1 2014-03-18 Jakub Jelinek <jakub@redhat.com>
4 * internal-fn.c: Include diagnostic-core.h.
5 (expand_BUILTIN_EXPECT): New function.
6 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
7 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
8 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
9 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
11 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
12 Revert 3 argument __builtin_expect code.
13 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
14 * gimple-fold.c (gimple_fold_call): Likewise.
15 * tree.h (fold_builtin_expect): New prototype.
16 * builtins.c (build_builtin_expect_predicate): Add predictor
17 argument, if non-NULL, create 3 argument __builtin_expect.
18 (fold_builtin_expect): No longer static. Add ARG2 argument,
19 pass it through to build_builtin_expect_predicate.
20 (fold_builtin_2): Adjust caller.
21 (fold_builtin_3): Handle BUILT_IN_EXPECT.
22 * internal-fn.def (BUILTIN_EXPECT): New.
24 2014-03-18 Tobias Burnus <burnus@net-b.de>
27 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
28 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
29 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
31 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
34 * predict.c (combine_predictions_for_bb): Fix up formatting.
35 (expr_expected_value_1, expr_expected_value): Add predictor argument,
36 fill what it points to if non-NULL.
37 (tree_predict_by_opcode): Adjust caller, use the predictor.
38 * predict.def (PRED_COMPARE_AND_SWAP): Add.
40 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
42 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
43 proper constant for the store mode.
45 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
47 * symtab.c (change_decl_assembler_name): Fix transparent alias
50 2014-03-16 Renlin Li <Renlin.Li@arm.com>
52 * config/aarch64/aarch64.c: Correct the comments about the
55 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
57 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
58 check for GF_OMP_FOR_KIND_FOR.
60 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
62 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
63 ymm and zmm register names.
65 2014-03-17 Jakub Jelinek <jakub@redhat.com>
68 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
69 note creation for the 2010-08-31 changes.
71 2014-03-17 Marek Polacek <polacek@redhat.com>
74 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
75 as -fno-tree-loop-vectorize.
76 (expand_omp_simd): Likewise.
78 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
80 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
81 (eligible_for_call_delay): New prototype.
82 * config/sparc/sparc.c (tls_call_delay): Rename into...
83 (eligible_for_call_delay): ...this. Return false if the instruction
84 cannot be put in the delay slot of a branch.
85 (eligible_for_restore_insn): Simplify.
86 (eligible_for_return_delay): Return false if the instruction cannot be
87 put in the delay slot of a branch and simplify.
88 (eligible_for_sibcall_delay): Return false if the instruction cannot be
89 put in the delay slot of a branch.
90 * config/sparc/sparc.md (fix_ut699): New attribute.
91 (tls_call_delay): Delete.
92 (in_call_delay): Reimplement.
93 (eligible_for_sibcall_delay): Rename into...
94 (in_sibcall_delay): ...this.
95 (eligible_for_return_delay): Rename into...
96 (in_return_delay): ...this.
97 (in_branch_delay): Reimplement.
98 (in_uncond_branch_delay): Delete.
99 (in_annul_branch_delay): Delete.
101 2014-03-14 Richard Henderson <rth@redhat.com>
104 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
105 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
106 (*floathi<X87MODEF>2_i387_with_temp): Remove.
107 (floathi splitters): Remove.
108 (float<SWI48x>xf2): New pattern.
109 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
110 code that tried to handle DImode for 32-bit, but which was excluded
111 by the pattern's condition. Drop allocation of stack temporary.
112 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
113 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
114 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
115 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
116 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
117 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
118 (*float<SWI48><MODEF>2_sse_interunit): Remove.
119 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
120 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
121 (*float<SWI48x><X87MODEF>2_i387): Remove.
122 (all float _with_temp splitters): Remove.
123 (*float<SWI48x><MODEF>2_i387): New pattern.
124 (*float<SWI48><MODEF>2_sse): New pattern.
125 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
126 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
128 2014-03-14 Jakub Jelinek <jakub@redhat.com>
129 Marek Polacek <polacek@redhat.com>
132 * common.opt (dump_base_name_prefixed): New Variable.
133 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
134 if x_dump_base_name_prefixed is already set, set it at the end.
136 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
138 PR rtl-optimization/60508
139 * lra-constraints.c (get_reload_reg): Add new parameter
141 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
142 Pass the new parameter values.
144 2014-03-14 Richard Biener <rguenther@suse.de>
146 * common.opt: Revert unintented changes from r205065.
149 2014-03-14 Richard Biener <rguenther@suse.de>
152 * cfghooks.c (split_block): Properly adjust all loops the
153 block was a latch of.
155 2014-03-14 Martin Jambor <mjambor@suse.cz>
158 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
161 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
164 * config/avr/avr.c (avr_set_current_function): Pass function name
165 through default_strip_name_encoding before sanity checking instead
166 of skipping the first char of the assembler name.
168 2014-03-13 Richard Henderson <rth@redhat.com>
171 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
172 (ix86_force_to_memory, ix86_free_from_memory): Remove.
173 * config/i386/i386-protos.h: Likewise.
174 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
175 in the expander instead of a splitter.
176 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
177 any possibility of requiring a memory.
178 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
179 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
180 (fp branch splitters): Update for ix86_split_fp_branch.
181 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
182 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
183 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
184 (*fop_<MODEF>_2_i387): Remove f/r alternative.
185 (*fop_<MODEF>_3_i387): Likewise.
186 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
187 (splitters for the fop_* register patterns): Remove.
188 (fscalexf4_i387): Rename from *fscalexf4_i387.
189 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
191 2014-03-13 Jakub Jelinek <jakub@redhat.com>
193 PR tree-optimization/59779
194 * tree-dfa.c (get_ref_base_and_extent): Use double_int
195 type for bitsize and maxsize instead of HOST_WIDE_INT.
197 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
199 PR rtl-optimization/57320
200 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
201 the CFG after thread_prologue_and_epilogue_insns.
203 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
205 PR rtl-optimization/57189
206 * lra-constraints.c (process_alt_operands): Disfavor spilling
209 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
211 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
214 2014-03-13 Jakub Jelinek <jakub@redhat.com>
216 PR tree-optimization/59025
218 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
219 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
221 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
224 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
225 calls of avr_out_plus_1.
227 2014-03-13 Bin Cheng <bin.cheng@arm.com>
229 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
230 BB's single pred and update the father loop's latch info later.
232 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
234 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
238 (VEC_base): Likewise.
239 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
240 registers, we need to swap double words in little endian mode.
242 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
243 to be a container mode for 128-bit integer operations added in ISA
244 2.07. Unlike TImode and PTImode, the preferred register set is
245 the Altivec/VMX registers for the 128-bit operations.
247 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
249 (rs6000_split_128bit_ok_p): Likewise.
251 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
252 macros for creating ISA 2.07 normal and overloaded builtin
253 functions with 3 arguments.
254 (BU_P8V_OVERLOAD_3): Likewise.
255 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
256 for use as overloaded functions.
257 (VPERM_1TI_UNS): Likewise.
258 (VSEL_1TI): Likewise.
259 (VSEL_1TI_UNS): Likewise.
260 (ST_INTERNAL_1ti): Likewise.
261 (LD_INTERNAL_1ti): Likewise.
262 (XXSEL_1TI): Likewise.
263 (XXSEL_1TI_UNS): Likewise.
264 (VPERM_1TI): Likewise.
265 (VPERM_1TI_UNS): Likewise.
266 (XXPERMDI_1TI): Likewise.
268 (LXVD2X_V1TI): Likewise.
269 (STXVD2X_V1TI): Likewise.
270 (VEC_INIT_V1TI): Likewise.
271 (VEC_SET_V1TI): Likewise.
272 (VEC_EXT_V1TI): Likewise.
273 (EQV_V1TI): Likewise.
274 (NAND_V1TI): Likewise.
275 (ORC_V1TI): Likewise.
276 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
277 added in ISA 2.07. Add both normal 'altivec' builtins, and the
281 (VADDEUQM): Likewise.
282 (VADDECUQ): Likewise.
283 (VSUBEUQM): Likewise.
284 (VSUBECUQ): Likewise.
286 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
287 __int128_t and __uint128_t types.
288 (__uint128_type): Likewise.
289 (altivec_categorize_keyword): Add support for vector __int128_t,
290 vector __uint128_t, vector __int128, and vector unsigned __int128
291 as a container type for TImode operations that need to be done in
292 VSX/Altivec registers.
293 (rs6000_macro_to_expand): Likewise.
294 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
295 to support 128-bit integer instructions vaddcuq, vadduqm,
296 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
297 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
299 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
300 for V1TImode, and set up preferences to use VSX/Altivec registers.
301 Setup VSX reload handlers.
302 (rs6000_debug_reg_global): Likewise.
303 (rs6000_init_hard_regno_mode_ok): Likewise.
304 (rs6000_preferred_simd_mode): Likewise.
305 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
306 (easy_altivec_constant): Likewise.
307 (output_vec_const_move): Likewise.
308 (rs6000_expand_vector_set): Convert V1TImode set and extract to
310 (rs6000_expand_vector_extract): Likewise.
311 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
313 (rs6000_const_vec): Add support for V1TImode.
314 (rs6000_emit_le_vsx_load): Swap double words when loading or
315 storing TImode/V1TImode.
316 (rs6000_emit_le_vsx_store): Likewise.
317 (rs6000_emit_le_vsx_move): Likewise.
318 (rs6000_emit_move): Add support for V1TImode.
319 (altivec_expand_ld_builtin): Likewise.
320 (altivec_expand_st_builtin): Likewise.
321 (altivec_expand_vec_init_builtin): Likewise.
322 (altivec_expand_builtin): Likewise.
323 (rs6000_init_builtins): Add support for V1TImode type. Add
324 support for ISA 2.07 128-bit integer builtins. Define type names
325 for the VSX/Altivec vector types.
326 (altivec_init_builtins): Add support for overloaded vector
327 functions with V1TImode type.
328 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
329 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
331 (rs6000_split_128bit_ok_p): Likewise.
332 (rs6000_handle_altivec_attribute): Create V1TImode from vector
333 __int128_t and vector __uint128_t.
335 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
343 (VS_scalar): Likewise.
344 (VS_double): Likewise.
345 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
347 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
348 we support the ISA 2.07 128-bit integer arithmetic instructions.
349 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
350 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
351 and TImode types for use with the builtin functions.
352 (V1TI_type_node): Likewise.
353 (unsigned_V1TI_type_node): Likewise.
354 (intTI_type_internal_node): Likewise.
355 (uintTI_type_internal_node): Likewise.
357 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
358 128-bit builtin functions.
359 (UNSPEC_VADDEUQM): Likewise.
360 (UNSPEC_VADDECUQ): Likewise.
361 (UNSPEC_VSUBCUQ): Likewise.
362 (UNSPEC_VSUBEUQM): Likewise.
363 (UNSPEC_VSUBECUQ): Likewise.
364 (VM): Add V1TImode to vector mode iterators.
367 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
368 (altivec_vaddcuq): Likewise.
369 (altivec_vsubuqm): Likewise.
370 (altivec_vsubcuq): Likewise.
371 (altivec_vaddeuqm): Likewise.
372 (altivec_vaddecuq): Likewise.
373 (altivec_vsubeuqm): Likewise.
374 (altivec_vsubecuq): Likewise.
376 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
378 (BOOL_128): Likewise.
379 (BOOL_REGS_OUTPUT): Likewise.
380 (BOOL_REGS_OP1): Likewise.
381 (BOOL_REGS_OP2): Likewise.
382 (BOOL_REGS_UNARY): Likewise.
383 (BOOL_REGS_AND_CR0): Likewise.
385 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
386 128-bit integer builtin support.
387 (vec_vadduqm): Likewise.
388 (vec_vaddecuq): Likewise.
389 (vec_vaddeuqm): Likewise.
390 (vec_vsubecuq): Likewise.
391 (vec_vsubeuqm): Likewise.
392 (vec_vsubcuq): Likewise.
393 (vec_vsubuqm): Likewise.
395 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
396 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
397 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
398 128-bit integer add/subtract to ISA 2.07.
400 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
402 * config/arc/arc.c (arc_predicate_delay_insns):
403 Fix third argument passed to conditionalize_nonjump.
405 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
407 * config/aarch64/aarch64-builtins.c
408 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
409 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
410 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
411 instead of __builtin_lfloor.
412 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
414 2014-03-12 Jakub Jelinek <jakub@redhat.com>
416 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
417 (tree_ssa_ifcombine_bb_1): New function.
418 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
419 is an empty forwarder block to then_bb or vice versa and then_bb
420 and else_bb are effectively swapped.
422 2014-03-12 Christian Bruel <christian.bruel@st.com>
425 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
426 REG_CFA_DEF_CFA note.
427 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
428 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
430 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
432 PR tree-optimization/60454
433 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
435 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
437 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
438 Do not define target_cpu_default2 to generic.
439 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
440 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
441 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
443 2014-03-12 Jakub Jelinek <jakub@redhat.com>
444 Marc Glisse <marc.glisse@inria.fr>
446 PR tree-optimization/60502
447 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
448 instead of build_low_bits_mask.
450 2014-03-12 Jakub Jelinek <jakub@redhat.com>
453 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
454 if there are multiple uses, but op doesn't live on E edge.
455 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
456 clobber stmts before __builtin_unreachable.
458 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
460 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
461 hard_frame_pointer_rtx.
462 * cse.c (cse_insn): Remove volatile check.
463 * cselib.c (cselib_process_insn): Likewise.
464 * dse.c (scan_insn): Likewise.
466 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
468 * config/arc/arc.c (conditionalize_nonjump): New function,
470 (arc_ifcvt): ... this.
471 (arc_predicate_delay_insns): Use it.
473 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
475 * config/arc/predicates.md (extend_operand): During/after reload,
476 allow const_int_operand.
477 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
478 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
479 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
481 (umulsi3_highpart_i): Likewise.
483 2014-03-11 Richard Biener <rguenther@suse.de>
485 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
486 Add asserts to guard possible wrong-code bugs.
488 2014-03-11 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/60429
491 PR tree-optimization/60485
492 * tree-ssa-structalias.c (set_union_with_increment): Properly
493 take into account all fields that overlap the shifted vars.
494 (do_sd_constraint): Likewise.
495 (do_ds_constraint): Likewise.
496 (get_constraint_for_ptr_offset): Likewise.
498 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
500 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
501 (nios2_compute_frame_layout):
502 Add calculation of cfun->machine->fp_save_offset.
503 (nios2_expand_prologue): Correct setting of frame pointer register
505 (nios2_expand_epilogue): Update recovery of stack pointer from
506 frame pointer accordingly.
507 (nios2_initial_elimination_offset): Update calculation of offset
508 for eliminating to HARD_FRAME_POINTER_REGNUM.
510 2014-03-10 Jakub Jelinek <jakub@redhat.com>
513 * ipa.c (symtab_remove_unreachable_nodes): Don't call
514 cgraph_get_create_node on VAR_DECLs.
516 2014-03-10 Richard Biener <rguenther@suse.de>
519 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
521 2014-03-08 Douglas B Rupp <rupp@gnat.com>
523 * config/vms/vms.opt (vms_float_format): New variable.
525 2014-03-08 Tobias Burnus <burnus@net-b.de>
527 * doc/invoke.texi (-fcilkplus): Update implementation status.
529 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
530 Richard Biener <rguenther@suse.de>
532 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
533 consistently accross all TUs.
534 (run_gcc): Enable -fshort-double automatically at link at link-time
535 and disallow override.
537 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
540 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
541 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
542 if they can't be used.
544 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
546 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
547 for Solaris 11/x86 ld.
548 * configure: Regenerate.
550 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
552 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
553 (LIB_TLS_SPEC): Save as ld_tls_libs.
554 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
555 (HAVE_AS_IX86_TLSLDM): New test.
556 * configure, config.in: Regenerate.
557 * config/i386/i386.c (legitimize_tls_address): Fall back to
558 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
559 cannot support TLS_MODEL_LOCAL_DYNAMIC.
560 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
561 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
563 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
565 * common.opt (fira-loop-pressure): Mark as optimization.
567 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
569 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
570 an OpenMP mappable type.
572 2014-03-06 Matthias Klose <doko@ubuntu.com>
574 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
575 MULTILIB_OSDIRNAMES is not defined.
577 2014-03-06 Jakub Jelinek <jakub@redhat.com>
578 Meador Inge <meadori@codesourcery.com>
581 * config/arm/arm.c (arm_tls_symbol_p): Remove.
582 (arm_legitimize_address): Call legitimize_tls_address for any
583 arm_tls_referenced_p expression, handle constant addend. Call it
584 before testing for !TARGET_ARM.
585 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
587 2014-03-06 Richard Biener <rguenther@suse.de>
593 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
595 * tree-streamer.c (record_common_node): Assert we don't record
596 nodes with type double.
597 (preload_common_node): Skip type double, complex double and double
598 pointer since it is now frontend dependent due to fshort-double option.
600 2014-03-06 Richard Biener <rguenther@suse.de>
602 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
603 or -fno-lto is specified and the linker has full plugin support.
604 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
605 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
606 * lto-wrapper.c (merge_and_complain): Merge compile-time
608 (run_gcc): And pass it through to the link options.
610 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
614 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
616 * cselib.c (remove_useless_values): Skip to avoid quadratic
617 behavior if the condition moved from...
618 (cselib_process_insn): ... here holds.
620 2014-03-05 Jakub Jelinek <jakub@redhat.com>
623 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
624 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
627 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
628 (TM_H): Add x86-tune.def.
630 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
632 * config/aarch64/aarch64.c (generic_tunings):
633 Use cortexa57_extra_costs.
635 2014-03-05 Jakub Jelinek <jakub@redhat.com>
638 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
639 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
640 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
643 2014-03-04 Heiher <r@hev.cc>
645 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
646 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
648 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
650 * config/i386/predicates.md (const2356_operand): Change to ...
651 (const2367_operand): ... this.
652 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
654 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
655 (*avx512pf_scatterpf<mode>sf): Ditto.
656 (avx512pf_scatterpf<mode>df): Ditto.
657 (*avx512pf_scatterpf<mode>df_mask): Ditto.
658 (*avx512pf_scatterpf<mode>df): Ditto.
659 * config/i386/i386.c (ix86_expand_builtin): Update
660 incorrect hint operand error message.
662 2014-03-04 Richard Biener <rguenther@suse.de>
664 * lto-section-in.c (lto_get_section_data): Fix const cast.
666 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
668 * tree-streamer.c (record_common_node): Assert we don't record
669 nodes with type double.
670 (preload_common_node): Skip type double, complex double and double
671 pointer since it is now frontend dependent due to fshort-double option.
673 2014-03-04 Richard Biener <rguenther@suse.de>
676 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
677 (lto_input_toplevel_asms): Likewise.
678 * lto-section-in.c (lto_get_section_data): Instead do it here
681 2014-03-04 Richard Biener <rguenther@suse.de>
683 PR tree-optimization/60382
684 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
685 dead PHIs a reduction.
687 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
689 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
691 (_mm_prefetch): Move out of GCC target("sse") pragma.
692 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
693 GCC target("prfchw") pragma.
694 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
696 * config/i386/i386.c (ix86_option_override_internal): Enable
697 -mprfchw with -mprefetchwt1.
699 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
701 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
704 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
706 * opts.h (CL_PCH_IGNORE): Define.
707 * targhooks.c (option_affects_pch_p):
708 Return false for options that have CL_PCH_IGNORE set.
709 * opt-functions.awk: Process PchIgnore.
710 * doc/options.texi: Document PchIgnore.
712 * config/arc/arc.opt (misize): Add PchIgnore property.
714 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
716 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
717 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
718 constraint on constants to permit them being loaded into
719 GENERAL_REGS or BASE_REGS.
721 2014-03-03 Nick Clifton <nickc@redhat.com>
723 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
724 anti-cacnonical alternatives.
725 (negandhi3_real): New pattern.
726 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
728 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
730 * config/avr/avr-mcus.def: Remove atxmega16x1.
731 * config/avr/avr-tables.opt: Regenerate.
732 * config/avr/t-multilib: Regenerate.
733 * doc/avr-mmcu.texi: Regenerate.
735 2014-03-03 Tobias Grosser <tobias@grosser.es>
736 Mircea Namolaru <mircea.namolaru@inria.fr>
738 PR tree-optimization/58028
739 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
742 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
744 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
745 not handled by recognizers.
747 2014-03-03 Jakub Jelinek <jakub@redhat.com>
750 * function.c (expand_function_end): Don't emit
751 clobber_return_register sequence if clobber_after is a BARRIER.
752 * cfgexpand.c (construct_exit_block): Append instructions before
753 return_label to prev_bb.
755 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
757 * config/rs6000/constraints.md: Document reserved use of "wc".
759 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
762 * ipa.c (function_and_variable_visibility): When dissolving comdat
763 group, also set all symbols to local.
765 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
770 2013-12-14 Jan Hubicka <jh@suse.cz>
772 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
774 2014-03-02 Jon Beniston <jon@beniston.com>
780 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
781 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
782 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
783 (simple_return, *simple_return): New patterns
784 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
785 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
787 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
789 * dwarf2out.c (gen_subprogram_die): Tidy.
791 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
794 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
795 (*mov_t_msb_neg_negc): ... this new insn.
797 2014-02-28 Jason Merrill <jason@redhat.com>
800 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
803 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
806 * dwarf2out.c (decltype_auto_die): New static.
807 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
808 (gen_type_die_with_usage): Handle 'decltype(auto)'.
809 (is_cxx_auto): Likewise.
811 2014-02-28 Ian Bolton <ian.bolton@arm.com>
813 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
814 we are not using general regs only.
816 2014-02-28 Richard Biener <rguenther@suse.de>
819 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
820 previous fix and only allow to remove trivial pre-headers
821 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
822 (remove_forwarder_block): Properly update the latch of a loop.
824 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
827 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
828 (cselib_preserved_hash_table): New.
829 (preserve_constants_and_equivs): Move preserved vals to it.
830 (cselib_find_slot): Look it up first.
831 (cselib_init): Initialize it.
832 (cselib_finish): Release it.
833 (dump_cselib_table): Dump it.
835 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
838 * cselib.c (remove_useless_values): Skip to avoid quadratic
839 behavior if the condition moved from...
840 (cselib_process_insn): ... here holds.
842 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
845 * var-tracking.c (vt_initialize): Apply the same condition to
846 preserve the CFA base value.
848 2014-02-28 Joey Ye <joey.ye@arm.com>
851 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
852 if reload in progress or completed.
854 2014-02-28 Tobias Burnus <burnus@net-b.de>
857 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
860 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
862 * doc/tm.texi.in (Condition Code Status): Update documention for
863 relative locations of cc0-setter and cc0-user.
865 2014-02-27 Jeff Law <law@redhat.com>
867 PR rtl-optimization/52714
868 * combine.c (try_combine): When splitting an unrecognized PARALLEL
869 into two independent simple sets, if I3 is a jump, ensure the
870 pattern we place into I3 is a (set (pc) ...).
872 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
873 Jeff Law <law@redhat.com>
875 PR rtl-optimization/49847
876 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
877 are in different blocks.
878 * doc/tm.texi (Condition Code Status): Update documention for
879 relative locations of cc0-setter and cc0-user.
881 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
884 * lra.c (lra_emit_add): Check SUBREG too.
886 2014-02-27 Andreas Schwab <schwab@suse.de>
888 * config/m68k/m68k.c (m68k_option_override): Disable
889 -flive-range-shrinkage for classic m68k.
890 (m68k_override_options_after_change): Likewise.
892 2014-02-27 Marek Polacek <polacek@redhat.com>
895 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
896 -Wmaybe-uninitialized.
898 2014-02-27 Alan Modra <amodra@gmail.com>
901 * reload1.c (emit_input_reload_insns): When reload_override_in,
902 set old to rl->in_reg when rl->in_reg is a subreg.
904 2014-02-26 Richard Biener <rguenther@suse.de>
907 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
909 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
911 * common/config/i386/predicates.md (const1256_operand): Remove.
912 (const2356_operand): New.
913 (const_1_to_2_operand): Remove.
914 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
915 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
916 (*avx512pf_gatherpf<mode>sf): Ditto.
917 (avx512pf_gatherpf<mode>df): Ditto.
918 (*avx512pf_gatherpf<mode>df_mask): Ditto.
919 (*avx512pf_gatherpf<mode>df): Ditto.
920 (avx512pf_scatterpf<mode>sf): Ditto.
921 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
922 (*avx512pf_scatterpf<mode>sf): Ditto.
923 (avx512pf_scatterpf<mode>df): Ditto.
924 (*avx512pf_scatterpf<mode>df_mask): Ditto.
925 (*avx512pf_scatterpf<mode>df): Ditto.
926 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
928 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
930 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
931 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
932 (_mm512_mask_testn_epi64_mask): Move to ...
933 * config/i386/avx512cdintrin.h: Here.
934 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
935 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
936 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
937 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
938 TARGET_AVX512F from TARGET_AVX512CD.
940 2014-02-26 Richard Biener <rguenther@suse.de>
943 * ipa.c (walk_polymorphic_call_targets): Properly guard
944 call to inline_update_overall_summary.
946 2014-02-26 Bin Cheng <bin.cheng@arm.com>
949 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
950 and latches only if requested. Fix latch if it is removed.
951 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
952 LOOPS_HAVE_PREHEADERS.
954 2014-02-25 Andrew Pinski <apinski@cavium.com>
956 * builtins.c (expand_builtin_thread_pointer): Create a new target
957 when the target is NULL.
959 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
961 PR rtl-optimization/60317
962 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
963 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
964 * lra-assigns.c: Include params.h.
965 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
966 other reload pseudos considerations.
968 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
970 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
971 to use canonical form for nor<mode>3.
973 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
976 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
979 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
981 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
982 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
983 (ix86_handle_option): Handle OPT_mprefetchwt1.
984 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
985 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
987 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
988 OPTION_MASK_ISA_PREFETCHWT1.
989 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
990 (PTA_PREFETCHWT1): New.
991 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
992 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
993 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
994 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
995 (*prefetch_avx512pf_<mode>_: Change into ...
996 (*prefetch_prefetchwt1_<mode>: This.
997 * config/i386/i386.opt (mprefetchwt1): New.
998 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
999 (_mm_prefetch): Handle intent to write.
1000 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1002 2014-02-25 Richard Biener <rguenther@suse.de>
1005 * emit-rtl.c (mem_attrs_htab): Remove.
1006 (mem_attrs_htab_hash): Likewise.
1007 (mem_attrs_htab_eq): Likewise.
1008 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1009 (init_emit_once): Do not allocate mem_attrs_htab.
1011 2014-02-25 Richard Biener <rguenther@suse.de>
1014 * lto-opts.c (lto_write_options): Output non-explicit conservative
1015 -fwrapv, -fno-trapv and -fno-strict-overflow.
1016 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1017 (run_gcc): And pass them through.
1019 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1021 * sel-sched.c (calculate_new_fences): New parameter ptime.
1022 Calculate it as a maximum over all fence cycles.
1023 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1024 Print the final schedule timing when sched_verbose.
1026 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1028 PR rtl-optimization/60292
1029 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1030 bit fot the fence instruction.
1032 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1034 * calls.h: Fix typo in comment.
1036 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1038 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1039 adjusting offsetable addresses.
1041 2014-02-24 Guozhi Wei <carrot@google.com>
1043 * sparseset.h (sparseset_pop): Fix the wrong index.
1045 2014-02-24 Walter Lee <walt@tilera.com>
1047 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1048 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1050 * common/config/tilegx/tilegx-common.c
1051 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1052 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1054 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1055 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1056 (tilegx_gimplify_va_arg_expr): Handle big endian.
1057 (tilegx_expand_unaligned_load): Ditto.
1058 (tilegx_expand_unaligned_store): Ditto.
1059 (TARGET_RETURN_IN_MSB): New.
1060 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1061 (TARGET_ENDIAN_DEFAULT): New.
1062 (TARGET_BIG_ENDIAN): Handle big endian.
1063 (BYTES_BIG_ENDIAN): Ditto.
1064 (WORDS_BIG_ENDIAN): Ditto.
1065 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1068 * config/tilegx/tilegx.md (extv): Handle big endian.
1070 (insn_st<n>): Ditto.
1071 (insn_st<n>_add<bitsuffix>): Ditto.
1072 (insn_stnt<n>): Ditto.
1073 (insn_stnt<n>_add<bitsuffix>):Ditto.
1074 (vec_interleave_highv8qi): Handle big endian.
1075 (vec_interleave_highv8qi_be): New.
1076 (vec_interleave_highv8qi_le): New.
1077 (insn_v1int_h): Handle big endian.
1078 (vec_interleave_lowv8qi): Handle big endian.
1079 (vec_interleave_lowv8qi_be): New.
1080 (vec_interleave_lowv8qi_le): New.
1081 (insn_v1int_l): Handle big endian.
1082 (vec_interleave_highv4hi): Handle big endian.
1083 (vec_interleave_highv4hi_be): New.
1084 (vec_interleave_highv4hi_le): New.
1085 (insn_v2int_h): Handle big endian.
1086 (vec_interleave_lowv4hi): Handle big endian.
1087 (vec_interleave_lowv4hi_be): New.
1088 (vec_interleave_lowv4hi_le): New.
1089 (insn_v2int_l): Handle big endian.
1090 (vec_interleave_highv2si): Handle big endian.
1091 (vec_interleave_highv2si_be): New.
1092 (vec_interleave_highv2si_le): New.
1093 (insn_v4int_h): Handle big endian.
1094 (vec_interleave_lowv2si): Handle big endian.
1095 (vec_interleave_lowv2si_be): New.
1096 (vec_interleave_lowv2si_le): New.
1097 (insn_v4int_l): Handle big endian.
1098 * config/tilegx/tilegx.opt (mbig-endian): New option.
1099 (mlittle-endian): New option.
1100 * doc/install.texi: Document tilegxbe-linux.
1101 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1103 2014-02-24 Martin Jambor <mjambor@suse.cz>
1106 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1107 there are no parameter descriptors.
1109 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1111 PR rtl-optimization/60268
1112 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1113 initialization to ...
1114 (sched_rgn_init): ... here.
1115 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1117 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1119 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1122 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1124 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1127 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1129 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1130 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1132 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1134 * config/microblaze/predicates.md: Add cmp_op predicate.
1135 * config/microblaze/microblaze.md: Add branch_compare instruction
1136 which uses cmp_op predicate and emits cmp insn before branch.
1137 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1138 to microblaze_expand_conditional_branch and consolidate logic.
1139 (microblaze_expand_conditional_branch): emit branch_compare
1140 insn instead of handling cmp op separate from branch insn.
1142 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1144 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1147 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1149 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1150 define_insn with define_expand and new define_insn
1151 *altivec_lve<VI_char>x_internal.
1152 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1153 and new define_insn *altivec_stve<VI_char>x_internal.
1154 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1156 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1158 (altivec_expand_stvex_be): New function.
1160 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1162 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1163 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1164 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1165 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1167 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1170 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1171 instead of emit_move_insn.
1173 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1175 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1177 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1178 gen_altivec_vsumsws.
1180 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1182 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1183 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1184 (altivec_lvxl_<mode>): New define_expand incorporating
1185 -maltivec=be semantics where needed.
1186 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1187 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1188 semantics where needed.
1189 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1190 (altivec_stvx_<mode>): New define_expand incorporating
1191 -maltivec=be semantics where needed.
1192 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1193 VM2 iterator instead of V4SI.
1194 (altivec_stvxl_<mode>): New define_expand incorporating
1195 -maltivec=be semantics where needed.
1196 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1197 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1198 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1199 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1200 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1201 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1202 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1203 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1204 ALTIVEC_BUILTIN_STVXL.
1205 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1206 (altivec_expand_stvx_be): Likewise.
1207 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1208 (altivec_expand_lvx_be): Likewise.
1209 (altivec_expand_stvx_be): Likewise.
1210 (altivec_expand_builtin): Add cases for
1211 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1212 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1213 (altivec_init_builtins): Add definitions for
1214 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1215 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1217 2014-02-21 Catherine Moore <clm@codesourcery.com>
1219 * doc/invoke.texi (mvirt, mno-virt): Document.
1220 * config/mips/mips.opt (mvirt): New option.
1221 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1223 2014-02-21 Richard Biener <rguenther@suse.de>
1225 PR tree-optimization/60276
1226 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1227 (STMT_VINFO_MIN_NEG_DIST): New macro.
1228 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1229 STMT_VINFO_MIN_NEG_DIST.
1230 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1231 made for negative dependence distances still hold.
1233 2014-02-21 Richard Biener <rguenther@suse.de>
1236 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1237 DECL_INITIAL for globals not in the current function context.
1239 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1241 PR tree-optimization/56490
1242 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1243 * tree-ssa-uninit.c: Include params.h.
1244 (compute_control_dep_chain): Add num_calls argument, return false
1245 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1246 num_calls to recursive call.
1247 (find_predicates): Change dep_chain into normal array,
1248 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1249 variable and adjust compute_control_dep_chain caller.
1250 (find_def_preds): Likewise.
1252 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1254 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1255 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1257 2014-02-21 Nick Clifton <nickc@redhat.com>
1259 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1260 (pushhi1): Likewise.
1261 (popqi1): Add mode to pre_dec.
1264 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1266 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1267 mode for mask of V8SFmode permutation.
1269 2014-02-20 Richard Henderson <rth@redhat.com>
1272 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1273 a new pseudo for OLDVAL.
1275 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1278 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1279 gen_reg_rtx if d->testing_p.
1280 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1281 if d->testing_p and we will certainly return true.
1282 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1285 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1287 * emit-rtl.c (gen_reg_rtx): Assert that
1288 crtl->emit.regno_pointer_align_length is non-zero.
1290 2014-02-20 Richard Henderson <rth@redhat.com>
1293 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1294 on failure the store back into EXPECT.
1296 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1297 Sandra Loosemore <sandra@codesourcery.com>
1299 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1300 * config/nios2/nios2.c (nios2_function_profiler): Add
1301 -fPIC (flag_pic == 2) support.
1302 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1303 (nios2_large_offset_p): New function.
1304 (nios2_unspec_reloc_p): Move up position, update to use
1305 nios2_large_offset_p.
1306 (nios2_unspec_address): Remove function.
1307 (nios2_unspec_offset): New function.
1308 (nios2_large_got_address): New function.
1309 (nios2_got_address): Add large offset support.
1310 (nios2_legitimize_tls_address): Update usage of removed and new
1312 (nios2_symbol_binds_local_p): New function.
1313 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1314 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1315 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1316 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1317 processing for (const (unspec ...)).
1318 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1320 2014-02-20 Richard Biener <rguenther@suse.de>
1322 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1323 doing the substitution.
1324 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1326 2014-02-20 Martin Jambor <mjambor@suse.cz>
1329 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1330 info when checking whether lattices are bottom.
1332 2014-02-20 Richard Biener <rguenther@suse.de>
1335 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1338 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1341 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1342 parameter specifying the scaling.
1343 (inline_call): Update.
1344 (want_inline_recursively): Guard division by zero.
1345 (recursive_inlining): Update.
1346 * ipa-inline.h (clone_inlined_nodes): Update.
1348 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1351 * config/i386/i386.c (classify_argument): Pass structures of size
1352 64 bytes or less in register.
1354 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1355 Kirill Yukhin <kirill.yukhin@intel.com>
1357 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1358 (_mm_rcp28_round_ss): Ditto.
1359 (_mm_rsqrt28_round_sd): Ditto.
1360 (_mm_rsqrt28_round_ss): Ditto.
1361 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1362 (_mm_rcp14_round_ss): Ditto.
1363 (_mm_rsqrt14_round_sd): Ditto.
1364 (_mm_rsqrt14_round_ss): Ditto.
1365 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1366 the first input operand, get rid of match_dup.
1367 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1369 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1371 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1372 operand as the first input operand, set type attribute.
1373 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1375 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1376 operand as the first input operand, set type attribute.
1378 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1380 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1383 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1386 * config/i386/i386.c (construct_container): Remove TFmode check
1387 for X86_64_INTEGER_CLASS.
1389 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1392 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1393 only when -Wpsabi is enabled.
1395 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1398 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1399 passing arrays in registers are the same as for structs, so remove the
1400 special case for them.
1402 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1404 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1405 destination type, extract only the valid bits if the source type is not
1406 integral and has a different mode.
1408 2014-02-19 Richard Biener <rguenther@suse.de>
1411 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1414 2014-02-19 Richard Biener <rguenther@suse.de>
1417 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1418 (ipa_modify_call_arguments): Emit an argument load explicitely and
1419 preserve virtual SSA form there and for the replacement call.
1420 Do not update SSA form nor free dominance info.
1422 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1424 * ipa.c (function_and_variable_visibility): Also clear WEAK
1425 flag when disolving COMDAT_GROUP.
1427 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1429 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1430 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1432 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1433 do more sanity checks.
1434 (detect_type_change): Return true when giving up early.
1435 (compute_complex_assign_jump_func): Fix type parameter of
1436 ipa_set_ancestor_jf.
1437 (compute_complex_ancestor_jump_func): Likewise.
1438 (update_jump_functions_after_inlining): Fix updating of
1440 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1442 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1444 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1445 inline clones when edge disappears.
1447 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1450 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1451 Split 64-bit moves into 2 patterns. Do not allow the use of
1452 direct move for TDmode in little endian, since the decimal value
1453 has little endian bytes within a word, but the 64-bit pieces are
1454 ordered in a big endian fashion, and normal subreg's of TDmode are
1456 (mov<mode>_64bit_dm): Likewise.
1457 (movtd_64bit_nodm): Likewise.
1459 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1461 PR tree-optimization/60174
1462 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1463 statement of an SSA_NAME that occurs in an abnormal PHI node.
1465 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1468 * final.c (SEEN_BB): Remove.
1469 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1470 (final_scan_insn): Don't force_source_line on second
1471 NOTE_INSN_BASIC_BLOCK.
1473 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1476 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1477 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1478 (type_natural_mode): Warn ABI change when %zmm register is not
1479 available for AVX512F vector value passing.
1481 2014-02-18 Kai Tietz <ktietz@redhat.com>
1484 * config/i386/i386.c (ix86_expand_prologue): Use value in
1485 rax register as displacement when restoring %r10 or %rax.
1486 Fix wrong offset when restoring both registers.
1488 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1490 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1491 assertion with conditional return.
1493 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1494 Uros Bizjak <ubizjak@gmail.com>
1497 * config/i386/driver-i386.c (host_detect_local_cpu): If
1498 YMM state is not saved by the OS, also clear has_f16c. Move
1499 CPUID 0x80000001 handling before YMM state saving checking.
1501 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1503 PR rtl-optimization/58960
1504 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1505 factored out from ...
1506 (sched_init): ... here.
1507 (free_global_sched_pressure_data): New, factored out from ...
1508 (sched_finish): ... here.
1509 * sched-int.h (free_global_sched_pressure_data): Declare.
1510 * sched-rgn.c (nr_regions_initial): New static global.
1511 (haifa_find_rgns): Initialize it.
1512 (schedule_region): Disable sched-pressure for the newly
1515 2014-02-17 Richard Biener <rguenther@suse.de>
1517 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1518 release SSA defs of pattern stmts.
1520 2014-02-17 Richard Biener <rguenther@suse.de>
1522 * tree-inline.c (expand_call_inline): Release the virtual
1523 operand defined by the call we are about to inline.
1525 2014-02-17 Richard Biener <rguenther@suse.de>
1527 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1529 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1530 Ilya Tocar <ilya.tocar@intel.com>
1532 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1533 arguments order in builtin.
1534 (_mm512_permutexvar_epi64): Ditto.
1535 (_mm512_mask_permutexvar_epi64): Ditto
1536 (_mm512_maskz_permutexvar_epi32): Ditto
1537 (_mm512_permutexvar_epi32): Ditto
1538 (_mm512_mask_permutexvar_epi32): Ditto
1540 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1542 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1543 (p8_vmrgow): Likewise.
1545 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1547 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1550 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1553 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1554 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1555 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1556 using direct move instructions on ISA 2.07. Also adjust
1557 instruction length for 64-bit.
1558 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1559 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1561 2014-02-15 Alan Modra <amodra@gmail.com>
1565 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1566 find_replacement on parts of insn rtl that might be reloaded.
1568 2014-02-15 Richard Biener <rguenther@suse.de>
1570 PR tree-optimization/60183
1571 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1572 (tree_ssa_phiprop): Calculate and free post-dominators.
1574 2014-02-14 Jeff Law <law@redhat.com>
1576 PR rtl-optimization/60131
1577 * ree.c (get_extended_src_reg): New function.
1578 (combine_reaching_defs): Use it rather than assuming location of REG.
1579 (find_and_remove_re): Verify first operand of extension is
1580 a REG before adding the insns to the copy list.
1582 2014-02-14 Roland McGrath <mcgrathr@google.com>
1584 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1585 * configure: Regenerated.
1586 * config.in: Regenerated.
1587 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1588 instead of ASM_SHORT.
1590 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1591 Richard Earnshaw <rearnsha@arm.com>
1593 PR rtl-optimization/59535
1594 * lra-constraints.c (process_alt_operands): Encourage alternative
1595 when unassigned pseudo class is superset of the alternative class.
1596 (inherit_reload_reg): Don't inherit when optimizing for code size.
1597 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1598 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1599 modes not less than 4 for Thumb1.
1601 2014-02-14 Kyle McMartin <kyle@redhat.com>
1604 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1606 2014-02-14 Richard Biener <rguenther@suse.de>
1608 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1609 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1610 Do not strip INDIRECT_REFs.
1612 2014-02-14 Richard Biener <rguenther@suse.de>
1615 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1616 DECL_FUNCTION_SPECIFIC_TARGET.
1617 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1618 * tree-streamer-out.c (pack_ts_target_option): Remove.
1619 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1620 (write_ts_function_decl_tree_pointers): Do not stream
1621 DECL_FUNCTION_SPECIFIC_TARGET.
1622 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1623 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1624 (lto_input_ts_function_decl_tree_pointers): Do not stream
1625 DECL_FUNCTION_SPECIFIC_TARGET.
1627 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1629 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1630 (get_initial_def_for_induction, vectorizable_induction): Ignore
1631 debug stmts when looking for exit_phi.
1632 (vectorizable_live_operation): Fix up condition.
1634 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1636 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1637 nreverse() because it changes the content of original tree list.
1639 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1641 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1642 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1644 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1646 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1647 GNU coding standards.
1649 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1652 * dwarf2out.c (gen_subprogram_die): Don't call
1653 add_calling_convention_attribute if subr_die is old_die.
1655 2014-02-13 Sharad Singhai <singhai@google.com>
1657 * doc/optinfo.texi: Fix order of nodes.
1659 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1661 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1662 operands[2], not operands[3].
1664 2014-02-13 Richard Biener <rguenther@suse.de>
1667 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1668 mention the possibility of an in-tree build.
1669 (CLooG): Update recommended version to 0.18.1, mention the
1670 possibility of an in-tree build and clarify that the ISL
1671 bundled with CLooG does not work.
1673 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1676 * expr.c (compress_float_constant): If x is a hard register,
1677 extend into a pseudo and then move to x.
1679 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1681 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1682 caused by bad second argument to warning_at() with -mhotpatch and
1683 nested functions (e.g. with gfortran).
1685 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1687 * opts.c (option_name): Remove "enabled by default" rider.
1689 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1691 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1693 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1694 Uros Bizjak <ubizjak@gmail.com>
1697 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1698 * configure: Regenerated.
1700 2014-02-12 Richard Biener <rguenther@suse.de>
1702 * vec.c (vec_prefix::calculate_allocation): Move as
1703 inline variant to vec.h.
1704 (vec_prefix::calculate_allocation_1): New out-of-line version.
1705 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1706 (vec_prefix::m_has_auto_buf): Rename to ...
1707 (vec_prefix::m_using_auto_storage): ... this.
1708 (vec_prefix::calculate_allocation): Inline the easy cases
1709 and dispatch to calculate_allocation_1 which doesn't need the
1711 (va_heap::reserve): Use gcc_checking_assert.
1712 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1713 m_using_auto_storage.
1714 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1716 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1717 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1718 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1720 2014-02-12 Richard Biener <rguenther@suse.de>
1722 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1723 when we found a dependence.
1725 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1727 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1729 (maybe_fold_stmt): ... into this new function.
1730 * omp-low.c (lower_omp): Update comment.
1732 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1735 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1738 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1740 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1741 identifiers in comments.
1742 (cortexa53_extra_costs): Likewise.
1743 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1744 (cortexa7_extra_costs): Likewise.
1745 (cortexa12_extra_costs): Likewise.
1746 (cortexa15_extra_costs): Likewise.
1747 (v7m_extra_costs): Likewise.
1749 2014-02-12 Richard Biener <rguenther@suse.de>
1752 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1753 of posix_memalign being successful.
1754 (lower_stmt): Restrict lowering of posix_memalign to when
1755 -ftree-bit-ccp is enabled.
1757 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1759 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1761 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1763 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1765 PR rtl-optimization/60116
1766 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1767 other_insn once the combination has been validated.
1769 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1772 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1774 * ipa-devirt.c: Include demangle.h
1775 (odr_violation_reported): New static variable.
1776 (add_type_duplicate): Update odr_violations.
1777 (maybe_record_node): Add completep parameter; update it.
1778 (record_target_from_binfo): Add COMPLETEP parameter;
1779 update it as needed.
1780 (possible_polymorphic_call_targets_1): Likewise.
1781 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1782 rename FINAL to COMPLETE.
1783 (record_targets_from_bases): Sanity check we found the binfo;
1784 fix COMPLETEP updating.
1785 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1786 parameter, fix computing of COMPLETEP.
1787 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1788 at LTO time do demangling.
1789 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1790 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1792 (gimple_get_virt_method_for_binfo): Likewise.
1793 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1794 gimple_get_virt_method_for_vtable): Update prototypes.
1796 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1799 * genautomata.c (add_presence_absence): Fix typo with
1800 {final_}presence_list.
1802 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1805 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1806 for VSX/Altivec vectors that land in GPR registers.
1808 2014-02-11 Richard Henderson <rth@redhat.com>
1809 Jakub Jelinek <jakub@redhat.com>
1812 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1813 around drhs if type conversion to lacc->type is not useless.
1815 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1817 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1819 (cortex-a57.cortex-a53): Likewise.
1820 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1822 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1824 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1827 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1829 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1830 add_options_for_arm_vfp3.
1832 2014-02-11 Jeff Law <law@redhat.com>
1835 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1836 object with an undesirable mode.
1838 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1841 * config/i386/sol2-9.h: New file.
1842 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1843 *-*-solaris2.9*): Use it.
1845 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1847 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1848 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1850 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1852 * config/microblaze/microblaze.c: Extend mcpu version format
1854 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1856 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1858 2014-02-10 Richard Henderson <rth@redhat.com>
1861 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1862 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1863 ms-abi vs -mno-accumulate-outgoing-args.
1864 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1865 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1868 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1871 * cfgexpand.c (expand_asm_operands): Attach source location to
1872 ASM_INPUT rtx objects.
1873 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1875 2014-02-10 Nick Clifton <nickc@redhat.com>
1877 * config/mn10300/mn10300.c (popcount): New function.
1878 (mn10300_expand_prologue): Include saved registers in stack usage
1881 2014-02-10 Jeff Law <law@redhat.com>
1884 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1885 when changing the SET_DEST of a prior insn to avoid an input reload.
1887 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1889 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1890 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1891 -mcall-openbsd, or -mcall-linux.
1892 (CC1_ENDIAN_BIG_SPEC): Remove.
1893 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1894 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1895 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1896 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1897 and %cc1_endian_default.
1898 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1900 2014-02-10 Richard Biener <rguenther@suse.de>
1902 PR tree-optimization/60115
1903 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1904 MEM_REF handling. Properly verify that the accesses are not
1905 out of the objects bound.
1907 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1909 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1912 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1914 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1915 proper constants and fix formatting.
1916 (possible_polymorphic_call_targets): Fix formatting.
1918 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1919 Ilya Tocar <ilya.tocar@intel.com>
1921 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1922 (_mm512_loadu_epi32): Renamed into...
1923 (_mm512_loadu_si512): This.
1924 (_mm512_storeu_epi32): Renamed into...
1925 (_mm512_storeu_si512): This.
1926 (_mm512_maskz_ceil_ps): Removed.
1927 (_mm512_maskz_ceil_pd): Ditto.
1928 (_mm512_maskz_floor_ps): Ditto.
1929 (_mm512_maskz_floor_pd): Ditto.
1930 (_mm512_floor_round_ps): Ditto.
1931 (_mm512_floor_round_pd): Ditto.
1932 (_mm512_ceil_round_ps): Ditto.
1933 (_mm512_ceil_round_pd): Ditto.
1934 (_mm512_mask_floor_round_ps): Ditto.
1935 (_mm512_mask_floor_round_pd): Ditto.
1936 (_mm512_mask_ceil_round_ps): Ditto.
1937 (_mm512_mask_ceil_round_pd): Ditto.
1938 (_mm512_maskz_floor_round_ps): Ditto.
1939 (_mm512_maskz_floor_round_pd): Ditto.
1940 (_mm512_maskz_ceil_round_ps): Ditto.
1941 (_mm512_maskz_ceil_round_pd): Ditto.
1942 (_mm512_expand_pd): Ditto.
1943 (_mm512_expand_ps): Ditto.
1944 * config/i386/i386.c (ix86_builtins): Remove
1945 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1946 (bdesc_args): Ditto.
1947 * config/i386/predicates.md (const1256_operand): New.
1948 (const_1_to_2_operand): Ditto.
1949 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1950 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1951 (*avx512pf_gatherpf<mode>sf): Ditto.
1952 (avx512pf_gatherpf<mode>df): Ditto.
1953 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1954 (*avx512pf_gatherpf<mode>df): Ditto.
1955 (avx512pf_scatterpf<mode>sf): Ditto.
1956 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1957 (*avx512pf_scatterpf<mode>sf): Ditto.
1958 (avx512pf_scatterpf<mode>df): Ditto.
1959 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1960 (*avx512pf_scatterpf<mode>df): Ditto.
1961 (avx512f_expand<mode>): Removed.
1962 (<shift_insn><mode>3<mask_name>): Change predicate type.
1964 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1966 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1967 not at the end of datarefs vector use ordered_remove to avoid
1968 reordering datarefs vector.
1971 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1972 mark local addressable non-static vars as GOVD_PRIVATE
1973 instead of GOVD_LOCAL.
1974 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1975 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1979 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1980 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1981 assume_aligned or alloc_align attributes.
1982 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1983 arguments. Handle also assume_aligned and alloc_align attributes.
1984 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1985 calls to functions with assume_aligned or alloc_align attributes.
1986 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1988 2014-02-08 Terry Guo <terry.guo@arm.com>
1990 * doc/invoke.texi: Document ARM -march=armv7e-m.
1992 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1994 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1995 flag on __cilkrts_rethrow builtin.
1998 * ipa-cp.c (determine_versionability): Fail at -O0
1999 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2000 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2003 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2006 * tree-inline.c (copy_forbidden): Fail for
2007 __attribute__((optimize (0))) functions.
2009 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2011 * varpool.c: Include pointer-set.h.
2012 (varpool_remove_unreferenced_decls): Variables in other partitions
2013 will not be output; be however careful to not lose information
2016 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2018 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2019 lookup in the vtable constructor.
2021 2014-02-07 Jeff Law <law@redhat.com>
2024 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2025 define_insn_and_split.
2027 * ipa-inline.c (inline_small_functions): Fix typos.
2029 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2031 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2032 (s390_can_use_return_insn): Declare.
2033 * config/s390/s390.h (EPILOGUE_USES): Define.
2034 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2036 (s390_chunkify_start): Handle return JUMP_LABELs.
2037 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2038 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2039 (s390_can_use_return_insn): New functions.
2040 (s390_fix_long_loop_prediction): Handle conditional returns.
2041 (TARGET_SET_UP_BY_PROLOGUE): Define.
2042 * config/s390/s390.md (ANY_RETURN): New code iterator.
2043 (*creturn, *csimple_return, return, simple_return): New patterns.
2045 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2047 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2048 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2049 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2050 REG_CFA_RESTORE list when deciding not to restore a register.
2052 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2054 * config/s390/s390.c: Include tree-pass.h and context.h.
2055 (s390_early_mach): New function, split out from...
2056 (s390_emit_prologue): ...here.
2057 (pass_data_s390_early_mach): New pass structure.
2058 (pass_s390_early_mach): New class.
2059 (s390_option_override): Create and register early_mach pass.
2060 Move to end of file.
2062 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2064 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2065 to match for the exit block.
2067 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2069 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2070 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2071 Reject misaligned operands.
2073 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2075 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2077 2014-02-07 Richard Biener <rguenther@suse.de>
2080 * gimple-low.c (lower_builtin_posix_memalign): New function.
2081 (lower_stmt): Call it to lower posix_memalign in a way
2082 to make alignment info accessible.
2084 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2087 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2088 __builtin_setjmp_receiver.
2090 2014-02-07 Richard Biener <rguenther@suse.de>
2093 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2094 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2095 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2096 Handle BUILT_IN_POSIX_MEMALIGN.
2097 (find_func_clobbers): Likewise.
2098 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2099 (call_may_clobber_ref_p_1): Likewise.
2101 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2104 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2107 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2110 * lto-cgraph.c (lto_output_node): Use
2111 symtab_get_symbol_partitioning_class.
2112 (lto_output_varpool_node): likewise.
2113 (symtab_get_symbol_partitioning_class): Move here from
2115 * cgraph.h (symbol_partitioning_class): Likewise.
2116 (symtab_get_symbol_partitioning_class): Declare.
2118 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2120 * ggc.h (ggc_internal_cleared_alloc): New macro.
2121 * vec.h (vec_safe_copy): Handle memory stats.
2122 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2123 * target-globals.c (save_target_globals): Likewise.
2125 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2128 * expr.c (emit_move_resolve_push): Export; be bit more selective
2129 on when to clear alias set.
2130 * expr.h (emit_move_resolve_push): Declare.
2131 * function.h (struct function): Add tail_call_marked.
2132 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2133 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2134 * config/i386/i386.md (TImode move expander): De not call
2136 (FP push expanders): Preserve memory attributes.
2137 * config/i386/sse.md (push<mode>1): Remove.
2138 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2139 (ix86_expand_push): Remove.
2140 * config/i386/mmx.md (push<mode>1): Remove.
2142 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2144 PR rtl-optimization/60030
2145 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2146 lopart with paradoxical subreg before shifting it up by hprec.
2148 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2150 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2151 Remove extra newline at end of file.
2152 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2153 (arm_issue_rate): Handle cortexa57.
2154 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2155 (cortex-a57.cortex-a53): Likewise.
2157 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2160 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2161 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2163 (arm_expand_prologue): Adjust all callers.
2164 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2165 info, registers also at the lowest numbered registers side. Use
2166 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2170 * var-tracking.c (adjust_mems): Before adding a SET to
2171 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2173 2014-02-06 Alan Modra <amodra@gmail.com>
2176 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2177 change SDmode to DDmode when lra_in_progress.
2179 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2182 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2183 free_data_ref on the dr first, and before goto again also set dr
2184 to the next dr. For simd_lane_access, free old datarefs[i] before
2185 overwriting it. For get_vectype_for_scalar_type failure, don't
2186 free_data_ref if simd_lane_access.
2188 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2191 * tree.h (opts_for_fn): New inline function.
2192 (opt_for_fn): Define.
2193 * config/i386/i386.c (ix86_function_regparm): Use
2194 opt_for_fn (decl, optimize) instead of optimize.
2196 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2198 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2199 for SYMBOL_REF in large memory model.
2201 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2203 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2205 (cortex-a57): Likewise.
2206 (cortex-a57.cortex-a53): Likewise.
2208 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2209 Kugan Vivekanandarajah <kuganv@linaro.org>
2211 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2213 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2215 2014-02-06 Richard Biener <rguenther@suse.de>
2217 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2218 set_loop_copy and initialize_original_copy_tables.
2220 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2222 * config/aarch64/aarch64-simd.md
2223 (aarch64_ashr_simddi): Change QI to SI.
2225 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2226 Jakub Jelinek <jakub@redhat.com>
2229 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2232 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2234 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2235 CODE_FOR_altivec_vpku[hw]um to
2236 CODE_FOR_altivec_vpku[hw]um_direct.
2237 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2238 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2239 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2240 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2242 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2244 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2245 generation for -maltivec=be.
2246 (altivec_vsumsws): Simplify redundant test.
2248 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2250 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2251 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2252 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2253 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2254 gen_altivec_vpkuwum.
2255 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2257 (altivec_vpks<VI_char>ss): Likewise.
2258 (altivec_vpks<VI_char>us): Likewise.
2259 (altivec_vpku<VI_char>us): Likewise.
2260 (altivec_vpku<VI_char>um): Likewise.
2261 (altivec_vpku<VI_char>um_direct): New (copy of
2262 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2264 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2265 target is little endian and -maltivec=be is not specified.
2266 (*altivec_vupkhs<VU_char>_direct): New (copy of
2267 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2268 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2269 target is little endian and -maltivec=be is not specified.
2270 (*altivec_vupkls<VU_char>_direct): New (copy of
2271 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2272 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2273 little endian and -maltivec=be is not specified.
2274 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2275 little endian and -maltivec=be is not specified.
2277 2014-02-05 Richard Henderson <rth@redhat.com>
2280 * combine-stack-adj.c: Revert r206943.
2281 * sched-int.h (struct deps_desc): Add last_args_size.
2282 * sched-deps.c (init_deps): Initialize it.
2283 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2284 contain REG_ARGS_SIZE notes.
2286 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2288 * lto-cgraph.c (asm_nodes_output): Make global.
2289 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2290 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2291 (driver_handle_option): Handle OPT_fwpa.
2293 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2296 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2297 a comment typo and formatting issue. If odr_hash hasn't been
2298 created, return vNULL and set *completep to false.
2301 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2302 bb with no successors.
2304 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2307 * doc/invoke.texi (-march): Clarify documentation for ARM.
2311 2014-02-05 Richard Biener <rguenther@suse.de>
2313 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2314 when not vectorizing because of too many alias checks.
2315 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2316 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2318 2014-02-05 Nick Clifton <nickc@redhat.com>
2320 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2321 accept extended registers in any mode when compiling for the MN10300.
2323 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2325 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2326 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2327 sanitization attributes.
2328 (can_inline_edge_p): Likewise.
2329 (sanitize_attrs_match_for_inline_p): New function.
2331 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2333 * ipa-prop.c (detect_type_change): Shor circuit testing of
2334 type changes on THIS pointer.
2336 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2339 * config/pa/pa.c (legitimize_tls_address): Return original address
2340 if not passed a SYMBOL_REF rtx.
2341 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2343 (pa_emit_move_sequence): Simplify TLS source operands.
2344 (pa_legitimate_constant_p): Reject all TLS constants.
2345 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2346 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2348 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2350 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2351 groups when we know they are controlled by LTO.
2352 * varasm.c (default_binds_local_p_1): If object is in other partition,
2353 it will be resolved locally.
2355 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2357 * config/host-linux.c (linux_gt_pch_use_address): Don't
2358 use SSIZE_MAX because it is not always defined.
2360 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2363 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2364 threshold for pseudo splitting.
2365 (update_ebb_live_info): Process call argument hard registers and
2366 hard registers from insn definition too.
2367 (max_small_class_regs_num): New constant.
2368 (inherit_in_ebb): Update live hard regs through EBBs. Update
2369 reloads_num only for small register classes. Don't split for
2372 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2375 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2378 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2380 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2383 2014-02-04 Marek Polacek <polacek@redhat.com>
2385 * gdbinit.in (pel): Define.
2387 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2389 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2392 2014-02-04 Richard Biener <rguenther@suse.de>
2395 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2396 in function context local.
2397 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2398 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2399 similar to LTO_imported_decl_ref.
2401 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2403 PR tree-optimization/60002
2404 * cgraphclones.c (build_function_decl_skip_args): Clear
2407 PR tree-optimization/60023
2408 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2409 false to gsi_replace.
2410 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2411 has been in some EH region and vec_stmt could throw, add
2412 vec_stmt into the same EH region.
2413 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2414 has no lhs, ignore it.
2415 * internal-fn.c (expand_MASK_LOAD): Likewise.
2418 * tree-inline.c (copy_forbidden): Fail for
2419 __attribute__((optimize (0))) functions.
2422 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2423 is set, copy one less argument.
2424 (expand_simd_clones): Don't subtract clone_info->inbranch
2425 from simd_clone_struct_alloc argument.
2427 PR rtl-optimization/57915
2428 * recog.c (simplify_while_replacing): If all unary/binary/relational
2429 operation arguments are constant, attempt to simplify those.
2432 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2433 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2435 2014-02-04 Richard Biener <rguenther@suse.de>
2437 PR tree-optimization/60012
2438 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2439 TBAA disambiguation to all DDRs.
2441 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2444 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2445 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2447 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2450 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2452 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2454 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2455 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2457 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2460 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2461 to figure out targets of polymorphic calls with known decl.
2462 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2463 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2464 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2465 (get_polymorphic_call_info): ... here.
2466 (get_polymorphic_call_info_from_invariant): New function.
2468 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2470 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2471 lookup via vtable pointer; check for type consistency
2472 and turn inconsitent facts into UNREACHABLE.
2473 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2474 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2475 type inconsistent querries; return UNREACHABLE instead.
2477 2014-02-03 Richard Henderson <rth@twiddle.net>
2480 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2481 already processed this node.
2482 (normalize_one_pred_1): Pass along mark_set.
2483 (normalize_one_pred): Create and destroy a pointer_set_t.
2484 (normalize_one_pred_chain): Likewise.
2486 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2488 PR gcov-profile/58602
2489 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2491 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2494 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2495 -fno-devirtualize; try to devirtualize by the knowledge of
2496 virtual table pointer given by aggregate propagation.
2497 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2498 (ipa_print_node_jump_functions): Dump also offset that
2499 is relevant for polymorphic calls.
2500 (determine_known_aggregate_parts): Add arg_type parameter; use it
2501 instead of determining the type from pointer type.
2502 (ipa_compute_jump_functions_for_edge): Update call of
2503 determine_known_aggregate_parts.
2504 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2505 (gimple_get_virt_method_for_binfo): ... here; simplify using
2506 vtable_pointer_value_to_vtable.
2507 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2508 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2509 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2510 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2512 (vtable_pointer_value_to_binfo): ... here.
2513 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2515 2014-02-03 Teresa Johnson <tejohnson@google.com>
2517 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2518 redef of outer loop index variable.
2520 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2524 * doc/extend.texi (Function Attributes): Typo.
2526 2014-02-03 Cong Hou <congh@google.com>
2528 PR tree-optimization/60000
2529 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2530 if the vectorized statement is a store. A store statement can only
2531 appear at the end of pattern statements.
2533 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2535 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2536 (ix86_option_override_internal): Default long double to 64-bit for
2537 32-bit Bionic and to 128-bit for 64-bit Bionic.
2539 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2540 TARGET_LONG_DOUBLE_128 is true.
2541 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2543 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2544 (mlong-double-64): Negate -mlong-double-128.
2545 (mlong-double-128): New option.
2547 * config/i386/i386-c.c (ix86_target_macros): Define
2548 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2550 * doc/invoke.texi: Document -mlong-double-128.
2552 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2554 PR rtl-optimization/60024
2555 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2557 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2559 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2561 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2563 PR rtl-optimization/57662
2564 * sel-sched.c (code_motion_path_driver): Do not mark already not
2565 existing blocks in the visiting bitmap.
2567 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2569 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2570 on the insn being emitted.
2572 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2573 Will Deacon <will.deacon@arm.com>
2575 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2577 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2579 * config/arm/arm-tables.opt: Regenerate.
2581 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2583 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2584 for vector types other than V16QImode.
2585 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2586 define_expand, and call altivec_expand_vec_perm_le when producing
2587 code with little endian element order.
2588 (*altivec_vperm_<mode>_internal): New insn having previous
2589 behavior of altivec_vperm_<mode>.
2590 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2591 altivec_expand_vec_perm_le when producing code with little endian
2593 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2594 behavior of altivec_vperm_<mode>_uns.
2596 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2598 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2599 (altivec_vsumsws): Add handling for -maltivec=be with a little
2601 (altivec_vsumsws_direct): New.
2602 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2603 gen_altivec_vsumsws.
2605 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2607 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2608 vtable_pointer_value_to_binfo): New functions.
2609 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2610 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2612 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2614 * config/nios2/nios2.md (load_got_register): Initialize GOT
2615 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2616 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2618 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2620 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2621 preserverd by passthrough, do not propagate the type.
2623 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2625 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2626 (mips_atomic_assign_expand_fenv): New function.
2627 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2629 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2631 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2632 (__builtin_mips_set_fcsr): Likewise.
2633 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2634 MIPS_USI_FTYPE_VOID.
2635 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2636 (mips16_expand_set_fcsr): Likewise.
2637 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2638 (mips16_set_fcsr_stub): Likewise.
2639 (mips16_get_fcsr_one_only_stub): New class.
2640 (mips16_set_fcsr_one_only_stub): Likewise.
2641 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2642 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2643 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2644 (hard_float): New availability predicate.
2645 (mips_builtins): Add get_fcsr and set_fcsr.
2646 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2647 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2648 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2649 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2650 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2653 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2655 * config/mips/mips.c (mips_one_only_stub): New class.
2656 (mips_need_mips16_rdhwr_p): Replace with...
2657 (mips16_rdhwr_stub): ...this new variable.
2658 (mips16_stub_call_address): New function.
2659 (mips16_rdhwr_one_only_stub): New class.
2660 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2661 (mips_output_mips16_rdhwr): Delete.
2662 (mips_finish_stub): New function.
2663 (mips_code_end): Use it to handle rdhwr stubs.
2665 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2668 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2669 when calculating size of integer atomic types.
2671 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2673 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2675 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2677 PR tree-optimization/60003
2678 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2679 * profile.c (branch_prob): Use gimple_call_builtin_p
2680 to check for BUILT_IN_SETJMP_RECEIVER.
2681 * tree-inline.c (copy_bb): Call notice_special_calls.
2683 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2686 * lra-constraints.c (process_alt_operands): Update reload_sum only
2689 2014-01-31 Richard Henderson <rth@redhat.com>
2692 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2693 until after else_eh is processed.
2695 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2697 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2698 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2699 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2700 in smmintrin.h, remove them.
2701 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2702 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2703 * config/i386/i386.md (ROUND_SAE): Fix value.
2704 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2705 (const48_operand): New.
2706 * config/i386/subst.md (round), (round_expand): Use
2707 const_4_or_8_to_11_operand.
2708 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2710 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2712 * config/i386/constraints.md (Yk): Swap meaning with k.
2713 * config/i386/i386.md (movhi_internal): Change Yk to k.
2714 (movqi_internal): Ditto.
2715 (*k<logic><mode>): Ditto.
2718 (kandn<mode>): Ditto.
2719 (*<code>hi_1): Ditto.
2720 (*<code>qi_1): Ditto.
2721 (kxnor<mode>): Ditto.
2722 (kortestzhi): Ditto.
2723 (kortestchi): Ditto.
2725 (*one_cmplhi2_1): Ditto.
2726 (*one_cmplqi2_1): Ditto.
2727 * config/i386/sse.md (): Change k to Yk.
2728 (avx512f_load<mode>_mask): Ditto.
2729 (avx512f_blendm<mode>): Ditto.
2730 (avx512f_store<mode>_mask): Ditto.
2731 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2732 (avx512f_storedqu<mode>_mask): Ditto.
2733 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2735 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2736 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2737 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2738 (avx512f_maskcmp<mode>3): Ditto.
2739 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2740 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2741 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2742 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2743 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2744 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2745 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2746 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2747 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2748 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2749 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2750 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2751 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2752 (vec_extract_lo_<mode>_maskm): Ditto.
2753 (vec_extract_hi_<mode>_maskm): Ditto.
2754 (avx512f_vternlog<mode>_mask): Ditto.
2755 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2756 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2757 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2758 (avx512f_<code>v8div16qi2_mask): Ditto.
2759 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2760 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2761 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2762 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2763 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2764 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2765 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2766 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2767 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2768 (avx512cd_maskb_vec_dupv8di): Ditto.
2769 (avx512cd_maskw_vec_dupv16si): Ditto.
2770 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2771 (avx512f_vpermi2var<mode>3_mask): Ditto.
2772 (avx512f_vpermi2var<mode>3_mask): Ditto.
2773 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2774 (*avx512f_gathersi<mode>): Ditto.
2775 (*avx512f_gathersi<mode>_2): Ditto.
2776 (*avx512f_gatherdi<mode>): Ditto.
2777 (*avx512f_gatherdi<mode>_2): Ditto.
2778 (*avx512f_scattersi<mode>): Ditto.
2779 (*avx512f_scatterdi<mode>): Ditto.
2780 (avx512f_compress<mode>_mask): Ditto.
2781 (avx512f_compressstore<mode>_mask): Ditto.
2782 (avx512f_expand<mode>_mask): Ditto.
2783 * config/i386/subst.md (mask): Change k to Yk.
2784 (mask_scalar_merge): Ditto.
2787 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2789 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2791 2014-01-31 Richard Biener <rguenther@suse.de>
2794 * builtins.c (fold_builtin_memory_op): Make sure to not
2795 use a floating-point mode or a boolean or enumeral type for
2798 2014-01-30 DJ Delorie <dj@redhat.com>
2800 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2801 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2802 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2803 whenever main() has an epilogue.
2805 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2807 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2808 unused variable "field".
2809 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2810 (vsx_mergeh_<mode>): Likewise.
2811 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2812 (altivec_vmrghh): Likewise.
2813 (altivec_vmrghw): Likewise.
2814 (altivec_vmrglb): Likewise.
2815 (altivec_vmrglh): Likewise.
2816 (altivec_vmrglw): Likewise.
2817 (altivec_vspltb): Add missing uses.
2818 (altivec_vsplth): Likewise.
2819 (altivec_vspltw): Likewise.
2820 (altivec_vspltsf): Likewise.
2822 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2825 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2826 frame related instructions.
2828 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2830 PR rtl-optimization/59959
2831 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2832 any reload of register whose subreg is invalid.
2834 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2836 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2837 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2838 Add missing return type - void.
2840 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2842 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2843 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2844 remove element index adjustment for endian (now handled in vsx.md
2846 (altivec_expand_vec_perm_const): Use
2847 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2848 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2849 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2850 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2851 define_expand and a new define_insn *altivec_vspltb_internal;
2852 adjust for -maltivec=be on a little endian target.
2853 (altivec_vspltb_direct): New.
2854 (altivec_vsplth): Divide into a define_expand and a new
2855 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2856 little endian target.
2857 (altivec_vsplth_direct): New.
2858 (altivec_vspltw): Divide into a define_expand and a new
2859 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2860 little endian target.
2861 (altivec_vspltw_direct): New.
2862 (altivec_vspltsf): Divide into a define_expand and a new
2863 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2864 a little endian target.
2866 2014-01-30 Richard Biener <rguenther@suse.de>
2868 PR tree-optimization/59993
2869 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2870 can propagate form the earlier stmt and avoid the transform
2871 when the intermediate result is needed.
2873 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2875 * README.Portability: Fix typo.
2877 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2879 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2880 comparison_operator with ordered_comparison_operator.
2882 2014-01-30 Nick Clifton <nickc@redhat.com>
2884 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2885 Rename to mn10300_store_multiple_regs.
2886 * config/mn10300/mn10300.c: Likewise.
2887 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2888 store_multiple_regs.
2889 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2890 Call mn10300_store_multiple_regs.
2892 2014-01-30 Nick Clifton <nickc@redhat.com>
2893 DJ Delorie <dj@redhat.com>
2895 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2896 %fp 2 to keep registers after it properly word-aligned.
2897 (rl78_alloc_physical_registers_umul): Handle the case where both
2898 input operands are the same.
2900 2014-01-30 Richard Biener <rguenther@suse.de>
2902 PR tree-optimization/59903
2903 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2906 2014-01-30 Jason Merrill <jason@redhat.com>
2909 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2912 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2914 2014-01-30 Richard Biener <rguenther@suse.de>
2916 PR tree-optimization/59951
2917 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2919 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2922 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2923 SFmode to DFmode case.
2925 2014-01-29 DJ Delorie <dj@redhat.com>
2927 * config/msp430/msp430.opt (-minrt): New.
2928 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2930 (ENDFILE_SPEC): Likewise.
2932 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2934 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2935 (estimate_function_body_sizes): Use it.
2937 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2940 * dwarf2out.c (is_cxx_auto): New.
2941 (is_base_type): Use it.
2942 (gen_type_die_with_usage): Likewise.
2944 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2946 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2947 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2948 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2949 -maltivec=be with LE targets.
2950 (vsx_mergeh_<mode>): Likewise.
2951 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2952 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2953 (altivec_vmrghb): Replace with define_expand and new
2954 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2955 (altivec_vmrghb_direct): New define_insn.
2956 (altivec_vmrghh): Replace with define_expand and new
2957 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2958 (altivec_vmrghh_direct): New define_insn.
2959 (altivec_vmrghw): Replace with define_expand and new
2960 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2961 (altivec_vmrghw_direct): New define_insn.
2962 (*altivec_vmrghsf): Adjust for endianness.
2963 (altivec_vmrglb): Replace with define_expand and new
2964 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2965 (altivec_vmrglb_direct): New define_insn.
2966 (altivec_vmrglh): Replace with define_expand and new
2967 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2968 (altivec_vmrglh_direct): New define_insn.
2969 (altivec_vmrglw): Replace with define_expand and new
2970 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2971 (altivec_vmrglw_direct): New define_insn.
2972 (*altivec_vmrglsf): Adjust for endianness.
2973 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2974 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2975 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2976 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2977 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2978 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2979 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2980 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2982 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2984 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2985 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2988 2014-01-29 Richard Biener <rguenther@suse.de>
2990 PR tree-optimization/58742
2991 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2992 associate_pointerplus_align.
2993 (associate_pointerplus_diff): New function.
2994 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2995 and associate_pointerplus_diff.
2997 2014-01-29 Richard Biener <rguenther@suse.de>
2999 * lto-streamer.h (LTO_major_version): Bump to 3.
3000 (LTO_minor_version): Reset to 0.
3002 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3004 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3005 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3006 (arm_file_start): Generate correct asm header for armv7ve.
3007 * config/arm/bpabi.h: Add multilib support for armv7ve.
3008 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3009 and cortex-a15 to armv7ve.
3010 * config/arm/t-aprofile: Add multilib support for armv7ve.
3011 * doc/invoke.texi: Document -march=armv7ve.
3013 2014-01-29 Richard Biener <rguenther@suse.de>
3015 PR tree-optimization/58742
3016 * tree-ssa-forwprop.c (associate_plusminus): Return true
3017 if we changed sth, defer EH cleanup to ...
3018 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3019 (simplify_mult): New function.
3021 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3024 PR tree-optimization/59920
3025 * tree.c (build_common_builtin_nodes): Remove
3026 __builtin_setjmp_dispatcher initialization.
3027 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3028 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3029 instead of gsi_after_labels + manually skipping debug stmts.
3030 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3031 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3032 * tree-inline.c (copy_edges_for_bb): Remove
3033 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3034 argument. Ignore computed_goto_p stmts. Don't call
3035 make_abnormal_goto_edges. If a call might need abnormal edges
3036 for non-local gotos, see if it already has an edge to
3037 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3038 with true argument, don't do anything then, otherwise add
3039 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3040 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3042 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3043 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3044 (lower_stmt): Don't set data->calls_builtin_setjmp.
3045 (lower_builtin_setjmp): Adjust comment.
3046 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3047 * tree-cfg.c (found_computed_goto): Remove.
3048 (factor_computed_gotos): Remove.
3049 (make_goto_expr_edges): Return bool, true for computed gotos.
3050 Don't call make_abnormal_goto_edges.
3051 (build_gimple_cfg): Don't set found_computed_goto, don't call
3052 factor_computed_gotos.
3053 (computed_goto_p): No longer static.
3054 (make_blocks): Don't set found_computed_goto.
3055 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3056 (make_edges): If make_goto_expr_edges returns true, push bb
3057 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3058 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3059 vector. Record mapping between bbs and OpenMP regions if there
3060 are any, adjust make_gimple_omp_edges caller. Call
3061 handle_abnormal_edges.
3062 (make_abnormal_goto_edges): Remove.
3063 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3064 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3065 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3066 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3067 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3068 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3069 filling *region also set *region_idx to (*region)->entry->index.
3072 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3073 For REGs set ORIGINAL_REGNO.
3075 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3077 * doc/md.texi: Mention that a target shouldn't implement
3078 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3081 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3083 PR tree-optimization/59594
3084 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3085 a copy of the datarefs vector rather than the vector itself.
3087 2014-01-28 Jason Merrill <jason@redhat.com>
3090 * dwarf2out.c (auto_die): New static.
3091 (gen_type_die_with_usage): Handle C++1y 'auto'.
3092 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3095 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3098 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3099 (SPEC_X32): Likewise.
3100 (SPEC_64): Likewise.
3101 * config/i386/i386.c (ix86_option_override_internal): Turn off
3102 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3104 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3105 * config/i386/i386.h (TARGET_16BIT): New macro.
3106 (TARGET_16BIT_P): Likewise.
3107 * config/i386/i386.opt: Add m16.
3108 * doc/invoke.texi: Document -m16.
3110 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3112 PR preprocessor/59935
3113 * input.c (location_get_source_line): Bail out on when line number
3114 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3116 2014-01-28 Richard Biener <rguenther@suse.de>
3118 PR tree-optimization/58742
3119 * tree-ssa-forwprop.c (associate_plusminus): Handle
3120 pointer subtraction of the form (T)(P + A) - (T)P.
3122 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3124 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3127 2014-01-28 Richard Biener <rguenther@suse.de>
3130 2014-01-28 Richard Biener <rguenther@suse.de>
3132 PR rtl-optimization/45364
3133 PR rtl-optimization/59890
3134 * var-tracking.c (local_get_addr_clear_given_value): Handle
3135 already cleared slot.
3136 (val_reset): Handle not allocated local_get_addr_cache.
3137 (vt_find_locations): Use post-order on the inverted CFG.
3139 2014-01-28 Richard Biener <rguenther@suse.de>
3141 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3143 2014-01-28 Richard Biener <rguenther@suse.de>
3145 PR rtl-optimization/45364
3146 PR rtl-optimization/59890
3147 * var-tracking.c (local_get_addr_clear_given_value): Handle
3148 already cleared slot.
3149 (val_reset): Handle not allocated local_get_addr_cache.
3150 (vt_find_locations): Use post-order on the inverted CFG.
3152 2014-01-28 Alan Modra <amodra@gmail.com>
3154 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3155 * configure.ac <recursive call for build != host>: Define
3156 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3157 and LD_FOR_BUILD too.
3158 * configure: Regenerate.
3160 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3162 * config/i386/i386.c (get_builtin_code_for_version): Separate
3163 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3164 Broadwell from Haswell.
3166 2014-01-27 Steve Ellcey <sellcey@mips.com>
3168 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3169 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3170 * config/mips/mips.c (mips_option_override): Change setting
3172 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3173 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3174 Change from Mask to Var.
3176 2014-01-27 Jeff Law <law@redhat.com>
3178 * ipa-inline.c (inline_small_functions): Fix typo.
3180 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3182 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3183 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3184 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3185 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3186 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3187 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3188 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3189 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3190 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3191 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3192 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3193 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3194 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3195 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3196 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3197 (_mm512_storeu_epi64): Ditto.
3198 (_mm512_cmpge_epi32_mask): Ditto.
3199 (_mm512_cmpge_epu32_mask): Ditto.
3200 (_mm512_cmpge_epi64_mask): Ditto.
3201 (_mm512_cmpge_epu64_mask): Ditto.
3202 (_mm512_cmple_epi32_mask): Ditto.
3203 (_mm512_cmple_epu32_mask): Ditto.
3204 (_mm512_cmple_epi64_mask): Ditto.
3205 (_mm512_cmple_epu64_mask): Ditto.
3206 (_mm512_cmplt_epi32_mask): Ditto.
3207 (_mm512_cmplt_epu32_mask): Ditto.
3208 (_mm512_cmplt_epi64_mask): Ditto.
3209 (_mm512_cmplt_epu64_mask): Ditto.
3210 (_mm512_cmpneq_epi32_mask): Ditto.
3211 (_mm512_cmpneq_epu32_mask): Ditto.
3212 (_mm512_cmpneq_epi64_mask): Ditto.
3213 (_mm512_cmpneq_epu64_mask): Ditto.
3214 (_mm512_expand_pd): Ditto.
3215 (_mm512_expand_ps): Ditto.
3216 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3217 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3218 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3219 * config/i386/i386.c (ix86_builtins): Add
3220 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3221 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3222 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3223 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3224 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3225 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3226 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3227 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3228 IX86_BUILTIN_PMOVUSQW512_MEM.
3229 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3230 __builtin_ia32_pmovsqd512mem_mask,
3231 __builtin_ia32_pmovqd512mem_mask,
3232 __builtin_ia32_pmovusqw512mem_mask,
3233 __builtin_ia32_pmovsqw512mem_mask,
3234 __builtin_ia32_pmovqw512mem_mask,
3235 __builtin_ia32_pmovusdw512mem_mask,
3236 __builtin_ia32_pmovsdw512mem_mask,
3237 __builtin_ia32_pmovdw512mem_mask,
3238 __builtin_ia32_pmovqb512mem_mask,
3239 __builtin_ia32_pmovusqb512mem_mask,
3240 __builtin_ia32_pmovsqb512mem_mask,
3241 __builtin_ia32_pmovusdb512mem_mask,
3242 __builtin_ia32_pmovsdb512mem_mask,
3243 __builtin_ia32_pmovdb512mem_mask.
3244 (bdesc_args): Add __builtin_ia32_expanddf512,
3245 __builtin_ia32_expandsf512.
3246 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3247 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3248 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3249 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3250 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3251 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3252 (avx512f_<code>v8div16qi2_mask_store): This.
3253 (avx512f_expand<mode>): New.
3255 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3257 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3259 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3260 (_mm512_prefetch_i32scatter_pd): Ditto.
3261 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3262 (_mm512_prefetch_i64scatter_pd): Ditto.
3263 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3264 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3265 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3266 (_mm512_prefetch_i32scatter_ps): Ditto.
3267 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3268 (_mm512_prefetch_i64scatter_ps): Ditto.
3269 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3270 * config/i386/i386-builtin-types.def: Define
3271 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3272 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3273 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3274 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3275 IX86_BUILTIN_SCATTERPFQPD.
3276 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3277 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3278 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3279 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3280 __builtin_ia32_scatterpfqps.
3281 (ix86_expand_builtin): Expand new built-ins.
3282 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3283 fix memory access data type.
3284 (*avx512pf_gatherpf<mode>_mask): Ditto.
3285 (*avx512pf_gatherpf<mode>): Ditto.
3286 (avx512pf_scatterpf<mode>): Ditto.
3287 (*avx512pf_scatterpf<mode>_mask): Ditto.
3288 (*avx512pf_scatterpf<mode>): Ditto.
3289 (GATHER_SCATTER_SF_MEM_MODE): New.
3290 (avx512pf_gatherpf<mode>df): Ditto.
3291 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3292 (*avx512pf_scatterpf<mode>df): Ditto.
3294 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3297 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3298 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3301 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3303 * common/config/arm/arm-common.c
3304 (arm_rewrite_mcpu): Handle multiple names.
3306 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3308 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3310 * gimple-builder.h (create_gimple_tmp): Delete.
3312 2014-01-27 Christian Bruel <christian.bruel@st.com>
3314 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3317 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3319 * config/pa/pa.md (call): Generate indirect long calls to non-local
3320 functions when outputing 32-bit code.
3321 (call_value): Likewise except for special call to buggy powf function.
3323 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3324 portable runtime and PIC indirect calls.
3325 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3326 and PIC call sequences. Use ldo instead of blr to set return register
3327 in PIC call sequence.
3329 2014-01-25 Walter Lee <walt@tilera.com>
3331 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3332 avoid clobbering a live register.
3334 2014-01-25 Walter Lee <walt@tilera.com>
3336 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3337 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3338 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3339 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3341 2014-01-25 Walter Lee <walt@tilera.com>
3343 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3344 arguments on even registers.
3345 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3347 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3348 (BIGGEST_ALIGNMENT): Ditto.
3349 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3351 2014-01-25 Walter Lee <walt@tilera.com>
3353 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3354 insns before bundling.
3355 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3357 2014-01-25 Walter Lee <walt@tilera.com>
3359 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3360 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3361 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3363 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3365 * config/mips/constraints.md (kl): Delete.
3366 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3367 define expands, using...
3368 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3369 instructions for MIPS16.
3370 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3371 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3373 2014-01-25 Walter Lee <walt@tilera.com>
3375 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3379 2014-01-25 Walter Lee <walt@tilera.com>
3381 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3382 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3384 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3386 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3389 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3391 * print-rtl.c (in_call_function_usage): New var.
3392 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3393 EXPR_LIST mode as mode and not as reg note name.
3396 * cfgloopmanip.c (copy_loop_info): If
3397 loop->warned_aggressive_loop_optimizations, make sure
3398 the flag is set in target loop too.
3400 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3402 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3404 * builtins.def: Likewise.
3405 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3406 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3407 * ira.c (ira_setup_eliminable_regset): Likewise.
3408 * omp-low.c (gate_expand_omp): Likewise.
3409 (execute_lower_omp): Likewise.
3410 (diagnose_sb_0): Likewise.
3411 (gate_diagnose_omp_blocks): Likewise.
3412 (simd_clone_clauses_extract): Likewise.
3415 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3417 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3418 correction for little endian...
3419 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3422 2014-01-24 Jeff Law <law@redhat.com>
3424 PR tree-optimization/59919
3425 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3426 for non-returning calls.
3428 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3430 * common/config/aarch64/aarch64-common.c
3431 (aarch64_rewrite_mcpu): Handle multiple names.
3432 * config/aarch64/aarch64.h
3433 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3435 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3437 * input.c (add_file_to_cache_tab): Handle the case where fopen
3440 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3443 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3444 from push operand if code of push isn't PRE_DEC.
3446 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3449 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3450 -mquad-memory-atomic. Update -mquad-memory documentation to say
3451 it is only used for non-atomic loads/stores.
3453 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3454 -mquad-memory or -mquad-memory-atomic switches.
3456 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3457 -mquad-memory-atomic to ISA 2.07 support.
3459 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3460 to separate support of normal quad word memory operations (ldq, stq)
3461 from the atomic quad word memory operations.
3463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3464 support to separate non-atomic quad word operations from atomic
3465 quad word operations. Disable non-atomic quad word operations in
3466 little endian mode so that we don't have to swap words after the
3467 load and before the store.
3468 (quad_load_store_p): Add comment about atomic quad word support.
3469 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3470 options printed with -mdebug=reg.
3472 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3473 -mquad-memory-atomic as the test for whether we have quad word
3474 atomic instructions.
3475 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3476 or -mp8-vector are used, allow byte/half-word atomic operations.
3478 * config/rs6000/sync.md (load_lockedti): Insure that the address
3479 is a proper indexed or indirect address for the lqarx instruction.
3480 On little endian systems, swap the hi/lo registers after the lqarx
3482 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3483 insure the address is valid for the lqarx instruction.
3484 (store_conditionalti): Insure that the address is a proper indexed
3485 or indirect address for the stqcrx. instruction. On little endian
3486 systems, swap the hi/lo registers before doing the stqcrx.
3488 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3489 insure the address is valid for the stqcrx. instruction.
3491 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3492 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3493 type of quad memory support is available.
3495 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3498 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3499 there is a danger of looping.
3501 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3503 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3504 force flag_ira_loop_pressure if set via command line.
3506 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3508 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3509 (ashr_simd): New builtin handling DI mode.
3510 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3511 (aarch64_sshr_simddi): New match pattern.
3512 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3513 (vshrd_n_s64): Likewise.
3514 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3516 2014-01-23 Nick Clifton <nickc@redhat.com>
3518 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3519 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3520 favour of mcu specific scripts.
3521 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3524 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3525 Alex Velenko <Alex.Velenko@arm.com>
3527 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3528 (vaddv_s16): Likewise.
3529 (vaddv_s32): Likewise.
3530 (vaddv_u8): Likewise.
3531 (vaddv_u16): Likewise.
3532 (vaddv_u32): Likewise.
3533 (vaddvq_s8): Likewise.
3534 (vaddvq_s16): Likewise.
3535 (vaddvq_s32): Likewise.
3536 (vaddvq_s64): Likewise.
3537 (vaddvq_u8): Likewise.
3538 (vaddvq_u16): Likewise.
3539 (vaddvq_u32): Likewise.
3540 (vaddvq_u64): Likewise.
3541 (vaddv_f32): Likewise.
3542 (vaddvq_f32): Likewise.
3543 (vaddvq_f64): Likewise.
3544 (vmaxv_f32): Likewise.
3545 (vmaxv_s8): Likewise.
3546 (vmaxv_s16): Likewise.
3547 (vmaxv_s32): Likewise.
3548 (vmaxv_u8): Likewise.
3549 (vmaxv_u16): Likewise.
3550 (vmaxv_u32): Likewise.
3551 (vmaxvq_f32): Likewise.
3552 (vmaxvq_f64): Likewise.
3553 (vmaxvq_s8): Likewise.
3554 (vmaxvq_s16): Likewise.
3555 (vmaxvq_s32): Likewise.
3556 (vmaxvq_u8): Likewise.
3557 (vmaxvq_u16): Likewise.
3558 (vmaxvq_u32): Likewise.
3559 (vmaxnmv_f32): Likewise.
3560 (vmaxnmvq_f32): Likewise.
3561 (vmaxnmvq_f64): Likewise.
3562 (vminv_f32): Likewise.
3563 (vminv_s8): Likewise.
3564 (vminv_s16): Likewise.
3565 (vminv_s32): Likewise.
3566 (vminv_u8): Likewise.
3567 (vminv_u16): Likewise.
3568 (vminv_u32): Likewise.
3569 (vminvq_f32): Likewise.
3570 (vminvq_f64): Likewise.
3571 (vminvq_s8): Likewise.
3572 (vminvq_s16): Likewise.
3573 (vminvq_s32): Likewise.
3574 (vminvq_u8): Likewise.
3575 (vminvq_u16): Likewise.
3576 (vminvq_u32): Likewise.
3577 (vminnmv_f32): Likewise.
3578 (vminnmvq_f32): Likewise.
3579 (vminnmvq_f64): Likewise.
3581 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3583 * config/aarch64/aarch64-simd.md
3584 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3585 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3586 (*aarch64_mul3_elt<mode>): Likewise.
3587 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3588 (*aarch64_mul3_elt_to_64v2df): Likewise.
3589 (*aarch64_mla_elt<mode>): Likewise.
3590 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3591 (*aarch64_mls_elt<mode>): Likewise.
3592 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3593 (*aarch64_fma4_elt<mode>): Likewise.
3594 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3595 (*aarch64_fma4_elt_to_64v2df): Likewise.
3596 (*aarch64_fnma4_elt<mode>): Likewise.
3597 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3598 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3599 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3600 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3601 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3602 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3603 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3604 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3605 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3607 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3609 * config/aarch64/aarch64-simd.md
3610 (aarch64_be_checked_get_lane<mode>): New define_expand.
3611 * config/aarch64/aarch64-simd-builtins.def
3612 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3613 New builtin definition.
3614 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3615 Use new safe be builtin.
3617 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3619 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3621 (aarch64_be_st1<mode>): Likewise.
3622 (aarch_ld1<VALL:mode>): Define_expand modified.
3623 (aarch_st1<VALL:mode>): Likewise.
3624 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3625 (UNSPEC_ST1): Likewise.
3627 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3629 * config/microblaze/microblaze.md: Add trap insn and attribute
3631 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3633 PR preprocessor/58580
3634 * input.h (location_get_source_line): Take an additional line_size
3636 (void diagnostics_file_cache_fini): Declare new function.
3637 * input.c (struct fcache): New type.
3638 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3639 New static constants.
3640 (diagnostic_file_cache_init, total_lines_num)
3641 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3642 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3643 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3644 (get_next_line, read_next_line, goto_next_line, read_line_num):
3645 New static function definitions.
3646 (diagnostic_file_cache_fini): New function.
3647 (location_get_source_line): Take an additional output line_len
3648 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3650 * diagnostic.c (diagnostic_finish): Call
3651 diagnostic_file_cache_fini.
3652 (adjust_line): Take an additional input parameter for the length
3653 of the line, rather than calculating it with strlen.
3654 (diagnostic_show_locus): Adjust the use of
3655 location_get_source_line and adjust_line with respect to their new
3656 signature. While displaying a line now, do not stop at the first
3657 null byte. Rather, display the zero byte as a space and keep
3658 going until we reach the size of the line.
3659 * Makefile.in: Add vec.o to OBJS-libcommon
3661 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3662 Ilya Tocar <ilya.tocar@intel.com>
3664 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3665 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3666 (__builtin_ia32_kmov16): Ditto.
3667 * config/i386/i386.md (UNSPEC_KMOV): New.
3670 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3672 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3673 (_mm512_storeu_si512): Ditto.
3675 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3678 * rtl.h (get_referenced_operands): Declare.
3679 * recog.c (get_referenced_operands): New function.
3680 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3681 operands have been referenced when recording LO_SUM references.
3683 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3685 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3687 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3689 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3690 Enable for generic and recent AMD targets.
3692 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3694 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3695 ARG_SIZE note when adjustment was eliminated.
3697 2014-01-22 Jeff Law <law@redhat.com>
3699 PR tree-optimization/59597
3700 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3701 in file. Accept new argument REGISTERING and use it to modify
3702 dump output appropriately.
3703 (register_jump_thread): Corresponding changes.
3704 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3705 thread paths involving joiner blocks. Add code to dump cancelled
3706 jump threading paths.
3708 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3710 PR rtl-optimization/59477
3711 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3712 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3714 2014-01-22 Tom Tromey <tromey@redhat.com>
3716 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3718 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3720 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3722 PR rtl-optimization/59896
3723 * lra-constraints.c (process_alt_operands): Check unused note for
3724 matched operands of insn with no output reloads.
3726 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3728 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3729 (mips_move_from_gpr_cost): Likewise.
3731 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3733 PR rtl-optimization/59858
3734 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3735 ira_class_hard_regs_num.
3736 (process_alt_operands): Increase reject for dying matched operand.
3738 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3741 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3742 smaller than size, perform several stores or loads and stores
3743 at dst + count - size to store or copy all of size bytes, rather
3744 than just last modesize bytes.
3746 2014-01-20 DJ Delorie <dj@redhat.com>
3748 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3749 that CLOBBERs are REGs before propogating their values.
3751 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3754 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3755 (cgraph_inline_failed_type): New function.
3756 * cgraph.h (DEFCIFCODE): Add type.
3757 (cgraph_inline_failed_type_t): New enum.
3758 (cgraph_inline_failed_type): New prototype.
3759 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3760 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3761 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3762 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3763 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3764 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3765 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3766 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3767 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3768 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3769 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3770 OPTIMIZATION_MISMATCH.
3771 * tree-inline.c (expand_call_inline): Emit errors during
3772 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3774 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3777 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3778 mode attribute in insn output.
3780 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3782 * output.h (output_constant): Delete.
3783 * varasm.c (output_constant): Make private.
3785 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3787 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3789 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3792 * tree.h (fold_builtin_strcat): New prototype.
3793 * builtins.c (fold_builtin_strcat): No longer static. Add len
3794 argument, if non-NULL, don't call c_strlen. Optimize
3795 directly into __builtin_memcpy instead of __builtin_strcpy.
3796 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3797 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3799 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3801 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3802 for SImode_address_operand operands, having only a REG argument.
3804 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3806 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3807 loader name using mbig-endian.
3808 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3810 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3812 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3816 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3818 * config/aarch64/aarch64-protos.h
3819 (aarch64_cannot_change_mode_class_ptr): Declare.
3820 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3821 aarch64_cannot_change_mode_class_ptr): New.
3822 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3823 backend hook aarch64_cannot_change_mode_class.
3825 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3827 * common/config/aarch64/aarch64-common.c
3828 (aarch64_handle_option): Don't handle any option order logic here.
3829 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3830 selected_cpu, warn on architecture version mismatch.
3831 (aarch64_override_options): Fix parsing order for option strings.
3833 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3834 Iain Sandoe <iain@codesourcery.com>
3837 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3838 warning. Amend comment to reflect current functionality.
3840 2014-01-20 Richard Biener <rguenther@suse.de>
3843 * builtins.c (fold_builtin_strcat): Remove case better handled
3844 by tree-ssa-strlen.c.
3846 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3848 * config/aarch64/aarch64.opt
3849 (mcpu, march, mtune): Make case-insensitive.
3851 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3854 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3855 if operands[1] is a REG or ZERO_EXTEND of a REG.
3857 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3859 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3861 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3863 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3864 long non-pic millicode calls.
3866 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3868 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3870 2014-01-19 Kito Cheng <kito@0xlab.org>
3872 * builtins.c (expand_movstr): Check movstr expand done or fail.
3874 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3875 H.J. Lu <hongjiu.lu@intel.com>
3878 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3879 to DImode for zero-extended addresses.
3881 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3883 PR rtl-optimization/57763
3884 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3885 on the new indirect jump_insn and increment LABEL_NUSES (label).
3887 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3891 * config.gcc (x86_archs): New variable.
3892 (x86_64_archs): Likewise.
3893 (x86_cpus): Likewise.
3894 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3895 --with-arch/--with-cpu= options.
3896 Support --with-arch=/--with-cpu={nehalem,westmere,
3897 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3899 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3901 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3902 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3904 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3906 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3908 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3911 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3912 clear cpp_get_options (parse_in)->warn_unused_macros for
3913 ix86_target_macros_internal with cpp_define.
3915 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3917 * jump.c (delete_related_insns): Keep (use (insn))s.
3918 * reorg.c (redundant_insn): Check for barriers too.
3920 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3922 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3924 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3926 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3927 call to $$dyncall when TARGET_LONG_CALLS is true.
3929 2014-01-17 Jeff Law <law@redhat.com>
3931 * ree.c (combine_set_extension): Temporarily disable test for
3932 changing number of hard registers.
3934 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3937 * ipa-inline-analysis.c (inline_free_summary):
3938 Do not free summary of aliases.
3940 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3943 * gimplify.c (gimplify_expr): Use create_tmp_var
3944 instead of create_tmp_var_raw. If cond doesn't have
3945 integral type, don't add the IFN_ANNOTATE builtin at all.
3947 2014-01-17 Martin Jambor <mjambor@suse.cz>
3950 * ipa-cp.c (prev_edge_clone): New variable.
3951 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3952 Also resize prev_edge_clone vector.
3953 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3954 (ipcp_edge_removal_hook): New function.
3955 (ipcp_driver): Register ipcp_edge_removal_hook.
3957 2014-01-17 Andrew Pinski <apinski@cavium.com>
3958 Steve Ellcey <sellcey@mips.com>
3961 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3964 2014-01-17 Jeff Law <law@redhat.com>
3967 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3968 so that pass_ccp runs first.
3970 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3972 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3973 (ix86_adjust_cost): Use !TARGET_XXX.
3974 (do_reorder_for_imul): Likewise.
3975 (swap_top_of_ready_list): Likewise.
3976 (ix86_sched_reorder): Likewise.
3978 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3980 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3981 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3982 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3983 (intel_memset): New. Duplicate slm_memset.
3984 (intel_cost): New. Duplicate slm_cost.
3985 (m_INTEL): New macro.
3986 (processor_target_table): Add "intel".
3987 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3988 with PROCESSOR_INTEL for "intel".
3989 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3990 PROCESSOR_SILVERMONT.
3991 (ix86_issue_rate): Likewise.
3992 (ix86_adjust_cost): Likewise.
3993 (ia32_multipass_dfa_lookahead): Likewise.
3994 (swap_top_of_ready_list): Likewise.
3995 (ix86_sched_reorder): Likewise.
3996 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3997 instead of TARGET_OPT_AGU.
3998 * config/i386/i386.h (TARGET_INTEL): New.
3999 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4000 (processor_type): Add PROCESSOR_INTEL.
4001 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4002 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4004 2014-01-17 Marek Polacek <polacek@redhat.com>
4007 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4010 2014-01-17 Richard Biener <rguenther@suse.de>
4012 PR tree-optimization/46590
4013 * opts.c (default_options_table): Add entries for
4014 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4015 all enabled at -O1 but not for -Og.
4016 * common.opt (fbranch-count-reg): Remove Init(1).
4017 (fmove-loop-invariants): Likewise.
4018 (ftree-pta): Likewise.
4020 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4022 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4023 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4024 decls to at least the GCC 4.8 used alignments.
4027 * tree-nested.c (convert_nonlocal_reference_stmt,
4028 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4029 of GIMPLE_BIND stmts, adjust associated decls.
4031 2014-01-17 Richard Biener <rguenther@suse.de>
4033 PR tree-optimization/46590
4034 * vec.h (vec<>::bseach): New member function implementing
4035 binary search according to C89 bsearch.
4036 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4037 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4038 bitmap pointer again. Make accesses_in_loop a flat array.
4039 (mem_ref_obstack): New global.
4040 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4041 (mark_ref_stored): Likewise.
4042 (ref_indep_loop_p_2): Likewise.
4043 (set_ref_stored_in_loop): New helper function.
4044 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4045 (memref_free): Adjust.
4046 (record_mem_ref_loc): Simplify.
4047 (gather_mem_refs_stmt): Adjust.
4048 (sort_locs_in_loop_postorder_cmp): New function.
4049 (analyze_memory_references): Sort accesses_in_loop after
4050 loop postorder number.
4051 (find_ref_loc_in_loop_cmp): New function.
4052 (for_all_locs_in_loop): Find relevant cluster of locs in
4053 accesses_in_loop and iterate without recursion.
4054 (execute_sm): Avoid uninit warning.
4055 (struct ref_always_accessed): Simplify.
4056 (ref_always_accessed::operator ()): Likewise.
4057 (ref_always_accessed_p): Likewise.
4058 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4059 loop postorder numbers here.
4060 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4063 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4066 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4067 on decls for which assemble_alias has been called.
4069 2014-01-17 Nick Clifton <nickc@redhat.com>
4071 * config/msp430/msp430.opt: (mcpu): New option.
4072 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4073 (msp430_option_override): Parse target_cpu. If the MCU name
4074 matches a generic string, clear target_mcu.
4075 (msp430_attr): Allow numeric interrupt values up to 63.
4076 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4077 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4079 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4081 * config/msp430/msp430.md (popm): Use %J rather than %I.
4082 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4083 (addhi_cy_i): Use immediate_operand for operand 2.
4084 * doc/invoke.texi: Document -mcpu option.
4086 2014-01-17 Richard Biener <rguenther@suse.de>
4088 PR rtl-optimization/38518
4089 * df.h (df_analyze_loop): Declare.
4090 * df-core.c: Include cfgloop.h.
4091 (df_analyze_1): Split out main part of df_analyze.
4092 (df_analyze): Adjust.
4093 (loop_inverted_post_order_compute): New function.
4094 (loop_post_order_compute): Likewise.
4095 (df_analyze_loop): New function avoiding whole-function
4097 * loop-invariant.c (find_defs): Use df_analyze_loop.
4098 (find_invariants): Adjust.
4099 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4101 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4103 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4104 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4106 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4108 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4109 traversal when removing references.
4111 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4114 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4116 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4119 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4120 pushing a reload for an autoinc when we had previously reloaded an
4121 inner part of the address.
4123 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4125 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4127 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4128 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4129 when not giving up or versioning for alias only because of
4131 (vect_analyze_data_ref_dependences): Set to true.
4132 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4134 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4135 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4139 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4142 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4143 operand 0 predicate for gathers, use a new pseudo as subtarget.
4145 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4148 * lra-constraints.c (process_alt_operands): Add printing debug info.
4149 Check absence of input/output reloads for matched operands too.
4151 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4153 PR rtl-optimization/59835
4154 * ira.c (ira_init_register_move_cost): Increase cost for
4157 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4159 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4161 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4164 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4165 non-register objects. Use gen_(high/low)part more consistently.
4168 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4171 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4172 endian support, remove tests for WORDS_BIG_ENDIAN.
4173 (p8_mfvsrd_3_<mode>): Likewise.
4174 (reload_gpr_from_vsx<mode>): Likewise.
4175 (reload_gpr_from_vsxsf): Likewise.
4176 (p8_mfvsrd_4_disf): Likewise.
4178 2014-01-16 Richard Biener <rguenther@suse.de>
4180 PR rtl-optimization/46590
4181 * lcm.c (compute_antinout_edge): Use postorder iteration.
4182 (compute_laterin): Use inverted postorder iteration.
4184 2014-01-16 Nick Clifton <nickc@redhat.com>
4187 * varasm.c (output_constant): Return the number of bytes actually
4189 (output_constructor_array_range): Update the field size with the
4190 number of bytes emitted by output_constant.
4191 (output_constructor_regular_field): Likewise. Also do not
4192 complain if the total number of bytes emitted is now greater
4193 than the expected fieldpos.
4194 * output.h (output_constant): Update prototype and descriptive comment.
4196 2014-01-16 Marek Polacek <polacek@redhat.com>
4199 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4200 it is error_mark_node.
4202 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4204 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4205 VALID_AVX256_REG_OR_OI_MODE.
4207 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4209 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4210 current procedure should be profiled.
4212 2014-01-15 Andrew Pinski <apinski@cavium.com>
4214 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4215 of moving from/to the STACK_REG register class.
4217 2014-01-15 Richard Henderson <rth@redhat.com>
4220 * reginfo.c (global_regs_decl): Globalize.
4221 * rtl.h (global_regs_decl): Declare.
4222 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4223 reserved via global_regs.
4225 2014-01-15 Teresa Johnson <tejohnson@google.com>
4227 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4229 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4231 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4232 and vmulosh rather than call gen_vec_widen_smult_*.
4233 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4234 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4235 (vec_widen_smult_even_v16qi): Likewise.
4236 (vec_widen_umult_even_v8hi): Likewise.
4237 (vec_widen_smult_even_v8hi): Likewise.
4238 (vec_widen_umult_odd_v16qi): Likewise.
4239 (vec_widen_smult_odd_v16qi): Likewise.
4240 (vec_widen_umult_odd_v8hi): Likewise.
4241 (vec_widen_smult_odd_v8hi): Likewise.
4242 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4243 vmuloub rather than call gen_vec_widen_umult_*.
4244 (vec_widen_umult_lo_v16qi): Likewise.
4245 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4246 vmulosb rather than call gen_vec_widen_smult_*.
4247 (vec_widen_smult_lo_v16qi): Likewise.
4248 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4249 rather than call gen_vec_widen_umult_*.
4250 (vec_widen_umult_lo_v8hi): Likewise.
4251 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4252 rather than call gen_vec_widen_smult_*.
4253 (vec_widen_smult_lo_v8hi): Likewise.
4255 2014-01-15 Jeff Law <law@redhat.com>
4257 PR tree-optimization/59747
4258 * ree.c (find_and_remove_re): Properly handle case where a second
4259 eliminated extension requires widening a copy created for elimination
4260 of a prior extension.
4261 (combine_set_extension): Ensure that the number of hard regs needed
4262 for a destination register does not change when we widen it.
4264 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4266 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4267 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4268 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4269 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4270 (avr-*-rtems*): Likewise.
4271 (bfin*-rtems*): Likewise.
4272 (moxie-*-rtems*): Likewise.
4273 (h8300-*-rtems*): Likewise.
4274 (i[34567]86-*-rtems*): Likewise.
4275 (lm32-*-rtems*): Likewise.
4276 (m32r-*-rtems*): Likewise.
4277 (m68k-*-rtems*): Likewise.
4278 (microblaze*-*-rtems*): Likewise.
4279 (mips*-*-rtems*): Likewise.
4280 (powerpc-*-rtems*): Likewise.
4281 (sh-*-rtems*): Likewise.
4282 (sparc-*-rtems*): Likewise.
4283 (sparc64-*-rtems*): Likewise.
4284 (v850-*-rtems*): Likewise.
4285 (m32c-*-rtems*): Likewise.
4287 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4289 PR rtl-optimization/59511
4290 * ira.c (ira_init_register_move_cost): Use memory costs for some
4291 cases of register move cost calculations.
4292 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4293 instead of BB frequency.
4294 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4295 * lra-assigns.c (find_hard_regno_for): Ditto.
4297 2014-01-15 Richard Biener <rguenther@suse.de>
4299 PR tree-optimization/59822
4300 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4301 (vectorizable_load): Use it to hoist defs of uses of invariant
4302 loads out of the loop.
4304 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4305 Kugan Vivekanandarajah <kuganv@linaro.org>
4308 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4311 2014-01-15 Richard Biener <rguenther@suse.de>
4313 PR rtl-optimization/59802
4314 * lcm.c (compute_available): Use inverted postorder to seed
4315 the initial worklist.
4317 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4320 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4321 ADDR_REGS for invalid symrefs in non-PIC code.
4323 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4326 * builtins.c (determine_block_size): Initialize *probable_max_size
4327 even if len_rtx is CONST_INT.
4329 2014-01-14 Andrew Pinski <apinski@cavium.com>
4331 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4332 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4333 (cortexa53_tunings): Likewise.
4334 (aarch64_sched_issue_rate): New function.
4335 (TARGET_SCHED_ISSUE_RATE): Define.
4337 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4339 * ira-costs.c (find_costs_and_classes): Add missed
4340 ira_init_register_move_cost_if_necessary.
4342 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4345 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4347 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4350 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4351 to indicate if type is used for function return value. Warn ABI
4352 change if the vector mode isn't available for function return value.
4353 (ix86_function_arg_advance): Pass false to type_natural_mode.
4354 (ix86_function_arg): Likewise.
4355 (ix86_gimplify_va_arg): Likewise.
4356 (function_arg_32): Don't warn ABI change.
4357 (ix86_function_value): Pass true to type_natural_mode.
4358 (ix86_return_in_memory): Likewise.
4359 (ix86_struct_value_rtx): Removed.
4360 (TARGET_STRUCT_VALUE_RTX): Likewise.
4362 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4364 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4365 converting a conditional jump into a conditional return.
4367 2014-01-14 Richard Biener <rguenther@suse.de>
4369 PR tree-optimization/58921
4370 PR tree-optimization/59006
4371 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4372 hoisting invariant stmts.
4373 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4374 invariant loads on the preheader edge if possible.
4376 2014-01-14 Joey Ye <joey.ye@arm.com>
4378 * doc/plugin.texi (Building GCC plugins): Update to C++.
4380 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4382 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4383 (_mm_rcp28_round_ss): Ditto.
4384 (_mm_rsqrt28_round_sd): Ditto.
4385 (_mm_rsqrt28_round_ss): Ditto.
4386 (_mm_rcp28_sd): Ditto.
4387 (_mm_rcp28_ss): Ditto.
4388 (_mm_rsqrt28_sd): Ditto.
4389 (_mm_rsqrt28_ss): Ditto.
4390 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4391 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4392 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4393 (IX86_BUILTIN_RCP28SD): Ditto.
4394 (IX86_BUILTIN_RCP28SS): Ditto.
4395 (IX86_BUILTIN_RSQRT28SD): Ditto.
4396 (IX86_BUILTIN_RSQRT28SS): Ditto.
4397 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4398 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4399 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4400 (ix86_expand_special_args_builtin): Expand new FTYPE.
4401 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4402 (srcp14<mode>): Make insn unary.
4403 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4404 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4405 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4406 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4407 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4408 Fix rounding: make it SAE only.
4409 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4411 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4413 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4414 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4415 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4416 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4417 (round_saeonly_mask_scalar_operand4): Ditto.
4418 (round_saeonly_mask_scalar_op3): Ditto.
4419 (round_saeonly_mask_scalar_op4): Ditto.
4421 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4423 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4424 Implement -maltivec=be for vec_insert and vec_extract.
4426 2014-01-10 DJ Delorie <dj@redhat.com>
4428 * config/msp430/msp430.md (call_internal): Don't allow memory
4429 references with SP as the base register.
4430 (call_value_internal): Likewise.
4431 * config/msp430/constraints.md (Yc): New. For memory references
4432 that don't use SP as a base register.
4434 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4435 "an integer without a # prefix"
4436 * config/msp430/msp430.md (epilogue_helper): Use it.
4438 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4441 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4442 AVX512F gather builtins.
4443 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4444 on gather decls with INTEGER_TYPE masktype.
4445 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4446 directly into the builtin rather than hoisting it before loop.
4448 PR tree-optimization/59387
4449 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4450 (scev_const_prop): If folded_casts and type has undefined overflow,
4451 use force_gimple_operand instead of force_gimple_operand_gsi and
4452 for each added stmt if it is assign with
4453 arith_code_with_undefined_signed_overflow, call
4454 rewrite_to_defined_overflow.
4455 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4456 gimple-fold.h instead.
4457 (arith_code_with_undefined_signed_overflow,
4458 rewrite_to_defined_overflow): Moved to ...
4459 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4460 rewrite_to_defined_overflow): ... here. No longer static.
4461 Include gimplify-me.h.
4462 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4463 rewrite_to_defined_overflow): New prototypes.
4465 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4467 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4469 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4471 * builtins.c (get_object_alignment_2): Minor tweak.
4472 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4474 2014-01-13 Christian Bruel <christian.bruel@st.com>
4476 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4477 optimized non constant lengths.
4479 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4482 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4483 load as __atomic_load_N if possible.
4485 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4487 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4489 (rs6000_expand_builtin): Adjust call.
4491 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4494 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4495 * config/rs6000/rs6000.c: Include target-globals.h.
4496 (rs6000_set_current_function): Instead of doing target_reinit
4497 unconditionally, use save_target_globals_default_opts and
4498 restore_target_globals.
4500 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4502 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4503 (rs6000_expand_builtin): Handle mffs and mtfsf.
4504 (rs6000_init_builtins): Define mffs and mtfsf.
4505 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4506 (rs6000_mffs): New pattern.
4507 (rs6000_mtfsf): New pattern.
4509 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4511 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4512 Start narrowing with START. Apply candidate-use pair
4513 and check overall cost in narrowing.
4514 (iv_ca_prune): Pass new argument.
4516 2014-01-10 Jeff Law <law@redhat.com>
4519 * ree.c (combine_reaching_defs): Ensure the defining statement
4520 occurs before the extension when optimizing extensions with
4521 different source and destination hard registers.
4523 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4526 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4527 vtables into the type inheritance graph.
4529 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4531 PR rtl-optimization/59754
4532 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4533 modes in the REGNO != REGNO case.
4535 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4537 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4539 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4541 PR tree-optimization/59745
4542 * tree-predcom.c (tree_predictive_commoning_loop): Call
4543 free_affine_expand_cache if giving up because components is NULL.
4545 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4546 GC in payload of target_globals struct instead of allocating them on
4547 the heap and the larger structs separately using GC.
4548 * target-globals.h (struct target_globals): Make regs, hard_regs,
4549 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4550 of GTY((skip)) and change type to void *.
4551 (reset_target_globals): Cast loads from those fields to corresponding
4554 2014-01-10 Steve Ellcey <sellcey@mips.com>
4557 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4558 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4559 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4561 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4564 * aarch64-modes.def (CC_Zmode): New flags mode.
4565 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4566 represents an equality.
4567 (aarch64_get_condition_code): Handle CC_Zmode.
4568 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4570 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4572 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4573 extraction in good case.
4575 2014-01-10 Richard Biener <rguenther@suse.de>
4577 PR tree-optimization/59374
4578 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4579 checking after SLP discovery. Mark stmts not participating
4580 in any SLP instance properly.
4582 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4584 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4585 when handling a SET rtx.
4587 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4589 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4590 (cortex-a57): Likewise.
4591 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4593 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4595 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4596 non-iwmmxt builtins.
4598 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4602 * ipa-devirt.c record_target_from_binfo): Take as argument
4603 stack of binfos and lookup matching one for virtual inheritance.
4604 (possible_polymorphic_call_targets_1): Update.
4606 2014-01-10 Huacai Chen <chenhc@lemote.com>
4608 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4609 kernel strings for Loongson-2E/2F/3A.
4611 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4614 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4615 is_gimple_call before calling gimple_call_internal_p.
4617 2014-01-09 Steve Ellcey <sellcey@mips.com>
4619 * Makefile.in (TREE_FLOW_H): Remove.
4620 (TREE_SSA_H): Add file names from tree-flow.h.
4621 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4622 * tree.h: Remove tree-flow.h reference.
4623 * hash-table.h: Remove tree-flow.h reference.
4624 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4625 reference with tree-ssa-loop.h.
4627 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4629 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4630 default element-order behavior for -maltivec.
4631 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4633 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4634 when targeting big endian, at least for now.
4635 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4637 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4640 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4641 var satisfies use_register_for_decl, just take into account type
4642 alignment, rather than decl alignment.
4644 PR tree-optimization/59622
4645 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4646 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4647 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4648 Don't devirtualize for inplace at all. For targets.length () == 1,
4649 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4651 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4653 * config/i386/i386.md (cpu): Remove the unused btver1.
4655 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4657 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4659 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4662 * tree-core.h (struct target_globals): New forward declaration.
4663 (struct tree_target_option): Add globals field.
4664 * tree.h (TREE_TARGET_GLOBALS): Define.
4665 (prepare_target_option_nodes_for_pch): New prototype.
4666 * target-globals.h (struct target_globals): Define even if
4668 * tree.c (prepare_target_option_node_for_pch,
4669 prepare_target_option_nodes_for_pch): New functions.
4670 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4671 * config/i386/i386.c: Include target-globals.h.
4672 (ix86_set_current_function): Instead of doing target_reinit
4673 unconditionally, use save_target_globals_default_opts and
4674 restore_target_globals.
4676 2014-01-09 Richard Biener <rguenther@suse.de>
4678 PR tree-optimization/59715
4679 * tree-cfg.h (split_critical_edges): Declare.
4680 * tree-cfg.c (split_critical_edges): Export.
4681 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4683 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4685 * cfgexpand.c (expand_stack_vars): Optionally disable
4686 asan stack protection.
4687 (expand_used_vars): Likewise.
4688 (partition_stack_vars): Likewise.
4689 * asan.c (asan_emit_stack_protection): Optionally disable
4690 after return stack usage.
4691 (instrument_derefs): Optionally disable memory access instrumentation.
4692 (instrument_builtin_call): Likewise.
4693 (instrument_strlen_call): Likewise.
4694 (asan_protect_global): Optionally disable global variables protection.
4695 * doc/invoke.texi: Added doc for new options.
4696 * params.def: Added new options.
4697 * params.h: Likewise.
4699 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4701 PR rtl-optimization/59724
4702 * ifcvt.c (cond_exec_process_if_block): Don't call
4703 flow_find_head_matching_sequence with 0 longest_match.
4704 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4705 non-active insns if !stop_after.
4706 (try_head_merge_bb): Revert 2014-01-07 changes.
4708 2014-01-08 Jeff Law <law@redhat.com>
4710 * ree.c (get_sub_rtx): New function, extracted from...
4711 (merge_def_and_ext): Here.
4712 (combine_reaching_defs): Use get_sub_rtx.
4714 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4716 * cgraph.h (varpool_variable_node): Do not choke on null node.
4718 2014-01-08 Catherine Moore <clm@codesourcery.com>
4720 * config/mips/mips.md (simple_return): Attempt to use JRC
4722 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4724 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4726 PR rtl-optimization/59137
4727 * reorg.c (steal_delay_list_from_target): Call update_block for
4729 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4731 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4734 two duplicate entries.
4736 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4739 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4741 * config/mips/mips.c (mips_truncated_op_cost): New function.
4742 (mips_rtx_costs): Adjust test for BADDU.
4743 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4745 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4747 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4748 (*baddu_si): ...this new pattern.
4750 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4753 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4755 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4758 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4760 (expand_expr, expand_normal): Adjust.
4761 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4762 inner_reference_p. Use inner_reference_p to expand inner references.
4763 (store_expr): Adjust.
4764 * cfgexpand.c (expand_call_stmt): Adjust.
4766 2014-01-08 Rong Xu <xur@google.com>
4768 * gcov-io.c (gcov_var): Move from gcov-io.h.
4769 (gcov_position): Ditto.
4770 (gcov_is_error): Ditto.
4771 (gcov_rewrite): Ditto.
4772 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4773 only part to libgcc/libgcov.h.
4775 2014-01-08 Marek Polacek <polacek@redhat.com>
4778 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4780 2014-01-08 Marek Polacek <polacek@redhat.com>
4783 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4785 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4787 PR rtl-optimization/59649
4788 * stor-layout.c (get_mode_bounds): For BImode return
4789 0 and STORE_FLAG_VALUE.
4791 2014-01-08 Richard Biener <rguenther@suse.de>
4794 * gimple.h (is_gimple_builtin_call): Remove.
4795 (gimple_builtin_call_types_compatible_p): New.
4796 (gimple_call_builtin_p): New overload.
4797 * gimple.c (is_gimple_builtin_call): Remove.
4798 (validate_call): Rename to ...
4799 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4801 (validate_type): New static function.
4802 (gimple_call_builtin_p): New overload and adjust.
4803 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4804 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4805 (gimple_fold_stmt_to_constant_1): Likewise.
4806 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4808 2014-01-08 Richard Biener <rguenther@suse.de>
4811 * gimplify.c (gimplify_expr): Gimplify register-register type
4812 VIEW_CONVERT_EXPRs to separate stmts.
4814 2014-01-07 Jeff Law <law@redhat.com>
4817 * ree.c (combine_set_extension): Handle case where source
4818 and destination registers in an extension insn are different.
4819 (combine_reaching_defs): Allow source and destination registers
4820 in extension to be different under limited circumstances.
4821 (add_removable_extension): Remove restriction that the
4822 source and destination registers in the extension are the same.
4823 (find_and_remove_re): Emit a copy from the extension's
4824 destination to its source after the defining insn if
4825 the source and destination registers are different.
4828 * ifcvt.c (merge_if_block): If we are merging a block with more than
4829 one successor with a block with no successors, remove any BARRIER
4830 after the second block.
4832 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4834 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4836 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4839 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4840 for 14-bit register offsets when INT14_OK_STRICT is false.
4842 2014-01-07 Roland Stigge <stigge@antcom.de>
4843 Michael Meissner <meissner@linux.vnet.ibm.com>
4846 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4847 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4849 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4851 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4854 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4856 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4857 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4858 rtx is const0_rtx or not.
4860 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4863 * target-globals.c (save_target_globals): Remove this_fn_optab
4865 * toplev.c: Include optabs.h.
4866 (target_reinit): Temporarily restore the global options if another
4867 set of options are in force.
4869 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4871 PR rtl-optimization/58668
4872 * cfgcleanup.c (flow_find_cross_jump): Don't count
4873 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4874 to determine what is counted.
4875 (flow_find_head_matching_sequence): Use active_insn_p to determine
4877 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4879 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4880 determine what is counted.
4882 PR tree-optimization/59643
4883 * tree-predcom.c (split_data_refs_to_components): If one dr is
4884 read and one write, determine_offset fails and the write isn't
4885 in the bad component, just put the read into the bad component.
4887 2014-01-07 Mike Stump <mikestump@comcast.net>
4888 Jakub Jelinek <jakub@redhat.com>
4891 * tree-core.h (struct tree_optimization_option): Change optabs
4892 type from unsigned char * to void *.
4893 * optabs.c (init_tree_optimization_optabs): Adjust
4894 TREE_OPTIMIZATION_OPTABS initialization.
4896 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4899 * config/i386/i386.h (struct machine_function): Add
4900 no_drap_save_restore field.
4901 * config/i386/i386.c (ix86_save_reg): Use
4902 !cfun->machine->no_drap_save_restore instead of
4903 crtl->stack_realign_needed.
4904 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4905 this function clears frame_pointer_needed. Set
4906 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4907 and DRAP reg is needed.
4909 2014-01-06 Marek Polacek <polacek@redhat.com>
4912 * doc/implement-c.texi: Mention that other integer types are
4913 permitted as bit-field types in strictly conforming mode.
4915 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4917 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4920 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4922 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4924 2014-01-06 Martin Jambor <mjambor@suse.cz>
4927 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4929 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4931 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4935 * var-tracking.c (add_stores): Preserve the value of the source even if
4936 we don't record the store.
4938 2014-01-06 Terry Guo <terry.guo@arm.com>
4940 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4942 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4945 * config/darwin.c (darwin_function_section): Adjust return values to
4946 correspond to optimisation changes made in r206070.
4948 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4950 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4951 from prefetch_block tune setting.
4952 (nocona_cost): Correct size of prefetch block to 64.
4954 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4956 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4957 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4958 used to save the static chain register in the computation of the offset
4959 from which the FP registers need to be restored.
4961 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4963 PR tree-optimization/59519
4964 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4965 ICE if get_current_def (current_new_name) is already non-NULL, as long
4966 as it is a phi result of some other phi in *new_exit_bb that has
4969 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4970 or vmovdqu* for misaligned_operand.
4971 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4972 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4973 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4974 aligned_mem for AVX512F masked aligned load and store builtins and for
4977 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4979 PR tree-optimization/59651
4980 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4981 Address range for negative step should be added by TYPE_SIZE_UNIT.
4983 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4985 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4986 overlapping registers also for registers other than the stack pointer.
4988 2014-01-03 Marek Polacek <polacek@redhat.com>
4991 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4994 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4997 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5000 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5001 (push splitter): Use <P:MODE_SIZE> instead of
5002 GET_MODE_SIZE (<P:MODE>mode).
5003 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5004 (mov -1, reg peephole2): Likewise.
5005 * config/i386/sse.md (*mov<mode>_internal,
5006 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5007 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5008 *<code><mode>3, *andnot<mode>3<mask_name>,
5009 <mask_codefor><code><mode>3<mask_name>): Likewise.
5010 * config/i386/subst.md (mask_mode512bit_condition,
5011 sd_mask_mode512bit_condition): Likewise.
5013 2014-01-02 Xinliang David Li <davidxl@google.com>
5015 PR tree-optimization/59303
5016 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5017 (dump_predicates): Better output format.
5018 (pred_equal_p): New function.
5019 (is_neq_relop_p): Ditto.
5020 (is_neq_zero_form_p): Ditto.
5021 (pred_expr_equal_p): Ditto.
5022 (pred_neg_p): Ditto.
5023 (simplify_pred): Ditto.
5024 (simplify_preds_2): Ditto.
5025 (simplify_preds_3): Ditto.
5026 (simplify_preds_4): Ditto.
5027 (simplify_preds): Ditto.
5029 (push_to_worklist): Ditto.
5030 (get_pred_info_from_cmp): Ditto.
5031 (is_degenerated_phi): Ditto.
5032 (normalize_one_pred_1): Ditto.
5033 (normalize_one_pred): Ditto.
5034 (normalize_one_pred_chain): Ditto.
5035 (normalize_preds): Ditto.
5036 (normalize_cond_1): Remove function.
5037 (normalize_cond): Ditto.
5038 (is_gcond_subset_of): Ditto.
5039 (is_subset_of_any): Ditto.
5040 (is_or_set_subset_of): Ditto.
5041 (is_and_set_subset_of): Ditto.
5042 (is_norm_cond_subset_of): Ditto.
5043 (pred_chain_length_cmp): Ditto.
5044 (convert_control_dep_chain_into_preds): Type change.
5045 (find_predicates): Ditto.
5046 (find_def_preds): Ditto.
5047 (destroy_predicates_vecs): Ditto.
5048 (find_matching_predicates_in_rest_chains): Ditto.
5049 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5050 (is_pred_expr_subset): Ditto.
5051 (is_pred_chain_subset_of): Ditto.
5052 (is_included_in): Ditto.
5053 (is_superset_of): Ditto.
5055 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5057 Update copyright years.
5059 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5061 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5062 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5063 config/arc/arc.md, config/arc/arc.opt,
5064 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5065 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5066 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5067 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5068 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5069 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5070 vtable-verify.c, vtable-verify.h: Use the standard form for the
5073 2014-01-02 Tobias Burnus <burnus@net-b.de>
5075 * gcc.c (process_command): Update copyright notice dates.
5076 * gcov-dump.c: Ditto.
5078 * doc/cpp.texi: Bump @copying's copyright year.
5079 * doc/cppinternals.texi: Ditto.
5080 * doc/gcc.texi: Ditto.
5081 * doc/gccint.texi: Ditto.
5082 * doc/gcov.texi: Ditto.
5083 * doc/install.texi: Ditto.
5084 * doc/invoke.texi: Ditto.
5086 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5088 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5090 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5092 * config/i386/sse.md (*mov<mode>_internal): Guard
5093 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5095 PR rtl-optimization/59647
5096 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5097 new_rtx into UNSIGNED_FLOAT rtxes.
5099 Copyright (C) 2014 Free Software Foundation, Inc.
5101 Copying and distribution of this file, with or without modification,
5102 are permitted in any medium without royalty provided the copyright
5103 notice and this notice are preserved.