1 2014-03-18 Richard Biener <rguenther@suse.de>
3 * doc/lto.texi (fresolution): Fix typo.
5 2014-03-18 Richard Biener <rguenther@suse.de>
7 * doc/invoke.texi (flto): Update for changes in 4.9.
9 2014-03-18 Richard Biener <rguenther@suse.de>
11 * doc/loop.texi: Remove section on the removed lambda framework.
12 Update loop docs with recent changes in preserving loop structure.
14 2014-03-18 Richard Biener <rguenther@suse.de>
16 * doc/lto.texi (-fresolution): Document.
18 2014-03-18 Richard Biener <rguenther@suse.de>
20 * doc/contrib.texi: Adjust my name.
22 2014-03-18 Jakub Jelinek <jakub@redhat.com>
25 * internal-fn.c: Include diagnostic-core.h.
26 (expand_BUILTIN_EXPECT): New function.
27 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
28 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
29 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
30 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
32 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
33 Revert 3 argument __builtin_expect code.
34 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
35 * gimple-fold.c (gimple_fold_call): Likewise.
36 * tree.h (fold_builtin_expect): New prototype.
37 * builtins.c (build_builtin_expect_predicate): Add predictor
38 argument, if non-NULL, create 3 argument __builtin_expect.
39 (fold_builtin_expect): No longer static. Add ARG2 argument,
40 pass it through to build_builtin_expect_predicate.
41 (fold_builtin_2): Adjust caller.
42 (fold_builtin_3): Handle BUILT_IN_EXPECT.
43 * internal-fn.def (BUILTIN_EXPECT): New.
45 2014-03-18 Tobias Burnus <burnus@net-b.de>
48 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
49 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
50 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
52 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
55 * predict.c (combine_predictions_for_bb): Fix up formatting.
56 (expr_expected_value_1, expr_expected_value): Add predictor argument,
57 fill what it points to if non-NULL.
58 (tree_predict_by_opcode): Adjust caller, use the predictor.
59 * predict.def (PRED_COMPARE_AND_SWAP): Add.
61 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
63 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
64 proper constant for the store mode.
66 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
68 * symtab.c (change_decl_assembler_name): Fix transparent alias
71 2014-03-16 Renlin Li <Renlin.Li@arm.com>
73 * config/aarch64/aarch64.c: Correct the comments about the
76 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
78 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
79 check for GF_OMP_FOR_KIND_FOR.
81 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
83 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
84 ymm and zmm register names.
86 2014-03-17 Jakub Jelinek <jakub@redhat.com>
89 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
90 note creation for the 2010-08-31 changes.
92 2014-03-17 Marek Polacek <polacek@redhat.com>
95 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
96 as -fno-tree-loop-vectorize.
97 (expand_omp_simd): Likewise.
99 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
101 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
102 (eligible_for_call_delay): New prototype.
103 * config/sparc/sparc.c (tls_call_delay): Rename into...
104 (eligible_for_call_delay): ...this. Return false if the instruction
105 cannot be put in the delay slot of a branch.
106 (eligible_for_restore_insn): Simplify.
107 (eligible_for_return_delay): Return false if the instruction cannot be
108 put in the delay slot of a branch and simplify.
109 (eligible_for_sibcall_delay): Return false if the instruction cannot be
110 put in the delay slot of a branch.
111 * config/sparc/sparc.md (fix_ut699): New attribute.
112 (tls_call_delay): Delete.
113 (in_call_delay): Reimplement.
114 (eligible_for_sibcall_delay): Rename into...
115 (in_sibcall_delay): ...this.
116 (eligible_for_return_delay): Rename into...
117 (in_return_delay): ...this.
118 (in_branch_delay): Reimplement.
119 (in_uncond_branch_delay): Delete.
120 (in_annul_branch_delay): Delete.
122 2014-03-14 Richard Henderson <rth@redhat.com>
125 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
126 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
127 (*floathi<X87MODEF>2_i387_with_temp): Remove.
128 (floathi splitters): Remove.
129 (float<SWI48x>xf2): New pattern.
130 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
131 code that tried to handle DImode for 32-bit, but which was excluded
132 by the pattern's condition. Drop allocation of stack temporary.
133 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
134 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
135 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
136 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
137 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
138 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
139 (*float<SWI48><MODEF>2_sse_interunit): Remove.
140 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
141 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
142 (*float<SWI48x><X87MODEF>2_i387): Remove.
143 (all float _with_temp splitters): Remove.
144 (*float<SWI48x><MODEF>2_i387): New pattern.
145 (*float<SWI48><MODEF>2_sse): New pattern.
146 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
147 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
149 2014-03-14 Jakub Jelinek <jakub@redhat.com>
150 Marek Polacek <polacek@redhat.com>
153 * common.opt (dump_base_name_prefixed): New Variable.
154 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
155 if x_dump_base_name_prefixed is already set, set it at the end.
157 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
159 PR rtl-optimization/60508
160 * lra-constraints.c (get_reload_reg): Add new parameter
162 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
163 Pass the new parameter values.
165 2014-03-14 Richard Biener <rguenther@suse.de>
167 * common.opt: Revert unintented changes from r205065.
170 2014-03-14 Richard Biener <rguenther@suse.de>
173 * cfghooks.c (split_block): Properly adjust all loops the
174 block was a latch of.
176 2014-03-14 Martin Jambor <mjambor@suse.cz>
179 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
182 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
185 * config/avr/avr.c (avr_set_current_function): Pass function name
186 through default_strip_name_encoding before sanity checking instead
187 of skipping the first char of the assembler name.
189 2014-03-13 Richard Henderson <rth@redhat.com>
192 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
193 (ix86_force_to_memory, ix86_free_from_memory): Remove.
194 * config/i386/i386-protos.h: Likewise.
195 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
196 in the expander instead of a splitter.
197 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
198 any possibility of requiring a memory.
199 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
200 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
201 (fp branch splitters): Update for ix86_split_fp_branch.
202 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
203 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
204 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
205 (*fop_<MODEF>_2_i387): Remove f/r alternative.
206 (*fop_<MODEF>_3_i387): Likewise.
207 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
208 (splitters for the fop_* register patterns): Remove.
209 (fscalexf4_i387): Rename from *fscalexf4_i387.
210 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
212 2014-03-13 Jakub Jelinek <jakub@redhat.com>
214 PR tree-optimization/59779
215 * tree-dfa.c (get_ref_base_and_extent): Use double_int
216 type for bitsize and maxsize instead of HOST_WIDE_INT.
218 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
220 PR rtl-optimization/57320
221 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
222 the CFG after thread_prologue_and_epilogue_insns.
224 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
226 PR rtl-optimization/57189
227 * lra-constraints.c (process_alt_operands): Disfavor spilling
230 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
232 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
235 2014-03-13 Jakub Jelinek <jakub@redhat.com>
237 PR tree-optimization/59025
239 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
240 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
242 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
245 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
246 calls of avr_out_plus_1.
248 2014-03-13 Bin Cheng <bin.cheng@arm.com>
250 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
251 BB's single pred and update the father loop's latch info later.
253 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
255 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
259 (VEC_base): Likewise.
260 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
261 registers, we need to swap double words in little endian mode.
263 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
264 to be a container mode for 128-bit integer operations added in ISA
265 2.07. Unlike TImode and PTImode, the preferred register set is
266 the Altivec/VMX registers for the 128-bit operations.
268 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
270 (rs6000_split_128bit_ok_p): Likewise.
272 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
273 macros for creating ISA 2.07 normal and overloaded builtin
274 functions with 3 arguments.
275 (BU_P8V_OVERLOAD_3): Likewise.
276 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
277 for use as overloaded functions.
278 (VPERM_1TI_UNS): Likewise.
279 (VSEL_1TI): Likewise.
280 (VSEL_1TI_UNS): Likewise.
281 (ST_INTERNAL_1ti): Likewise.
282 (LD_INTERNAL_1ti): Likewise.
283 (XXSEL_1TI): Likewise.
284 (XXSEL_1TI_UNS): Likewise.
285 (VPERM_1TI): Likewise.
286 (VPERM_1TI_UNS): Likewise.
287 (XXPERMDI_1TI): Likewise.
289 (LXVD2X_V1TI): Likewise.
290 (STXVD2X_V1TI): Likewise.
291 (VEC_INIT_V1TI): Likewise.
292 (VEC_SET_V1TI): Likewise.
293 (VEC_EXT_V1TI): Likewise.
294 (EQV_V1TI): Likewise.
295 (NAND_V1TI): Likewise.
296 (ORC_V1TI): Likewise.
297 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
298 added in ISA 2.07. Add both normal 'altivec' builtins, and the
302 (VADDEUQM): Likewise.
303 (VADDECUQ): Likewise.
304 (VSUBEUQM): Likewise.
305 (VSUBECUQ): Likewise.
307 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
308 __int128_t and __uint128_t types.
309 (__uint128_type): Likewise.
310 (altivec_categorize_keyword): Add support for vector __int128_t,
311 vector __uint128_t, vector __int128, and vector unsigned __int128
312 as a container type for TImode operations that need to be done in
313 VSX/Altivec registers.
314 (rs6000_macro_to_expand): Likewise.
315 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
316 to support 128-bit integer instructions vaddcuq, vadduqm,
317 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
318 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
320 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
321 for V1TImode, and set up preferences to use VSX/Altivec registers.
322 Setup VSX reload handlers.
323 (rs6000_debug_reg_global): Likewise.
324 (rs6000_init_hard_regno_mode_ok): Likewise.
325 (rs6000_preferred_simd_mode): Likewise.
326 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
327 (easy_altivec_constant): Likewise.
328 (output_vec_const_move): Likewise.
329 (rs6000_expand_vector_set): Convert V1TImode set and extract to
331 (rs6000_expand_vector_extract): Likewise.
332 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
334 (rs6000_const_vec): Add support for V1TImode.
335 (rs6000_emit_le_vsx_load): Swap double words when loading or
336 storing TImode/V1TImode.
337 (rs6000_emit_le_vsx_store): Likewise.
338 (rs6000_emit_le_vsx_move): Likewise.
339 (rs6000_emit_move): Add support for V1TImode.
340 (altivec_expand_ld_builtin): Likewise.
341 (altivec_expand_st_builtin): Likewise.
342 (altivec_expand_vec_init_builtin): Likewise.
343 (altivec_expand_builtin): Likewise.
344 (rs6000_init_builtins): Add support for V1TImode type. Add
345 support for ISA 2.07 128-bit integer builtins. Define type names
346 for the VSX/Altivec vector types.
347 (altivec_init_builtins): Add support for overloaded vector
348 functions with V1TImode type.
349 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
350 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
352 (rs6000_split_128bit_ok_p): Likewise.
353 (rs6000_handle_altivec_attribute): Create V1TImode from vector
354 __int128_t and vector __uint128_t.
356 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
364 (VS_scalar): Likewise.
365 (VS_double): Likewise.
366 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
368 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
369 we support the ISA 2.07 128-bit integer arithmetic instructions.
370 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
371 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
372 and TImode types for use with the builtin functions.
373 (V1TI_type_node): Likewise.
374 (unsigned_V1TI_type_node): Likewise.
375 (intTI_type_internal_node): Likewise.
376 (uintTI_type_internal_node): Likewise.
378 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
379 128-bit builtin functions.
380 (UNSPEC_VADDEUQM): Likewise.
381 (UNSPEC_VADDECUQ): Likewise.
382 (UNSPEC_VSUBCUQ): Likewise.
383 (UNSPEC_VSUBEUQM): Likewise.
384 (UNSPEC_VSUBECUQ): Likewise.
385 (VM): Add V1TImode to vector mode iterators.
388 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
389 (altivec_vaddcuq): Likewise.
390 (altivec_vsubuqm): Likewise.
391 (altivec_vsubcuq): Likewise.
392 (altivec_vaddeuqm): Likewise.
393 (altivec_vaddecuq): Likewise.
394 (altivec_vsubeuqm): Likewise.
395 (altivec_vsubecuq): Likewise.
397 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
399 (BOOL_128): Likewise.
400 (BOOL_REGS_OUTPUT): Likewise.
401 (BOOL_REGS_OP1): Likewise.
402 (BOOL_REGS_OP2): Likewise.
403 (BOOL_REGS_UNARY): Likewise.
404 (BOOL_REGS_AND_CR0): Likewise.
406 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
407 128-bit integer builtin support.
408 (vec_vadduqm): Likewise.
409 (vec_vaddecuq): Likewise.
410 (vec_vaddeuqm): Likewise.
411 (vec_vsubecuq): Likewise.
412 (vec_vsubeuqm): Likewise.
413 (vec_vsubcuq): Likewise.
414 (vec_vsubuqm): Likewise.
416 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
417 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
418 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
419 128-bit integer add/subtract to ISA 2.07.
421 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
423 * config/arc/arc.c (arc_predicate_delay_insns):
424 Fix third argument passed to conditionalize_nonjump.
426 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
428 * config/aarch64/aarch64-builtins.c
429 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
430 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
431 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
432 instead of __builtin_lfloor.
433 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
435 2014-03-12 Jakub Jelinek <jakub@redhat.com>
437 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
438 (tree_ssa_ifcombine_bb_1): New function.
439 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
440 is an empty forwarder block to then_bb or vice versa and then_bb
441 and else_bb are effectively swapped.
443 2014-03-12 Christian Bruel <christian.bruel@st.com>
446 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
447 REG_CFA_DEF_CFA note.
448 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
449 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
451 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
453 PR tree-optimization/60454
454 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
456 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
458 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
459 Do not define target_cpu_default2 to generic.
460 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
461 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
462 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
464 2014-03-12 Jakub Jelinek <jakub@redhat.com>
465 Marc Glisse <marc.glisse@inria.fr>
467 PR tree-optimization/60502
468 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
469 instead of build_low_bits_mask.
471 2014-03-12 Jakub Jelinek <jakub@redhat.com>
474 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
475 if there are multiple uses, but op doesn't live on E edge.
476 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
477 clobber stmts before __builtin_unreachable.
479 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
481 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
482 hard_frame_pointer_rtx.
483 * cse.c (cse_insn): Remove volatile check.
484 * cselib.c (cselib_process_insn): Likewise.
485 * dse.c (scan_insn): Likewise.
487 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
489 * config/arc/arc.c (conditionalize_nonjump): New function,
491 (arc_ifcvt): ... this.
492 (arc_predicate_delay_insns): Use it.
494 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
496 * config/arc/predicates.md (extend_operand): During/after reload,
497 allow const_int_operand.
498 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
499 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
500 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
502 (umulsi3_highpart_i): Likewise.
504 2014-03-11 Richard Biener <rguenther@suse.de>
506 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
507 Add asserts to guard possible wrong-code bugs.
509 2014-03-11 Richard Biener <rguenther@suse.de>
511 PR tree-optimization/60429
512 PR tree-optimization/60485
513 * tree-ssa-structalias.c (set_union_with_increment): Properly
514 take into account all fields that overlap the shifted vars.
515 (do_sd_constraint): Likewise.
516 (do_ds_constraint): Likewise.
517 (get_constraint_for_ptr_offset): Likewise.
519 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
521 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
522 (nios2_compute_frame_layout):
523 Add calculation of cfun->machine->fp_save_offset.
524 (nios2_expand_prologue): Correct setting of frame pointer register
526 (nios2_expand_epilogue): Update recovery of stack pointer from
527 frame pointer accordingly.
528 (nios2_initial_elimination_offset): Update calculation of offset
529 for eliminating to HARD_FRAME_POINTER_REGNUM.
531 2014-03-10 Jakub Jelinek <jakub@redhat.com>
534 * ipa.c (symtab_remove_unreachable_nodes): Don't call
535 cgraph_get_create_node on VAR_DECLs.
537 2014-03-10 Richard Biener <rguenther@suse.de>
540 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
542 2014-03-08 Douglas B Rupp <rupp@gnat.com>
544 * config/vms/vms.opt (vms_float_format): New variable.
546 2014-03-08 Tobias Burnus <burnus@net-b.de>
548 * doc/invoke.texi (-fcilkplus): Update implementation status.
550 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
551 Richard Biener <rguenther@suse.de>
553 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
554 consistently accross all TUs.
555 (run_gcc): Enable -fshort-double automatically at link at link-time
556 and disallow override.
558 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
561 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
562 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
563 if they can't be used.
565 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
567 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
568 for Solaris 11/x86 ld.
569 * configure: Regenerate.
571 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
573 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
574 (LIB_TLS_SPEC): Save as ld_tls_libs.
575 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
576 (HAVE_AS_IX86_TLSLDM): New test.
577 * configure, config.in: Regenerate.
578 * config/i386/i386.c (legitimize_tls_address): Fall back to
579 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
580 cannot support TLS_MODEL_LOCAL_DYNAMIC.
581 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
582 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
584 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
586 * common.opt (fira-loop-pressure): Mark as optimization.
588 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
590 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
591 an OpenMP mappable type.
593 2014-03-06 Matthias Klose <doko@ubuntu.com>
595 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
596 MULTILIB_OSDIRNAMES is not defined.
598 2014-03-06 Jakub Jelinek <jakub@redhat.com>
599 Meador Inge <meadori@codesourcery.com>
602 * config/arm/arm.c (arm_tls_symbol_p): Remove.
603 (arm_legitimize_address): Call legitimize_tls_address for any
604 arm_tls_referenced_p expression, handle constant addend. Call it
605 before testing for !TARGET_ARM.
606 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
608 2014-03-06 Richard Biener <rguenther@suse.de>
614 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
616 * tree-streamer.c (record_common_node): Assert we don't record
617 nodes with type double.
618 (preload_common_node): Skip type double, complex double and double
619 pointer since it is now frontend dependent due to fshort-double option.
621 2014-03-06 Richard Biener <rguenther@suse.de>
623 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
624 or -fno-lto is specified and the linker has full plugin support.
625 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
626 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
627 * lto-wrapper.c (merge_and_complain): Merge compile-time
629 (run_gcc): And pass it through to the link options.
631 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
635 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
637 * cselib.c (remove_useless_values): Skip to avoid quadratic
638 behavior if the condition moved from...
639 (cselib_process_insn): ... here holds.
641 2014-03-05 Jakub Jelinek <jakub@redhat.com>
644 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
645 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
648 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
649 (TM_H): Add x86-tune.def.
651 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
653 * config/aarch64/aarch64.c (generic_tunings):
654 Use cortexa57_extra_costs.
656 2014-03-05 Jakub Jelinek <jakub@redhat.com>
659 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
660 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
661 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
664 2014-03-04 Heiher <r@hev.cc>
666 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
667 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
669 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
671 * config/i386/predicates.md (const2356_operand): Change to ...
672 (const2367_operand): ... this.
673 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
675 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
676 (*avx512pf_scatterpf<mode>sf): Ditto.
677 (avx512pf_scatterpf<mode>df): Ditto.
678 (*avx512pf_scatterpf<mode>df_mask): Ditto.
679 (*avx512pf_scatterpf<mode>df): Ditto.
680 * config/i386/i386.c (ix86_expand_builtin): Update
681 incorrect hint operand error message.
683 2014-03-04 Richard Biener <rguenther@suse.de>
685 * lto-section-in.c (lto_get_section_data): Fix const cast.
687 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
689 * tree-streamer.c (record_common_node): Assert we don't record
690 nodes with type double.
691 (preload_common_node): Skip type double, complex double and double
692 pointer since it is now frontend dependent due to fshort-double option.
694 2014-03-04 Richard Biener <rguenther@suse.de>
697 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
698 (lto_input_toplevel_asms): Likewise.
699 * lto-section-in.c (lto_get_section_data): Instead do it here
702 2014-03-04 Richard Biener <rguenther@suse.de>
704 PR tree-optimization/60382
705 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
706 dead PHIs a reduction.
708 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
710 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
712 (_mm_prefetch): Move out of GCC target("sse") pragma.
713 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
714 GCC target("prfchw") pragma.
715 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
717 * config/i386/i386.c (ix86_option_override_internal): Enable
718 -mprfchw with -mprefetchwt1.
720 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
722 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
725 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
727 * opts.h (CL_PCH_IGNORE): Define.
728 * targhooks.c (option_affects_pch_p):
729 Return false for options that have CL_PCH_IGNORE set.
730 * opt-functions.awk: Process PchIgnore.
731 * doc/options.texi: Document PchIgnore.
733 * config/arc/arc.opt (misize): Add PchIgnore property.
735 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
737 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
738 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
739 constraint on constants to permit them being loaded into
740 GENERAL_REGS or BASE_REGS.
742 2014-03-03 Nick Clifton <nickc@redhat.com>
744 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
745 anti-cacnonical alternatives.
746 (negandhi3_real): New pattern.
747 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
749 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
751 * config/avr/avr-mcus.def: Remove atxmega16x1.
752 * config/avr/avr-tables.opt: Regenerate.
753 * config/avr/t-multilib: Regenerate.
754 * doc/avr-mmcu.texi: Regenerate.
756 2014-03-03 Tobias Grosser <tobias@grosser.es>
757 Mircea Namolaru <mircea.namolaru@inria.fr>
759 PR tree-optimization/58028
760 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
763 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
765 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
766 not handled by recognizers.
768 2014-03-03 Jakub Jelinek <jakub@redhat.com>
771 * function.c (expand_function_end): Don't emit
772 clobber_return_register sequence if clobber_after is a BARRIER.
773 * cfgexpand.c (construct_exit_block): Append instructions before
774 return_label to prev_bb.
776 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
778 * config/rs6000/constraints.md: Document reserved use of "wc".
780 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
783 * ipa.c (function_and_variable_visibility): When dissolving comdat
784 group, also set all symbols to local.
786 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
791 2013-12-14 Jan Hubicka <jh@suse.cz>
793 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
795 2014-03-02 Jon Beniston <jon@beniston.com>
801 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
802 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
803 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
804 (simple_return, *simple_return): New patterns
805 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
806 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
808 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
810 * dwarf2out.c (gen_subprogram_die): Tidy.
812 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
815 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
816 (*mov_t_msb_neg_negc): ... this new insn.
818 2014-02-28 Jason Merrill <jason@redhat.com>
821 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
824 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
827 * dwarf2out.c (decltype_auto_die): New static.
828 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
829 (gen_type_die_with_usage): Handle 'decltype(auto)'.
830 (is_cxx_auto): Likewise.
832 2014-02-28 Ian Bolton <ian.bolton@arm.com>
834 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
835 we are not using general regs only.
837 2014-02-28 Richard Biener <rguenther@suse.de>
840 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
841 previous fix and only allow to remove trivial pre-headers
842 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
843 (remove_forwarder_block): Properly update the latch of a loop.
845 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
848 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
849 (cselib_preserved_hash_table): New.
850 (preserve_constants_and_equivs): Move preserved vals to it.
851 (cselib_find_slot): Look it up first.
852 (cselib_init): Initialize it.
853 (cselib_finish): Release it.
854 (dump_cselib_table): Dump it.
856 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
859 * cselib.c (remove_useless_values): Skip to avoid quadratic
860 behavior if the condition moved from...
861 (cselib_process_insn): ... here holds.
863 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
866 * var-tracking.c (vt_initialize): Apply the same condition to
867 preserve the CFA base value.
869 2014-02-28 Joey Ye <joey.ye@arm.com>
872 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
873 if reload in progress or completed.
875 2014-02-28 Tobias Burnus <burnus@net-b.de>
878 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
881 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
883 * doc/tm.texi.in (Condition Code Status): Update documention for
884 relative locations of cc0-setter and cc0-user.
886 2014-02-27 Jeff Law <law@redhat.com>
888 PR rtl-optimization/52714
889 * combine.c (try_combine): When splitting an unrecognized PARALLEL
890 into two independent simple sets, if I3 is a jump, ensure the
891 pattern we place into I3 is a (set (pc) ...).
893 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
894 Jeff Law <law@redhat.com>
896 PR rtl-optimization/49847
897 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
898 are in different blocks.
899 * doc/tm.texi (Condition Code Status): Update documention for
900 relative locations of cc0-setter and cc0-user.
902 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
905 * lra.c (lra_emit_add): Check SUBREG too.
907 2014-02-27 Andreas Schwab <schwab@suse.de>
909 * config/m68k/m68k.c (m68k_option_override): Disable
910 -flive-range-shrinkage for classic m68k.
911 (m68k_override_options_after_change): Likewise.
913 2014-02-27 Marek Polacek <polacek@redhat.com>
916 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
917 -Wmaybe-uninitialized.
919 2014-02-27 Alan Modra <amodra@gmail.com>
922 * reload1.c (emit_input_reload_insns): When reload_override_in,
923 set old to rl->in_reg when rl->in_reg is a subreg.
925 2014-02-26 Richard Biener <rguenther@suse.de>
928 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
930 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
932 * common/config/i386/predicates.md (const1256_operand): Remove.
933 (const2356_operand): New.
934 (const_1_to_2_operand): Remove.
935 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
936 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
937 (*avx512pf_gatherpf<mode>sf): Ditto.
938 (avx512pf_gatherpf<mode>df): Ditto.
939 (*avx512pf_gatherpf<mode>df_mask): Ditto.
940 (*avx512pf_gatherpf<mode>df): Ditto.
941 (avx512pf_scatterpf<mode>sf): Ditto.
942 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
943 (*avx512pf_scatterpf<mode>sf): Ditto.
944 (avx512pf_scatterpf<mode>df): Ditto.
945 (*avx512pf_scatterpf<mode>df_mask): Ditto.
946 (*avx512pf_scatterpf<mode>df): Ditto.
947 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
949 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
951 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
952 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
953 (_mm512_mask_testn_epi64_mask): Move to ...
954 * config/i386/avx512cdintrin.h: Here.
955 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
956 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
957 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
958 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
959 TARGET_AVX512F from TARGET_AVX512CD.
961 2014-02-26 Richard Biener <rguenther@suse.de>
964 * ipa.c (walk_polymorphic_call_targets): Properly guard
965 call to inline_update_overall_summary.
967 2014-02-26 Bin Cheng <bin.cheng@arm.com>
970 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
971 and latches only if requested. Fix latch if it is removed.
972 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
973 LOOPS_HAVE_PREHEADERS.
975 2014-02-25 Andrew Pinski <apinski@cavium.com>
977 * builtins.c (expand_builtin_thread_pointer): Create a new target
978 when the target is NULL.
980 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
982 PR rtl-optimization/60317
983 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
984 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
985 * lra-assigns.c: Include params.h.
986 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
987 other reload pseudos considerations.
989 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
991 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
992 to use canonical form for nor<mode>3.
994 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
997 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1000 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1002 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1003 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1004 (ix86_handle_option): Handle OPT_mprefetchwt1.
1005 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1006 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1008 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1009 OPTION_MASK_ISA_PREFETCHWT1.
1010 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1011 (PTA_PREFETCHWT1): New.
1012 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1013 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1014 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1015 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1016 (*prefetch_avx512pf_<mode>_: Change into ...
1017 (*prefetch_prefetchwt1_<mode>: This.
1018 * config/i386/i386.opt (mprefetchwt1): New.
1019 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1020 (_mm_prefetch): Handle intent to write.
1021 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1023 2014-02-25 Richard Biener <rguenther@suse.de>
1026 * emit-rtl.c (mem_attrs_htab): Remove.
1027 (mem_attrs_htab_hash): Likewise.
1028 (mem_attrs_htab_eq): Likewise.
1029 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1030 (init_emit_once): Do not allocate mem_attrs_htab.
1032 2014-02-25 Richard Biener <rguenther@suse.de>
1035 * lto-opts.c (lto_write_options): Output non-explicit conservative
1036 -fwrapv, -fno-trapv and -fno-strict-overflow.
1037 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1038 (run_gcc): And pass them through.
1040 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1042 * sel-sched.c (calculate_new_fences): New parameter ptime.
1043 Calculate it as a maximum over all fence cycles.
1044 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1045 Print the final schedule timing when sched_verbose.
1047 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1049 PR rtl-optimization/60292
1050 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1051 bit fot the fence instruction.
1053 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1055 * calls.h: Fix typo in comment.
1057 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1059 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1060 adjusting offsetable addresses.
1062 2014-02-24 Guozhi Wei <carrot@google.com>
1064 * sparseset.h (sparseset_pop): Fix the wrong index.
1066 2014-02-24 Walter Lee <walt@tilera.com>
1068 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1069 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1071 * common/config/tilegx/tilegx-common.c
1072 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1073 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1075 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1076 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1077 (tilegx_gimplify_va_arg_expr): Handle big endian.
1078 (tilegx_expand_unaligned_load): Ditto.
1079 (tilegx_expand_unaligned_store): Ditto.
1080 (TARGET_RETURN_IN_MSB): New.
1081 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1082 (TARGET_ENDIAN_DEFAULT): New.
1083 (TARGET_BIG_ENDIAN): Handle big endian.
1084 (BYTES_BIG_ENDIAN): Ditto.
1085 (WORDS_BIG_ENDIAN): Ditto.
1086 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1089 * config/tilegx/tilegx.md (extv): Handle big endian.
1091 (insn_st<n>): Ditto.
1092 (insn_st<n>_add<bitsuffix>): Ditto.
1093 (insn_stnt<n>): Ditto.
1094 (insn_stnt<n>_add<bitsuffix>):Ditto.
1095 (vec_interleave_highv8qi): Handle big endian.
1096 (vec_interleave_highv8qi_be): New.
1097 (vec_interleave_highv8qi_le): New.
1098 (insn_v1int_h): Handle big endian.
1099 (vec_interleave_lowv8qi): Handle big endian.
1100 (vec_interleave_lowv8qi_be): New.
1101 (vec_interleave_lowv8qi_le): New.
1102 (insn_v1int_l): Handle big endian.
1103 (vec_interleave_highv4hi): Handle big endian.
1104 (vec_interleave_highv4hi_be): New.
1105 (vec_interleave_highv4hi_le): New.
1106 (insn_v2int_h): Handle big endian.
1107 (vec_interleave_lowv4hi): Handle big endian.
1108 (vec_interleave_lowv4hi_be): New.
1109 (vec_interleave_lowv4hi_le): New.
1110 (insn_v2int_l): Handle big endian.
1111 (vec_interleave_highv2si): Handle big endian.
1112 (vec_interleave_highv2si_be): New.
1113 (vec_interleave_highv2si_le): New.
1114 (insn_v4int_h): Handle big endian.
1115 (vec_interleave_lowv2si): Handle big endian.
1116 (vec_interleave_lowv2si_be): New.
1117 (vec_interleave_lowv2si_le): New.
1118 (insn_v4int_l): Handle big endian.
1119 * config/tilegx/tilegx.opt (mbig-endian): New option.
1120 (mlittle-endian): New option.
1121 * doc/install.texi: Document tilegxbe-linux.
1122 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1124 2014-02-24 Martin Jambor <mjambor@suse.cz>
1127 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1128 there are no parameter descriptors.
1130 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1132 PR rtl-optimization/60268
1133 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1134 initialization to ...
1135 (sched_rgn_init): ... here.
1136 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1138 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1140 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1143 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1145 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1148 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1150 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1151 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1153 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1155 * config/microblaze/predicates.md: Add cmp_op predicate.
1156 * config/microblaze/microblaze.md: Add branch_compare instruction
1157 which uses cmp_op predicate and emits cmp insn before branch.
1158 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1159 to microblaze_expand_conditional_branch and consolidate logic.
1160 (microblaze_expand_conditional_branch): emit branch_compare
1161 insn instead of handling cmp op separate from branch insn.
1163 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1165 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1168 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1170 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1171 define_insn with define_expand and new define_insn
1172 *altivec_lve<VI_char>x_internal.
1173 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1174 and new define_insn *altivec_stve<VI_char>x_internal.
1175 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1177 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1179 (altivec_expand_stvex_be): New function.
1181 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1183 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1184 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1185 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1186 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1188 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1191 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1192 instead of emit_move_insn.
1194 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1196 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1198 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1199 gen_altivec_vsumsws.
1201 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1203 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1204 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1205 (altivec_lvxl_<mode>): New define_expand incorporating
1206 -maltivec=be semantics where needed.
1207 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1208 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1209 semantics where needed.
1210 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1211 (altivec_stvx_<mode>): New define_expand incorporating
1212 -maltivec=be semantics where needed.
1213 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1214 VM2 iterator instead of V4SI.
1215 (altivec_stvxl_<mode>): New define_expand incorporating
1216 -maltivec=be semantics where needed.
1217 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1218 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1219 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1220 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1221 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1222 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1223 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1224 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1225 ALTIVEC_BUILTIN_STVXL.
1226 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1227 (altivec_expand_stvx_be): Likewise.
1228 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1229 (altivec_expand_lvx_be): Likewise.
1230 (altivec_expand_stvx_be): Likewise.
1231 (altivec_expand_builtin): Add cases for
1232 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1233 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1234 (altivec_init_builtins): Add definitions for
1235 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1236 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1238 2014-02-21 Catherine Moore <clm@codesourcery.com>
1240 * doc/invoke.texi (mvirt, mno-virt): Document.
1241 * config/mips/mips.opt (mvirt): New option.
1242 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1244 2014-02-21 Richard Biener <rguenther@suse.de>
1246 PR tree-optimization/60276
1247 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1248 (STMT_VINFO_MIN_NEG_DIST): New macro.
1249 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1250 STMT_VINFO_MIN_NEG_DIST.
1251 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1252 made for negative dependence distances still hold.
1254 2014-02-21 Richard Biener <rguenther@suse.de>
1257 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1258 DECL_INITIAL for globals not in the current function context.
1260 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1262 PR tree-optimization/56490
1263 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1264 * tree-ssa-uninit.c: Include params.h.
1265 (compute_control_dep_chain): Add num_calls argument, return false
1266 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1267 num_calls to recursive call.
1268 (find_predicates): Change dep_chain into normal array,
1269 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1270 variable and adjust compute_control_dep_chain caller.
1271 (find_def_preds): Likewise.
1273 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1275 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1276 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1278 2014-02-21 Nick Clifton <nickc@redhat.com>
1280 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1281 (pushhi1): Likewise.
1282 (popqi1): Add mode to pre_dec.
1285 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1287 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1288 mode for mask of V8SFmode permutation.
1290 2014-02-20 Richard Henderson <rth@redhat.com>
1293 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1294 a new pseudo for OLDVAL.
1296 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1299 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1300 gen_reg_rtx if d->testing_p.
1301 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1302 if d->testing_p and we will certainly return true.
1303 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1306 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1308 * emit-rtl.c (gen_reg_rtx): Assert that
1309 crtl->emit.regno_pointer_align_length is non-zero.
1311 2014-02-20 Richard Henderson <rth@redhat.com>
1314 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1315 on failure the store back into EXPECT.
1317 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1318 Sandra Loosemore <sandra@codesourcery.com>
1320 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1321 * config/nios2/nios2.c (nios2_function_profiler): Add
1322 -fPIC (flag_pic == 2) support.
1323 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1324 (nios2_large_offset_p): New function.
1325 (nios2_unspec_reloc_p): Move up position, update to use
1326 nios2_large_offset_p.
1327 (nios2_unspec_address): Remove function.
1328 (nios2_unspec_offset): New function.
1329 (nios2_large_got_address): New function.
1330 (nios2_got_address): Add large offset support.
1331 (nios2_legitimize_tls_address): Update usage of removed and new
1333 (nios2_symbol_binds_local_p): New function.
1334 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1335 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1336 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1337 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1338 processing for (const (unspec ...)).
1339 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1341 2014-02-20 Richard Biener <rguenther@suse.de>
1343 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1344 doing the substitution.
1345 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1347 2014-02-20 Martin Jambor <mjambor@suse.cz>
1350 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1351 info when checking whether lattices are bottom.
1353 2014-02-20 Richard Biener <rguenther@suse.de>
1356 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1359 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1362 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1363 parameter specifying the scaling.
1364 (inline_call): Update.
1365 (want_inline_recursively): Guard division by zero.
1366 (recursive_inlining): Update.
1367 * ipa-inline.h (clone_inlined_nodes): Update.
1369 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1372 * config/i386/i386.c (classify_argument): Pass structures of size
1373 64 bytes or less in register.
1375 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1376 Kirill Yukhin <kirill.yukhin@intel.com>
1378 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1379 (_mm_rcp28_round_ss): Ditto.
1380 (_mm_rsqrt28_round_sd): Ditto.
1381 (_mm_rsqrt28_round_ss): Ditto.
1382 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1383 (_mm_rcp14_round_ss): Ditto.
1384 (_mm_rsqrt14_round_sd): Ditto.
1385 (_mm_rsqrt14_round_ss): Ditto.
1386 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1387 the first input operand, get rid of match_dup.
1388 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1390 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1392 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1393 operand as the first input operand, set type attribute.
1394 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1396 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1397 operand as the first input operand, set type attribute.
1399 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1401 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1404 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1407 * config/i386/i386.c (construct_container): Remove TFmode check
1408 for X86_64_INTEGER_CLASS.
1410 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1413 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1414 only when -Wpsabi is enabled.
1416 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1419 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1420 passing arrays in registers are the same as for structs, so remove the
1421 special case for them.
1423 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1425 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1426 destination type, extract only the valid bits if the source type is not
1427 integral and has a different mode.
1429 2014-02-19 Richard Biener <rguenther@suse.de>
1432 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1435 2014-02-19 Richard Biener <rguenther@suse.de>
1438 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1439 (ipa_modify_call_arguments): Emit an argument load explicitely and
1440 preserve virtual SSA form there and for the replacement call.
1441 Do not update SSA form nor free dominance info.
1443 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1445 * ipa.c (function_and_variable_visibility): Also clear WEAK
1446 flag when disolving COMDAT_GROUP.
1448 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1450 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1451 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1453 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1454 do more sanity checks.
1455 (detect_type_change): Return true when giving up early.
1456 (compute_complex_assign_jump_func): Fix type parameter of
1457 ipa_set_ancestor_jf.
1458 (compute_complex_ancestor_jump_func): Likewise.
1459 (update_jump_functions_after_inlining): Fix updating of
1461 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1463 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1465 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1466 inline clones when edge disappears.
1468 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1471 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1472 Split 64-bit moves into 2 patterns. Do not allow the use of
1473 direct move for TDmode in little endian, since the decimal value
1474 has little endian bytes within a word, but the 64-bit pieces are
1475 ordered in a big endian fashion, and normal subreg's of TDmode are
1477 (mov<mode>_64bit_dm): Likewise.
1478 (movtd_64bit_nodm): Likewise.
1480 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1482 PR tree-optimization/60174
1483 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1484 statement of an SSA_NAME that occurs in an abnormal PHI node.
1486 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1489 * final.c (SEEN_BB): Remove.
1490 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1491 (final_scan_insn): Don't force_source_line on second
1492 NOTE_INSN_BASIC_BLOCK.
1494 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1497 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1498 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1499 (type_natural_mode): Warn ABI change when %zmm register is not
1500 available for AVX512F vector value passing.
1502 2014-02-18 Kai Tietz <ktietz@redhat.com>
1505 * config/i386/i386.c (ix86_expand_prologue): Use value in
1506 rax register as displacement when restoring %r10 or %rax.
1507 Fix wrong offset when restoring both registers.
1509 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1511 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1512 assertion with conditional return.
1514 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1515 Uros Bizjak <ubizjak@gmail.com>
1518 * config/i386/driver-i386.c (host_detect_local_cpu): If
1519 YMM state is not saved by the OS, also clear has_f16c. Move
1520 CPUID 0x80000001 handling before YMM state saving checking.
1522 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1524 PR rtl-optimization/58960
1525 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1526 factored out from ...
1527 (sched_init): ... here.
1528 (free_global_sched_pressure_data): New, factored out from ...
1529 (sched_finish): ... here.
1530 * sched-int.h (free_global_sched_pressure_data): Declare.
1531 * sched-rgn.c (nr_regions_initial): New static global.
1532 (haifa_find_rgns): Initialize it.
1533 (schedule_region): Disable sched-pressure for the newly
1536 2014-02-17 Richard Biener <rguenther@suse.de>
1538 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1539 release SSA defs of pattern stmts.
1541 2014-02-17 Richard Biener <rguenther@suse.de>
1543 * tree-inline.c (expand_call_inline): Release the virtual
1544 operand defined by the call we are about to inline.
1546 2014-02-17 Richard Biener <rguenther@suse.de>
1548 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1550 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1551 Ilya Tocar <ilya.tocar@intel.com>
1553 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1554 arguments order in builtin.
1555 (_mm512_permutexvar_epi64): Ditto.
1556 (_mm512_mask_permutexvar_epi64): Ditto
1557 (_mm512_maskz_permutexvar_epi32): Ditto
1558 (_mm512_permutexvar_epi32): Ditto
1559 (_mm512_mask_permutexvar_epi32): Ditto
1561 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1563 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1564 (p8_vmrgow): Likewise.
1566 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1568 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1571 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1574 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1575 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1576 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1577 using direct move instructions on ISA 2.07. Also adjust
1578 instruction length for 64-bit.
1579 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1580 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1582 2014-02-15 Alan Modra <amodra@gmail.com>
1586 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1587 find_replacement on parts of insn rtl that might be reloaded.
1589 2014-02-15 Richard Biener <rguenther@suse.de>
1591 PR tree-optimization/60183
1592 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1593 (tree_ssa_phiprop): Calculate and free post-dominators.
1595 2014-02-14 Jeff Law <law@redhat.com>
1597 PR rtl-optimization/60131
1598 * ree.c (get_extended_src_reg): New function.
1599 (combine_reaching_defs): Use it rather than assuming location of REG.
1600 (find_and_remove_re): Verify first operand of extension is
1601 a REG before adding the insns to the copy list.
1603 2014-02-14 Roland McGrath <mcgrathr@google.com>
1605 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1606 * configure: Regenerated.
1607 * config.in: Regenerated.
1608 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1609 instead of ASM_SHORT.
1611 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1612 Richard Earnshaw <rearnsha@arm.com>
1614 PR rtl-optimization/59535
1615 * lra-constraints.c (process_alt_operands): Encourage alternative
1616 when unassigned pseudo class is superset of the alternative class.
1617 (inherit_reload_reg): Don't inherit when optimizing for code size.
1618 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1619 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1620 modes not less than 4 for Thumb1.
1622 2014-02-14 Kyle McMartin <kyle@redhat.com>
1625 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1627 2014-02-14 Richard Biener <rguenther@suse.de>
1629 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1630 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1631 Do not strip INDIRECT_REFs.
1633 2014-02-14 Richard Biener <rguenther@suse.de>
1636 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1637 DECL_FUNCTION_SPECIFIC_TARGET.
1638 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1639 * tree-streamer-out.c (pack_ts_target_option): Remove.
1640 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1641 (write_ts_function_decl_tree_pointers): Do not stream
1642 DECL_FUNCTION_SPECIFIC_TARGET.
1643 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1644 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1645 (lto_input_ts_function_decl_tree_pointers): Do not stream
1646 DECL_FUNCTION_SPECIFIC_TARGET.
1648 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1650 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1651 (get_initial_def_for_induction, vectorizable_induction): Ignore
1652 debug stmts when looking for exit_phi.
1653 (vectorizable_live_operation): Fix up condition.
1655 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1657 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1658 nreverse() because it changes the content of original tree list.
1660 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1662 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1663 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1665 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1667 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1668 GNU coding standards.
1670 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1673 * dwarf2out.c (gen_subprogram_die): Don't call
1674 add_calling_convention_attribute if subr_die is old_die.
1676 2014-02-13 Sharad Singhai <singhai@google.com>
1678 * doc/optinfo.texi: Fix order of nodes.
1680 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1682 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1683 operands[2], not operands[3].
1685 2014-02-13 Richard Biener <rguenther@suse.de>
1688 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1689 mention the possibility of an in-tree build.
1690 (CLooG): Update recommended version to 0.18.1, mention the
1691 possibility of an in-tree build and clarify that the ISL
1692 bundled with CLooG does not work.
1694 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1697 * expr.c (compress_float_constant): If x is a hard register,
1698 extend into a pseudo and then move to x.
1700 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1702 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1703 caused by bad second argument to warning_at() with -mhotpatch and
1704 nested functions (e.g. with gfortran).
1706 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1708 * opts.c (option_name): Remove "enabled by default" rider.
1710 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1712 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1714 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1715 Uros Bizjak <ubizjak@gmail.com>
1718 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1719 * configure: Regenerated.
1721 2014-02-12 Richard Biener <rguenther@suse.de>
1723 * vec.c (vec_prefix::calculate_allocation): Move as
1724 inline variant to vec.h.
1725 (vec_prefix::calculate_allocation_1): New out-of-line version.
1726 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1727 (vec_prefix::m_has_auto_buf): Rename to ...
1728 (vec_prefix::m_using_auto_storage): ... this.
1729 (vec_prefix::calculate_allocation): Inline the easy cases
1730 and dispatch to calculate_allocation_1 which doesn't need the
1732 (va_heap::reserve): Use gcc_checking_assert.
1733 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1734 m_using_auto_storage.
1735 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1737 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1738 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1739 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1741 2014-02-12 Richard Biener <rguenther@suse.de>
1743 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1744 when we found a dependence.
1746 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1748 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1750 (maybe_fold_stmt): ... into this new function.
1751 * omp-low.c (lower_omp): Update comment.
1753 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1756 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1759 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1761 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1762 identifiers in comments.
1763 (cortexa53_extra_costs): Likewise.
1764 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1765 (cortexa7_extra_costs): Likewise.
1766 (cortexa12_extra_costs): Likewise.
1767 (cortexa15_extra_costs): Likewise.
1768 (v7m_extra_costs): Likewise.
1770 2014-02-12 Richard Biener <rguenther@suse.de>
1773 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1774 of posix_memalign being successful.
1775 (lower_stmt): Restrict lowering of posix_memalign to when
1776 -ftree-bit-ccp is enabled.
1778 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1780 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1782 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1784 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1786 PR rtl-optimization/60116
1787 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1788 other_insn once the combination has been validated.
1790 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1793 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1795 * ipa-devirt.c: Include demangle.h
1796 (odr_violation_reported): New static variable.
1797 (add_type_duplicate): Update odr_violations.
1798 (maybe_record_node): Add completep parameter; update it.
1799 (record_target_from_binfo): Add COMPLETEP parameter;
1800 update it as needed.
1801 (possible_polymorphic_call_targets_1): Likewise.
1802 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1803 rename FINAL to COMPLETE.
1804 (record_targets_from_bases): Sanity check we found the binfo;
1805 fix COMPLETEP updating.
1806 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1807 parameter, fix computing of COMPLETEP.
1808 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1809 at LTO time do demangling.
1810 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1811 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1813 (gimple_get_virt_method_for_binfo): Likewise.
1814 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1815 gimple_get_virt_method_for_vtable): Update prototypes.
1817 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1820 * genautomata.c (add_presence_absence): Fix typo with
1821 {final_}presence_list.
1823 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1826 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1827 for VSX/Altivec vectors that land in GPR registers.
1829 2014-02-11 Richard Henderson <rth@redhat.com>
1830 Jakub Jelinek <jakub@redhat.com>
1833 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1834 around drhs if type conversion to lacc->type is not useless.
1836 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1838 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1840 (cortex-a57.cortex-a53): Likewise.
1841 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1843 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1845 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1848 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1850 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1851 add_options_for_arm_vfp3.
1853 2014-02-11 Jeff Law <law@redhat.com>
1856 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1857 object with an undesirable mode.
1859 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1862 * config/i386/sol2-9.h: New file.
1863 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1864 *-*-solaris2.9*): Use it.
1866 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1868 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1869 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1871 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1873 * config/microblaze/microblaze.c: Extend mcpu version format
1875 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1877 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1879 2014-02-10 Richard Henderson <rth@redhat.com>
1882 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1883 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1884 ms-abi vs -mno-accumulate-outgoing-args.
1885 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1886 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1889 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1892 * cfgexpand.c (expand_asm_operands): Attach source location to
1893 ASM_INPUT rtx objects.
1894 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1896 2014-02-10 Nick Clifton <nickc@redhat.com>
1898 * config/mn10300/mn10300.c (popcount): New function.
1899 (mn10300_expand_prologue): Include saved registers in stack usage
1902 2014-02-10 Jeff Law <law@redhat.com>
1905 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1906 when changing the SET_DEST of a prior insn to avoid an input reload.
1908 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1910 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1911 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1912 -mcall-openbsd, or -mcall-linux.
1913 (CC1_ENDIAN_BIG_SPEC): Remove.
1914 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1915 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1916 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1917 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1918 and %cc1_endian_default.
1919 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1921 2014-02-10 Richard Biener <rguenther@suse.de>
1923 PR tree-optimization/60115
1924 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1925 MEM_REF handling. Properly verify that the accesses are not
1926 out of the objects bound.
1928 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1930 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1933 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1935 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1936 proper constants and fix formatting.
1937 (possible_polymorphic_call_targets): Fix formatting.
1939 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1940 Ilya Tocar <ilya.tocar@intel.com>
1942 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1943 (_mm512_loadu_epi32): Renamed into...
1944 (_mm512_loadu_si512): This.
1945 (_mm512_storeu_epi32): Renamed into...
1946 (_mm512_storeu_si512): This.
1947 (_mm512_maskz_ceil_ps): Removed.
1948 (_mm512_maskz_ceil_pd): Ditto.
1949 (_mm512_maskz_floor_ps): Ditto.
1950 (_mm512_maskz_floor_pd): Ditto.
1951 (_mm512_floor_round_ps): Ditto.
1952 (_mm512_floor_round_pd): Ditto.
1953 (_mm512_ceil_round_ps): Ditto.
1954 (_mm512_ceil_round_pd): Ditto.
1955 (_mm512_mask_floor_round_ps): Ditto.
1956 (_mm512_mask_floor_round_pd): Ditto.
1957 (_mm512_mask_ceil_round_ps): Ditto.
1958 (_mm512_mask_ceil_round_pd): Ditto.
1959 (_mm512_maskz_floor_round_ps): Ditto.
1960 (_mm512_maskz_floor_round_pd): Ditto.
1961 (_mm512_maskz_ceil_round_ps): Ditto.
1962 (_mm512_maskz_ceil_round_pd): Ditto.
1963 (_mm512_expand_pd): Ditto.
1964 (_mm512_expand_ps): Ditto.
1965 * config/i386/i386.c (ix86_builtins): Remove
1966 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1967 (bdesc_args): Ditto.
1968 * config/i386/predicates.md (const1256_operand): New.
1969 (const_1_to_2_operand): Ditto.
1970 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1971 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1972 (*avx512pf_gatherpf<mode>sf): Ditto.
1973 (avx512pf_gatherpf<mode>df): Ditto.
1974 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1975 (*avx512pf_gatherpf<mode>df): Ditto.
1976 (avx512pf_scatterpf<mode>sf): Ditto.
1977 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1978 (*avx512pf_scatterpf<mode>sf): Ditto.
1979 (avx512pf_scatterpf<mode>df): Ditto.
1980 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1981 (*avx512pf_scatterpf<mode>df): Ditto.
1982 (avx512f_expand<mode>): Removed.
1983 (<shift_insn><mode>3<mask_name>): Change predicate type.
1985 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1987 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1988 not at the end of datarefs vector use ordered_remove to avoid
1989 reordering datarefs vector.
1992 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1993 mark local addressable non-static vars as GOVD_PRIVATE
1994 instead of GOVD_LOCAL.
1995 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1996 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2000 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2001 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2002 assume_aligned or alloc_align attributes.
2003 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2004 arguments. Handle also assume_aligned and alloc_align attributes.
2005 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2006 calls to functions with assume_aligned or alloc_align attributes.
2007 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2009 2014-02-08 Terry Guo <terry.guo@arm.com>
2011 * doc/invoke.texi: Document ARM -march=armv7e-m.
2013 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2015 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2016 flag on __cilkrts_rethrow builtin.
2019 * ipa-cp.c (determine_versionability): Fail at -O0
2020 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2021 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2024 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2027 * tree-inline.c (copy_forbidden): Fail for
2028 __attribute__((optimize (0))) functions.
2030 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2032 * varpool.c: Include pointer-set.h.
2033 (varpool_remove_unreferenced_decls): Variables in other partitions
2034 will not be output; be however careful to not lose information
2037 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2039 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2040 lookup in the vtable constructor.
2042 2014-02-07 Jeff Law <law@redhat.com>
2045 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2046 define_insn_and_split.
2048 * ipa-inline.c (inline_small_functions): Fix typos.
2050 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2052 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2053 (s390_can_use_return_insn): Declare.
2054 * config/s390/s390.h (EPILOGUE_USES): Define.
2055 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2057 (s390_chunkify_start): Handle return JUMP_LABELs.
2058 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2059 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2060 (s390_can_use_return_insn): New functions.
2061 (s390_fix_long_loop_prediction): Handle conditional returns.
2062 (TARGET_SET_UP_BY_PROLOGUE): Define.
2063 * config/s390/s390.md (ANY_RETURN): New code iterator.
2064 (*creturn, *csimple_return, return, simple_return): New patterns.
2066 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2068 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2069 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2070 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2071 REG_CFA_RESTORE list when deciding not to restore a register.
2073 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2075 * config/s390/s390.c: Include tree-pass.h and context.h.
2076 (s390_early_mach): New function, split out from...
2077 (s390_emit_prologue): ...here.
2078 (pass_data_s390_early_mach): New pass structure.
2079 (pass_s390_early_mach): New class.
2080 (s390_option_override): Create and register early_mach pass.
2081 Move to end of file.
2083 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2085 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2086 to match for the exit block.
2088 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2090 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2091 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2092 Reject misaligned operands.
2094 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2096 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2098 2014-02-07 Richard Biener <rguenther@suse.de>
2101 * gimple-low.c (lower_builtin_posix_memalign): New function.
2102 (lower_stmt): Call it to lower posix_memalign in a way
2103 to make alignment info accessible.
2105 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2108 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2109 __builtin_setjmp_receiver.
2111 2014-02-07 Richard Biener <rguenther@suse.de>
2114 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2115 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2116 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2117 Handle BUILT_IN_POSIX_MEMALIGN.
2118 (find_func_clobbers): Likewise.
2119 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2120 (call_may_clobber_ref_p_1): Likewise.
2122 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2125 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2128 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2131 * lto-cgraph.c (lto_output_node): Use
2132 symtab_get_symbol_partitioning_class.
2133 (lto_output_varpool_node): likewise.
2134 (symtab_get_symbol_partitioning_class): Move here from
2136 * cgraph.h (symbol_partitioning_class): Likewise.
2137 (symtab_get_symbol_partitioning_class): Declare.
2139 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2141 * ggc.h (ggc_internal_cleared_alloc): New macro.
2142 * vec.h (vec_safe_copy): Handle memory stats.
2143 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2144 * target-globals.c (save_target_globals): Likewise.
2146 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2149 * expr.c (emit_move_resolve_push): Export; be bit more selective
2150 on when to clear alias set.
2151 * expr.h (emit_move_resolve_push): Declare.
2152 * function.h (struct function): Add tail_call_marked.
2153 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2154 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2155 * config/i386/i386.md (TImode move expander): De not call
2157 (FP push expanders): Preserve memory attributes.
2158 * config/i386/sse.md (push<mode>1): Remove.
2159 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2160 (ix86_expand_push): Remove.
2161 * config/i386/mmx.md (push<mode>1): Remove.
2163 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2165 PR rtl-optimization/60030
2166 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2167 lopart with paradoxical subreg before shifting it up by hprec.
2169 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2171 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2172 Remove extra newline at end of file.
2173 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2174 (arm_issue_rate): Handle cortexa57.
2175 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2176 (cortex-a57.cortex-a53): Likewise.
2178 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2181 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2182 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2184 (arm_expand_prologue): Adjust all callers.
2185 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2186 info, registers also at the lowest numbered registers side. Use
2187 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2191 * var-tracking.c (adjust_mems): Before adding a SET to
2192 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2194 2014-02-06 Alan Modra <amodra@gmail.com>
2197 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2198 change SDmode to DDmode when lra_in_progress.
2200 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2203 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2204 free_data_ref on the dr first, and before goto again also set dr
2205 to the next dr. For simd_lane_access, free old datarefs[i] before
2206 overwriting it. For get_vectype_for_scalar_type failure, don't
2207 free_data_ref if simd_lane_access.
2209 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2212 * tree.h (opts_for_fn): New inline function.
2213 (opt_for_fn): Define.
2214 * config/i386/i386.c (ix86_function_regparm): Use
2215 opt_for_fn (decl, optimize) instead of optimize.
2217 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2219 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2220 for SYMBOL_REF in large memory model.
2222 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2224 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2226 (cortex-a57): Likewise.
2227 (cortex-a57.cortex-a53): Likewise.
2229 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2230 Kugan Vivekanandarajah <kuganv@linaro.org>
2232 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2234 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2236 2014-02-06 Richard Biener <rguenther@suse.de>
2238 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2239 set_loop_copy and initialize_original_copy_tables.
2241 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2243 * config/aarch64/aarch64-simd.md
2244 (aarch64_ashr_simddi): Change QI to SI.
2246 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2247 Jakub Jelinek <jakub@redhat.com>
2250 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2253 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2255 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2256 CODE_FOR_altivec_vpku[hw]um to
2257 CODE_FOR_altivec_vpku[hw]um_direct.
2258 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2259 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2260 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2261 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2263 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2265 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2266 generation for -maltivec=be.
2267 (altivec_vsumsws): Simplify redundant test.
2269 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2271 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2272 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2273 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2274 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2275 gen_altivec_vpkuwum.
2276 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2278 (altivec_vpks<VI_char>ss): Likewise.
2279 (altivec_vpks<VI_char>us): Likewise.
2280 (altivec_vpku<VI_char>us): Likewise.
2281 (altivec_vpku<VI_char>um): Likewise.
2282 (altivec_vpku<VI_char>um_direct): New (copy of
2283 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2285 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2286 target is little endian and -maltivec=be is not specified.
2287 (*altivec_vupkhs<VU_char>_direct): New (copy of
2288 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2289 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2290 target is little endian and -maltivec=be is not specified.
2291 (*altivec_vupkls<VU_char>_direct): New (copy of
2292 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2293 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2294 little endian and -maltivec=be is not specified.
2295 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2296 little endian and -maltivec=be is not specified.
2298 2014-02-05 Richard Henderson <rth@redhat.com>
2301 * combine-stack-adj.c: Revert r206943.
2302 * sched-int.h (struct deps_desc): Add last_args_size.
2303 * sched-deps.c (init_deps): Initialize it.
2304 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2305 contain REG_ARGS_SIZE notes.
2307 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2309 * lto-cgraph.c (asm_nodes_output): Make global.
2310 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2311 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2312 (driver_handle_option): Handle OPT_fwpa.
2314 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2317 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2318 a comment typo and formatting issue. If odr_hash hasn't been
2319 created, return vNULL and set *completep to false.
2322 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2323 bb with no successors.
2325 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2328 * doc/invoke.texi (-march): Clarify documentation for ARM.
2332 2014-02-05 Richard Biener <rguenther@suse.de>
2334 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2335 when not vectorizing because of too many alias checks.
2336 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2337 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2339 2014-02-05 Nick Clifton <nickc@redhat.com>
2341 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2342 accept extended registers in any mode when compiling for the MN10300.
2344 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2346 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2347 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2348 sanitization attributes.
2349 (can_inline_edge_p): Likewise.
2350 (sanitize_attrs_match_for_inline_p): New function.
2352 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2354 * ipa-prop.c (detect_type_change): Shor circuit testing of
2355 type changes on THIS pointer.
2357 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2360 * config/pa/pa.c (legitimize_tls_address): Return original address
2361 if not passed a SYMBOL_REF rtx.
2362 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2364 (pa_emit_move_sequence): Simplify TLS source operands.
2365 (pa_legitimate_constant_p): Reject all TLS constants.
2366 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2367 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2369 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2371 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2372 groups when we know they are controlled by LTO.
2373 * varasm.c (default_binds_local_p_1): If object is in other partition,
2374 it will be resolved locally.
2376 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2378 * config/host-linux.c (linux_gt_pch_use_address): Don't
2379 use SSIZE_MAX because it is not always defined.
2381 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2384 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2385 threshold for pseudo splitting.
2386 (update_ebb_live_info): Process call argument hard registers and
2387 hard registers from insn definition too.
2388 (max_small_class_regs_num): New constant.
2389 (inherit_in_ebb): Update live hard regs through EBBs. Update
2390 reloads_num only for small register classes. Don't split for
2393 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2396 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2399 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2401 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2404 2014-02-04 Marek Polacek <polacek@redhat.com>
2406 * gdbinit.in (pel): Define.
2408 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2410 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2413 2014-02-04 Richard Biener <rguenther@suse.de>
2416 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2417 in function context local.
2418 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2419 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2420 similar to LTO_imported_decl_ref.
2422 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2424 PR tree-optimization/60002
2425 * cgraphclones.c (build_function_decl_skip_args): Clear
2428 PR tree-optimization/60023
2429 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2430 false to gsi_replace.
2431 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2432 has been in some EH region and vec_stmt could throw, add
2433 vec_stmt into the same EH region.
2434 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2435 has no lhs, ignore it.
2436 * internal-fn.c (expand_MASK_LOAD): Likewise.
2439 * tree-inline.c (copy_forbidden): Fail for
2440 __attribute__((optimize (0))) functions.
2443 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2444 is set, copy one less argument.
2445 (expand_simd_clones): Don't subtract clone_info->inbranch
2446 from simd_clone_struct_alloc argument.
2448 PR rtl-optimization/57915
2449 * recog.c (simplify_while_replacing): If all unary/binary/relational
2450 operation arguments are constant, attempt to simplify those.
2453 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2454 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2456 2014-02-04 Richard Biener <rguenther@suse.de>
2458 PR tree-optimization/60012
2459 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2460 TBAA disambiguation to all DDRs.
2462 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2465 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2466 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2468 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2471 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2473 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2475 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2476 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2478 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2481 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2482 to figure out targets of polymorphic calls with known decl.
2483 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2484 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2485 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2486 (get_polymorphic_call_info): ... here.
2487 (get_polymorphic_call_info_from_invariant): New function.
2489 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2491 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2492 lookup via vtable pointer; check for type consistency
2493 and turn inconsitent facts into UNREACHABLE.
2494 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2495 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2496 type inconsistent querries; return UNREACHABLE instead.
2498 2014-02-03 Richard Henderson <rth@twiddle.net>
2501 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2502 already processed this node.
2503 (normalize_one_pred_1): Pass along mark_set.
2504 (normalize_one_pred): Create and destroy a pointer_set_t.
2505 (normalize_one_pred_chain): Likewise.
2507 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2509 PR gcov-profile/58602
2510 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2512 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2515 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2516 -fno-devirtualize; try to devirtualize by the knowledge of
2517 virtual table pointer given by aggregate propagation.
2518 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2519 (ipa_print_node_jump_functions): Dump also offset that
2520 is relevant for polymorphic calls.
2521 (determine_known_aggregate_parts): Add arg_type parameter; use it
2522 instead of determining the type from pointer type.
2523 (ipa_compute_jump_functions_for_edge): Update call of
2524 determine_known_aggregate_parts.
2525 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2526 (gimple_get_virt_method_for_binfo): ... here; simplify using
2527 vtable_pointer_value_to_vtable.
2528 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2529 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2530 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2531 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2533 (vtable_pointer_value_to_binfo): ... here.
2534 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2536 2014-02-03 Teresa Johnson <tejohnson@google.com>
2538 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2539 redef of outer loop index variable.
2541 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2545 * doc/extend.texi (Function Attributes): Typo.
2547 2014-02-03 Cong Hou <congh@google.com>
2549 PR tree-optimization/60000
2550 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2551 if the vectorized statement is a store. A store statement can only
2552 appear at the end of pattern statements.
2554 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2556 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2557 (ix86_option_override_internal): Default long double to 64-bit for
2558 32-bit Bionic and to 128-bit for 64-bit Bionic.
2560 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2561 TARGET_LONG_DOUBLE_128 is true.
2562 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2564 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2565 (mlong-double-64): Negate -mlong-double-128.
2566 (mlong-double-128): New option.
2568 * config/i386/i386-c.c (ix86_target_macros): Define
2569 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2571 * doc/invoke.texi: Document -mlong-double-128.
2573 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2575 PR rtl-optimization/60024
2576 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2578 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2580 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2582 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2584 PR rtl-optimization/57662
2585 * sel-sched.c (code_motion_path_driver): Do not mark already not
2586 existing blocks in the visiting bitmap.
2588 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2590 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2591 on the insn being emitted.
2593 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2594 Will Deacon <will.deacon@arm.com>
2596 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2598 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2600 * config/arm/arm-tables.opt: Regenerate.
2602 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2604 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2605 for vector types other than V16QImode.
2606 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2607 define_expand, and call altivec_expand_vec_perm_le when producing
2608 code with little endian element order.
2609 (*altivec_vperm_<mode>_internal): New insn having previous
2610 behavior of altivec_vperm_<mode>.
2611 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2612 altivec_expand_vec_perm_le when producing code with little endian
2614 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2615 behavior of altivec_vperm_<mode>_uns.
2617 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2619 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2620 (altivec_vsumsws): Add handling for -maltivec=be with a little
2622 (altivec_vsumsws_direct): New.
2623 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2624 gen_altivec_vsumsws.
2626 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2628 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2629 vtable_pointer_value_to_binfo): New functions.
2630 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2631 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2633 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2635 * config/nios2/nios2.md (load_got_register): Initialize GOT
2636 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2637 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2639 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2641 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2642 preserverd by passthrough, do not propagate the type.
2644 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2646 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2647 (mips_atomic_assign_expand_fenv): New function.
2648 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2650 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2652 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2653 (__builtin_mips_set_fcsr): Likewise.
2654 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2655 MIPS_USI_FTYPE_VOID.
2656 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2657 (mips16_expand_set_fcsr): Likewise.
2658 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2659 (mips16_set_fcsr_stub): Likewise.
2660 (mips16_get_fcsr_one_only_stub): New class.
2661 (mips16_set_fcsr_one_only_stub): Likewise.
2662 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2663 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2664 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2665 (hard_float): New availability predicate.
2666 (mips_builtins): Add get_fcsr and set_fcsr.
2667 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2668 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2669 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2670 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2671 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2674 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2676 * config/mips/mips.c (mips_one_only_stub): New class.
2677 (mips_need_mips16_rdhwr_p): Replace with...
2678 (mips16_rdhwr_stub): ...this new variable.
2679 (mips16_stub_call_address): New function.
2680 (mips16_rdhwr_one_only_stub): New class.
2681 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2682 (mips_output_mips16_rdhwr): Delete.
2683 (mips_finish_stub): New function.
2684 (mips_code_end): Use it to handle rdhwr stubs.
2686 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2689 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2690 when calculating size of integer atomic types.
2692 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2694 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2696 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2698 PR tree-optimization/60003
2699 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2700 * profile.c (branch_prob): Use gimple_call_builtin_p
2701 to check for BUILT_IN_SETJMP_RECEIVER.
2702 * tree-inline.c (copy_bb): Call notice_special_calls.
2704 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2707 * lra-constraints.c (process_alt_operands): Update reload_sum only
2710 2014-01-31 Richard Henderson <rth@redhat.com>
2713 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2714 until after else_eh is processed.
2716 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2718 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2719 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2720 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2721 in smmintrin.h, remove them.
2722 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2723 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2724 * config/i386/i386.md (ROUND_SAE): Fix value.
2725 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2726 (const48_operand): New.
2727 * config/i386/subst.md (round), (round_expand): Use
2728 const_4_or_8_to_11_operand.
2729 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2731 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2733 * config/i386/constraints.md (Yk): Swap meaning with k.
2734 * config/i386/i386.md (movhi_internal): Change Yk to k.
2735 (movqi_internal): Ditto.
2736 (*k<logic><mode>): Ditto.
2739 (kandn<mode>): Ditto.
2740 (*<code>hi_1): Ditto.
2741 (*<code>qi_1): Ditto.
2742 (kxnor<mode>): Ditto.
2743 (kortestzhi): Ditto.
2744 (kortestchi): Ditto.
2746 (*one_cmplhi2_1): Ditto.
2747 (*one_cmplqi2_1): Ditto.
2748 * config/i386/sse.md (): Change k to Yk.
2749 (avx512f_load<mode>_mask): Ditto.
2750 (avx512f_blendm<mode>): Ditto.
2751 (avx512f_store<mode>_mask): Ditto.
2752 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2753 (avx512f_storedqu<mode>_mask): Ditto.
2754 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2756 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2757 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2758 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2759 (avx512f_maskcmp<mode>3): Ditto.
2760 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2761 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2762 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2763 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2764 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2765 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2766 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2767 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2768 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2769 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2770 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2771 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2772 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2773 (vec_extract_lo_<mode>_maskm): Ditto.
2774 (vec_extract_hi_<mode>_maskm): Ditto.
2775 (avx512f_vternlog<mode>_mask): Ditto.
2776 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2777 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2778 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2779 (avx512f_<code>v8div16qi2_mask): Ditto.
2780 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2781 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2782 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2783 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2784 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2785 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2786 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2787 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2788 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2789 (avx512cd_maskb_vec_dupv8di): Ditto.
2790 (avx512cd_maskw_vec_dupv16si): Ditto.
2791 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2792 (avx512f_vpermi2var<mode>3_mask): Ditto.
2793 (avx512f_vpermi2var<mode>3_mask): Ditto.
2794 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2795 (*avx512f_gathersi<mode>): Ditto.
2796 (*avx512f_gathersi<mode>_2): Ditto.
2797 (*avx512f_gatherdi<mode>): Ditto.
2798 (*avx512f_gatherdi<mode>_2): Ditto.
2799 (*avx512f_scattersi<mode>): Ditto.
2800 (*avx512f_scatterdi<mode>): Ditto.
2801 (avx512f_compress<mode>_mask): Ditto.
2802 (avx512f_compressstore<mode>_mask): Ditto.
2803 (avx512f_expand<mode>_mask): Ditto.
2804 * config/i386/subst.md (mask): Change k to Yk.
2805 (mask_scalar_merge): Ditto.
2808 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2810 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2812 2014-01-31 Richard Biener <rguenther@suse.de>
2815 * builtins.c (fold_builtin_memory_op): Make sure to not
2816 use a floating-point mode or a boolean or enumeral type for
2819 2014-01-30 DJ Delorie <dj@redhat.com>
2821 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2822 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2823 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2824 whenever main() has an epilogue.
2826 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2828 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2829 unused variable "field".
2830 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2831 (vsx_mergeh_<mode>): Likewise.
2832 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2833 (altivec_vmrghh): Likewise.
2834 (altivec_vmrghw): Likewise.
2835 (altivec_vmrglb): Likewise.
2836 (altivec_vmrglh): Likewise.
2837 (altivec_vmrglw): Likewise.
2838 (altivec_vspltb): Add missing uses.
2839 (altivec_vsplth): Likewise.
2840 (altivec_vspltw): Likewise.
2841 (altivec_vspltsf): Likewise.
2843 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2846 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2847 frame related instructions.
2849 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2851 PR rtl-optimization/59959
2852 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2853 any reload of register whose subreg is invalid.
2855 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2857 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2858 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2859 Add missing return type - void.
2861 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2863 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2864 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2865 remove element index adjustment for endian (now handled in vsx.md
2867 (altivec_expand_vec_perm_const): Use
2868 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2869 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2870 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2871 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2872 define_expand and a new define_insn *altivec_vspltb_internal;
2873 adjust for -maltivec=be on a little endian target.
2874 (altivec_vspltb_direct): New.
2875 (altivec_vsplth): Divide into a define_expand and a new
2876 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2877 little endian target.
2878 (altivec_vsplth_direct): New.
2879 (altivec_vspltw): Divide into a define_expand and a new
2880 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2881 little endian target.
2882 (altivec_vspltw_direct): New.
2883 (altivec_vspltsf): Divide into a define_expand and a new
2884 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2885 a little endian target.
2887 2014-01-30 Richard Biener <rguenther@suse.de>
2889 PR tree-optimization/59993
2890 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2891 can propagate form the earlier stmt and avoid the transform
2892 when the intermediate result is needed.
2894 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2896 * README.Portability: Fix typo.
2898 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2900 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2901 comparison_operator with ordered_comparison_operator.
2903 2014-01-30 Nick Clifton <nickc@redhat.com>
2905 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2906 Rename to mn10300_store_multiple_regs.
2907 * config/mn10300/mn10300.c: Likewise.
2908 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2909 store_multiple_regs.
2910 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2911 Call mn10300_store_multiple_regs.
2913 2014-01-30 Nick Clifton <nickc@redhat.com>
2914 DJ Delorie <dj@redhat.com>
2916 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2917 %fp 2 to keep registers after it properly word-aligned.
2918 (rl78_alloc_physical_registers_umul): Handle the case where both
2919 input operands are the same.
2921 2014-01-30 Richard Biener <rguenther@suse.de>
2923 PR tree-optimization/59903
2924 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2927 2014-01-30 Jason Merrill <jason@redhat.com>
2930 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2933 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2935 2014-01-30 Richard Biener <rguenther@suse.de>
2937 PR tree-optimization/59951
2938 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2940 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2943 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2944 SFmode to DFmode case.
2946 2014-01-29 DJ Delorie <dj@redhat.com>
2948 * config/msp430/msp430.opt (-minrt): New.
2949 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2951 (ENDFILE_SPEC): Likewise.
2953 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2955 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2956 (estimate_function_body_sizes): Use it.
2958 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2961 * dwarf2out.c (is_cxx_auto): New.
2962 (is_base_type): Use it.
2963 (gen_type_die_with_usage): Likewise.
2965 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2967 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2968 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2969 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2970 -maltivec=be with LE targets.
2971 (vsx_mergeh_<mode>): Likewise.
2972 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2973 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2974 (altivec_vmrghb): Replace with define_expand and new
2975 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2976 (altivec_vmrghb_direct): New define_insn.
2977 (altivec_vmrghh): Replace with define_expand and new
2978 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2979 (altivec_vmrghh_direct): New define_insn.
2980 (altivec_vmrghw): Replace with define_expand and new
2981 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2982 (altivec_vmrghw_direct): New define_insn.
2983 (*altivec_vmrghsf): Adjust for endianness.
2984 (altivec_vmrglb): Replace with define_expand and new
2985 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2986 (altivec_vmrglb_direct): New define_insn.
2987 (altivec_vmrglh): Replace with define_expand and new
2988 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2989 (altivec_vmrglh_direct): New define_insn.
2990 (altivec_vmrglw): Replace with define_expand and new
2991 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2992 (altivec_vmrglw_direct): New define_insn.
2993 (*altivec_vmrglsf): Adjust for endianness.
2994 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2995 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2996 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2997 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2998 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2999 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3000 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3001 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3003 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3005 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3006 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3009 2014-01-29 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/58742
3012 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3013 associate_pointerplus_align.
3014 (associate_pointerplus_diff): New function.
3015 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3016 and associate_pointerplus_diff.
3018 2014-01-29 Richard Biener <rguenther@suse.de>
3020 * lto-streamer.h (LTO_major_version): Bump to 3.
3021 (LTO_minor_version): Reset to 0.
3023 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3025 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3026 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3027 (arm_file_start): Generate correct asm header for armv7ve.
3028 * config/arm/bpabi.h: Add multilib support for armv7ve.
3029 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3030 and cortex-a15 to armv7ve.
3031 * config/arm/t-aprofile: Add multilib support for armv7ve.
3032 * doc/invoke.texi: Document -march=armv7ve.
3034 2014-01-29 Richard Biener <rguenther@suse.de>
3036 PR tree-optimization/58742
3037 * tree-ssa-forwprop.c (associate_plusminus): Return true
3038 if we changed sth, defer EH cleanup to ...
3039 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3040 (simplify_mult): New function.
3042 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3045 PR tree-optimization/59920
3046 * tree.c (build_common_builtin_nodes): Remove
3047 __builtin_setjmp_dispatcher initialization.
3048 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3049 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3050 instead of gsi_after_labels + manually skipping debug stmts.
3051 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3052 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3053 * tree-inline.c (copy_edges_for_bb): Remove
3054 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3055 argument. Ignore computed_goto_p stmts. Don't call
3056 make_abnormal_goto_edges. If a call might need abnormal edges
3057 for non-local gotos, see if it already has an edge to
3058 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3059 with true argument, don't do anything then, otherwise add
3060 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3061 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3063 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3064 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3065 (lower_stmt): Don't set data->calls_builtin_setjmp.
3066 (lower_builtin_setjmp): Adjust comment.
3067 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3068 * tree-cfg.c (found_computed_goto): Remove.
3069 (factor_computed_gotos): Remove.
3070 (make_goto_expr_edges): Return bool, true for computed gotos.
3071 Don't call make_abnormal_goto_edges.
3072 (build_gimple_cfg): Don't set found_computed_goto, don't call
3073 factor_computed_gotos.
3074 (computed_goto_p): No longer static.
3075 (make_blocks): Don't set found_computed_goto.
3076 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3077 (make_edges): If make_goto_expr_edges returns true, push bb
3078 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3079 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3080 vector. Record mapping between bbs and OpenMP regions if there
3081 are any, adjust make_gimple_omp_edges caller. Call
3082 handle_abnormal_edges.
3083 (make_abnormal_goto_edges): Remove.
3084 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3085 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3086 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3087 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3088 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3089 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3090 filling *region also set *region_idx to (*region)->entry->index.
3093 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3094 For REGs set ORIGINAL_REGNO.
3096 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3098 * doc/md.texi: Mention that a target shouldn't implement
3099 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3102 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3104 PR tree-optimization/59594
3105 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3106 a copy of the datarefs vector rather than the vector itself.
3108 2014-01-28 Jason Merrill <jason@redhat.com>
3111 * dwarf2out.c (auto_die): New static.
3112 (gen_type_die_with_usage): Handle C++1y 'auto'.
3113 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3116 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3119 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3120 (SPEC_X32): Likewise.
3121 (SPEC_64): Likewise.
3122 * config/i386/i386.c (ix86_option_override_internal): Turn off
3123 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3125 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3126 * config/i386/i386.h (TARGET_16BIT): New macro.
3127 (TARGET_16BIT_P): Likewise.
3128 * config/i386/i386.opt: Add m16.
3129 * doc/invoke.texi: Document -m16.
3131 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3133 PR preprocessor/59935
3134 * input.c (location_get_source_line): Bail out on when line number
3135 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3137 2014-01-28 Richard Biener <rguenther@suse.de>
3139 PR tree-optimization/58742
3140 * tree-ssa-forwprop.c (associate_plusminus): Handle
3141 pointer subtraction of the form (T)(P + A) - (T)P.
3143 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3145 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3148 2014-01-28 Richard Biener <rguenther@suse.de>
3151 2014-01-28 Richard Biener <rguenther@suse.de>
3153 PR rtl-optimization/45364
3154 PR rtl-optimization/59890
3155 * var-tracking.c (local_get_addr_clear_given_value): Handle
3156 already cleared slot.
3157 (val_reset): Handle not allocated local_get_addr_cache.
3158 (vt_find_locations): Use post-order on the inverted CFG.
3160 2014-01-28 Richard Biener <rguenther@suse.de>
3162 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3164 2014-01-28 Richard Biener <rguenther@suse.de>
3166 PR rtl-optimization/45364
3167 PR rtl-optimization/59890
3168 * var-tracking.c (local_get_addr_clear_given_value): Handle
3169 already cleared slot.
3170 (val_reset): Handle not allocated local_get_addr_cache.
3171 (vt_find_locations): Use post-order on the inverted CFG.
3173 2014-01-28 Alan Modra <amodra@gmail.com>
3175 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3176 * configure.ac <recursive call for build != host>: Define
3177 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3178 and LD_FOR_BUILD too.
3179 * configure: Regenerate.
3181 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3183 * config/i386/i386.c (get_builtin_code_for_version): Separate
3184 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3185 Broadwell from Haswell.
3187 2014-01-27 Steve Ellcey <sellcey@mips.com>
3189 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3190 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3191 * config/mips/mips.c (mips_option_override): Change setting
3193 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3194 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3195 Change from Mask to Var.
3197 2014-01-27 Jeff Law <law@redhat.com>
3199 * ipa-inline.c (inline_small_functions): Fix typo.
3201 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3203 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3204 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3205 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3206 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3207 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3208 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3209 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3210 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3211 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3212 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3213 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3214 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3215 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3216 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3217 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3218 (_mm512_storeu_epi64): Ditto.
3219 (_mm512_cmpge_epi32_mask): Ditto.
3220 (_mm512_cmpge_epu32_mask): Ditto.
3221 (_mm512_cmpge_epi64_mask): Ditto.
3222 (_mm512_cmpge_epu64_mask): Ditto.
3223 (_mm512_cmple_epi32_mask): Ditto.
3224 (_mm512_cmple_epu32_mask): Ditto.
3225 (_mm512_cmple_epi64_mask): Ditto.
3226 (_mm512_cmple_epu64_mask): Ditto.
3227 (_mm512_cmplt_epi32_mask): Ditto.
3228 (_mm512_cmplt_epu32_mask): Ditto.
3229 (_mm512_cmplt_epi64_mask): Ditto.
3230 (_mm512_cmplt_epu64_mask): Ditto.
3231 (_mm512_cmpneq_epi32_mask): Ditto.
3232 (_mm512_cmpneq_epu32_mask): Ditto.
3233 (_mm512_cmpneq_epi64_mask): Ditto.
3234 (_mm512_cmpneq_epu64_mask): Ditto.
3235 (_mm512_expand_pd): Ditto.
3236 (_mm512_expand_ps): Ditto.
3237 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3238 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3239 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3240 * config/i386/i386.c (ix86_builtins): Add
3241 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3242 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3243 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3244 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3245 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3246 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3247 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3248 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3249 IX86_BUILTIN_PMOVUSQW512_MEM.
3250 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3251 __builtin_ia32_pmovsqd512mem_mask,
3252 __builtin_ia32_pmovqd512mem_mask,
3253 __builtin_ia32_pmovusqw512mem_mask,
3254 __builtin_ia32_pmovsqw512mem_mask,
3255 __builtin_ia32_pmovqw512mem_mask,
3256 __builtin_ia32_pmovusdw512mem_mask,
3257 __builtin_ia32_pmovsdw512mem_mask,
3258 __builtin_ia32_pmovdw512mem_mask,
3259 __builtin_ia32_pmovqb512mem_mask,
3260 __builtin_ia32_pmovusqb512mem_mask,
3261 __builtin_ia32_pmovsqb512mem_mask,
3262 __builtin_ia32_pmovusdb512mem_mask,
3263 __builtin_ia32_pmovsdb512mem_mask,
3264 __builtin_ia32_pmovdb512mem_mask.
3265 (bdesc_args): Add __builtin_ia32_expanddf512,
3266 __builtin_ia32_expandsf512.
3267 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3268 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3269 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3270 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3271 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3272 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3273 (avx512f_<code>v8div16qi2_mask_store): This.
3274 (avx512f_expand<mode>): New.
3276 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3278 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3280 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3281 (_mm512_prefetch_i32scatter_pd): Ditto.
3282 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3283 (_mm512_prefetch_i64scatter_pd): Ditto.
3284 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3285 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3286 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3287 (_mm512_prefetch_i32scatter_ps): Ditto.
3288 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3289 (_mm512_prefetch_i64scatter_ps): Ditto.
3290 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3291 * config/i386/i386-builtin-types.def: Define
3292 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3293 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3294 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3295 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3296 IX86_BUILTIN_SCATTERPFQPD.
3297 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3298 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3299 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3300 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3301 __builtin_ia32_scatterpfqps.
3302 (ix86_expand_builtin): Expand new built-ins.
3303 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3304 fix memory access data type.
3305 (*avx512pf_gatherpf<mode>_mask): Ditto.
3306 (*avx512pf_gatherpf<mode>): Ditto.
3307 (avx512pf_scatterpf<mode>): Ditto.
3308 (*avx512pf_scatterpf<mode>_mask): Ditto.
3309 (*avx512pf_scatterpf<mode>): Ditto.
3310 (GATHER_SCATTER_SF_MEM_MODE): New.
3311 (avx512pf_gatherpf<mode>df): Ditto.
3312 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3313 (*avx512pf_scatterpf<mode>df): Ditto.
3315 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3318 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3319 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3322 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3324 * common/config/arm/arm-common.c
3325 (arm_rewrite_mcpu): Handle multiple names.
3327 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3329 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3331 * gimple-builder.h (create_gimple_tmp): Delete.
3333 2014-01-27 Christian Bruel <christian.bruel@st.com>
3335 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3338 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3340 * config/pa/pa.md (call): Generate indirect long calls to non-local
3341 functions when outputing 32-bit code.
3342 (call_value): Likewise except for special call to buggy powf function.
3344 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3345 portable runtime and PIC indirect calls.
3346 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3347 and PIC call sequences. Use ldo instead of blr to set return register
3348 in PIC call sequence.
3350 2014-01-25 Walter Lee <walt@tilera.com>
3352 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3353 avoid clobbering a live register.
3355 2014-01-25 Walter Lee <walt@tilera.com>
3357 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3358 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3359 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3360 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3362 2014-01-25 Walter Lee <walt@tilera.com>
3364 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3365 arguments on even registers.
3366 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3368 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3369 (BIGGEST_ALIGNMENT): Ditto.
3370 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3372 2014-01-25 Walter Lee <walt@tilera.com>
3374 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3375 insns before bundling.
3376 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3378 2014-01-25 Walter Lee <walt@tilera.com>
3380 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3381 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3382 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3384 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3386 * config/mips/constraints.md (kl): Delete.
3387 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3388 define expands, using...
3389 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3390 instructions for MIPS16.
3391 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3392 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3394 2014-01-25 Walter Lee <walt@tilera.com>
3396 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3400 2014-01-25 Walter Lee <walt@tilera.com>
3402 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3403 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3405 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3407 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3410 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3412 * print-rtl.c (in_call_function_usage): New var.
3413 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3414 EXPR_LIST mode as mode and not as reg note name.
3417 * cfgloopmanip.c (copy_loop_info): If
3418 loop->warned_aggressive_loop_optimizations, make sure
3419 the flag is set in target loop too.
3421 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3423 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3425 * builtins.def: Likewise.
3426 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3427 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3428 * ira.c (ira_setup_eliminable_regset): Likewise.
3429 * omp-low.c (gate_expand_omp): Likewise.
3430 (execute_lower_omp): Likewise.
3431 (diagnose_sb_0): Likewise.
3432 (gate_diagnose_omp_blocks): Likewise.
3433 (simd_clone_clauses_extract): Likewise.
3436 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3438 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3439 correction for little endian...
3440 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3443 2014-01-24 Jeff Law <law@redhat.com>
3445 PR tree-optimization/59919
3446 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3447 for non-returning calls.
3449 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3451 * common/config/aarch64/aarch64-common.c
3452 (aarch64_rewrite_mcpu): Handle multiple names.
3453 * config/aarch64/aarch64.h
3454 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3456 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3458 * input.c (add_file_to_cache_tab): Handle the case where fopen
3461 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3464 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3465 from push operand if code of push isn't PRE_DEC.
3467 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3470 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3471 -mquad-memory-atomic. Update -mquad-memory documentation to say
3472 it is only used for non-atomic loads/stores.
3474 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3475 -mquad-memory or -mquad-memory-atomic switches.
3477 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3478 -mquad-memory-atomic to ISA 2.07 support.
3480 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3481 to separate support of normal quad word memory operations (ldq, stq)
3482 from the atomic quad word memory operations.
3484 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3485 support to separate non-atomic quad word operations from atomic
3486 quad word operations. Disable non-atomic quad word operations in
3487 little endian mode so that we don't have to swap words after the
3488 load and before the store.
3489 (quad_load_store_p): Add comment about atomic quad word support.
3490 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3491 options printed with -mdebug=reg.
3493 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3494 -mquad-memory-atomic as the test for whether we have quad word
3495 atomic instructions.
3496 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3497 or -mp8-vector are used, allow byte/half-word atomic operations.
3499 * config/rs6000/sync.md (load_lockedti): Insure that the address
3500 is a proper indexed or indirect address for the lqarx instruction.
3501 On little endian systems, swap the hi/lo registers after the lqarx
3503 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3504 insure the address is valid for the lqarx instruction.
3505 (store_conditionalti): Insure that the address is a proper indexed
3506 or indirect address for the stqcrx. instruction. On little endian
3507 systems, swap the hi/lo registers before doing the stqcrx.
3509 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3510 insure the address is valid for the stqcrx. instruction.
3512 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3513 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3514 type of quad memory support is available.
3516 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3519 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3520 there is a danger of looping.
3522 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3525 force flag_ira_loop_pressure if set via command line.
3527 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3529 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3530 (ashr_simd): New builtin handling DI mode.
3531 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3532 (aarch64_sshr_simddi): New match pattern.
3533 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3534 (vshrd_n_s64): Likewise.
3535 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3537 2014-01-23 Nick Clifton <nickc@redhat.com>
3539 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3540 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3541 favour of mcu specific scripts.
3542 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3545 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3546 Alex Velenko <Alex.Velenko@arm.com>
3548 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3549 (vaddv_s16): Likewise.
3550 (vaddv_s32): Likewise.
3551 (vaddv_u8): Likewise.
3552 (vaddv_u16): Likewise.
3553 (vaddv_u32): Likewise.
3554 (vaddvq_s8): Likewise.
3555 (vaddvq_s16): Likewise.
3556 (vaddvq_s32): Likewise.
3557 (vaddvq_s64): Likewise.
3558 (vaddvq_u8): Likewise.
3559 (vaddvq_u16): Likewise.
3560 (vaddvq_u32): Likewise.
3561 (vaddvq_u64): Likewise.
3562 (vaddv_f32): Likewise.
3563 (vaddvq_f32): Likewise.
3564 (vaddvq_f64): Likewise.
3565 (vmaxv_f32): Likewise.
3566 (vmaxv_s8): Likewise.
3567 (vmaxv_s16): Likewise.
3568 (vmaxv_s32): Likewise.
3569 (vmaxv_u8): Likewise.
3570 (vmaxv_u16): Likewise.
3571 (vmaxv_u32): Likewise.
3572 (vmaxvq_f32): Likewise.
3573 (vmaxvq_f64): Likewise.
3574 (vmaxvq_s8): Likewise.
3575 (vmaxvq_s16): Likewise.
3576 (vmaxvq_s32): Likewise.
3577 (vmaxvq_u8): Likewise.
3578 (vmaxvq_u16): Likewise.
3579 (vmaxvq_u32): Likewise.
3580 (vmaxnmv_f32): Likewise.
3581 (vmaxnmvq_f32): Likewise.
3582 (vmaxnmvq_f64): Likewise.
3583 (vminv_f32): Likewise.
3584 (vminv_s8): Likewise.
3585 (vminv_s16): Likewise.
3586 (vminv_s32): Likewise.
3587 (vminv_u8): Likewise.
3588 (vminv_u16): Likewise.
3589 (vminv_u32): Likewise.
3590 (vminvq_f32): Likewise.
3591 (vminvq_f64): Likewise.
3592 (vminvq_s8): Likewise.
3593 (vminvq_s16): Likewise.
3594 (vminvq_s32): Likewise.
3595 (vminvq_u8): Likewise.
3596 (vminvq_u16): Likewise.
3597 (vminvq_u32): Likewise.
3598 (vminnmv_f32): Likewise.
3599 (vminnmvq_f32): Likewise.
3600 (vminnmvq_f64): Likewise.
3602 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3604 * config/aarch64/aarch64-simd.md
3605 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3606 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3607 (*aarch64_mul3_elt<mode>): Likewise.
3608 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3609 (*aarch64_mul3_elt_to_64v2df): Likewise.
3610 (*aarch64_mla_elt<mode>): Likewise.
3611 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3612 (*aarch64_mls_elt<mode>): Likewise.
3613 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3614 (*aarch64_fma4_elt<mode>): Likewise.
3615 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3616 (*aarch64_fma4_elt_to_64v2df): Likewise.
3617 (*aarch64_fnma4_elt<mode>): Likewise.
3618 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3619 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3620 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3621 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3622 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3623 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3624 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3625 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3626 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3628 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3630 * config/aarch64/aarch64-simd.md
3631 (aarch64_be_checked_get_lane<mode>): New define_expand.
3632 * config/aarch64/aarch64-simd-builtins.def
3633 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3634 New builtin definition.
3635 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3636 Use new safe be builtin.
3638 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3640 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3642 (aarch64_be_st1<mode>): Likewise.
3643 (aarch_ld1<VALL:mode>): Define_expand modified.
3644 (aarch_st1<VALL:mode>): Likewise.
3645 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3646 (UNSPEC_ST1): Likewise.
3648 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3650 * config/microblaze/microblaze.md: Add trap insn and attribute
3652 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3654 PR preprocessor/58580
3655 * input.h (location_get_source_line): Take an additional line_size
3657 (void diagnostics_file_cache_fini): Declare new function.
3658 * input.c (struct fcache): New type.
3659 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3660 New static constants.
3661 (diagnostic_file_cache_init, total_lines_num)
3662 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3663 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3664 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3665 (get_next_line, read_next_line, goto_next_line, read_line_num):
3666 New static function definitions.
3667 (diagnostic_file_cache_fini): New function.
3668 (location_get_source_line): Take an additional output line_len
3669 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3671 * diagnostic.c (diagnostic_finish): Call
3672 diagnostic_file_cache_fini.
3673 (adjust_line): Take an additional input parameter for the length
3674 of the line, rather than calculating it with strlen.
3675 (diagnostic_show_locus): Adjust the use of
3676 location_get_source_line and adjust_line with respect to their new
3677 signature. While displaying a line now, do not stop at the first
3678 null byte. Rather, display the zero byte as a space and keep
3679 going until we reach the size of the line.
3680 * Makefile.in: Add vec.o to OBJS-libcommon
3682 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3683 Ilya Tocar <ilya.tocar@intel.com>
3685 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3686 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3687 (__builtin_ia32_kmov16): Ditto.
3688 * config/i386/i386.md (UNSPEC_KMOV): New.
3691 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3693 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3694 (_mm512_storeu_si512): Ditto.
3696 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3699 * rtl.h (get_referenced_operands): Declare.
3700 * recog.c (get_referenced_operands): New function.
3701 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3702 operands have been referenced when recording LO_SUM references.
3704 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3706 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3708 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3710 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3711 Enable for generic and recent AMD targets.
3713 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3715 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3716 ARG_SIZE note when adjustment was eliminated.
3718 2014-01-22 Jeff Law <law@redhat.com>
3720 PR tree-optimization/59597
3721 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3722 in file. Accept new argument REGISTERING and use it to modify
3723 dump output appropriately.
3724 (register_jump_thread): Corresponding changes.
3725 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3726 thread paths involving joiner blocks. Add code to dump cancelled
3727 jump threading paths.
3729 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3731 PR rtl-optimization/59477
3732 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3733 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3735 2014-01-22 Tom Tromey <tromey@redhat.com>
3737 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3739 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3741 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3743 PR rtl-optimization/59896
3744 * lra-constraints.c (process_alt_operands): Check unused note for
3745 matched operands of insn with no output reloads.
3747 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3749 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3750 (mips_move_from_gpr_cost): Likewise.
3752 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3754 PR rtl-optimization/59858
3755 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3756 ira_class_hard_regs_num.
3757 (process_alt_operands): Increase reject for dying matched operand.
3759 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3762 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3763 smaller than size, perform several stores or loads and stores
3764 at dst + count - size to store or copy all of size bytes, rather
3765 than just last modesize bytes.
3767 2014-01-20 DJ Delorie <dj@redhat.com>
3769 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3770 that CLOBBERs are REGs before propogating their values.
3772 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3775 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3776 (cgraph_inline_failed_type): New function.
3777 * cgraph.h (DEFCIFCODE): Add type.
3778 (cgraph_inline_failed_type_t): New enum.
3779 (cgraph_inline_failed_type): New prototype.
3780 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3781 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3782 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3783 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3784 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3785 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3786 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3787 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3788 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3789 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3790 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3791 OPTIMIZATION_MISMATCH.
3792 * tree-inline.c (expand_call_inline): Emit errors during
3793 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3795 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3798 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3799 mode attribute in insn output.
3801 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3803 * output.h (output_constant): Delete.
3804 * varasm.c (output_constant): Make private.
3806 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3808 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3810 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3813 * tree.h (fold_builtin_strcat): New prototype.
3814 * builtins.c (fold_builtin_strcat): No longer static. Add len
3815 argument, if non-NULL, don't call c_strlen. Optimize
3816 directly into __builtin_memcpy instead of __builtin_strcpy.
3817 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3818 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3820 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3822 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3823 for SImode_address_operand operands, having only a REG argument.
3825 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3827 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3828 loader name using mbig-endian.
3829 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3831 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3833 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3837 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3839 * config/aarch64/aarch64-protos.h
3840 (aarch64_cannot_change_mode_class_ptr): Declare.
3841 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3842 aarch64_cannot_change_mode_class_ptr): New.
3843 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3844 backend hook aarch64_cannot_change_mode_class.
3846 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3848 * common/config/aarch64/aarch64-common.c
3849 (aarch64_handle_option): Don't handle any option order logic here.
3850 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3851 selected_cpu, warn on architecture version mismatch.
3852 (aarch64_override_options): Fix parsing order for option strings.
3854 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3855 Iain Sandoe <iain@codesourcery.com>
3858 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3859 warning. Amend comment to reflect current functionality.
3861 2014-01-20 Richard Biener <rguenther@suse.de>
3864 * builtins.c (fold_builtin_strcat): Remove case better handled
3865 by tree-ssa-strlen.c.
3867 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3869 * config/aarch64/aarch64.opt
3870 (mcpu, march, mtune): Make case-insensitive.
3872 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3875 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3876 if operands[1] is a REG or ZERO_EXTEND of a REG.
3878 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3880 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3882 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3884 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3885 long non-pic millicode calls.
3887 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3889 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3891 2014-01-19 Kito Cheng <kito@0xlab.org>
3893 * builtins.c (expand_movstr): Check movstr expand done or fail.
3895 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3896 H.J. Lu <hongjiu.lu@intel.com>
3899 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3900 to DImode for zero-extended addresses.
3902 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3904 PR rtl-optimization/57763
3905 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3906 on the new indirect jump_insn and increment LABEL_NUSES (label).
3908 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3912 * config.gcc (x86_archs): New variable.
3913 (x86_64_archs): Likewise.
3914 (x86_cpus): Likewise.
3915 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3916 --with-arch/--with-cpu= options.
3917 Support --with-arch=/--with-cpu={nehalem,westmere,
3918 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3920 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3922 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3923 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3925 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3927 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3929 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3932 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3933 clear cpp_get_options (parse_in)->warn_unused_macros for
3934 ix86_target_macros_internal with cpp_define.
3936 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3938 * jump.c (delete_related_insns): Keep (use (insn))s.
3939 * reorg.c (redundant_insn): Check for barriers too.
3941 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3943 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3945 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3947 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3948 call to $$dyncall when TARGET_LONG_CALLS is true.
3950 2014-01-17 Jeff Law <law@redhat.com>
3952 * ree.c (combine_set_extension): Temporarily disable test for
3953 changing number of hard registers.
3955 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3958 * ipa-inline-analysis.c (inline_free_summary):
3959 Do not free summary of aliases.
3961 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3964 * gimplify.c (gimplify_expr): Use create_tmp_var
3965 instead of create_tmp_var_raw. If cond doesn't have
3966 integral type, don't add the IFN_ANNOTATE builtin at all.
3968 2014-01-17 Martin Jambor <mjambor@suse.cz>
3971 * ipa-cp.c (prev_edge_clone): New variable.
3972 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3973 Also resize prev_edge_clone vector.
3974 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3975 (ipcp_edge_removal_hook): New function.
3976 (ipcp_driver): Register ipcp_edge_removal_hook.
3978 2014-01-17 Andrew Pinski <apinski@cavium.com>
3979 Steve Ellcey <sellcey@mips.com>
3982 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3985 2014-01-17 Jeff Law <law@redhat.com>
3988 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3989 so that pass_ccp runs first.
3991 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3993 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3994 (ix86_adjust_cost): Use !TARGET_XXX.
3995 (do_reorder_for_imul): Likewise.
3996 (swap_top_of_ready_list): Likewise.
3997 (ix86_sched_reorder): Likewise.
3999 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4001 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4002 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4003 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4004 (intel_memset): New. Duplicate slm_memset.
4005 (intel_cost): New. Duplicate slm_cost.
4006 (m_INTEL): New macro.
4007 (processor_target_table): Add "intel".
4008 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4009 with PROCESSOR_INTEL for "intel".
4010 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4011 PROCESSOR_SILVERMONT.
4012 (ix86_issue_rate): Likewise.
4013 (ix86_adjust_cost): Likewise.
4014 (ia32_multipass_dfa_lookahead): Likewise.
4015 (swap_top_of_ready_list): Likewise.
4016 (ix86_sched_reorder): Likewise.
4017 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4018 instead of TARGET_OPT_AGU.
4019 * config/i386/i386.h (TARGET_INTEL): New.
4020 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4021 (processor_type): Add PROCESSOR_INTEL.
4022 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4023 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4025 2014-01-17 Marek Polacek <polacek@redhat.com>
4028 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4031 2014-01-17 Richard Biener <rguenther@suse.de>
4033 PR tree-optimization/46590
4034 * opts.c (default_options_table): Add entries for
4035 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4036 all enabled at -O1 but not for -Og.
4037 * common.opt (fbranch-count-reg): Remove Init(1).
4038 (fmove-loop-invariants): Likewise.
4039 (ftree-pta): Likewise.
4041 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4043 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4044 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4045 decls to at least the GCC 4.8 used alignments.
4048 * tree-nested.c (convert_nonlocal_reference_stmt,
4049 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4050 of GIMPLE_BIND stmts, adjust associated decls.
4052 2014-01-17 Richard Biener <rguenther@suse.de>
4054 PR tree-optimization/46590
4055 * vec.h (vec<>::bseach): New member function implementing
4056 binary search according to C89 bsearch.
4057 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4058 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4059 bitmap pointer again. Make accesses_in_loop a flat array.
4060 (mem_ref_obstack): New global.
4061 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4062 (mark_ref_stored): Likewise.
4063 (ref_indep_loop_p_2): Likewise.
4064 (set_ref_stored_in_loop): New helper function.
4065 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4066 (memref_free): Adjust.
4067 (record_mem_ref_loc): Simplify.
4068 (gather_mem_refs_stmt): Adjust.
4069 (sort_locs_in_loop_postorder_cmp): New function.
4070 (analyze_memory_references): Sort accesses_in_loop after
4071 loop postorder number.
4072 (find_ref_loc_in_loop_cmp): New function.
4073 (for_all_locs_in_loop): Find relevant cluster of locs in
4074 accesses_in_loop and iterate without recursion.
4075 (execute_sm): Avoid uninit warning.
4076 (struct ref_always_accessed): Simplify.
4077 (ref_always_accessed::operator ()): Likewise.
4078 (ref_always_accessed_p): Likewise.
4079 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4080 loop postorder numbers here.
4081 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4084 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4087 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4088 on decls for which assemble_alias has been called.
4090 2014-01-17 Nick Clifton <nickc@redhat.com>
4092 * config/msp430/msp430.opt: (mcpu): New option.
4093 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4094 (msp430_option_override): Parse target_cpu. If the MCU name
4095 matches a generic string, clear target_mcu.
4096 (msp430_attr): Allow numeric interrupt values up to 63.
4097 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4098 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4100 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4102 * config/msp430/msp430.md (popm): Use %J rather than %I.
4103 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4104 (addhi_cy_i): Use immediate_operand for operand 2.
4105 * doc/invoke.texi: Document -mcpu option.
4107 2014-01-17 Richard Biener <rguenther@suse.de>
4109 PR rtl-optimization/38518
4110 * df.h (df_analyze_loop): Declare.
4111 * df-core.c: Include cfgloop.h.
4112 (df_analyze_1): Split out main part of df_analyze.
4113 (df_analyze): Adjust.
4114 (loop_inverted_post_order_compute): New function.
4115 (loop_post_order_compute): Likewise.
4116 (df_analyze_loop): New function avoiding whole-function
4118 * loop-invariant.c (find_defs): Use df_analyze_loop.
4119 (find_invariants): Adjust.
4120 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4122 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4124 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4125 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4127 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4129 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4130 traversal when removing references.
4132 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4135 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4137 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4140 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4141 pushing a reload for an autoinc when we had previously reloaded an
4142 inner part of the address.
4144 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4146 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4148 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4149 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4150 when not giving up or versioning for alias only because of
4152 (vect_analyze_data_ref_dependences): Set to true.
4153 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4155 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4156 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4160 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4163 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4164 operand 0 predicate for gathers, use a new pseudo as subtarget.
4166 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4169 * lra-constraints.c (process_alt_operands): Add printing debug info.
4170 Check absence of input/output reloads for matched operands too.
4172 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4174 PR rtl-optimization/59835
4175 * ira.c (ira_init_register_move_cost): Increase cost for
4178 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4180 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4182 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4185 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4186 non-register objects. Use gen_(high/low)part more consistently.
4189 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4192 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4193 endian support, remove tests for WORDS_BIG_ENDIAN.
4194 (p8_mfvsrd_3_<mode>): Likewise.
4195 (reload_gpr_from_vsx<mode>): Likewise.
4196 (reload_gpr_from_vsxsf): Likewise.
4197 (p8_mfvsrd_4_disf): Likewise.
4199 2014-01-16 Richard Biener <rguenther@suse.de>
4201 PR rtl-optimization/46590
4202 * lcm.c (compute_antinout_edge): Use postorder iteration.
4203 (compute_laterin): Use inverted postorder iteration.
4205 2014-01-16 Nick Clifton <nickc@redhat.com>
4208 * varasm.c (output_constant): Return the number of bytes actually
4210 (output_constructor_array_range): Update the field size with the
4211 number of bytes emitted by output_constant.
4212 (output_constructor_regular_field): Likewise. Also do not
4213 complain if the total number of bytes emitted is now greater
4214 than the expected fieldpos.
4215 * output.h (output_constant): Update prototype and descriptive comment.
4217 2014-01-16 Marek Polacek <polacek@redhat.com>
4220 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4221 it is error_mark_node.
4223 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4225 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4226 VALID_AVX256_REG_OR_OI_MODE.
4228 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4230 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4231 current procedure should be profiled.
4233 2014-01-15 Andrew Pinski <apinski@cavium.com>
4235 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4236 of moving from/to the STACK_REG register class.
4238 2014-01-15 Richard Henderson <rth@redhat.com>
4241 * reginfo.c (global_regs_decl): Globalize.
4242 * rtl.h (global_regs_decl): Declare.
4243 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4244 reserved via global_regs.
4246 2014-01-15 Teresa Johnson <tejohnson@google.com>
4248 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4250 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4252 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4253 and vmulosh rather than call gen_vec_widen_smult_*.
4254 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4255 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4256 (vec_widen_smult_even_v16qi): Likewise.
4257 (vec_widen_umult_even_v8hi): Likewise.
4258 (vec_widen_smult_even_v8hi): Likewise.
4259 (vec_widen_umult_odd_v16qi): Likewise.
4260 (vec_widen_smult_odd_v16qi): Likewise.
4261 (vec_widen_umult_odd_v8hi): Likewise.
4262 (vec_widen_smult_odd_v8hi): Likewise.
4263 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4264 vmuloub rather than call gen_vec_widen_umult_*.
4265 (vec_widen_umult_lo_v16qi): Likewise.
4266 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4267 vmulosb rather than call gen_vec_widen_smult_*.
4268 (vec_widen_smult_lo_v16qi): Likewise.
4269 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4270 rather than call gen_vec_widen_umult_*.
4271 (vec_widen_umult_lo_v8hi): Likewise.
4272 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4273 rather than call gen_vec_widen_smult_*.
4274 (vec_widen_smult_lo_v8hi): Likewise.
4276 2014-01-15 Jeff Law <law@redhat.com>
4278 PR tree-optimization/59747
4279 * ree.c (find_and_remove_re): Properly handle case where a second
4280 eliminated extension requires widening a copy created for elimination
4281 of a prior extension.
4282 (combine_set_extension): Ensure that the number of hard regs needed
4283 for a destination register does not change when we widen it.
4285 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4287 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4288 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4289 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4290 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4291 (avr-*-rtems*): Likewise.
4292 (bfin*-rtems*): Likewise.
4293 (moxie-*-rtems*): Likewise.
4294 (h8300-*-rtems*): Likewise.
4295 (i[34567]86-*-rtems*): Likewise.
4296 (lm32-*-rtems*): Likewise.
4297 (m32r-*-rtems*): Likewise.
4298 (m68k-*-rtems*): Likewise.
4299 (microblaze*-*-rtems*): Likewise.
4300 (mips*-*-rtems*): Likewise.
4301 (powerpc-*-rtems*): Likewise.
4302 (sh-*-rtems*): Likewise.
4303 (sparc-*-rtems*): Likewise.
4304 (sparc64-*-rtems*): Likewise.
4305 (v850-*-rtems*): Likewise.
4306 (m32c-*-rtems*): Likewise.
4308 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4310 PR rtl-optimization/59511
4311 * ira.c (ira_init_register_move_cost): Use memory costs for some
4312 cases of register move cost calculations.
4313 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4314 instead of BB frequency.
4315 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4316 * lra-assigns.c (find_hard_regno_for): Ditto.
4318 2014-01-15 Richard Biener <rguenther@suse.de>
4320 PR tree-optimization/59822
4321 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4322 (vectorizable_load): Use it to hoist defs of uses of invariant
4323 loads out of the loop.
4325 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4326 Kugan Vivekanandarajah <kuganv@linaro.org>
4329 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4332 2014-01-15 Richard Biener <rguenther@suse.de>
4334 PR rtl-optimization/59802
4335 * lcm.c (compute_available): Use inverted postorder to seed
4336 the initial worklist.
4338 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4341 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4342 ADDR_REGS for invalid symrefs in non-PIC code.
4344 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4347 * builtins.c (determine_block_size): Initialize *probable_max_size
4348 even if len_rtx is CONST_INT.
4350 2014-01-14 Andrew Pinski <apinski@cavium.com>
4352 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4353 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4354 (cortexa53_tunings): Likewise.
4355 (aarch64_sched_issue_rate): New function.
4356 (TARGET_SCHED_ISSUE_RATE): Define.
4358 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4360 * ira-costs.c (find_costs_and_classes): Add missed
4361 ira_init_register_move_cost_if_necessary.
4363 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4366 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4368 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4371 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4372 to indicate if type is used for function return value. Warn ABI
4373 change if the vector mode isn't available for function return value.
4374 (ix86_function_arg_advance): Pass false to type_natural_mode.
4375 (ix86_function_arg): Likewise.
4376 (ix86_gimplify_va_arg): Likewise.
4377 (function_arg_32): Don't warn ABI change.
4378 (ix86_function_value): Pass true to type_natural_mode.
4379 (ix86_return_in_memory): Likewise.
4380 (ix86_struct_value_rtx): Removed.
4381 (TARGET_STRUCT_VALUE_RTX): Likewise.
4383 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4385 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4386 converting a conditional jump into a conditional return.
4388 2014-01-14 Richard Biener <rguenther@suse.de>
4390 PR tree-optimization/58921
4391 PR tree-optimization/59006
4392 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4393 hoisting invariant stmts.
4394 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4395 invariant loads on the preheader edge if possible.
4397 2014-01-14 Joey Ye <joey.ye@arm.com>
4399 * doc/plugin.texi (Building GCC plugins): Update to C++.
4401 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4403 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4404 (_mm_rcp28_round_ss): Ditto.
4405 (_mm_rsqrt28_round_sd): Ditto.
4406 (_mm_rsqrt28_round_ss): Ditto.
4407 (_mm_rcp28_sd): Ditto.
4408 (_mm_rcp28_ss): Ditto.
4409 (_mm_rsqrt28_sd): Ditto.
4410 (_mm_rsqrt28_ss): Ditto.
4411 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4412 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4413 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4414 (IX86_BUILTIN_RCP28SD): Ditto.
4415 (IX86_BUILTIN_RCP28SS): Ditto.
4416 (IX86_BUILTIN_RSQRT28SD): Ditto.
4417 (IX86_BUILTIN_RSQRT28SS): Ditto.
4418 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4419 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4420 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4421 (ix86_expand_special_args_builtin): Expand new FTYPE.
4422 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4423 (srcp14<mode>): Make insn unary.
4424 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4425 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4426 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4427 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4428 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4429 Fix rounding: make it SAE only.
4430 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4432 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4434 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4435 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4436 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4437 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4438 (round_saeonly_mask_scalar_operand4): Ditto.
4439 (round_saeonly_mask_scalar_op3): Ditto.
4440 (round_saeonly_mask_scalar_op4): Ditto.
4442 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4444 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4445 Implement -maltivec=be for vec_insert and vec_extract.
4447 2014-01-10 DJ Delorie <dj@redhat.com>
4449 * config/msp430/msp430.md (call_internal): Don't allow memory
4450 references with SP as the base register.
4451 (call_value_internal): Likewise.
4452 * config/msp430/constraints.md (Yc): New. For memory references
4453 that don't use SP as a base register.
4455 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4456 "an integer without a # prefix"
4457 * config/msp430/msp430.md (epilogue_helper): Use it.
4459 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4462 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4463 AVX512F gather builtins.
4464 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4465 on gather decls with INTEGER_TYPE masktype.
4466 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4467 directly into the builtin rather than hoisting it before loop.
4469 PR tree-optimization/59387
4470 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4471 (scev_const_prop): If folded_casts and type has undefined overflow,
4472 use force_gimple_operand instead of force_gimple_operand_gsi and
4473 for each added stmt if it is assign with
4474 arith_code_with_undefined_signed_overflow, call
4475 rewrite_to_defined_overflow.
4476 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4477 gimple-fold.h instead.
4478 (arith_code_with_undefined_signed_overflow,
4479 rewrite_to_defined_overflow): Moved to ...
4480 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4481 rewrite_to_defined_overflow): ... here. No longer static.
4482 Include gimplify-me.h.
4483 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4484 rewrite_to_defined_overflow): New prototypes.
4486 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4488 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4490 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4492 * builtins.c (get_object_alignment_2): Minor tweak.
4493 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4495 2014-01-13 Christian Bruel <christian.bruel@st.com>
4497 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4498 optimized non constant lengths.
4500 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4503 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4504 load as __atomic_load_N if possible.
4506 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4508 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4510 (rs6000_expand_builtin): Adjust call.
4512 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4515 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4516 * config/rs6000/rs6000.c: Include target-globals.h.
4517 (rs6000_set_current_function): Instead of doing target_reinit
4518 unconditionally, use save_target_globals_default_opts and
4519 restore_target_globals.
4521 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4523 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4524 (rs6000_expand_builtin): Handle mffs and mtfsf.
4525 (rs6000_init_builtins): Define mffs and mtfsf.
4526 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4527 (rs6000_mffs): New pattern.
4528 (rs6000_mtfsf): New pattern.
4530 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4532 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4533 Start narrowing with START. Apply candidate-use pair
4534 and check overall cost in narrowing.
4535 (iv_ca_prune): Pass new argument.
4537 2014-01-10 Jeff Law <law@redhat.com>
4540 * ree.c (combine_reaching_defs): Ensure the defining statement
4541 occurs before the extension when optimizing extensions with
4542 different source and destination hard registers.
4544 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4547 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4548 vtables into the type inheritance graph.
4550 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4552 PR rtl-optimization/59754
4553 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4554 modes in the REGNO != REGNO case.
4556 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4558 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4560 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4562 PR tree-optimization/59745
4563 * tree-predcom.c (tree_predictive_commoning_loop): Call
4564 free_affine_expand_cache if giving up because components is NULL.
4566 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4567 GC in payload of target_globals struct instead of allocating them on
4568 the heap and the larger structs separately using GC.
4569 * target-globals.h (struct target_globals): Make regs, hard_regs,
4570 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4571 of GTY((skip)) and change type to void *.
4572 (reset_target_globals): Cast loads from those fields to corresponding
4575 2014-01-10 Steve Ellcey <sellcey@mips.com>
4578 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4579 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4580 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4582 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4585 * aarch64-modes.def (CC_Zmode): New flags mode.
4586 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4587 represents an equality.
4588 (aarch64_get_condition_code): Handle CC_Zmode.
4589 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4591 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4593 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4594 extraction in good case.
4596 2014-01-10 Richard Biener <rguenther@suse.de>
4598 PR tree-optimization/59374
4599 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4600 checking after SLP discovery. Mark stmts not participating
4601 in any SLP instance properly.
4603 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4605 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4606 when handling a SET rtx.
4608 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4610 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4611 (cortex-a57): Likewise.
4612 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4614 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4616 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4617 non-iwmmxt builtins.
4619 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4623 * ipa-devirt.c record_target_from_binfo): Take as argument
4624 stack of binfos and lookup matching one for virtual inheritance.
4625 (possible_polymorphic_call_targets_1): Update.
4627 2014-01-10 Huacai Chen <chenhc@lemote.com>
4629 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4630 kernel strings for Loongson-2E/2F/3A.
4632 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4635 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4636 is_gimple_call before calling gimple_call_internal_p.
4638 2014-01-09 Steve Ellcey <sellcey@mips.com>
4640 * Makefile.in (TREE_FLOW_H): Remove.
4641 (TREE_SSA_H): Add file names from tree-flow.h.
4642 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4643 * tree.h: Remove tree-flow.h reference.
4644 * hash-table.h: Remove tree-flow.h reference.
4645 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4646 reference with tree-ssa-loop.h.
4648 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4650 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4651 default element-order behavior for -maltivec.
4652 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4654 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4655 when targeting big endian, at least for now.
4656 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4658 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4661 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4662 var satisfies use_register_for_decl, just take into account type
4663 alignment, rather than decl alignment.
4665 PR tree-optimization/59622
4666 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4667 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4668 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4669 Don't devirtualize for inplace at all. For targets.length () == 1,
4670 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4672 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4674 * config/i386/i386.md (cpu): Remove the unused btver1.
4676 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4678 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4680 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4683 * tree-core.h (struct target_globals): New forward declaration.
4684 (struct tree_target_option): Add globals field.
4685 * tree.h (TREE_TARGET_GLOBALS): Define.
4686 (prepare_target_option_nodes_for_pch): New prototype.
4687 * target-globals.h (struct target_globals): Define even if
4689 * tree.c (prepare_target_option_node_for_pch,
4690 prepare_target_option_nodes_for_pch): New functions.
4691 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4692 * config/i386/i386.c: Include target-globals.h.
4693 (ix86_set_current_function): Instead of doing target_reinit
4694 unconditionally, use save_target_globals_default_opts and
4695 restore_target_globals.
4697 2014-01-09 Richard Biener <rguenther@suse.de>
4699 PR tree-optimization/59715
4700 * tree-cfg.h (split_critical_edges): Declare.
4701 * tree-cfg.c (split_critical_edges): Export.
4702 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4704 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4706 * cfgexpand.c (expand_stack_vars): Optionally disable
4707 asan stack protection.
4708 (expand_used_vars): Likewise.
4709 (partition_stack_vars): Likewise.
4710 * asan.c (asan_emit_stack_protection): Optionally disable
4711 after return stack usage.
4712 (instrument_derefs): Optionally disable memory access instrumentation.
4713 (instrument_builtin_call): Likewise.
4714 (instrument_strlen_call): Likewise.
4715 (asan_protect_global): Optionally disable global variables protection.
4716 * doc/invoke.texi: Added doc for new options.
4717 * params.def: Added new options.
4718 * params.h: Likewise.
4720 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4722 PR rtl-optimization/59724
4723 * ifcvt.c (cond_exec_process_if_block): Don't call
4724 flow_find_head_matching_sequence with 0 longest_match.
4725 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4726 non-active insns if !stop_after.
4727 (try_head_merge_bb): Revert 2014-01-07 changes.
4729 2014-01-08 Jeff Law <law@redhat.com>
4731 * ree.c (get_sub_rtx): New function, extracted from...
4732 (merge_def_and_ext): Here.
4733 (combine_reaching_defs): Use get_sub_rtx.
4735 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4737 * cgraph.h (varpool_variable_node): Do not choke on null node.
4739 2014-01-08 Catherine Moore <clm@codesourcery.com>
4741 * config/mips/mips.md (simple_return): Attempt to use JRC
4743 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4745 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4747 PR rtl-optimization/59137
4748 * reorg.c (steal_delay_list_from_target): Call update_block for
4750 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4752 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4754 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4755 two duplicate entries.
4757 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4760 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4762 * config/mips/mips.c (mips_truncated_op_cost): New function.
4763 (mips_rtx_costs): Adjust test for BADDU.
4764 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4766 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4768 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4769 (*baddu_si): ...this new pattern.
4771 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4774 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4776 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4779 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4781 (expand_expr, expand_normal): Adjust.
4782 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4783 inner_reference_p. Use inner_reference_p to expand inner references.
4784 (store_expr): Adjust.
4785 * cfgexpand.c (expand_call_stmt): Adjust.
4787 2014-01-08 Rong Xu <xur@google.com>
4789 * gcov-io.c (gcov_var): Move from gcov-io.h.
4790 (gcov_position): Ditto.
4791 (gcov_is_error): Ditto.
4792 (gcov_rewrite): Ditto.
4793 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4794 only part to libgcc/libgcov.h.
4796 2014-01-08 Marek Polacek <polacek@redhat.com>
4799 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4801 2014-01-08 Marek Polacek <polacek@redhat.com>
4804 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4806 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4808 PR rtl-optimization/59649
4809 * stor-layout.c (get_mode_bounds): For BImode return
4810 0 and STORE_FLAG_VALUE.
4812 2014-01-08 Richard Biener <rguenther@suse.de>
4815 * gimple.h (is_gimple_builtin_call): Remove.
4816 (gimple_builtin_call_types_compatible_p): New.
4817 (gimple_call_builtin_p): New overload.
4818 * gimple.c (is_gimple_builtin_call): Remove.
4819 (validate_call): Rename to ...
4820 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4822 (validate_type): New static function.
4823 (gimple_call_builtin_p): New overload and adjust.
4824 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4825 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4826 (gimple_fold_stmt_to_constant_1): Likewise.
4827 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4829 2014-01-08 Richard Biener <rguenther@suse.de>
4832 * gimplify.c (gimplify_expr): Gimplify register-register type
4833 VIEW_CONVERT_EXPRs to separate stmts.
4835 2014-01-07 Jeff Law <law@redhat.com>
4838 * ree.c (combine_set_extension): Handle case where source
4839 and destination registers in an extension insn are different.
4840 (combine_reaching_defs): Allow source and destination registers
4841 in extension to be different under limited circumstances.
4842 (add_removable_extension): Remove restriction that the
4843 source and destination registers in the extension are the same.
4844 (find_and_remove_re): Emit a copy from the extension's
4845 destination to its source after the defining insn if
4846 the source and destination registers are different.
4849 * ifcvt.c (merge_if_block): If we are merging a block with more than
4850 one successor with a block with no successors, remove any BARRIER
4851 after the second block.
4853 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4855 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4857 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4860 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4861 for 14-bit register offsets when INT14_OK_STRICT is false.
4863 2014-01-07 Roland Stigge <stigge@antcom.de>
4864 Michael Meissner <meissner@linux.vnet.ibm.com>
4867 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4868 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4870 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4872 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4875 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4877 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4878 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4879 rtx is const0_rtx or not.
4881 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4884 * target-globals.c (save_target_globals): Remove this_fn_optab
4886 * toplev.c: Include optabs.h.
4887 (target_reinit): Temporarily restore the global options if another
4888 set of options are in force.
4890 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4892 PR rtl-optimization/58668
4893 * cfgcleanup.c (flow_find_cross_jump): Don't count
4894 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4895 to determine what is counted.
4896 (flow_find_head_matching_sequence): Use active_insn_p to determine
4898 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4900 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4901 determine what is counted.
4903 PR tree-optimization/59643
4904 * tree-predcom.c (split_data_refs_to_components): If one dr is
4905 read and one write, determine_offset fails and the write isn't
4906 in the bad component, just put the read into the bad component.
4908 2014-01-07 Mike Stump <mikestump@comcast.net>
4909 Jakub Jelinek <jakub@redhat.com>
4912 * tree-core.h (struct tree_optimization_option): Change optabs
4913 type from unsigned char * to void *.
4914 * optabs.c (init_tree_optimization_optabs): Adjust
4915 TREE_OPTIMIZATION_OPTABS initialization.
4917 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4920 * config/i386/i386.h (struct machine_function): Add
4921 no_drap_save_restore field.
4922 * config/i386/i386.c (ix86_save_reg): Use
4923 !cfun->machine->no_drap_save_restore instead of
4924 crtl->stack_realign_needed.
4925 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4926 this function clears frame_pointer_needed. Set
4927 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4928 and DRAP reg is needed.
4930 2014-01-06 Marek Polacek <polacek@redhat.com>
4933 * doc/implement-c.texi: Mention that other integer types are
4934 permitted as bit-field types in strictly conforming mode.
4936 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4938 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4941 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4943 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4945 2014-01-06 Martin Jambor <mjambor@suse.cz>
4948 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4950 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4952 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4956 * var-tracking.c (add_stores): Preserve the value of the source even if
4957 we don't record the store.
4959 2014-01-06 Terry Guo <terry.guo@arm.com>
4961 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4963 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4966 * config/darwin.c (darwin_function_section): Adjust return values to
4967 correspond to optimisation changes made in r206070.
4969 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4971 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4972 from prefetch_block tune setting.
4973 (nocona_cost): Correct size of prefetch block to 64.
4975 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4977 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4978 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4979 used to save the static chain register in the computation of the offset
4980 from which the FP registers need to be restored.
4982 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4984 PR tree-optimization/59519
4985 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4986 ICE if get_current_def (current_new_name) is already non-NULL, as long
4987 as it is a phi result of some other phi in *new_exit_bb that has
4990 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4991 or vmovdqu* for misaligned_operand.
4992 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4993 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4994 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4995 aligned_mem for AVX512F masked aligned load and store builtins and for
4998 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5000 PR tree-optimization/59651
5001 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5002 Address range for negative step should be added by TYPE_SIZE_UNIT.
5004 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5006 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5007 overlapping registers also for registers other than the stack pointer.
5009 2014-01-03 Marek Polacek <polacek@redhat.com>
5012 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5015 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5018 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5021 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5022 (push splitter): Use <P:MODE_SIZE> instead of
5023 GET_MODE_SIZE (<P:MODE>mode).
5024 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5025 (mov -1, reg peephole2): Likewise.
5026 * config/i386/sse.md (*mov<mode>_internal,
5027 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5028 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5029 *<code><mode>3, *andnot<mode>3<mask_name>,
5030 <mask_codefor><code><mode>3<mask_name>): Likewise.
5031 * config/i386/subst.md (mask_mode512bit_condition,
5032 sd_mask_mode512bit_condition): Likewise.
5034 2014-01-02 Xinliang David Li <davidxl@google.com>
5036 PR tree-optimization/59303
5037 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5038 (dump_predicates): Better output format.
5039 (pred_equal_p): New function.
5040 (is_neq_relop_p): Ditto.
5041 (is_neq_zero_form_p): Ditto.
5042 (pred_expr_equal_p): Ditto.
5043 (pred_neg_p): Ditto.
5044 (simplify_pred): Ditto.
5045 (simplify_preds_2): Ditto.
5046 (simplify_preds_3): Ditto.
5047 (simplify_preds_4): Ditto.
5048 (simplify_preds): Ditto.
5050 (push_to_worklist): Ditto.
5051 (get_pred_info_from_cmp): Ditto.
5052 (is_degenerated_phi): Ditto.
5053 (normalize_one_pred_1): Ditto.
5054 (normalize_one_pred): Ditto.
5055 (normalize_one_pred_chain): Ditto.
5056 (normalize_preds): Ditto.
5057 (normalize_cond_1): Remove function.
5058 (normalize_cond): Ditto.
5059 (is_gcond_subset_of): Ditto.
5060 (is_subset_of_any): Ditto.
5061 (is_or_set_subset_of): Ditto.
5062 (is_and_set_subset_of): Ditto.
5063 (is_norm_cond_subset_of): Ditto.
5064 (pred_chain_length_cmp): Ditto.
5065 (convert_control_dep_chain_into_preds): Type change.
5066 (find_predicates): Ditto.
5067 (find_def_preds): Ditto.
5068 (destroy_predicates_vecs): Ditto.
5069 (find_matching_predicates_in_rest_chains): Ditto.
5070 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5071 (is_pred_expr_subset): Ditto.
5072 (is_pred_chain_subset_of): Ditto.
5073 (is_included_in): Ditto.
5074 (is_superset_of): Ditto.
5076 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5078 Update copyright years.
5080 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5082 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5083 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5084 config/arc/arc.md, config/arc/arc.opt,
5085 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5086 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5087 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5088 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5089 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5090 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5091 vtable-verify.c, vtable-verify.h: Use the standard form for the
5094 2014-01-02 Tobias Burnus <burnus@net-b.de>
5096 * gcc.c (process_command): Update copyright notice dates.
5097 * gcov-dump.c: Ditto.
5099 * doc/cpp.texi: Bump @copying's copyright year.
5100 * doc/cppinternals.texi: Ditto.
5101 * doc/gcc.texi: Ditto.
5102 * doc/gccint.texi: Ditto.
5103 * doc/gcov.texi: Ditto.
5104 * doc/install.texi: Ditto.
5105 * doc/invoke.texi: Ditto.
5107 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5109 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5111 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5113 * config/i386/sse.md (*mov<mode>_internal): Guard
5114 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5116 PR rtl-optimization/59647
5117 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5118 new_rtx into UNSIGNED_FLOAT rtxes.
5120 Copyright (C) 2014 Free Software Foundation, Inc.
5122 Copying and distribution of this file, with or without modification,
5123 are permitted in any medium without royalty provided the copyright
5124 notice and this notice are preserved.