073 (*avx512pf_gatherpf<mode>df): Ditto.
2074 (avx512pf_scatterpf<mode>sf): Ditto.
2075 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2076 (*avx512pf_scatterpf<mode>sf): Ditto.
2077 (avx512pf_scatterpf<mode>df): Ditto.
2078 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2079 (*avx512pf_scatterpf<mode>df): Ditto.
2080 (avx512f_expand<mode>): Removed.
2081 (<shift_insn><mode>3<mask_name>): Change predicate type.
2083 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2085 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2086 not at the end of datarefs vector use ordered_remove to avoid
2087 reordering datarefs vector.
2090 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2091 mark local addressable non-static vars as GOVD_PRIVATE
2092 instead of GOVD_LOCAL.
2093 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2094 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2098 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2099 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2100 assume_aligned or alloc_align attributes.
2101 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2102 arguments. Handle also assume_aligned and alloc_align attributes.
2103 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2104 calls to functions with assume_aligned or alloc_align attributes.
2105 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2107 2014-02-08 Terry Guo <terry.guo@arm.com>
2109 * doc/invoke.texi: Document ARM -march=armv7e-m.
2111 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2113 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2114 flag on __cilkrts_rethrow builtin.
2117 * ipa-cp.c (determine_versionability): Fail at -O0
2118 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2119 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2122 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2125 * tree-inline.c (copy_forbidden): Fail for
2126 __attribute__((optimize (0))) functions.
2128 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2130 * varpool.c: Include pointer-set.h.
2131 (varpool_remove_unreferenced_decls): Variables in other partitions
2132 will not be output; be however careful to not lose information
2135 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2137 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2138 lookup in the vtable constructor.
2140 2014-02-07 Jeff Law <law@redhat.com>
2143 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2144 define_insn_and_split.
2146 * ipa-inline.c (inline_small_functions): Fix typos.
2148 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2150 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2151 (s390_can_use_return_insn): Declare.
2152 * config/s390/s390.h (EPILOGUE_USES): Define.
2153 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2155 (s390_chunkify_start): Handle return JUMP_LABELs.
2156 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2157 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2158 (s390_can_use_return_insn): New functions.
2159 (s390_fix_long_loop_prediction): Handle conditional returns.
2160 (TARGET_SET_UP_BY_PROLOGUE): Define.
2161 * config/s390/s390.md (ANY_RETURN): New code iterator.
2162 (*creturn, *csimple_return, return, simple_return): New patterns.
2164 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2166 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2167 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2168 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2169 REG_CFA_RESTORE list when deciding not to restore a register.
2171 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2173 * config/s390/s390.c: Include tree-pass.h and context.h.
2174 (s390_early_mach): New function, split out from...
2175 (s390_emit_prologue): ...here.
2176 (pass_data_s390_early_mach): New pass structure.
2177 (pass_s390_early_mach): New class.
2178 (s390_option_override): Create and register early_mach pass.
2179 Move to end of file.
2181 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2183 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2184 to match for the exit block.
2186 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2188 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2189 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2190 Reject misaligned operands.
2192 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2194 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2196 2014-02-07 Richard Biener <rguenther@suse.de>
2199 * gimple-low.c (lower_builtin_posix_memalign): New function.
2200 (lower_stmt): Call it to lower posix_memalign in a way
2201 to make alignment info accessible.
2203 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2206 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2207 __builtin_setjmp_receiver.
2209 2014-02-07 Richard Biener <rguenther@suse.de>
2212 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2213 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2214 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2215 Handle BUILT_IN_POSIX_MEMALIGN.
2216 (find_func_clobbers): Likewise.
2217 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2218 (call_may_clobber_ref_p_1): Likewise.
2220 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2223 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2226 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2229 * lto-cgraph.c (lto_output_node): Use
2230 symtab_get_symbol_partitioning_class.
2231 (lto_output_varpool_node): likewise.
2232 (symtab_get_symbol_partitioning_class): Move here from
2234 * cgraph.h (symbol_partitioning_class): Likewise.
2235 (symtab_get_symbol_partitioning_class): Declare.
2237 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2239 * ggc.h (ggc_internal_cleared_alloc): New macro.
2240 * vec.h (vec_safe_copy): Handle memory stats.
2241 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2242 * target-globals.c (save_target_globals): Likewise.
2244 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2247 * expr.c (emit_move_resolve_push): Export; be bit more selective
2248 on when to clear alias set.
2249 * expr.h (emit_move_resolve_push): Declare.
2250 * function.h (struct function): Add tail_call_marked.
2251 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2252 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2253 * config/i386/i386.md (TImode move expander): De not call
2255 (FP push expanders): Preserve memory attributes.
2256 * config/i386/sse.md (push<mode>1): Remove.
2257 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2258 (ix86_expand_push): Remove.
2259 * config/i386/mmx.md (push<mode>1): Remove.
2261 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2263 PR rtl-optimization/60030
2264 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2265 lopart with paradoxical subreg before shifting it up by hprec.
2267 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2269 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2270 Remove extra newline at end of file.
2271 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2272 (arm_issue_rate): Handle cortexa57.
2273 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2274 (cortex-a57.cortex-a53): Likewise.
2276 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2279 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2280 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2282 (arm_expand_prologue): Adjust all callers.
2283 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2284 info, registers also at the lowest numbered registers side. Use
2285 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2289 * var-tracking.c (adjust_mems): Before adding a SET to
2290 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2292 2014-02-06 Alan Modra <amodra@gmail.com>
2295 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2296 change SDmode to DDmode when lra_in_progress.
2298 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2301 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2302 free_data_ref on the dr first, and before goto again also set dr
2303 to the next dr. For simd_lane_access, free old datarefs[i] before
2304 overwriting it. For get_vectype_for_scalar_type failure, don't
2305 free_data_ref if simd_lane_access.
2307 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2310 * tree.h (opts_for_fn): New inline function.
2311 (opt_for_fn): Define.
2312 * config/i386/i386.c (ix86_function_regparm): Use
2313 opt_for_fn (decl, optimize) instead of optimize.
2315 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2317 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2318 for SYMBOL_REF in large memory model.
2320 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2322 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2324 (cortex-a57): Likewise.
2325 (cortex-a57.cortex-a53): Likewise.
2327 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2328 Kugan Vivekanandarajah <kuganv@linaro.org>
2330 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2332 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2334 2014-02-06 Richard Biener <rguenther@suse.de>
2336 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2337 set_loop_copy and initialize_original_copy_tables.
2339 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2341 * config/aarch64/aarch64-simd.md
2342 (aarch64_ashr_simddi): Change QI to SI.
2344 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2345 Jakub Jelinek <jakub@redhat.com>
2348 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2351 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2353 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2354 CODE_FOR_altivec_vpku[hw]um to
2355 CODE_FOR_altivec_vpku[hw]um_direct.
2356 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2357 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2358 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2359 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2361 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2363 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2364 generation for -maltivec=be.
2365 (altivec_vsumsws): Simplify redundant test.
2367 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2369 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2370 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2371 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2372 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2373 gen_altivec_vpkuwum.
2374 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2376 (altivec_vpks<VI_char>ss): Likewise.
2377 (altivec_vpks<VI_char>us): Likewise.
2378 (altivec_vpku<VI_char>us): Likewise.
2379 (altivec_vpku<VI_char>um): Likewise.
2380 (altivec_vpku<VI_char>um_direct): New (copy of
2381 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2383 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2384 target is little endian and -maltivec=be is not specified.
2385 (*altivec_vupkhs<VU_char>_direct): New (copy of
2386 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2387 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2388 target is little endian and -maltivec=be is not specified.
2389 (*altivec_vupkls<VU_char>_direct): New (copy of
2390 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2391 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2392 little endian and -maltivec=be is not specified.
2393 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2394 little endian and -maltivec=be is not specified.
2396 2014-02-05 Richard Henderson <rth@redhat.com>
2399 * combine-stack-adj.c: Revert r206943.
2400 * sched-int.h (struct deps_desc): Add last_args_size.
2401 * sched-deps.c (init_deps): Initialize it.
2402 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2403 contain REG_ARGS_SIZE notes.
2405 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2407 * lto-cgraph.c (asm_nodes_output): Make global.
2408 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2409 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2410 (driver_handle_option): Handle OPT_fwpa.
2412 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2415 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2416 a comment typo and formatting issue. If odr_hash hasn't been
2417 created, return vNULL and set *completep to false.
2420 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2421 bb with no successors.
2423 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2426 * doc/invoke.texi (-march): Clarify documentation for ARM.
2430 2014-02-05 Richard Biener <rguenther@suse.de>
2432 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2433 when not vectorizing because of too many alias checks.
2434 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2435 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2437 2014-02-05 Nick Clifton <nickc@redhat.com>
2439 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2440 accept extended registers in any mode when compiling for the MN10300.
2442 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2444 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2445 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2446 sanitization attributes.
2447 (can_inline_edge_p): Likewise.
2448 (sanitize_attrs_match_for_inline_p): New function.
2450 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2452 * ipa-prop.c (detect_type_change): Shor circuit testing of
2453 type changes on THIS pointer.
2455 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2458 * config/pa/pa.c (legitimize_tls_address): Return original address
2459 if not passed a SYMBOL_REF rtx.
2460 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2462 (pa_emit_move_sequence): Simplify TLS source operands.
2463 (pa_legitimate_constant_p): Reject all TLS constants.
2464 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2465 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2467 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2469 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2470 groups when we know they are controlled by LTO.
2471 * varasm.c (default_binds_local_p_1): If object is in other partition,
2472 it will be resolved locally.
2474 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2476 * config/host-linux.c (linux_gt_pch_use_address): Don't
2477 use SSIZE_MAX because it is not always defined.
2479 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2482 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2483 threshold for pseudo splitting.
2484 (update_ebb_live_info): Process call argument hard registers and
2485 hard registers from insn definition too.
2486 (max_small_class_regs_num): New constant.
2487 (inherit_in_ebb): Update live hard regs through EBBs. Update
2488 reloads_num only for small register classes. Don't split for
2491 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2494 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2497 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2499 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2502 2014-02-04 Marek Polacek <polacek@redhat.com>
2504 * gdbinit.in (pel): Define.
2506 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2508 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2511 2014-02-04 Richard Biener <rguenther@suse.de>
2514 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2515 in function context local.
2516 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2517 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2518 similar to LTO_imported_decl_ref.
2520 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2522 PR tree-optimization/60002
2523 * cgraphclones.c (build_function_decl_skip_args): Clear
2526 PR tree-optimization/60023
2527 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2528 false to gsi_replace.
2529 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2530 has been in some EH region and vec_stmt could throw, add
2531 vec_stmt into the same EH region.
2532 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2533 has no lhs, ignore it.
2534 * internal-fn.c (expand_MASK_LOAD): Likewise.
2537 * tree-inline.c (copy_forbidden): Fail for
2538 __attribute__((optimize (0))) functions.
2541 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2542 is set, copy one less argument.
2543 (expand_simd_clones): Don't subtract clone_info->inbranch
2544 from simd_clone_struct_alloc argument.
2546 PR rtl-optimization/57915
2547 * recog.c (simplify_while_replacing): If all unary/binary/relational
2548 operation arguments are constant, attempt to simplify those.
2551 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2552 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2554 2014-02-04 Richard Biener <rguenther@suse.de>
2556 PR tree-optimization/60012
2557 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2558 TBAA disambiguation to all DDRs.
2560 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2563 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2564 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2566 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2569 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2571 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2573 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2574 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2576 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2579 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2580 to figure out targets of polymorphic calls with known decl.
2581 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2582 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2583 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2584 (get_polymorphic_call_info): ... here.
2585 (get_polymorphic_call_info_from_invariant): New function.
2587 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2589 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2590 lookup via vtable pointer; check for type consistency
2591 and turn inconsitent facts into UNREACHABLE.
2592 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2593 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2594 type inconsistent querries; return UNREACHABLE instead.
2596 2014-02-03 Richard Henderson <rth@twiddle.net>
2599 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2600 already processed this node.
2601 (normalize_one_pred_1): Pass along mark_set.
2602 (normalize_one_pred): Create and destroy a pointer_set_t.
2603 (normalize_one_pred_chain): Likewise.
2605 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2607 PR gcov-profile/58602
2608 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2610 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2613 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2614 -fno-devirtualize; try to devirtualize by the knowledge of
2615 virtual table pointer given by aggregate propagation.
2616 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2617 (ipa_print_node_jump_functions): Dump also offset that
2618 is relevant for polymorphic calls.
2619 (determine_known_aggregate_parts): Add arg_type parameter; use it
2620 instead of determining the type from pointer type.
2621 (ipa_compute_jump_functions_for_edge): Update call of
2622 determine_known_aggregate_parts.
2623 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2624 (gimple_get_virt_method_for_binfo): ... here; simplify using
2625 vtable_pointer_value_to_vtable.
2626 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2627 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2628 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2629 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2631 (vtable_pointer_value_to_binfo): ... here.
2632 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2634 2014-02-03 Teresa Johnson <tejohnson@google.com>
2636 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2637 redef of outer loop index variable.
2639 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2643 * doc/extend.texi (Function Attributes): Typo.
2645 2014-02-03 Cong Hou <congh@google.com>
2647 PR tree-optimization/60000
2648 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2649 if the vectorized statement is a store. A store statement can only
2650 appear at the end of pattern statements.
2652 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2654 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2655 (ix86_option_override_internal): Default long double to 64-bit for
2656 32-bit Bionic and to 128-bit for 64-bit Bionic.
2658 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2659 TARGET_LONG_DOUBLE_128 is true.
2660 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2662 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2663 (mlong-double-64): Negate -mlong-double-128.
2664 (mlong-double-128): New option.
2666 * config/i386/i386-c.c (ix86_target_macros): Define
2667 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2669 * doc/invoke.texi: Document -mlong-double-128.
2671 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2673 PR rtl-optimization/60024
2674 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2676 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2678 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2680 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2682 PR rtl-optimization/57662
2683 * sel-sched.c (code_motion_path_driver): Do not mark already not
2684 existing blocks in the visiting bitmap.
2686 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2688 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2689 on the insn being emitted.
2691 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2692 Will Deacon <will.deacon@arm.com>
2694 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2696 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2698 * config/arm/arm-tables.opt: Regenerate.
2700 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2702 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2703 for vector types other than V16QImode.
2704 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2705 define_expand, and call altivec_expand_vec_perm_le when producing
2706 code with little endian element order.
2707 (*altivec_vperm_<mode>_internal): New insn having previous
2708 behavior of altivec_vperm_<mode>.
2709 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2710 altivec_expand_vec_perm_le when producing code with little endian
2712 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2713 behavior of altivec_vperm_<mode>_uns.
2715 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2717 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2718 (altivec_vsumsws): Add handling for -maltivec=be with a little
2720 (altivec_vsumsws_direct): New.
2721 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2722 gen_altivec_vsumsws.
2724 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2726 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2727 vtable_pointer_value_to_binfo): New functions.
2728 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2729 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2731 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2733 * config/nios2/nios2.md (load_got_register): Initialize GOT
2734 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2735 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2737 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2739 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2740 preserverd by passthrough, do not propagate the type.
2742 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2744 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2745 (mips_atomic_assign_expand_fenv): New function.
2746 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2748 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2750 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2751 (__builtin_mips_set_fcsr): Likewise.
2752 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2753 MIPS_USI_FTYPE_VOID.
2754 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2755 (mips16_expand_set_fcsr): Likewise.
2756 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2757 (mips16_set_fcsr_stub): Likewise.
2758 (mips16_get_fcsr_one_only_stub): New class.
2759 (mips16_set_fcsr_one_only_stub): Likewise.
2760 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2761 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2762 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2763 (hard_float): New availability predicate.
2764 (mips_builtins): Add get_fcsr and set_fcsr.
2765 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2766 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2767 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2768 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2769 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2772 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2774 * config/mips/mips.c (mips_one_only_stub): New class.
2775 (mips_need_mips16_rdhwr_p): Replace with...
2776 (mips16_rdhwr_stub): ...this new variable.
2777 (mips16_stub_call_address): New function.
2778 (mips16_rdhwr_one_only_stub): New class.
2779 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2780 (mips_output_mips16_rdhwr): Delete.
2781 (mips_finish_stub): New function.
2782 (mips_code_end): Use it to handle rdhwr stubs.
2784 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2787 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2788 when calculating size of integer atomic types.
2790 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2792 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2794 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2796 PR tree-optimization/60003
2797 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2798 * profile.c (branch_prob): Use gimple_call_builtin_p
2799 to check for BUILT_IN_SETJMP_RECEIVER.
2800 * tree-inline.c (copy_bb): Call notice_special_calls.
2802 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2805 * lra-constraints.c (process_alt_operands): Update reload_sum only
2808 2014-01-31 Richard Henderson <rth@redhat.com>
2811 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2812 until after else_eh is processed.
2814 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2816 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2817 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2818 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2819 in smmintrin.h, remove them.
2820 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2821 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2822 * config/i386/i386.md (ROUND_SAE): Fix value.
2823 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2824 (const48_operand): New.
2825 * config/i386/subst.md (round), (round_expand): Use
2826 const_4_or_8_to_11_operand.
2827 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2829 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2831 * config/i386/constraints.md (Yk): Swap meaning with k.
2832 * config/i386/i386.md (movhi_internal): Change Yk to k.
2833 (movqi_internal): Ditto.
2834 (*k<logic><mode>): Ditto.
2837 (kandn<mode>): Ditto.
2838 (*<code>hi_1): Ditto.
2839 (*<code>qi_1): Ditto.
2840 (kxnor<mode>): Ditto.
2841 (kortestzhi): Ditto.
2842 (kortestchi): Ditto.
2844 (*one_cmplhi2_1): Ditto.
2845 (*one_cmplqi2_1): Ditto.
2846 * config/i386/sse.md (): Change k to Yk.
2847 (avx512f_load<mode>_mask): Ditto.
2848 (avx512f_blendm<mode>): Ditto.
2849 (avx512f_store<mode>_mask): Ditto.
2850 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2851 (avx512f_storedqu<mode>_mask): Ditto.
2852 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2854 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2855 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2856 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2857 (avx512f_maskcmp<mode>3): Ditto.
2858 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2859 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2860 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2861 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2862 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2863 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2864 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2865 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2866 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2867 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2868 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2869 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2870 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2871 (vec_extract_lo_<mode>_maskm): Ditto.
2872 (vec_extract_hi_<mode>_maskm): Ditto.
2873 (avx512f_vternlog<mode>_mask): Ditto.
2874 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2875 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2876 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2877 (avx512f_<code>v8div16qi2_mask): Ditto.
2878 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2879 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2880 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2881 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2882 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2883 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2884 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2885 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2886 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2887 (avx512cd_maskb_vec_dupv8di): Ditto.
2888 (avx512cd_maskw_vec_dupv16si): Ditto.
2889 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2890 (avx512f_vpermi2var<mode>3_mask): Ditto.
2891 (avx512f_vpermi2var<mode>3_mask): Ditto.
2892 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2893 (*avx512f_gathersi<mode>): Ditto.
2894 (*avx512f_gathersi<mode>_2): Ditto.
2895 (*avx512f_gatherdi<mode>): Ditto.
2896 (*avx512f_gatherdi<mode>_2): Ditto.
2897 (*avx512f_scattersi<mode>): Ditto.
2898 (*avx512f_scatterdi<mode>): Ditto.
2899 (avx512f_compress<mode>_mask): Ditto.
2900 (avx512f_compressstore<mode>_mask): Ditto.
2901 (avx512f_expand<mode>_mask): Ditto.
2902 * config/i386/subst.md (mask): Change k to Yk.
2903 (mask_scalar_merge): Ditto.
2906 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2908 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2910 2014-01-31 Richard Biener <rguenther@suse.de>
2913 * builtins.c (fold_builtin_memory_op): Make sure to not
2914 use a floating-point mode or a boolean or enumeral type for
2917 2014-01-30 DJ Delorie <dj@redhat.com>
2919 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2920 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2921 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2922 whenever main() has an epilogue.
2924 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2926 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2927 unused variable "field".
2928 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2929 (vsx_mergeh_<mode>): Likewise.
2930 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2931 (altivec_vmrghh): Likewise.
2932 (altivec_vmrghw): Likewise.
2933 (altivec_vmrglb): Likewise.
2934 (altivec_vmrglh): Likewise.
2935 (altivec_vmrglw): Likewise.
2936 (altivec_vspltb): Add missing uses.
2937 (altivec_vsplth): Likewise.
2938 (altivec_vspltw): Likewise.
2939 (altivec_vspltsf): Likewise.
2941 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2944 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2945 frame related instructions.
2947 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2949 PR rtl-optimization/59959
2950 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2951 any reload of register whose subreg is invalid.
2953 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2955 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2956 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2957 Add missing return type - void.
2959 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2961 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2962 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2963 remove element index adjustment for endian (now handled in vsx.md
2965 (altivec_expand_vec_perm_const): Use
2966 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2967 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2968 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2969 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2970 define_expand and a new define_insn *altivec_vspltb_internal;
2971 adjust for -maltivec=be on a little endian target.
2972 (altivec_vspltb_direct): New.
2973 (altivec_vsplth): Divide into a define_expand and a new
2974 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2975 little endian target.
2976 (altivec_vsplth_direct): New.
2977 (altivec_vspltw): Divide into a define_expand and a new
2978 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2979 little endian target.
2980 (altivec_vspltw_direct): New.
2981 (altivec_vspltsf): Divide into a define_expand and a new
2982 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2983 a little endian target.
2985 2014-01-30 Richard Biener <rguenther@suse.de>
2987 PR tree-optimization/59993
2988 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2989 can propagate form the earlier stmt and avoid the transform
2990 when the intermediate result is needed.
2992 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2994 * README.Portability: Fix typo.
2996 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2998 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2999 comparison_operator with ordered_comparison_operator.
3001 2014-01-30 Nick Clifton <nickc@redhat.com>
3003 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3004 Rename to mn10300_store_multiple_regs.
3005 * config/mn10300/mn10300.c: Likewise.
3006 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3007 store_multiple_regs.
3008 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3009 Call mn10300_store_multiple_regs.
3011 2014-01-30 Nick Clifton <nickc@redhat.com>
3012 DJ Delorie <dj@redhat.com>
3014 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3015 %fp 2 to keep registers after it properly word-aligned.
3016 (rl78_alloc_physical_registers_umul): Handle the case where both
3017 input operands are the same.
3019 2014-01-30 Richard Biener <rguenther@suse.de>
3021 PR tree-optimization/59903
3022 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3025 2014-01-30 Jason Merrill <jason@redhat.com>
3028 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3031 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3033 2014-01-30 Richard Biener <rguenther@suse.de>
3035 PR tree-optimization/59951
3036 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3038 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3041 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3042 SFmode to DFmode case.
3044 2014-01-29 DJ Delorie <dj@redhat.com>
3046 * config/msp430/msp430.opt (-minrt): New.
3047 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3049 (ENDFILE_SPEC): Likewise.
3051 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3053 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3054 (estimate_function_body_sizes): Use it.
3056 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3059 * dwarf2out.c (is_cxx_auto): New.
3060 (is_base_type): Use it.
3061 (gen_type_die_with_usage): Likewise.
3063 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3065 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3066 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3067 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3068 -maltivec=be with LE targets.
3069 (vsx_mergeh_<mode>): Likewise.
3070 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3071 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3072 (altivec_vmrghb): Replace with define_expand and new
3073 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3074 (altivec_vmrghb_direct): New define_insn.
3075 (altivec_vmrghh): Replace with define_expand and new
3076 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3077 (altivec_vmrghh_direct): New define_insn.
3078 (altivec_vmrghw): Replace with define_expand and new
3079 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3080 (altivec_vmrghw_direct): New define_insn.
3081 (*altivec_vmrghsf): Adjust for endianness.
3082 (altivec_vmrglb): Replace with define_expand and new
3083 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3084 (altivec_vmrglb_direct): New define_insn.
3085 (altivec_vmrglh): Replace with define_expand and new
3086 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3087 (altivec_vmrglh_direct): New define_insn.
3088 (altivec_vmrglw): Replace with define_expand and new
3089 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3090 (altivec_vmrglw_direct): New define_insn.
3091 (*altivec_vmrglsf): Adjust for endianness.
3092 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3093 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3094 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3095 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3096 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3097 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3098 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3099 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3101 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3103 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3104 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3107 2014-01-29 Richard Biener <rguenther@suse.de>
3109 PR tree-optimization/58742
3110 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3111 associate_pointerplus_align.
3112 (associate_pointerplus_diff): New function.
3113 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3114 and associate_pointerplus_diff.
3116 2014-01-29 Richard Biener <rguenther@suse.de>
3118 * lto-streamer.h (LTO_major_version): Bump to 3.
3119 (LTO_minor_version): Reset to 0.
3121 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3123 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3124 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3125 (arm_file_start): Generate correct asm header for armv7ve.
3126 * config/arm/bpabi.h: Add multilib support for armv7ve.
3127 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3128 and cortex-a15 to armv7ve.
3129 * config/arm/t-aprofile: Add multilib support for armv7ve.
3130 * doc/invoke.texi: Document -march=armv7ve.
3132 2014-01-29 Richard Biener <rguenther@suse.de>
3134 PR tree-optimization/58742
3135 * tree-ssa-forwprop.c (associate_plusminus): Return true
3136 if we changed sth, defer EH cleanup to ...
3137 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3138 (simplify_mult): New function.
3140 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3143 PR tree-optimization/59920
3144 * tree.c (build_common_builtin_nodes): Remove
3145 __builtin_setjmp_dispatcher initialization.
3146 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3147 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3148 instead of gsi_after_labels + manually skipping debug stmts.
3149 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3150 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3151 * tree-inline.c (copy_edges_for_bb): Remove
3152 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3153 argument. Ignore computed_goto_p stmts. Don't call
3154 make_abnormal_goto_edges. If a call might need abnormal edges
3155 for non-local gotos, see if it already has an edge to
3156 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3157 with true argument, don't do anything then, otherwise add
3158 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3159 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3161 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3162 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3163 (lower_stmt): Don't set data->calls_builtin_setjmp.
3164 (lower_builtin_setjmp): Adjust comment.
3165 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3166 * tree-cfg.c (found_computed_goto): Remove.
3167 (factor_computed_gotos): Remove.
3168 (make_goto_expr_edges): Return bool, true for computed gotos.
3169 Don't call make_abnormal_goto_edges.
3170 (build_gimple_cfg): Don't set found_computed_goto, don't call
3171 factor_computed_gotos.
3172 (computed_goto_p): No longer static.
3173 (make_blocks): Don't set found_computed_goto.
3174 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3175 (make_edges): If make_goto_expr_edges returns true, push bb
3176 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3177 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3178 vector. Record mapping between bbs and OpenMP regions if there
3179 are any, adjust make_gimple_omp_edges caller. Call
3180 handle_abnormal_edges.
3181 (make_abnormal_goto_edges): Remove.
3182 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3183 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3184 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3185 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3186 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3187 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3188 filling *region also set *region_idx to (*region)->entry->index.
3191 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3192 For REGs set ORIGINAL_REGNO.
3194 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3196 * doc/md.texi: Mention that a target shouldn't implement
3197 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3200 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3202 PR tree-optimization/59594
3203 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3204 a copy of the datarefs vector rather than the vector itself.
3206 2014-01-28 Jason Merrill <jason@redhat.com>
3209 * dwarf2out.c (auto_die): New static.
3210 (gen_type_die_with_usage): Handle C++1y 'auto'.
3211 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3214 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3217 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3218 (SPEC_X32): Likewise.
3219 (SPEC_64): Likewise.
3220 * config/i386/i386.c (ix86_option_override_internal): Turn off
3221 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3223 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3224 * config/i386/i386.h (TARGET_16BIT): New macro.
3225 (TARGET_16BIT_P): Likewise.
3226 * config/i386/i386.opt: Add m16.
3227 * doc/invoke.texi: Document -m16.
3229 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3231 PR preprocessor/59935
3232 * input.c (location_get_source_line): Bail out on when line number
3233 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3235 2014-01-28 Richard Biener <rguenther@suse.de>
3237 PR tree-optimization/58742
3238 * tree-ssa-forwprop.c (associate_plusminus): Handle
3239 pointer subtraction of the form (T)(P + A) - (T)P.
3241 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3243 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3246 2014-01-28 Richard Biener <rguenther@suse.de>
3249 2014-01-28 Richard Biener <rguenther@suse.de>
3251 PR rtl-optimization/45364
3252 PR rtl-optimization/59890
3253 * var-tracking.c (local_get_addr_clear_given_value): Handle
3254 already cleared slot.
3255 (val_reset): Handle not allocated local_get_addr_cache.
3256 (vt_find_locations): Use post-order on the inverted CFG.
3258 2014-01-28 Richard Biener <rguenther@suse.de>
3260 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3262 2014-01-28 Richard Biener <rguenther@suse.de>
3264 PR rtl-optimization/45364
3265 PR rtl-optimization/59890
3266 * var-tracking.c (local_get_addr_clear_given_value): Handle
3267 already cleared slot.
3268 (val_reset): Handle not allocated local_get_addr_cache.
3269 (vt_find_locations): Use post-order on the inverted CFG.
3271 2014-01-28 Alan Modra <amodra@gmail.com>
3273 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3274 * configure.ac <recursive call for build != host>: Define
3275 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3276 and LD_FOR_BUILD too.
3277 * configure: Regenerate.
3279 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3281 * config/i386/i386.c (get_builtin_code_for_version): Separate
3282 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3283 Broadwell from Haswell.
3285 2014-01-27 Steve Ellcey <sellcey@mips.com>
3287 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3288 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3289 * config/mips/mips.c (mips_option_override): Change setting
3291 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3292 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3293 Change from Mask to Var.
3295 2014-01-27 Jeff Law <law@redhat.com>
3297 * ipa-inline.c (inline_small_functions): Fix typo.
3299 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3301 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3302 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3303 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3304 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3305 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3306 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3307 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3308 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3309 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3310 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3311 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3312 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3313 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3314 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3315 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3316 (_mm512_storeu_epi64): Ditto.
3317 (_mm512_cmpge_epi32_mask): Ditto.
3318 (_mm512_cmpge_epu32_mask): Ditto.
3319 (_mm512_cmpge_epi64_mask): Ditto.
3320 (_mm512_cmpge_epu64_mask): Ditto.
3321 (_mm512_cmple_epi32_mask): Ditto.
3322 (_mm512_cmple_epu32_mask): Ditto.
3323 (_mm512_cmple_epi64_mask): Ditto.
3324 (_mm512_cmple_epu64_mask): Ditto.
3325 (_mm512_cmplt_epi32_mask): Ditto.
3326 (_mm512_cmplt_epu32_mask): Ditto.
3327 (_mm512_cmplt_epi64_mask): Ditto.
3328 (_mm512_cmplt_epu64_mask): Ditto.
3329 (_mm512_cmpneq_epi32_mask): Ditto.
3330 (_mm512_cmpneq_epu32_mask): Ditto.
3331 (_mm512_cmpneq_epi64_mask): Ditto.
3332 (_mm512_cmpneq_epu64_mask): Ditto.
3333 (_mm512_expand_pd): Ditto.
3334 (_mm512_expand_ps): Ditto.
3335 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3336 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3337 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3338 * config/i386/i386.c (ix86_builtins): Add
3339 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3340 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3341 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3342 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3343 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3344 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3345 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3346 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3347 IX86_BUILTIN_PMOVUSQW512_MEM.
3348 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3349 __builtin_ia32_pmovsqd512mem_mask,
3350 __builtin_ia32_pmovqd512mem_mask,
3351 __builtin_ia32_pmovusqw512mem_mask,
3352 __builtin_ia32_pmovsqw512mem_mask,
3353 __builtin_ia32_pmovqw512mem_mask,
3354 __builtin_ia32_pmovusdw512mem_mask,
3355 __builtin_ia32_pmovsdw512mem_mask,
3356 __builtin_ia32_pmovdw512mem_mask,
3357 __builtin_ia32_pmovqb512mem_mask,
3358 __builtin_ia32_pmovusqb512mem_mask,
3359 __builtin_ia32_pmovsqb512mem_mask,
3360 __builtin_ia32_pmovusdb512mem_mask,
3361 __builtin_ia32_pmovsdb512mem_mask,
3362 __builtin_ia32_pmovdb512mem_mask.
3363 (bdesc_args): Add __builtin_ia32_expanddf512,
3364 __builtin_ia32_expandsf512.
3365 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3366 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3367 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3368 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3369 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3370 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3371 (avx512f_<code>v8div16qi2_mask_store): This.
3372 (avx512f_expand<mode>): New.
3374 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3376 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3378 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3379 (_mm512_prefetch_i32scatter_pd): Ditto.
3380 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3381 (_mm512_prefetch_i64scatter_pd): Ditto.
3382 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3383 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3384 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3385 (_mm512_prefetch_i32scatter_ps): Ditto.
3386 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3387 (_mm512_prefetch_i64scatter_ps): Ditto.
3388 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3389 * config/i386/i386-builtin-types.def: Define
3390 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3391 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3392 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3393 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3394 IX86_BUILTIN_SCATTERPFQPD.
3395 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3396 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3397 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3398 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3399 __builtin_ia32_scatterpfqps.
3400 (ix86_expand_builtin): Expand new built-ins.
3401 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3402 fix memory access data type.
3403 (*avx512pf_gatherpf<mode>_mask): Ditto.
3404 (*avx512pf_gatherpf<mode>): Ditto.
3405 (avx512pf_scatterpf<mode>): Ditto.
3406 (*avx512pf_scatterpf<mode>_mask): Ditto.
3407 (*avx512pf_scatterpf<mode>): Ditto.
3408 (GATHER_SCATTER_SF_MEM_MODE): New.
3409 (avx512pf_gatherpf<mode>df): Ditto.
3410 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3411 (*avx512pf_scatterpf<mode>df): Ditto.
3413 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3416 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3417 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3420 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3422 * common/config/arm/arm-common.c
3423 (arm_rewrite_mcpu): Handle multiple names.
3425 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3427 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3429 * gimple-builder.h (create_gimple_tmp): Delete.
3431 2014-01-27 Christian Bruel <christian.bruel@st.com>
3433 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3436 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3438 * config/pa/pa.md (call): Generate indirect long calls to non-local
3439 functions when outputing 32-bit code.
3440 (call_value): Likewise except for special call to buggy powf function.
3442 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3443 portable runtime and PIC indirect calls.
3444 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3445 and PIC call sequences. Use ldo instead of blr to set return register
3446 in PIC call sequence.
3448 2014-01-25 Walter Lee <walt@tilera.com>
3450 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3451 avoid clobbering a live register.
3453 2014-01-25 Walter Lee <walt@tilera.com>
3455 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3456 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3457 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3458 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3460 2014-01-25 Walter Lee <walt@tilera.com>
3462 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3463 arguments on even registers.
3464 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3466 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3467 (BIGGEST_ALIGNMENT): Ditto.
3468 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3470 2014-01-25 Walter Lee <walt@tilera.com>
3472 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3473 insns before bundling.
3474 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3476 2014-01-25 Walter Lee <walt@tilera.com>
3478 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3479 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3480 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3482 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3484 * config/mips/constraints.md (kl): Delete.
3485 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3486 define expands, using...
3487 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3488 instructions for MIPS16.
3489 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3490 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3492 2014-01-25 Walter Lee <walt@tilera.com>
3494 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3498 2014-01-25 Walter Lee <walt@tilera.com>
3500 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3501 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3503 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3505 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3508 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3510 * print-rtl.c (in_call_function_usage): New var.
3511 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3512 EXPR_LIST mode as mode and not as reg note name.
3515 * cfgloopmanip.c (copy_loop_info): If
3516 loop->warned_aggressive_loop_optimizations, make sure
3517 the flag is set in target loop too.
3519 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3521 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3523 * builtins.def: Likewise.
3524 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3525 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3526 * ira.c (ira_setup_eliminable_regset): Likewise.
3527 * omp-low.c (gate_expand_omp): Likewise.
3528 (execute_lower_omp): Likewise.
3529 (diagnose_sb_0): Likewise.
3530 (gate_diagnose_omp_blocks): Likewise.
3531 (simd_clone_clauses_extract): Likewise.
3534 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3536 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3537 correction for little endian...
3538 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3541 2014-01-24 Jeff Law <law@redhat.com>
3543 PR tree-optimization/59919
3544 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3545 for non-returning calls.
3547 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3549 * common/config/aarch64/aarch64-common.c
3550 (aarch64_rewrite_mcpu): Handle multiple names.
3551 * config/aarch64/aarch64.h
3552 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3554 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3556 * input.c (add_file_to_cache_tab): Handle the case where fopen
3559 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3562 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3563 from push operand if code of push isn't PRE_DEC.
3565 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3568 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3569 -mquad-memory-atomic. Update -mquad-memory documentation to say
3570 it is only used for non-atomic loads/stores.
3572 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3573 -mquad-memory or -mquad-memory-atomic switches.
3575 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3576 -mquad-memory-atomic to ISA 2.07 support.
3578 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3579 to separate support of normal quad word memory operations (ldq, stq)
3580 from the atomic quad word memory operations.
3582 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3583 support to separate non-atomic quad word operations from atomic
3584 quad word operations. Disable non-atomic quad word operations in
3585 little endian mode so that we don't have to swap words after the
3586 load and before the store.
3587 (quad_load_store_p): Add comment about atomic quad word support.
3588 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3589 options printed with -mdebug=reg.
3591 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3592 -mquad-memory-atomic as the test for whether we have quad word
3593 atomic instructions.
3594 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3595 or -mp8-vector are used, allow byte/half-word atomic operations.
3597 * config/rs6000/sync.md (load_lockedti): Insure that the address
3598 is a proper indexed or indirect address for the lqarx instruction.
3599 On little endian systems, swap the hi/lo registers after the lqarx
3601 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3602 insure the address is valid for the lqarx instruction.
3603 (store_conditionalti): Insure that the address is a proper indexed
3604 or indirect address for the stqcrx. instruction. On little endian
3605 systems, swap the hi/lo registers before doing the stqcrx.
3607 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3608 insure the address is valid for the stqcrx. instruction.
3610 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3611 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3612 type of quad memory support is available.
3614 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3617 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3618 there is a danger of looping.
3620 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3622 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3623 force flag_ira_loop_pressure if set via command line.
3625 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3627 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3628 (ashr_simd): New builtin handling DI mode.
3629 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3630 (aarch64_sshr_simddi): New match pattern.
3631 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3632 (vshrd_n_s64): Likewise.
3633 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3635 2014-01-23 Nick Clifton <nickc@redhat.com>
3637 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3638 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3639 favour of mcu specific scripts.
3640 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3643 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3644 Alex Velenko <Alex.Velenko@arm.com>
3646 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3647 (vaddv_s16): Likewise.
3648 (vaddv_s32): Likewise.
3649 (vaddv_u8): Likewise.
3650 (vaddv_u16): Likewise.
3651 (vaddv_u32): Likewise.
3652 (vaddvq_s8): Likewise.
3653 (vaddvq_s16): Likewise.
3654 (vaddvq_s32): Likewise.
3655 (vaddvq_s64): Likewise.
3656 (vaddvq_u8): Likewise.
3657 (vaddvq_u16): Likewise.
3658 (vaddvq_u32): Likewise.
3659 (vaddvq_u64): Likewise.
3660 (vaddv_f32): Likewise.
3661 (vaddvq_f32): Likewise.
3662 (vaddvq_f64): Likewise.
3663 (vmaxv_f32): Likewise.
3664 (vmaxv_s8): Likewise.
3665 (vmaxv_s16): Likewise.
3666 (vmaxv_s32): Likewise.
3667 (vmaxv_u8): Likewise.
3668 (vmaxv_u16): Likewise.
3669 (vmaxv_u32): Likewise.
3670 (vmaxvq_f32): Likewise.
3671 (vmaxvq_f64): Likewise.
3672 (vmaxvq_s8): Likewise.
3673 (vmaxvq_s16): Likewise.
3674 (vmaxvq_s32): Likewise.
3675 (vmaxvq_u8): Likewise.
3676 (vmaxvq_u16): Likewise.
3677 (vmaxvq_u32): Likewise.
3678 (vmaxnmv_f32): Likewise.
3679 (vmaxnmvq_f32): Likewise.
3680 (vmaxnmvq_f64): Likewise.
3681 (vminv_f32): Likewise.
3682 (vminv_s8): Likewise.
3683 (vminv_s16): Likewise.
3684 (vminv_s32): Likewise.
3685 (vminv_u8): Likewise.
3686 (vminv_u16): Likewise.
3687 (vminv_u32): Likewise.
3688 (vminvq_f32): Likewise.
3689 (vminvq_f64): Likewise.
3690 (vminvq_s8): Likewise.
3691 (vminvq_s16): Likewise.
3692 (vminvq_s32): Likewise.
3693 (vminvq_u8): Likewise.
3694 (vminvq_u16): Likewise.
3695 (vminvq_u32): Likewise.
3696 (vminnmv_f32): Likewise.
3697 (vminnmvq_f32): Likewise.
3698 (vminnmvq_f64): Likewise.
3700 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3702 * config/aarch64/aarch64-simd.md
3703 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3704 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3705 (*aarch64_mul3_elt<mode>): Likewise.
3706 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3707 (*aarch64_mul3_elt_to_64v2df): Likewise.
3708 (*aarch64_mla_elt<mode>): Likewise.
3709 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3710 (*aarch64_mls_elt<mode>): Likewise.
3711 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3712 (*aarch64_fma4_elt<mode>): Likewise.
3713 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3714 (*aarch64_fma4_elt_to_64v2df): Likewise.
3715 (*aarch64_fnma4_elt<mode>): Likewise.
3716 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3717 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3718 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3719 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3720 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3721 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3722 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3723 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3724 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3726 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3728 * config/aarch64/aarch64-simd.md
3729 (aarch64_be_checked_get_lane<mode>): New define_expand.
3730 * config/aarch64/aarch64-simd-builtins.def
3731 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3732 New builtin definition.
3733 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3734 Use new safe be builtin.
3736 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3738 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3740 (aarch64_be_st1<mode>): Likewise.
3741 (aarch_ld1<VALL:mode>): Define_expand modified.
3742 (aarch_st1<VALL:mode>): Likewise.
3743 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3744 (UNSPEC_ST1): Likewise.
3746 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3748 * config/microblaze/microblaze.md: Add trap insn and attribute
3750 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3752 PR preprocessor/58580
3753 * input.h (location_get_source_line): Take an additional line_size
3755 (void diagnostics_file_cache_fini): Declare new function.
3756 * input.c (struct fcache): New type.
3757 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3758 New static constants.
3759 (diagnostic_file_cache_init, total_lines_num)
3760 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3761 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3762 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3763 (get_next_line, read_next_line, goto_next_line, read_line_num):
3764 New static function definitions.
3765 (diagnostic_file_cache_fini): New function.
3766 (location_get_source_line): Take an additional output line_len
3767 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3769 * diagnostic.c (diagnostic_finish): Call
3770 diagnostic_file_cache_fini.
3771 (adjust_line): Take an additional input parameter for the length
3772 of the line, rather than calculating it with strlen.
3773 (diagnostic_show_locus): Adjust the use of
3774 location_get_source_line and adjust_line with respect to their new
3775 signature. While displaying a line now, do not stop at the first
3776 null byte. Rather, display the zero byte as a space and keep
3777 going until we reach the size of the line.
3778 * Makefile.in: Add vec.o to OBJS-libcommon
3780 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3781 Ilya Tocar <ilya.tocar@intel.com>
3783 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3784 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3785 (__builtin_ia32_kmov16): Ditto.
3786 * config/i386/i386.md (UNSPEC_KMOV): New.
3789 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3791 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3792 (_mm512_storeu_si512): Ditto.
3794 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3797 * rtl.h (get_referenced_operands): Declare.
3798 * recog.c (get_referenced_operands): New function.
3799 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3800 operands have been referenced when recording LO_SUM references.
3802 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3804 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3806 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3808 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3809 Enable for generic and recent AMD targets.
3811 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3813 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3814 ARG_SIZE note when adjustment was eliminated.
3816 2014-01-22 Jeff Law <law@redhat.com>
3818 PR tree-optimization/59597
3819 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3820 in file. Accept new argument REGISTERING and use it to modify
3821 dump output appropriately.
3822 (register_jump_thread): Corresponding changes.
3823 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3824 thread paths involving joiner blocks. Add code to dump cancelled
3825 jump threading paths.
3827 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3829 PR rtl-optimization/59477
3830 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3831 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3833 2014-01-22 Tom Tromey <tromey@redhat.com>
3835 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3837 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3839 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3841 PR rtl-optimization/59896
3842 * lra-constraints.c (process_alt_operands): Check unused note for
3843 matched operands of insn with no output reloads.
3845 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3847 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3848 (mips_move_from_gpr_cost): Likewise.
3850 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3852 PR rtl-optimization/59858
3853 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3854 ira_class_hard_regs_num.
3855 (process_alt_operands): Increase reject for dying matched operand.
3857 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3860 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3861 smaller than size, perform several stores or loads and stores
3862 at dst + count - size to store or copy all of size bytes, rather
3863 than just last modesize bytes.
3865 2014-01-20 DJ Delorie <dj@redhat.com>
3867 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3868 that CLOBBERs are REGs before propogating their values.
3870 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3873 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3874 (cgraph_inline_failed_type): New function.
3875 * cgraph.h (DEFCIFCODE): Add type.
3876 (cgraph_inline_failed_type_t): New enum.
3877 (cgraph_inline_failed_type): New prototype.
3878 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3879 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3880 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3881 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3882 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3883 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3884 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3885 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3886 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3887 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3888 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3889 OPTIMIZATION_MISMATCH.
3890 * tree-inline.c (expand_call_inline): Emit errors during
3891 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3893 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3896 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3897 mode attribute in insn output.
3899 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3901 * output.h (output_constant): Delete.
3902 * varasm.c (output_constant): Make private.
3904 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3906 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3908 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3911 * tree.h (fold_builtin_strcat): New prototype.
3912 * builtins.c (fold_builtin_strcat): No longer static. Add len
3913 argument, if non-NULL, don't call c_strlen. Optimize
3914 directly into __builtin_memcpy instead of __builtin_strcpy.
3915 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3916 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3918 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3920 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3921 for SImode_address_operand operands, having only a REG argument.
3923 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3925 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3926 loader name using mbig-endian.
3927 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3929 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3931 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3935 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3937 * config/aarch64/aarch64-protos.h
3938 (aarch64_cannot_change_mode_class_ptr): Declare.
3939 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3940 aarch64_cannot_change_mode_class_ptr): New.
3941 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3942 backend hook aarch64_cannot_change_mode_class.
3944 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3946 * common/config/aarch64/aarch64-common.c
3947 (aarch64_handle_option): Don't handle any option order logic here.
3948 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3949 selected_cpu, warn on architecture version mismatch.
3950 (aarch64_override_options): Fix parsing order for option strings.
3952 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3953 Iain Sandoe <iain@codesourcery.com>
3956 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3957 warning. Amend comment to reflect current functionality.
3959 2014-01-20 Richard Biener <rguenther@suse.de>
3962 * builtins.c (fold_builtin_strcat): Remove case better handled
3963 by tree-ssa-strlen.c.
3965 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3967 * config/aarch64/aarch64.opt
3968 (mcpu, march, mtune): Make case-insensitive.
3970 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3973 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3974 if operands[1] is a REG or ZERO_EXTEND of a REG.
3976 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3978 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3980 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3982 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3983 long non-pic millicode calls.
3985 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3987 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3989 2014-01-19 Kito Cheng <kito@0xlab.org>
3991 * builtins.c (expand_movstr): Check movstr expand done or fail.
3993 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3994 H.J. Lu <hongjiu.lu@intel.com>
3997 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3998 to DImode for zero-extended addresses.
4000 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4002 PR rtl-optimization/57763
4003 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4004 on the new indirect jump_insn and increment LABEL_NUSES (label).
4006 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4010 * config.gcc (x86_archs): New variable.
4011 (x86_64_archs): Likewise.
4012 (x86_cpus): Likewise.
4013 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4014 --with-arch/--with-cpu= options.
4015 Support --with-arch=/--with-cpu={nehalem,westmere,
4016 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4018 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4020 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4021 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4023 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4025 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4027 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4030 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4031 clear cpp_get_options (parse_in)->warn_unused_macros for
4032 ix86_target_macros_internal with cpp_define.
4034 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4036 * jump.c (delete_related_insns): Keep (use (insn))s.
4037 * reorg.c (redundant_insn): Check for barriers too.
4039 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4041 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4043 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4045 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4046 call to $$dyncall when TARGET_LONG_CALLS is true.
4048 2014-01-17 Jeff Law <law@redhat.com>
4050 * ree.c (combine_set_extension): Temporarily disable test for
4051 changing number of hard registers.
4053 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4056 * ipa-inline-analysis.c (inline_free_summary):
4057 Do not free summary of aliases.
4059 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4062 * gimplify.c (gimplify_expr): Use create_tmp_var
4063 instead of create_tmp_var_raw. If cond doesn't have
4064 integral type, don't add the IFN_ANNOTATE builtin at all.
4066 2014-01-17 Martin Jambor <mjambor@suse.cz>
4069 * ipa-cp.c (prev_edge_clone): New variable.
4070 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4071 Also resize prev_edge_clone vector.
4072 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4073 (ipcp_edge_removal_hook): New function.
4074 (ipcp_driver): Register ipcp_edge_removal_hook.
4076 2014-01-17 Andrew Pinski <apinski@cavium.com>
4077 Steve Ellcey <sellcey@mips.com>
4080 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4083 2014-01-17 Jeff Law <law@redhat.com>
4086 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4087 so that pass_ccp runs first.
4089 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4091 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4092 (ix86_adjust_cost): Use !TARGET_XXX.
4093 (do_reorder_for_imul): Likewise.
4094 (swap_top_of_ready_list): Likewise.
4095 (ix86_sched_reorder): Likewise.
4097 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4099 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4100 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4101 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4102 (intel_memset): New. Duplicate slm_memset.
4103 (intel_cost): New. Duplicate slm_cost.
4104 (m_INTEL): New macro.
4105 (processor_target_table): Add "intel".
4106 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4107 with PROCESSOR_INTEL for "intel".
4108 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4109 PROCESSOR_SILVERMONT.
4110 (ix86_issue_rate): Likewise.
4111 (ix86_adjust_cost): Likewise.
4112 (ia32_multipass_dfa_lookahead): Likewise.
4113 (swap_top_of_ready_list): Likewise.
4114 (ix86_sched_reorder): Likewise.
4115 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4116 instead of TARGET_OPT_AGU.
4117 * config/i386/i386.h (TARGET_INTEL): New.
4118 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4119 (processor_type): Add PROCESSOR_INTEL.
4120 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4121 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4123 2014-01-17 Marek Polacek <polacek@redhat.com>
4126 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4129 2014-01-17 Richard Biener <rguenther@suse.de>
4131 PR tree-optimization/46590
4132 * opts.c (default_options_table): Add entries for
4133 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4134 all enabled at -O1 but not for -Og.
4135 * common.opt (fbranch-count-reg): Remove Init(1).
4136 (fmove-loop-invariants): Likewise.
4137 (ftree-pta): Likewise.
4139 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4141 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4142 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4143 decls to at least the GCC 4.8 used alignments.
4146 * tree-nested.c (convert_nonlocal_reference_stmt,
4147 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4148 of GIMPLE_BIND stmts, adjust associated decls.
4150 2014-01-17 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/46590
4153 * vec.h (vec<>::bseach): New member function implementing
4154 binary search according to C89 bsearch.
4155 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4156 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4157 bitmap pointer again. Make accesses_in_loop a flat array.
4158 (mem_ref_obstack): New global.
4159 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4160 (mark_ref_stored): Likewise.
4161 (ref_indep_loop_p_2): Likewise.
4162 (set_ref_stored_in_loop): New helper function.
4163 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4164 (memref_free): Adjust.
4165 (record_mem_ref_loc): Simplify.
4166 (gather_mem_refs_stmt): Adjust.
4167 (sort_locs_in_loop_postorder_cmp): New function.
4168 (analyze_memory_references): Sort accesses_in_loop after
4169 loop postorder number.
4170 (find_ref_loc_in_loop_cmp): New function.
4171 (for_all_locs_in_loop): Find relevant cluster of locs in
4172 accesses_in_loop and iterate without recursion.
4173 (execute_sm): Avoid uninit warning.
4174 (struct ref_always_accessed): Simplify.
4175 (ref_always_accessed::operator ()): Likewise.
4176 (ref_always_accessed_p): Likewise.
4177 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4178 loop postorder numbers here.
4179 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4182 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4185 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4186 on decls for which assemble_alias has been called.
4188 2014-01-17 Nick Clifton <nickc@redhat.com>
4190 * config/msp430/msp430.opt: (mcpu): New option.
4191 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4192 (msp430_option_override): Parse target_cpu. If the MCU name
4193 matches a generic string, clear target_mcu.
4194 (msp430_attr): Allow numeric interrupt values up to 63.
4195 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4196 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4198 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4200 * config/msp430/msp430.md (popm): Use %J rather than %I.
4201 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4202 (addhi_cy_i): Use immediate_operand for operand 2.
4203 * doc/invoke.texi: Document -mcpu option.
4205 2014-01-17 Richard Biener <rguenther@suse.de>
4207 PR rtl-optimization/38518
4208 * df.h (df_analyze_loop): Declare.
4209 * df-core.c: Include cfgloop.h.
4210 (df_analyze_1): Split out main part of df_analyze.
4211 (df_analyze): Adjust.
4212 (loop_inverted_post_order_compute): New function.
4213 (loop_post_order_compute): Likewise.
4214 (df_analyze_loop): New function avoiding whole-function
4216 * loop-invariant.c (find_defs): Use df_analyze_loop.
4217 (find_invariants): Adjust.
4218 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4220 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4222 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4223 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4225 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4227 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4228 traversal when removing references.
4230 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4233 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4235 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4238 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4239 pushing a reload for an autoinc when we had previously reloaded an
4240 inner part of the address.
4242 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4244 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4246 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4247 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4248 when not giving up or versioning for alias only because of
4250 (vect_analyze_data_ref_dependences): Set to true.
4251 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4253 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4254 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4258 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4261 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4262 operand 0 predicate for gathers, use a new pseudo as subtarget.
4264 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4267 * lra-constraints.c (process_alt_operands): Add printing debug info.
4268 Check absence of input/output reloads for matched operands too.
4270 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4272 PR rtl-optimization/59835
4273 * ira.c (ira_init_register_move_cost): Increase cost for
4276 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4278 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4280 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4283 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4284 non-register objects. Use gen_(high/low)part more consistently.
4287 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4290 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4291 endian support, remove tests for WORDS_BIG_ENDIAN.
4292 (p8_mfvsrd_3_<mode>): Likewise.
4293 (reload_gpr_from_vsx<mode>): Likewise.
4294 (reload_gpr_from_vsxsf): Likewise.
4295 (p8_mfvsrd_4_disf): Likewise.
4297 2014-01-16 Richard Biener <rguenther@suse.de>
4299 PR rtl-optimization/46590
4300 * lcm.c (compute_antinout_edge): Use postorder iteration.
4301 (compute_laterin): Use inverted postorder iteration.
4303 2014-01-16 Nick Clifton <nickc@redhat.com>
4306 * varasm.c (output_constant): Return the number of bytes actually
4308 (output_constructor_array_range): Update the field size with the
4309 number of bytes emitted by output_constant.
4310 (output_constructor_regular_field): Likewise. Also do not
4311 complain if the total number of bytes emitted is now greater
4312 than the expected fieldpos.
4313 * output.h (output_constant): Update prototype and descriptive comment.
4315 2014-01-16 Marek Polacek <polacek@redhat.com>
4318 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4319 it is error_mark_node.
4321 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4323 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4324 VALID_AVX256_REG_OR_OI_MODE.
4326 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4328 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4329 current procedure should be profiled.
4331 2014-01-15 Andrew Pinski <apinski@cavium.com>
4333 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4334 of moving from/to the STACK_REG register class.
4336 2014-01-15 Richard Henderson <rth@redhat.com>
4339 * reginfo.c (global_regs_decl): Globalize.
4340 * rtl.h (global_regs_decl): Declare.
4341 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4342 reserved via global_regs.
4344 2014-01-15 Teresa Johnson <tejohnson@google.com>
4346 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4348 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4350 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4351 and vmulosh rather than call gen_vec_widen_smult_*.
4352 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4353 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4354 (vec_widen_smult_even_v16qi): Likewise.
4355 (vec_widen_umult_even_v8hi): Likewise.
4356 (vec_widen_smult_even_v8hi): Likewise.
4357 (vec_widen_umult_odd_v16qi): Likewise.
4358 (vec_widen_smult_odd_v16qi): Likewise.
4359 (vec_widen_umult_odd_v8hi): Likewise.
4360 (vec_widen_smult_odd_v8hi): Likewise.
4361 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4362 vmuloub rather than call gen_vec_widen_umult_*.
4363 (vec_widen_umult_lo_v16qi): Likewise.
4364 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4365 vmulosb rather than call gen_vec_widen_smult_*.
4366 (vec_widen_smult_lo_v16qi): Likewise.
4367 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4368 rather than call gen_vec_widen_umult_*.
4369 (vec_widen_umult_lo_v8hi): Likewise.
4370 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4371 rather than call gen_vec_widen_smult_*.
4372 (vec_widen_smult_lo_v8hi): Likewise.
4374 2014-01-15 Jeff Law <law@redhat.com>
4376 PR tree-optimization/59747
4377 * ree.c (find_and_remove_re): Properly handle case where a second
4378 eliminated extension requires widening a copy created for elimination
4379 of a prior extension.
4380 (combine_set_extension): Ensure that the number of hard regs needed
4381 for a destination register does not change when we widen it.
4383 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4385 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4386 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4387 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4388 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4389 (avr-*-rtems*): Likewise.
4390 (bfin*-rtems*): Likewise.
4391 (moxie-*-rtems*): Likewise.
4392 (h8300-*-rtems*): Likewise.
4393 (i[34567]86-*-rtems*): Likewise.
4394 (lm32-*-rtems*): Likewise.
4395 (m32r-*-rtems*): Likewise.
4396 (m68k-*-rtems*): Likewise.
4397 (microblaze*-*-rtems*): Likewise.
4398 (mips*-*-rtems*): Likewise.
4399 (powerpc-*-rtems*): Likewise.
4400 (sh-*-rtems*): Likewise.
4401 (sparc-*-rtems*): Likewise.
4402 (sparc64-*-rtems*): Likewise.
4403 (v850-*-rtems*): Likewise.
4404 (m32c-*-rtems*): Likewise.
4406 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4408 PR rtl-optimization/59511
4409 * ira.c (ira_init_register_move_cost): Use memory costs for some
4410 cases of register move cost calculations.
4411 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4412 instead of BB frequency.
4413 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4414 * lra-assigns.c (find_hard_regno_for): Ditto.
4416 2014-01-15 Richard Biener <rguenther@suse.de>
4418 PR tree-optimization/59822
4419 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4420 (vectorizable_load): Use it to hoist defs of uses of invariant
4421 loads out of the loop.
4423 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4424 Kugan Vivekanandarajah <kuganv@linaro.org>
4427 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4430 2014-01-15 Richard Biener <rguenther@suse.de>
4432 PR rtl-optimization/59802
4433 * lcm.c (compute_available): Use inverted postorder to seed
4434 the initial worklist.
4436 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4439 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4440 ADDR_REGS for invalid symrefs in non-PIC code.
4442 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4445 * builtins.c (determine_block_size): Initialize *probable_max_size
4446 even if len_rtx is CONST_INT.
4448 2014-01-14 Andrew Pinski <apinski@cavium.com>
4450 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4451 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4452 (cortexa53_tunings): Likewise.
4453 (aarch64_sched_issue_rate): New function.
4454 (TARGET_SCHED_ISSUE_RATE): Define.
4456 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4458 * ira-costs.c (find_costs_and_classes): Add missed
4459 ira_init_register_move_cost_if_necessary.
4461 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4464 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4466 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4469 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4470 to indicate if type is used for function return value. Warn ABI
4471 change if the vector mode isn't available for function return value.
4472 (ix86_function_arg_advance): Pass false to type_natural_mode.
4473 (ix86_function_arg): Likewise.
4474 (ix86_gimplify_va_arg): Likewise.
4475 (function_arg_32): Don't warn ABI change.
4476 (ix86_function_value): Pass true to type_natural_mode.
4477 (ix86_return_in_memory): Likewise.
4478 (ix86_struct_value_rtx): Removed.
4479 (TARGET_STRUCT_VALUE_RTX): Likewise.
4481 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4483 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4484 converting a conditional jump into a conditional return.
4486 2014-01-14 Richard Biener <rguenther@suse.de>
4488 PR tree-optimization/58921
4489 PR tree-optimization/59006
4490 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4491 hoisting invariant stmts.
4492 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4493 invariant loads on the preheader edge if possible.
4495 2014-01-14 Joey Ye <joey.ye@arm.com>
4497 * doc/plugin.texi (Building GCC plugins): Update to C++.
4499 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4501 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4502 (_mm_rcp28_round_ss): Ditto.
4503 (_mm_rsqrt28_round_sd): Ditto.
4504 (_mm_rsqrt28_round_ss): Ditto.
4505 (_mm_rcp28_sd): Ditto.
4506 (_mm_rcp28_ss): Ditto.
4507 (_mm_rsqrt28_sd): Ditto.
4508 (_mm_rsqrt28_ss): Ditto.
4509 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4510 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4511 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4512 (IX86_BUILTIN_RCP28SD): Ditto.
4513 (IX86_BUILTIN_RCP28SS): Ditto.
4514 (IX86_BUILTIN_RSQRT28SD): Ditto.
4515 (IX86_BUILTIN_RSQRT28SS): Ditto.
4516 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4517 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4518 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4519 (ix86_expand_special_args_builtin): Expand new FTYPE.
4520 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4521 (srcp14<mode>): Make insn unary.
4522 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4523 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4524 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4525 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4526 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4527 Fix rounding: make it SAE only.
4528 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4530 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4532 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4533 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4534 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4535 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4536 (round_saeonly_mask_scalar_operand4): Ditto.
4537 (round_saeonly_mask_scalar_op3): Ditto.
4538 (round_saeonly_mask_scalar_op4): Ditto.
4540 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4542 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4543 Implement -maltivec=be for vec_insert and vec_extract.
4545 2014-01-10 DJ Delorie <dj@redhat.com>
4547 * config/msp430/msp430.md (call_internal): Don't allow memory
4548 references with SP as the base register.
4549 (call_value_internal): Likewise.
4550 * config/msp430/constraints.md (Yc): New. For memory references
4551 that don't use SP as a base register.
4553 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4554 "an integer without a # prefix"
4555 * config/msp430/msp430.md (epilogue_helper): Use it.
4557 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4560 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4561 AVX512F gather builtins.
4562 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4563 on gather decls with INTEGER_TYPE masktype.
4564 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4565 directly into the builtin rather than hoisting it before loop.
4567 PR tree-optimization/59387
4568 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4569 (scev_const_prop): If folded_casts and type has undefined overflow,
4570 use force_gimple_operand instead of force_gimple_operand_gsi and
4571 for each added stmt if it is assign with
4572 arith_code_with_undefined_signed_overflow, call
4573 rewrite_to_defined_overflow.
4574 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4575 gimple-fold.h instead.
4576 (arith_code_with_undefined_signed_overflow,
4577 rewrite_to_defined_overflow): Moved to ...
4578 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4579 rewrite_to_defined_overflow): ... here. No longer static.
4580 Include gimplify-me.h.
4581 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4582 rewrite_to_defined_overflow): New prototypes.
4584 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4586 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4588 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4590 * builtins.c (get_object_alignment_2): Minor tweak.
4591 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4593 2014-01-13 Christian Bruel <christian.bruel@st.com>
4595 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4596 optimized non constant lengths.
4598 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4601 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4602 load as __atomic_load_N if possible.
4604 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4606 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4608 (rs6000_expand_builtin): Adjust call.
4610 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4613 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4614 * config/rs6000/rs6000.c: Include target-globals.h.
4615 (rs6000_set_current_function): Instead of doing target_reinit
4616 unconditionally, use save_target_globals_default_opts and
4617 restore_target_globals.
4619 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4621 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4622 (rs6000_expand_builtin): Handle mffs and mtfsf.
4623 (rs6000_init_builtins): Define mffs and mtfsf.
4624 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4625 (rs6000_mffs): New pattern.
4626 (rs6000_mtfsf): New pattern.
4628 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4630 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4631 Start narrowing with START. Apply candidate-use pair
4632 and check overall cost in narrowing.
4633 (iv_ca_prune): Pass new argument.
4635 2014-01-10 Jeff Law <law@redhat.com>
4638 * ree.c (combine_reaching_defs): Ensure the defining statement
4639 occurs before the extension when optimizing extensions with
4640 different source and destination hard registers.
4642 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4645 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4646 vtables into the type inheritance graph.
4648 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4650 PR rtl-optimization/59754
4651 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4652 modes in the REGNO != REGNO case.
4654 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4656 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4658 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4660 PR tree-optimization/59745
4661 * tree-predcom.c (tree_predictive_commoning_loop): Call
4662 free_affine_expand_cache if giving up because components is NULL.
4664 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4665 GC in payload of target_globals struct instead of allocating them on
4666 the heap and the larger structs separately using GC.
4667 * target-globals.h (struct target_globals): Make regs, hard_regs,
4668 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4669 of GTY((skip)) and change type to void *.
4670 (reset_target_globals): Cast loads from those fields to corresponding
4673 2014-01-10 Steve Ellcey <sellcey@mips.com>
4676 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4677 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4678 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4680 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4683 * aarch64-modes.def (CC_Zmode): New flags mode.
4684 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4685 represents an equality.
4686 (aarch64_get_condition_code): Handle CC_Zmode.
4687 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4689 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4691 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4692 extraction in good case.
4694 2014-01-10 Richard Biener <rguenther@suse.de>
4696 PR tree-optimization/59374
4697 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4698 checking after SLP discovery. Mark stmts not participating
4699 in any SLP instance properly.
4701 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4703 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4704 when handling a SET rtx.
4706 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4708 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4709 (cortex-a57): Likewise.
4710 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4712 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4714 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4715 non-iwmmxt builtins.
4717 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4721 * ipa-devirt.c record_target_from_binfo): Take as argument
4722 stack of binfos and lookup matching one for virtual inheritance.
4723 (possible_polymorphic_call_targets_1): Update.
4725 2014-01-10 Huacai Chen <chenhc@lemote.com>
4727 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4728 kernel strings for Loongson-2E/2F/3A.
4730 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4733 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4734 is_gimple_call before calling gimple_call_internal_p.
4736 2014-01-09 Steve Ellcey <sellcey@mips.com>
4738 * Makefile.in (TREE_FLOW_H): Remove.
4739 (TREE_SSA_H): Add file names from tree-flow.h.
4740 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4741 * tree.h: Remove tree-flow.h reference.
4742 * hash-table.h: Remove tree-flow.h reference.
4743 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4744 reference with tree-ssa-loop.h.
4746 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4748 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4749 default element-order behavior for -maltivec.
4750 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4751 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4752 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4753 when targeting big endian, at least for now.
4754 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4756 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4759 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4760 var satisfies use_register_for_decl, just take into account type
4761 alignment, rather than decl alignment.
4763 PR tree-optimization/59622
4764 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4765 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4766 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4767 Don't devirtualize for inplace at all. For targets.length () == 1,
4768 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4770 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4772 * config/i386/i386.md (cpu): Remove the unused btver1.
4774 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4776 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4778 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4781 * tree-core.h (struct target_globals): New forward declaration.
4782 (struct tree_target_option): Add globals field.
4783 * tree.h (TREE_TARGET_GLOBALS): Define.
4784 (prepare_target_option_nodes_for_pch): New prototype.
4785 * target-globals.h (struct target_globals): Define even if
4787 * tree.c (prepare_target_option_node_for_pch,
4788 prepare_target_option_nodes_for_pch): New functions.
4789 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4790 * config/i386/i386.c: Include target-globals.h.
4791 (ix86_set_current_function): Instead of doing target_reinit
4792 unconditionally, use save_target_globals_default_opts and
4793 restore_target_globals.
4795 2014-01-09 Richard Biener <rguenther@suse.de>
4797 PR tree-optimization/59715
4798 * tree-cfg.h (split_critical_edges): Declare.
4799 * tree-cfg.c (split_critical_edges): Export.
4800 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4802 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4804 * cfgexpand.c (expand_stack_vars): Optionally disable
4805 asan stack protection.
4806 (expand_used_vars): Likewise.
4807 (partition_stack_vars): Likewise.
4808 * asan.c (asan_emit_stack_protection): Optionally disable
4809 after return stack usage.
4810 (instrument_derefs): Optionally disable memory access instrumentation.
4811 (instrument_builtin_call): Likewise.
4812 (instrument_strlen_call): Likewise.
4813 (asan_protect_global): Optionally disable global variables protection.
4814 * doc/invoke.texi: Added doc for new options.
4815 * params.def: Added new options.
4816 * params.h: Likewise.
4818 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4820 PR rtl-optimization/59724
4821 * ifcvt.c (cond_exec_process_if_block): Don't call
4822 flow_find_head_matching_sequence with 0 longest_match.
4823 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4824 non-active insns if !stop_after.
4825 (try_head_merge_bb): Revert 2014-01-07 changes.
4827 2014-01-08 Jeff Law <law@redhat.com>
4829 * ree.c (get_sub_rtx): New function, extracted from...
4830 (merge_def_and_ext): Here.
4831 (combine_reaching_defs): Use get_sub_rtx.
4833 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4835 * cgraph.h (varpool_variable_node): Do not choke on null node.
4837 2014-01-08 Catherine Moore <clm@codesourcery.com>
4839 * config/mips/mips.md (simple_return): Attempt to use JRC
4841 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4843 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4845 PR rtl-optimization/59137
4846 * reorg.c (steal_delay_list_from_target): Call update_block for
4848 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4850 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4852 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4853 two duplicate entries.
4855 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4858 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4860 * config/mips/mips.c (mips_truncated_op_cost): New function.
4861 (mips_rtx_costs): Adjust test for BADDU.
4862 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4864 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4866 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4867 (*baddu_si): ...this new pattern.
4869 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4872 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4874 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4877 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4879 (expand_expr, expand_normal): Adjust.
4880 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4881 inner_reference_p. Use inner_reference_p to expand inner references.
4882 (store_expr): Adjust.
4883 * cfgexpand.c (expand_call_stmt): Adjust.
4885 2014-01-08 Rong Xu <xur@google.com>
4887 * gcov-io.c (gcov_var): Move from gcov-io.h.
4888 (gcov_position): Ditto.
4889 (gcov_is_error): Ditto.
4890 (gcov_rewrite): Ditto.
4891 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4892 only part to libgcc/libgcov.h.
4894 2014-01-08 Marek Polacek <polacek@redhat.com>
4897 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4899 2014-01-08 Marek Polacek <polacek@redhat.com>
4902 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4904 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4906 PR rtl-optimization/59649
4907 * stor-layout.c (get_mode_bounds): For BImode return
4908 0 and STORE_FLAG_VALUE.
4910 2014-01-08 Richard Biener <rguenther@suse.de>
4913 * gimple.h (is_gimple_builtin_call): Remove.
4914 (gimple_builtin_call_types_compatible_p): New.
4915 (gimple_call_builtin_p): New overload.
4916 * gimple.c (is_gimple_builtin_call): Remove.
4917 (validate_call): Rename to ...
4918 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4920 (validate_type): New static function.
4921 (gimple_call_builtin_p): New overload and adjust.
4922 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4923 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4924 (gimple_fold_stmt_to_constant_1): Likewise.
4925 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4927 2014-01-08 Richard Biener <rguenther@suse.de>
4930 * gimplify.c (gimplify_expr): Gimplify register-register type
4931 VIEW_CONVERT_EXPRs to separate stmts.
4933 2014-01-07 Jeff Law <law@redhat.com>
4936 * ree.c (combine_set_extension): Handle case where source
4937 and destination registers in an extension insn are different.
4938 (combine_reaching_defs): Allow source and destination registers
4939 in extension to be different under limited circumstances.
4940 (add_removable_extension): Remove restriction that the
4941 source and destination registers in the extension are the same.
4942 (find_and_remove_re): Emit a copy from the extension's
4943 destination to its source after the defining insn if
4944 the source and destination registers are different.
4947 * ifcvt.c (merge_if_block): If we are merging a block with more than
4948 one successor with a block with no successors, remove any BARRIER
4949 after the second block.
4951 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4953 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4955 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4958 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4959 for 14-bit register offsets when INT14_OK_STRICT is false.
4961 2014-01-07 Roland Stigge <stigge@antcom.de>
4962 Michael Meissner <meissner@linux.vnet.ibm.com>
4965 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4966 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4968 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4970 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4973 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4975 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4976 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4977 rtx is const0_rtx or not.
4979 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4982 * target-globals.c (save_target_globals): Remove this_fn_optab
4984 * toplev.c: Include optabs.h.
4985 (target_reinit): Temporarily restore the global options if another
4986 set of options are in force.
4988 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4990 PR rtl-optimization/58668
4991 * cfgcleanup.c (flow_find_cross_jump): Don't count
4992 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4993 to determine what is counted.
4994 (flow_find_head_matching_sequence): Use active_insn_p to determine
4996 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4998 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4999 determine what is counted.
5001 PR tree-optimization/59643
5002 * tree-predcom.c (split_data_refs_to_components): If one dr is
5003 read and one write, determine_offset fails and the write isn't
5004 in the bad component, just put the read into the bad component.
5006 2014-01-07 Mike Stump <mikestump@comcast.net>
5007 Jakub Jelinek <jakub@redhat.com>
5010 * tree-core.h (struct tree_optimization_option): Change optabs
5011 type from unsigned char * to void *.
5012 * optabs.c (init_tree_optimization_optabs): Adjust
5013 TREE_OPTIMIZATION_OPTABS initialization.
5015 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5018 * config/i386/i386.h (struct machine_function): Add
5019 no_drap_save_restore field.
5020 * config/i386/i386.c (ix86_save_reg): Use
5021 !cfun->machine->no_drap_save_restore instead of
5022 crtl->stack_realign_needed.
5023 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5024 this function clears frame_pointer_needed. Set
5025 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5026 and DRAP reg is needed.
5028 2014-01-06 Marek Polacek <polacek@redhat.com>
5031 * doc/implement-c.texi: Mention that other integer types are
5032 permitted as bit-field types in strictly conforming mode.
5034 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5036 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5039 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5041 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5043 2014-01-06 Martin Jambor <mjambor@suse.cz>
5046 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5048 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5050 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5054 * var-tracking.c (add_stores): Preserve the value of the source even if
5055 we don't record the store.
5057 2014-01-06 Terry Guo <terry.guo@arm.com>
5059 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5061 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5064 * config/darwin.c (darwin_function_section): Adjust return values to
5065 correspond to optimisation changes made in r206070.
5067 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5069 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5070 from prefetch_block tune setting.
5071 (nocona_cost): Correct size of prefetch block to 64.
5073 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5075 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5076 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5077 used to save the static chain register in the computation of the offset
5078 from which the FP registers need to be restored.
5080 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5082 PR tree-optimization/59519
5083 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5084 ICE if get_current_def (current_new_name) is already non-NULL, as long
5085 as it is a phi result of some other phi in *new_exit_bb that has
5088 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5089 or vmovdqu* for misaligned_operand.
5090 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5091 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5092 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5093 aligned_mem for AVX512F masked aligned load and store builtins and for
5096 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5098 PR tree-optimization/59651
5099 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5100 Address range for negative step should be added by TYPE_SIZE_UNIT.
5102 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5104 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5105 overlapping registers also for registers other than the stack pointer.
5107 2014-01-03 Marek Polacek <polacek@redhat.com>
5110 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5113 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5116 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5119 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5120 (push splitter): Use <P:MODE_SIZE> instead of
5121 GET_MODE_SIZE (<P:MODE>mode).
5122 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5123 (mov -1, reg peephole2): Likewise.
5124 * config/i386/sse.md (*mov<mode>_internal,
5125 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5126 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5127 *<code><mode>3, *andnot<mode>3<mask_name>,
5128 <mask_codefor><code><mode>3<mask_name>): Likewise.
5129 * config/i386/subst.md (mask_mode512bit_condition,
5130 sd_mask_mode512bit_condition): Likewise.
5132 2014-01-02 Xinliang David Li <davidxl@google.com>
5134 PR tree-optimization/59303
5135 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5136 (dump_predicates): Better output format.
5137 (pred_equal_p): New function.
5138 (is_neq_relop_p): Ditto.
5139 (is_neq_zero_form_p): Ditto.
5140 (pred_expr_equal_p): Ditto.
5141 (pred_neg_p): Ditto.
5142 (simplify_pred): Ditto.
5143 (simplify_preds_2): Ditto.
5144 (simplify_preds_3): Ditto.
5145 (simplify_preds_4): Ditto.
5146 (simplify_preds): Ditto.
5148 (push_to_worklist): Ditto.
5149 (get_pred_info_from_cmp): Ditto.
5150 (is_degenerated_phi): Ditto.
5151 (normalize_one_pred_1): Ditto.
5152 (normalize_one_pred): Ditto.
5153 (normalize_one_pred_chain): Ditto.
5154 (normalize_preds): Ditto.
5155 (normalize_cond_1): Remove function.
5156 (normalize_cond): Ditto.
5157 (is_gcond_subset_of): Ditto.
5158 (is_subset_of_any): Ditto.
5159 (is_or_set_subset_of): Ditto.
5160 (is_and_set_subset_of): Ditto.
5161 (is_norm_cond_subset_of): Ditto.
5162 (pred_chain_length_cmp): Ditto.
5163 (convert_control_dep_chain_into_preds): Type change.
5164 (find_predicates): Ditto.
5165 (find_def_preds): Ditto.
5166 (destroy_predicates_vecs): Ditto.
5167 (find_matching_predicates_in_rest_chains): Ditto.
5168 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5169 (is_pred_expr_subset): Ditto.
5170 (is_pred_chain_subset_of): Ditto.
5171 (is_included_in): Ditto.
5172 (is_superset_of): Ditto.
5174 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5176 Update copyright years.
5178 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5180 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5181 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5182 config/arc/arc.md, config/arc/arc.opt,
5183 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5184 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5185 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5186 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5187 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5188 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5189 vtable-verify.c, vtable-verify.h: Use the standard form for the
5192 2014-01-02 Tobias Burnus <burnus@net-b.de>
5194 * gcc.c (process_command): Update copyright notice dates.
5195 * gcov-dump.c: Ditto.
5197 * doc/cpp.texi: Bump @copying's copyright year.
5198 * doc/cppinternals.texi: Ditto.
5199 * doc/gcc.texi: Ditto.
5200 * doc/gccint.texi: Ditto.
5201 * doc/gcov.texi: Ditto.
5202 * doc/install.texi: Ditto.
5203 * doc/invoke.texi: Ditto.
5205 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5207 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5209 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5211 * config/i386/sse.md (*mov<mode>_internal): Guard
5212 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5214 PR rtl-optimization/59647
5215 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5216 new_rtx into UNSIGNED_FLOAT rtxes.
5218 Copyright (C) 2014 Free Software Foundation, Inc.
5220 Copying and distribution of this file, with or without modification,
5221 are permitted in any medium without royalty provided the copyright
5222 notice and this notice are preserved.