1 2014-03-18 Richard Biener <rguenther@suse.de>
3 * doc/loop.texi: Remove section on the removed lambda framework.
4 Update loop docs with recent changes in preserving loop structure.
6 2014-03-18 Richard Biener <rguenther@suse.de>
8 * doc/lto.texi (-fresolution): Document.
10 2014-03-18 Richard Biener <rguenther@suse.de>
12 * doc/contrib.texi: Adjust my name.
14 2014-03-18 Jakub Jelinek <jakub@redhat.com>
17 * internal-fn.c: Include diagnostic-core.h.
18 (expand_BUILTIN_EXPECT): New function.
19 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
20 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
21 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
22 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
24 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
25 Revert 3 argument __builtin_expect code.
26 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
27 * gimple-fold.c (gimple_fold_call): Likewise.
28 * tree.h (fold_builtin_expect): New prototype.
29 * builtins.c (build_builtin_expect_predicate): Add predictor
30 argument, if non-NULL, create 3 argument __builtin_expect.
31 (fold_builtin_expect): No longer static. Add ARG2 argument,
32 pass it through to build_builtin_expect_predicate.
33 (fold_builtin_2): Adjust caller.
34 (fold_builtin_3): Handle BUILT_IN_EXPECT.
35 * internal-fn.def (BUILTIN_EXPECT): New.
37 2014-03-18 Tobias Burnus <burnus@net-b.de>
40 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
41 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
42 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
44 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
47 * predict.c (combine_predictions_for_bb): Fix up formatting.
48 (expr_expected_value_1, expr_expected_value): Add predictor argument,
49 fill what it points to if non-NULL.
50 (tree_predict_by_opcode): Adjust caller, use the predictor.
51 * predict.def (PRED_COMPARE_AND_SWAP): Add.
53 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
55 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
56 proper constant for the store mode.
58 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
60 * symtab.c (change_decl_assembler_name): Fix transparent alias
63 2014-03-16 Renlin Li <Renlin.Li@arm.com>
65 * config/aarch64/aarch64.c: Correct the comments about the
68 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
70 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
71 check for GF_OMP_FOR_KIND_FOR.
73 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
75 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
76 ymm and zmm register names.
78 2014-03-17 Jakub Jelinek <jakub@redhat.com>
81 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
82 note creation for the 2010-08-31 changes.
84 2014-03-17 Marek Polacek <polacek@redhat.com>
87 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
88 as -fno-tree-loop-vectorize.
89 (expand_omp_simd): Likewise.
91 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
93 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
94 (eligible_for_call_delay): New prototype.
95 * config/sparc/sparc.c (tls_call_delay): Rename into...
96 (eligible_for_call_delay): ...this. Return false if the instruction
97 cannot be put in the delay slot of a branch.
98 (eligible_for_restore_insn): Simplify.
99 (eligible_for_return_delay): Return false if the instruction cannot be
100 put in the delay slot of a branch and simplify.
101 (eligible_for_sibcall_delay): Return false if the instruction cannot be
102 put in the delay slot of a branch.
103 * config/sparc/sparc.md (fix_ut699): New attribute.
104 (tls_call_delay): Delete.
105 (in_call_delay): Reimplement.
106 (eligible_for_sibcall_delay): Rename into...
107 (in_sibcall_delay): ...this.
108 (eligible_for_return_delay): Rename into...
109 (in_return_delay): ...this.
110 (in_branch_delay): Reimplement.
111 (in_uncond_branch_delay): Delete.
112 (in_annul_branch_delay): Delete.
114 2014-03-14 Richard Henderson <rth@redhat.com>
117 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
118 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
119 (*floathi<X87MODEF>2_i387_with_temp): Remove.
120 (floathi splitters): Remove.
121 (float<SWI48x>xf2): New pattern.
122 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
123 code that tried to handle DImode for 32-bit, but which was excluded
124 by the pattern's condition. Drop allocation of stack temporary.
125 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
126 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
127 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
128 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
129 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
130 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
131 (*float<SWI48><MODEF>2_sse_interunit): Remove.
132 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
133 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
134 (*float<SWI48x><X87MODEF>2_i387): Remove.
135 (all float _with_temp splitters): Remove.
136 (*float<SWI48x><MODEF>2_i387): New pattern.
137 (*float<SWI48><MODEF>2_sse): New pattern.
138 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
139 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
141 2014-03-14 Jakub Jelinek <jakub@redhat.com>
142 Marek Polacek <polacek@redhat.com>
145 * common.opt (dump_base_name_prefixed): New Variable.
146 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
147 if x_dump_base_name_prefixed is already set, set it at the end.
149 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
151 PR rtl-optimization/60508
152 * lra-constraints.c (get_reload_reg): Add new parameter
154 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
155 Pass the new parameter values.
157 2014-03-14 Richard Biener <rguenther@suse.de>
159 * common.opt: Revert unintented changes from r205065.
162 2014-03-14 Richard Biener <rguenther@suse.de>
165 * cfghooks.c (split_block): Properly adjust all loops the
166 block was a latch of.
168 2014-03-14 Martin Jambor <mjambor@suse.cz>
171 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
174 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
177 * config/avr/avr.c (avr_set_current_function): Pass function name
178 through default_strip_name_encoding before sanity checking instead
179 of skipping the first char of the assembler name.
181 2014-03-13 Richard Henderson <rth@redhat.com>
184 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
185 (ix86_force_to_memory, ix86_free_from_memory): Remove.
186 * config/i386/i386-protos.h: Likewise.
187 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
188 in the expander instead of a splitter.
189 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
190 any possibility of requiring a memory.
191 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
192 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
193 (fp branch splitters): Update for ix86_split_fp_branch.
194 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
195 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
196 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
197 (*fop_<MODEF>_2_i387): Remove f/r alternative.
198 (*fop_<MODEF>_3_i387): Likewise.
199 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
200 (splitters for the fop_* register patterns): Remove.
201 (fscalexf4_i387): Rename from *fscalexf4_i387.
202 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
204 2014-03-13 Jakub Jelinek <jakub@redhat.com>
206 PR tree-optimization/59779
207 * tree-dfa.c (get_ref_base_and_extent): Use double_int
208 type for bitsize and maxsize instead of HOST_WIDE_INT.
210 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
212 PR rtl-optimization/57320
213 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
214 the CFG after thread_prologue_and_epilogue_insns.
216 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
218 PR rtl-optimization/57189
219 * lra-constraints.c (process_alt_operands): Disfavor spilling
222 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
224 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
227 2014-03-13 Jakub Jelinek <jakub@redhat.com>
229 PR tree-optimization/59025
231 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
232 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
234 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
237 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
238 calls of avr_out_plus_1.
240 2014-03-13 Bin Cheng <bin.cheng@arm.com>
242 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
243 BB's single pred and update the father loop's latch info later.
245 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
247 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
251 (VEC_base): Likewise.
252 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
253 registers, we need to swap double words in little endian mode.
255 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
256 to be a container mode for 128-bit integer operations added in ISA
257 2.07. Unlike TImode and PTImode, the preferred register set is
258 the Altivec/VMX registers for the 128-bit operations.
260 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
262 (rs6000_split_128bit_ok_p): Likewise.
264 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
265 macros for creating ISA 2.07 normal and overloaded builtin
266 functions with 3 arguments.
267 (BU_P8V_OVERLOAD_3): Likewise.
268 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
269 for use as overloaded functions.
270 (VPERM_1TI_UNS): Likewise.
271 (VSEL_1TI): Likewise.
272 (VSEL_1TI_UNS): Likewise.
273 (ST_INTERNAL_1ti): Likewise.
274 (LD_INTERNAL_1ti): Likewise.
275 (XXSEL_1TI): Likewise.
276 (XXSEL_1TI_UNS): Likewise.
277 (VPERM_1TI): Likewise.
278 (VPERM_1TI_UNS): Likewise.
279 (XXPERMDI_1TI): Likewise.
281 (LXVD2X_V1TI): Likewise.
282 (STXVD2X_V1TI): Likewise.
283 (VEC_INIT_V1TI): Likewise.
284 (VEC_SET_V1TI): Likewise.
285 (VEC_EXT_V1TI): Likewise.
286 (EQV_V1TI): Likewise.
287 (NAND_V1TI): Likewise.
288 (ORC_V1TI): Likewise.
289 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
290 added in ISA 2.07. Add both normal 'altivec' builtins, and the
294 (VADDEUQM): Likewise.
295 (VADDECUQ): Likewise.
296 (VSUBEUQM): Likewise.
297 (VSUBECUQ): Likewise.
299 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
300 __int128_t and __uint128_t types.
301 (__uint128_type): Likewise.
302 (altivec_categorize_keyword): Add support for vector __int128_t,
303 vector __uint128_t, vector __int128, and vector unsigned __int128
304 as a container type for TImode operations that need to be done in
305 VSX/Altivec registers.
306 (rs6000_macro_to_expand): Likewise.
307 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
308 to support 128-bit integer instructions vaddcuq, vadduqm,
309 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
310 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
312 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
313 for V1TImode, and set up preferences to use VSX/Altivec registers.
314 Setup VSX reload handlers.
315 (rs6000_debug_reg_global): Likewise.
316 (rs6000_init_hard_regno_mode_ok): Likewise.
317 (rs6000_preferred_simd_mode): Likewise.
318 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
319 (easy_altivec_constant): Likewise.
320 (output_vec_const_move): Likewise.
321 (rs6000_expand_vector_set): Convert V1TImode set and extract to
323 (rs6000_expand_vector_extract): Likewise.
324 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
326 (rs6000_const_vec): Add support for V1TImode.
327 (rs6000_emit_le_vsx_load): Swap double words when loading or
328 storing TImode/V1TImode.
329 (rs6000_emit_le_vsx_store): Likewise.
330 (rs6000_emit_le_vsx_move): Likewise.
331 (rs6000_emit_move): Add support for V1TImode.
332 (altivec_expand_ld_builtin): Likewise.
333 (altivec_expand_st_builtin): Likewise.
334 (altivec_expand_vec_init_builtin): Likewise.
335 (altivec_expand_builtin): Likewise.
336 (rs6000_init_builtins): Add support for V1TImode type. Add
337 support for ISA 2.07 128-bit integer builtins. Define type names
338 for the VSX/Altivec vector types.
339 (altivec_init_builtins): Add support for overloaded vector
340 functions with V1TImode type.
341 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
342 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
344 (rs6000_split_128bit_ok_p): Likewise.
345 (rs6000_handle_altivec_attribute): Create V1TImode from vector
346 __int128_t and vector __uint128_t.
348 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
356 (VS_scalar): Likewise.
357 (VS_double): Likewise.
358 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
360 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
361 we support the ISA 2.07 128-bit integer arithmetic instructions.
362 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
363 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
364 and TImode types for use with the builtin functions.
365 (V1TI_type_node): Likewise.
366 (unsigned_V1TI_type_node): Likewise.
367 (intTI_type_internal_node): Likewise.
368 (uintTI_type_internal_node): Likewise.
370 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
371 128-bit builtin functions.
372 (UNSPEC_VADDEUQM): Likewise.
373 (UNSPEC_VADDECUQ): Likewise.
374 (UNSPEC_VSUBCUQ): Likewise.
375 (UNSPEC_VSUBEUQM): Likewise.
376 (UNSPEC_VSUBECUQ): Likewise.
377 (VM): Add V1TImode to vector mode iterators.
380 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
381 (altivec_vaddcuq): Likewise.
382 (altivec_vsubuqm): Likewise.
383 (altivec_vsubcuq): Likewise.
384 (altivec_vaddeuqm): Likewise.
385 (altivec_vaddecuq): Likewise.
386 (altivec_vsubeuqm): Likewise.
387 (altivec_vsubecuq): Likewise.
389 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
391 (BOOL_128): Likewise.
392 (BOOL_REGS_OUTPUT): Likewise.
393 (BOOL_REGS_OP1): Likewise.
394 (BOOL_REGS_OP2): Likewise.
395 (BOOL_REGS_UNARY): Likewise.
396 (BOOL_REGS_AND_CR0): Likewise.
398 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
399 128-bit integer builtin support.
400 (vec_vadduqm): Likewise.
401 (vec_vaddecuq): Likewise.
402 (vec_vaddeuqm): Likewise.
403 (vec_vsubecuq): Likewise.
404 (vec_vsubeuqm): Likewise.
405 (vec_vsubcuq): Likewise.
406 (vec_vsubuqm): Likewise.
408 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
409 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
410 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
411 128-bit integer add/subtract to ISA 2.07.
413 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
415 * config/arc/arc.c (arc_predicate_delay_insns):
416 Fix third argument passed to conditionalize_nonjump.
418 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
420 * config/aarch64/aarch64-builtins.c
421 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
422 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
423 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
424 instead of __builtin_lfloor.
425 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
427 2014-03-12 Jakub Jelinek <jakub@redhat.com>
429 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
430 (tree_ssa_ifcombine_bb_1): New function.
431 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
432 is an empty forwarder block to then_bb or vice versa and then_bb
433 and else_bb are effectively swapped.
435 2014-03-12 Christian Bruel <christian.bruel@st.com>
438 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
439 REG_CFA_DEF_CFA note.
440 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
441 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
443 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
445 PR tree-optimization/60454
446 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
448 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
450 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
451 Do not define target_cpu_default2 to generic.
452 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
453 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
454 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
456 2014-03-12 Jakub Jelinek <jakub@redhat.com>
457 Marc Glisse <marc.glisse@inria.fr>
459 PR tree-optimization/60502
460 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
461 instead of build_low_bits_mask.
463 2014-03-12 Jakub Jelinek <jakub@redhat.com>
466 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
467 if there are multiple uses, but op doesn't live on E edge.
468 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
469 clobber stmts before __builtin_unreachable.
471 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
473 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
474 hard_frame_pointer_rtx.
475 * cse.c (cse_insn): Remove volatile check.
476 * cselib.c (cselib_process_insn): Likewise.
477 * dse.c (scan_insn): Likewise.
479 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
481 * config/arc/arc.c (conditionalize_nonjump): New function,
483 (arc_ifcvt): ... this.
484 (arc_predicate_delay_insns): Use it.
486 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
488 * config/arc/predicates.md (extend_operand): During/after reload,
489 allow const_int_operand.
490 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
491 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
492 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
494 (umulsi3_highpart_i): Likewise.
496 2014-03-11 Richard Biener <rguenther@suse.de>
498 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
499 Add asserts to guard possible wrong-code bugs.
501 2014-03-11 Richard Biener <rguenther@suse.de>
503 PR tree-optimization/60429
504 PR tree-optimization/60485
505 * tree-ssa-structalias.c (set_union_with_increment): Properly
506 take into account all fields that overlap the shifted vars.
507 (do_sd_constraint): Likewise.
508 (do_ds_constraint): Likewise.
509 (get_constraint_for_ptr_offset): Likewise.
511 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
513 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
514 (nios2_compute_frame_layout):
515 Add calculation of cfun->machine->fp_save_offset.
516 (nios2_expand_prologue): Correct setting of frame pointer register
518 (nios2_expand_epilogue): Update recovery of stack pointer from
519 frame pointer accordingly.
520 (nios2_initial_elimination_offset): Update calculation of offset
521 for eliminating to HARD_FRAME_POINTER_REGNUM.
523 2014-03-10 Jakub Jelinek <jakub@redhat.com>
526 * ipa.c (symtab_remove_unreachable_nodes): Don't call
527 cgraph_get_create_node on VAR_DECLs.
529 2014-03-10 Richard Biener <rguenther@suse.de>
532 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
534 2014-03-08 Douglas B Rupp <rupp@gnat.com>
536 * config/vms/vms.opt (vms_float_format): New variable.
538 2014-03-08 Tobias Burnus <burnus@net-b.de>
540 * doc/invoke.texi (-fcilkplus): Update implementation status.
542 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
543 Richard Biener <rguenther@suse.de>
545 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
546 consistently accross all TUs.
547 (run_gcc): Enable -fshort-double automatically at link at link-time
548 and disallow override.
550 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
553 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
554 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
555 if they can't be used.
557 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
559 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
560 for Solaris 11/x86 ld.
561 * configure: Regenerate.
563 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
565 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
566 (LIB_TLS_SPEC): Save as ld_tls_libs.
567 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
568 (HAVE_AS_IX86_TLSLDM): New test.
569 * configure, config.in: Regenerate.
570 * config/i386/i386.c (legitimize_tls_address): Fall back to
571 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
572 cannot support TLS_MODEL_LOCAL_DYNAMIC.
573 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
574 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
576 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
578 * common.opt (fira-loop-pressure): Mark as optimization.
580 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
582 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
583 an OpenMP mappable type.
585 2014-03-06 Matthias Klose <doko@ubuntu.com>
587 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
588 MULTILIB_OSDIRNAMES is not defined.
590 2014-03-06 Jakub Jelinek <jakub@redhat.com>
591 Meador Inge <meadori@codesourcery.com>
594 * config/arm/arm.c (arm_tls_symbol_p): Remove.
595 (arm_legitimize_address): Call legitimize_tls_address for any
596 arm_tls_referenced_p expression, handle constant addend. Call it
597 before testing for !TARGET_ARM.
598 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
600 2014-03-06 Richard Biener <rguenther@suse.de>
606 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
608 * tree-streamer.c (record_common_node): Assert we don't record
609 nodes with type double.
610 (preload_common_node): Skip type double, complex double and double
611 pointer since it is now frontend dependent due to fshort-double option.
613 2014-03-06 Richard Biener <rguenther@suse.de>
615 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
616 or -fno-lto is specified and the linker has full plugin support.
617 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
618 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
619 * lto-wrapper.c (merge_and_complain): Merge compile-time
621 (run_gcc): And pass it through to the link options.
623 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
627 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
629 * cselib.c (remove_useless_values): Skip to avoid quadratic
630 behavior if the condition moved from...
631 (cselib_process_insn): ... here holds.
633 2014-03-05 Jakub Jelinek <jakub@redhat.com>
636 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
637 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
640 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
641 (TM_H): Add x86-tune.def.
643 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
645 * config/aarch64/aarch64.c (generic_tunings):
646 Use cortexa57_extra_costs.
648 2014-03-05 Jakub Jelinek <jakub@redhat.com>
651 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
652 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
653 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
656 2014-03-04 Heiher <r@hev.cc>
658 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
659 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
661 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
663 * config/i386/predicates.md (const2356_operand): Change to ...
664 (const2367_operand): ... this.
665 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
667 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
668 (*avx512pf_scatterpf<mode>sf): Ditto.
669 (avx512pf_scatterpf<mode>df): Ditto.
670 (*avx512pf_scatterpf<mode>df_mask): Ditto.
671 (*avx512pf_scatterpf<mode>df): Ditto.
672 * config/i386/i386.c (ix86_expand_builtin): Update
673 incorrect hint operand error message.
675 2014-03-04 Richard Biener <rguenther@suse.de>
677 * lto-section-in.c (lto_get_section_data): Fix const cast.
679 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
681 * tree-streamer.c (record_common_node): Assert we don't record
682 nodes with type double.
683 (preload_common_node): Skip type double, complex double and double
684 pointer since it is now frontend dependent due to fshort-double option.
686 2014-03-04 Richard Biener <rguenther@suse.de>
689 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
690 (lto_input_toplevel_asms): Likewise.
691 * lto-section-in.c (lto_get_section_data): Instead do it here
694 2014-03-04 Richard Biener <rguenther@suse.de>
696 PR tree-optimization/60382
697 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
698 dead PHIs a reduction.
700 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
702 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
704 (_mm_prefetch): Move out of GCC target("sse") pragma.
705 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
706 GCC target("prfchw") pragma.
707 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
709 * config/i386/i386.c (ix86_option_override_internal): Enable
710 -mprfchw with -mprefetchwt1.
712 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
714 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
717 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
719 * opts.h (CL_PCH_IGNORE): Define.
720 * targhooks.c (option_affects_pch_p):
721 Return false for options that have CL_PCH_IGNORE set.
722 * opt-functions.awk: Process PchIgnore.
723 * doc/options.texi: Document PchIgnore.
725 * config/arc/arc.opt (misize): Add PchIgnore property.
727 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
729 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
730 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
731 constraint on constants to permit them being loaded into
732 GENERAL_REGS or BASE_REGS.
734 2014-03-03 Nick Clifton <nickc@redhat.com>
736 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
737 anti-cacnonical alternatives.
738 (negandhi3_real): New pattern.
739 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
741 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
743 * config/avr/avr-mcus.def: Remove atxmega16x1.
744 * config/avr/avr-tables.opt: Regenerate.
745 * config/avr/t-multilib: Regenerate.
746 * doc/avr-mmcu.texi: Regenerate.
748 2014-03-03 Tobias Grosser <tobias@grosser.es>
749 Mircea Namolaru <mircea.namolaru@inria.fr>
751 PR tree-optimization/58028
752 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
755 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
757 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
758 not handled by recognizers.
760 2014-03-03 Jakub Jelinek <jakub@redhat.com>
763 * function.c (expand_function_end): Don't emit
764 clobber_return_register sequence if clobber_after is a BARRIER.
765 * cfgexpand.c (construct_exit_block): Append instructions before
766 return_label to prev_bb.
768 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
770 * config/rs6000/constraints.md: Document reserved use of "wc".
772 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
775 * ipa.c (function_and_variable_visibility): When dissolving comdat
776 group, also set all symbols to local.
778 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
783 2013-12-14 Jan Hubicka <jh@suse.cz>
785 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
787 2014-03-02 Jon Beniston <jon@beniston.com>
793 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
794 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
795 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
796 (simple_return, *simple_return): New patterns
797 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
798 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
800 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
802 * dwarf2out.c (gen_subprogram_die): Tidy.
804 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
807 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
808 (*mov_t_msb_neg_negc): ... this new insn.
810 2014-02-28 Jason Merrill <jason@redhat.com>
813 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
816 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
819 * dwarf2out.c (decltype_auto_die): New static.
820 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
821 (gen_type_die_with_usage): Handle 'decltype(auto)'.
822 (is_cxx_auto): Likewise.
824 2014-02-28 Ian Bolton <ian.bolton@arm.com>
826 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
827 we are not using general regs only.
829 2014-02-28 Richard Biener <rguenther@suse.de>
832 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
833 previous fix and only allow to remove trivial pre-headers
834 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
835 (remove_forwarder_block): Properly update the latch of a loop.
837 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
840 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
841 (cselib_preserved_hash_table): New.
842 (preserve_constants_and_equivs): Move preserved vals to it.
843 (cselib_find_slot): Look it up first.
844 (cselib_init): Initialize it.
845 (cselib_finish): Release it.
846 (dump_cselib_table): Dump it.
848 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
851 * cselib.c (remove_useless_values): Skip to avoid quadratic
852 behavior if the condition moved from...
853 (cselib_process_insn): ... here holds.
855 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
858 * var-tracking.c (vt_initialize): Apply the same condition to
859 preserve the CFA base value.
861 2014-02-28 Joey Ye <joey.ye@arm.com>
864 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
865 if reload in progress or completed.
867 2014-02-28 Tobias Burnus <burnus@net-b.de>
870 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
873 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
875 * doc/tm.texi.in (Condition Code Status): Update documention for
876 relative locations of cc0-setter and cc0-user.
878 2014-02-27 Jeff Law <law@redhat.com>
880 PR rtl-optimization/52714
881 * combine.c (try_combine): When splitting an unrecognized PARALLEL
882 into two independent simple sets, if I3 is a jump, ensure the
883 pattern we place into I3 is a (set (pc) ...).
885 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
886 Jeff Law <law@redhat.com>
888 PR rtl-optimization/49847
889 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
890 are in different blocks.
891 * doc/tm.texi (Condition Code Status): Update documention for
892 relative locations of cc0-setter and cc0-user.
894 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
897 * lra.c (lra_emit_add): Check SUBREG too.
899 2014-02-27 Andreas Schwab <schwab@suse.de>
901 * config/m68k/m68k.c (m68k_option_override): Disable
902 -flive-range-shrinkage for classic m68k.
903 (m68k_override_options_after_change): Likewise.
905 2014-02-27 Marek Polacek <polacek@redhat.com>
908 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
909 -Wmaybe-uninitialized.
911 2014-02-27 Alan Modra <amodra@gmail.com>
914 * reload1.c (emit_input_reload_insns): When reload_override_in,
915 set old to rl->in_reg when rl->in_reg is a subreg.
917 2014-02-26 Richard Biener <rguenther@suse.de>
920 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
922 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
924 * common/config/i386/predicates.md (const1256_operand): Remove.
925 (const2356_operand): New.
926 (const_1_to_2_operand): Remove.
927 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
928 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
929 (*avx512pf_gatherpf<mode>sf): Ditto.
930 (avx512pf_gatherpf<mode>df): Ditto.
931 (*avx512pf_gatherpf<mode>df_mask): Ditto.
932 (*avx512pf_gatherpf<mode>df): Ditto.
933 (avx512pf_scatterpf<mode>sf): Ditto.
934 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
935 (*avx512pf_scatterpf<mode>sf): Ditto.
936 (avx512pf_scatterpf<mode>df): Ditto.
937 (*avx512pf_scatterpf<mode>df_mask): Ditto.
938 (*avx512pf_scatterpf<mode>df): Ditto.
939 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
941 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
943 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
944 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
945 (_mm512_mask_testn_epi64_mask): Move to ...
946 * config/i386/avx512cdintrin.h: Here.
947 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
948 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
949 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
950 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
951 TARGET_AVX512F from TARGET_AVX512CD.
953 2014-02-26 Richard Biener <rguenther@suse.de>
956 * ipa.c (walk_polymorphic_call_targets): Properly guard
957 call to inline_update_overall_summary.
959 2014-02-26 Bin Cheng <bin.cheng@arm.com>
962 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
963 and latches only if requested. Fix latch if it is removed.
964 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
965 LOOPS_HAVE_PREHEADERS.
967 2014-02-25 Andrew Pinski <apinski@cavium.com>
969 * builtins.c (expand_builtin_thread_pointer): Create a new target
970 when the target is NULL.
972 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
974 PR rtl-optimization/60317
975 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
976 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
977 * lra-assigns.c: Include params.h.
978 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
979 other reload pseudos considerations.
981 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
983 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
984 to use canonical form for nor<mode>3.
986 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
989 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
992 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
994 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
995 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
996 (ix86_handle_option): Handle OPT_mprefetchwt1.
997 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
998 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1000 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1001 OPTION_MASK_ISA_PREFETCHWT1.
1002 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1003 (PTA_PREFETCHWT1): New.
1004 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1005 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1006 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1007 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1008 (*prefetch_avx512pf_<mode>_: Change into ...
1009 (*prefetch_prefetchwt1_<mode>: This.
1010 * config/i386/i386.opt (mprefetchwt1): New.
1011 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1012 (_mm_prefetch): Handle intent to write.
1013 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1015 2014-02-25 Richard Biener <rguenther@suse.de>
1018 * emit-rtl.c (mem_attrs_htab): Remove.
1019 (mem_attrs_htab_hash): Likewise.
1020 (mem_attrs_htab_eq): Likewise.
1021 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1022 (init_emit_once): Do not allocate mem_attrs_htab.
1024 2014-02-25 Richard Biener <rguenther@suse.de>
1027 * lto-opts.c (lto_write_options): Output non-explicit conservative
1028 -fwrapv, -fno-trapv and -fno-strict-overflow.
1029 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1030 (run_gcc): And pass them through.
1032 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1034 * sel-sched.c (calculate_new_fences): New parameter ptime.
1035 Calculate it as a maximum over all fence cycles.
1036 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1037 Print the final schedule timing when sched_verbose.
1039 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1041 PR rtl-optimization/60292
1042 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1043 bit fot the fence instruction.
1045 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1047 * calls.h: Fix typo in comment.
1049 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1051 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1052 adjusting offsetable addresses.
1054 2014-02-24 Guozhi Wei <carrot@google.com>
1056 * sparseset.h (sparseset_pop): Fix the wrong index.
1058 2014-02-24 Walter Lee <walt@tilera.com>
1060 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1061 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1063 * common/config/tilegx/tilegx-common.c
1064 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1065 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1067 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1068 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1069 (tilegx_gimplify_va_arg_expr): Handle big endian.
1070 (tilegx_expand_unaligned_load): Ditto.
1071 (tilegx_expand_unaligned_store): Ditto.
1072 (TARGET_RETURN_IN_MSB): New.
1073 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1074 (TARGET_ENDIAN_DEFAULT): New.
1075 (TARGET_BIG_ENDIAN): Handle big endian.
1076 (BYTES_BIG_ENDIAN): Ditto.
1077 (WORDS_BIG_ENDIAN): Ditto.
1078 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1081 * config/tilegx/tilegx.md (extv): Handle big endian.
1083 (insn_st<n>): Ditto.
1084 (insn_st<n>_add<bitsuffix>): Ditto.
1085 (insn_stnt<n>): Ditto.
1086 (insn_stnt<n>_add<bitsuffix>):Ditto.
1087 (vec_interleave_highv8qi): Handle big endian.
1088 (vec_interleave_highv8qi_be): New.
1089 (vec_interleave_highv8qi_le): New.
1090 (insn_v1int_h): Handle big endian.
1091 (vec_interleave_lowv8qi): Handle big endian.
1092 (vec_interleave_lowv8qi_be): New.
1093 (vec_interleave_lowv8qi_le): New.
1094 (insn_v1int_l): Handle big endian.
1095 (vec_interleave_highv4hi): Handle big endian.
1096 (vec_interleave_highv4hi_be): New.
1097 (vec_interleave_highv4hi_le): New.
1098 (insn_v2int_h): Handle big endian.
1099 (vec_interleave_lowv4hi): Handle big endian.
1100 (vec_interleave_lowv4hi_be): New.
1101 (vec_interleave_lowv4hi_le): New.
1102 (insn_v2int_l): Handle big endian.
1103 (vec_interleave_highv2si): Handle big endian.
1104 (vec_interleave_highv2si_be): New.
1105 (vec_interleave_highv2si_le): New.
1106 (insn_v4int_h): Handle big endian.
1107 (vec_interleave_lowv2si): Handle big endian.
1108 (vec_interleave_lowv2si_be): New.
1109 (vec_interleave_lowv2si_le): New.
1110 (insn_v4int_l): Handle big endian.
1111 * config/tilegx/tilegx.opt (mbig-endian): New option.
1112 (mlittle-endian): New option.
1113 * doc/install.texi: Document tilegxbe-linux.
1114 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1116 2014-02-24 Martin Jambor <mjambor@suse.cz>
1119 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1120 there are no parameter descriptors.
1122 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1124 PR rtl-optimization/60268
1125 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1126 initialization to ...
1127 (sched_rgn_init): ... here.
1128 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1130 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1132 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1135 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1137 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1140 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1142 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1143 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1145 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1147 * config/microblaze/predicates.md: Add cmp_op predicate.
1148 * config/microblaze/microblaze.md: Add branch_compare instruction
1149 which uses cmp_op predicate and emits cmp insn before branch.
1150 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1151 to microblaze_expand_conditional_branch and consolidate logic.
1152 (microblaze_expand_conditional_branch): emit branch_compare
1153 insn instead of handling cmp op separate from branch insn.
1155 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1157 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1160 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1162 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1163 define_insn with define_expand and new define_insn
1164 *altivec_lve<VI_char>x_internal.
1165 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1166 and new define_insn *altivec_stve<VI_char>x_internal.
1167 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1169 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1171 (altivec_expand_stvex_be): New function.
1173 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1175 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1176 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1177 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1178 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1180 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1183 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1184 instead of emit_move_insn.
1186 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1188 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1190 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1191 gen_altivec_vsumsws.
1193 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1195 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1196 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1197 (altivec_lvxl_<mode>): New define_expand incorporating
1198 -maltivec=be semantics where needed.
1199 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1200 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1201 semantics where needed.
1202 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1203 (altivec_stvx_<mode>): New define_expand incorporating
1204 -maltivec=be semantics where needed.
1205 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1206 VM2 iterator instead of V4SI.
1207 (altivec_stvxl_<mode>): New define_expand incorporating
1208 -maltivec=be semantics where needed.
1209 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1210 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1211 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1212 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1213 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1214 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1215 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1216 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1217 ALTIVEC_BUILTIN_STVXL.
1218 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1219 (altivec_expand_stvx_be): Likewise.
1220 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1221 (altivec_expand_lvx_be): Likewise.
1222 (altivec_expand_stvx_be): Likewise.
1223 (altivec_expand_builtin): Add cases for
1224 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1225 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1226 (altivec_init_builtins): Add definitions for
1227 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1228 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1230 2014-02-21 Catherine Moore <clm@codesourcery.com>
1232 * doc/invoke.texi (mvirt, mno-virt): Document.
1233 * config/mips/mips.opt (mvirt): New option.
1234 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1236 2014-02-21 Richard Biener <rguenther@suse.de>
1238 PR tree-optimization/60276
1239 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1240 (STMT_VINFO_MIN_NEG_DIST): New macro.
1241 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1242 STMT_VINFO_MIN_NEG_DIST.
1243 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1244 made for negative dependence distances still hold.
1246 2014-02-21 Richard Biener <rguenther@suse.de>
1249 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1250 DECL_INITIAL for globals not in the current function context.
1252 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1254 PR tree-optimization/56490
1255 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1256 * tree-ssa-uninit.c: Include params.h.
1257 (compute_control_dep_chain): Add num_calls argument, return false
1258 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1259 num_calls to recursive call.
1260 (find_predicates): Change dep_chain into normal array,
1261 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1262 variable and adjust compute_control_dep_chain caller.
1263 (find_def_preds): Likewise.
1265 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1267 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1268 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1270 2014-02-21 Nick Clifton <nickc@redhat.com>
1272 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1273 (pushhi1): Likewise.
1274 (popqi1): Add mode to pre_dec.
1277 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1279 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1280 mode for mask of V8SFmode permutation.
1282 2014-02-20 Richard Henderson <rth@redhat.com>
1285 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1286 a new pseudo for OLDVAL.
1288 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1291 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1292 gen_reg_rtx if d->testing_p.
1293 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1294 if d->testing_p and we will certainly return true.
1295 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1298 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1300 * emit-rtl.c (gen_reg_rtx): Assert that
1301 crtl->emit.regno_pointer_align_length is non-zero.
1303 2014-02-20 Richard Henderson <rth@redhat.com>
1306 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1307 on failure the store back into EXPECT.
1309 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1310 Sandra Loosemore <sandra@codesourcery.com>
1312 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1313 * config/nios2/nios2.c (nios2_function_profiler): Add
1314 -fPIC (flag_pic == 2) support.
1315 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1316 (nios2_large_offset_p): New function.
1317 (nios2_unspec_reloc_p): Move up position, update to use
1318 nios2_large_offset_p.
1319 (nios2_unspec_address): Remove function.
1320 (nios2_unspec_offset): New function.
1321 (nios2_large_got_address): New function.
1322 (nios2_got_address): Add large offset support.
1323 (nios2_legitimize_tls_address): Update usage of removed and new
1325 (nios2_symbol_binds_local_p): New function.
1326 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1327 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1328 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1329 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1330 processing for (const (unspec ...)).
1331 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1333 2014-02-20 Richard Biener <rguenther@suse.de>
1335 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1336 doing the substitution.
1337 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1339 2014-02-20 Martin Jambor <mjambor@suse.cz>
1342 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1343 info when checking whether lattices are bottom.
1345 2014-02-20 Richard Biener <rguenther@suse.de>
1348 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1351 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1354 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1355 parameter specifying the scaling.
1356 (inline_call): Update.
1357 (want_inline_recursively): Guard division by zero.
1358 (recursive_inlining): Update.
1359 * ipa-inline.h (clone_inlined_nodes): Update.
1361 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1364 * config/i386/i386.c (classify_argument): Pass structures of size
1365 64 bytes or less in register.
1367 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1368 Kirill Yukhin <kirill.yukhin@intel.com>
1370 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1371 (_mm_rcp28_round_ss): Ditto.
1372 (_mm_rsqrt28_round_sd): Ditto.
1373 (_mm_rsqrt28_round_ss): Ditto.
1374 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1375 (_mm_rcp14_round_ss): Ditto.
1376 (_mm_rsqrt14_round_sd): Ditto.
1377 (_mm_rsqrt14_round_ss): Ditto.
1378 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1379 the first input operand, get rid of match_dup.
1380 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1382 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1384 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1385 operand as the first input operand, set type attribute.
1386 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1388 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1389 operand as the first input operand, set type attribute.
1391 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1393 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1396 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1399 * config/i386/i386.c (construct_container): Remove TFmode check
1400 for X86_64_INTEGER_CLASS.
1402 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1405 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1406 only when -Wpsabi is enabled.
1408 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1411 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1412 passing arrays in registers are the same as for structs, so remove the
1413 special case for them.
1415 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1417 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1418 destination type, extract only the valid bits if the source type is not
1419 integral and has a different mode.
1421 2014-02-19 Richard Biener <rguenther@suse.de>
1424 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1427 2014-02-19 Richard Biener <rguenther@suse.de>
1430 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1431 (ipa_modify_call_arguments): Emit an argument load explicitely and
1432 preserve virtual SSA form there and for the replacement call.
1433 Do not update SSA form nor free dominance info.
1435 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1437 * ipa.c (function_and_variable_visibility): Also clear WEAK
1438 flag when disolving COMDAT_GROUP.
1440 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1442 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1443 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1445 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1446 do more sanity checks.
1447 (detect_type_change): Return true when giving up early.
1448 (compute_complex_assign_jump_func): Fix type parameter of
1449 ipa_set_ancestor_jf.
1450 (compute_complex_ancestor_jump_func): Likewise.
1451 (update_jump_functions_after_inlining): Fix updating of
1453 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1455 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1457 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1458 inline clones when edge disappears.
1460 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1463 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1464 Split 64-bit moves into 2 patterns. Do not allow the use of
1465 direct move for TDmode in little endian, since the decimal value
1466 has little endian bytes within a word, but the 64-bit pieces are
1467 ordered in a big endian fashion, and normal subreg's of TDmode are
1469 (mov<mode>_64bit_dm): Likewise.
1470 (movtd_64bit_nodm): Likewise.
1472 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1474 PR tree-optimization/60174
1475 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1476 statement of an SSA_NAME that occurs in an abnormal PHI node.
1478 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1481 * final.c (SEEN_BB): Remove.
1482 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1483 (final_scan_insn): Don't force_source_line on second
1484 NOTE_INSN_BASIC_BLOCK.
1486 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1489 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1490 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1491 (type_natural_mode): Warn ABI change when %zmm register is not
1492 available for AVX512F vector value passing.
1494 2014-02-18 Kai Tietz <ktietz@redhat.com>
1497 * config/i386/i386.c (ix86_expand_prologue): Use value in
1498 rax register as displacement when restoring %r10 or %rax.
1499 Fix wrong offset when restoring both registers.
1501 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1503 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1504 assertion with conditional return.
1506 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1507 Uros Bizjak <ubizjak@gmail.com>
1510 * config/i386/driver-i386.c (host_detect_local_cpu): If
1511 YMM state is not saved by the OS, also clear has_f16c. Move
1512 CPUID 0x80000001 handling before YMM state saving checking.
1514 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1516 PR rtl-optimization/58960
1517 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1518 factored out from ...
1519 (sched_init): ... here.
1520 (free_global_sched_pressure_data): New, factored out from ...
1521 (sched_finish): ... here.
1522 * sched-int.h (free_global_sched_pressure_data): Declare.
1523 * sched-rgn.c (nr_regions_initial): New static global.
1524 (haifa_find_rgns): Initialize it.
1525 (schedule_region): Disable sched-pressure for the newly
1528 2014-02-17 Richard Biener <rguenther@suse.de>
1530 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1531 release SSA defs of pattern stmts.
1533 2014-02-17 Richard Biener <rguenther@suse.de>
1535 * tree-inline.c (expand_call_inline): Release the virtual
1536 operand defined by the call we are about to inline.
1538 2014-02-17 Richard Biener <rguenther@suse.de>
1540 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1542 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1543 Ilya Tocar <ilya.tocar@intel.com>
1545 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1546 arguments order in builtin.
1547 (_mm512_permutexvar_epi64): Ditto.
1548 (_mm512_mask_permutexvar_epi64): Ditto
1549 (_mm512_maskz_permutexvar_epi32): Ditto
1550 (_mm512_permutexvar_epi32): Ditto
1551 (_mm512_mask_permutexvar_epi32): Ditto
1553 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1555 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1556 (p8_vmrgow): Likewise.
1558 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1560 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1563 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1566 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1567 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1568 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1569 using direct move instructions on ISA 2.07. Also adjust
1570 instruction length for 64-bit.
1571 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1572 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1574 2014-02-15 Alan Modra <amodra@gmail.com>
1578 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1579 find_replacement on parts of insn rtl that might be reloaded.
1581 2014-02-15 Richard Biener <rguenther@suse.de>
1583 PR tree-optimization/60183
1584 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1585 (tree_ssa_phiprop): Calculate and free post-dominators.
1587 2014-02-14 Jeff Law <law@redhat.com>
1589 PR rtl-optimization/60131
1590 * ree.c (get_extended_src_reg): New function.
1591 (combine_reaching_defs): Use it rather than assuming location of REG.
1592 (find_and_remove_re): Verify first operand of extension is
1593 a REG before adding the insns to the copy list.
1595 2014-02-14 Roland McGrath <mcgrathr@google.com>
1597 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1598 * configure: Regenerated.
1599 * config.in: Regenerated.
1600 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1601 instead of ASM_SHORT.
1603 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1604 Richard Earnshaw <rearnsha@arm.com>
1606 PR rtl-optimization/59535
1607 * lra-constraints.c (process_alt_operands): Encourage alternative
1608 when unassigned pseudo class is superset of the alternative class.
1609 (inherit_reload_reg): Don't inherit when optimizing for code size.
1610 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1611 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1612 modes not less than 4 for Thumb1.
1614 2014-02-14 Kyle McMartin <kyle@redhat.com>
1617 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1619 2014-02-14 Richard Biener <rguenther@suse.de>
1621 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1622 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1623 Do not strip INDIRECT_REFs.
1625 2014-02-14 Richard Biener <rguenther@suse.de>
1628 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1629 DECL_FUNCTION_SPECIFIC_TARGET.
1630 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1631 * tree-streamer-out.c (pack_ts_target_option): Remove.
1632 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1633 (write_ts_function_decl_tree_pointers): Do not stream
1634 DECL_FUNCTION_SPECIFIC_TARGET.
1635 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1636 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1637 (lto_input_ts_function_decl_tree_pointers): Do not stream
1638 DECL_FUNCTION_SPECIFIC_TARGET.
1640 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1642 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1643 (get_initial_def_for_induction, vectorizable_induction): Ignore
1644 debug stmts when looking for exit_phi.
1645 (vectorizable_live_operation): Fix up condition.
1647 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1649 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1650 nreverse() because it changes the content of original tree list.
1652 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1654 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1655 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1657 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1659 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1660 GNU coding standards.
1662 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1665 * dwarf2out.c (gen_subprogram_die): Don't call
1666 add_calling_convention_attribute if subr_die is old_die.
1668 2014-02-13 Sharad Singhai <singhai@google.com>
1670 * doc/optinfo.texi: Fix order of nodes.
1672 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1674 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1675 operands[2], not operands[3].
1677 2014-02-13 Richard Biener <rguenther@suse.de>
1680 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1681 mention the possibility of an in-tree build.
1682 (CLooG): Update recommended version to 0.18.1, mention the
1683 possibility of an in-tree build and clarify that the ISL
1684 bundled with CLooG does not work.
1686 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1689 * expr.c (compress_float_constant): If x is a hard register,
1690 extend into a pseudo and then move to x.
1692 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1694 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1695 caused by bad second argument to warning_at() with -mhotpatch and
1696 nested functions (e.g. with gfortran).
1698 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1700 * opts.c (option_name): Remove "enabled by default" rider.
1702 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1704 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1706 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1707 Uros Bizjak <ubizjak@gmail.com>
1710 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1711 * configure: Regenerated.
1713 2014-02-12 Richard Biener <rguenther@suse.de>
1715 * vec.c (vec_prefix::calculate_allocation): Move as
1716 inline variant to vec.h.
1717 (vec_prefix::calculate_allocation_1): New out-of-line version.
1718 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1719 (vec_prefix::m_has_auto_buf): Rename to ...
1720 (vec_prefix::m_using_auto_storage): ... this.
1721 (vec_prefix::calculate_allocation): Inline the easy cases
1722 and dispatch to calculate_allocation_1 which doesn't need the
1724 (va_heap::reserve): Use gcc_checking_assert.
1725 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1726 m_using_auto_storage.
1727 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1729 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1730 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1731 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1733 2014-02-12 Richard Biener <rguenther@suse.de>
1735 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1736 when we found a dependence.
1738 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1740 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1742 (maybe_fold_stmt): ... into this new function.
1743 * omp-low.c (lower_omp): Update comment.
1745 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1748 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1751 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1753 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1754 identifiers in comments.
1755 (cortexa53_extra_costs): Likewise.
1756 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1757 (cortexa7_extra_costs): Likewise.
1758 (cortexa12_extra_costs): Likewise.
1759 (cortexa15_extra_costs): Likewise.
1760 (v7m_extra_costs): Likewise.
1762 2014-02-12 Richard Biener <rguenther@suse.de>
1765 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1766 of posix_memalign being successful.
1767 (lower_stmt): Restrict lowering of posix_memalign to when
1768 -ftree-bit-ccp is enabled.
1770 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1772 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1774 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1776 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1778 PR rtl-optimization/60116
1779 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1780 other_insn once the combination has been validated.
1782 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1785 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1787 * ipa-devirt.c: Include demangle.h
1788 (odr_violation_reported): New static variable.
1789 (add_type_duplicate): Update odr_violations.
1790 (maybe_record_node): Add completep parameter; update it.
1791 (record_target_from_binfo): Add COMPLETEP parameter;
1792 update it as needed.
1793 (possible_polymorphic_call_targets_1): Likewise.
1794 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1795 rename FINAL to COMPLETE.
1796 (record_targets_from_bases): Sanity check we found the binfo;
1797 fix COMPLETEP updating.
1798 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1799 parameter, fix computing of COMPLETEP.
1800 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1801 at LTO time do demangling.
1802 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1803 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1805 (gimple_get_virt_method_for_binfo): Likewise.
1806 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1807 gimple_get_virt_method_for_vtable): Update prototypes.
1809 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1812 * genautomata.c (add_presence_absence): Fix typo with
1813 {final_}presence_list.
1815 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1818 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1819 for VSX/Altivec vectors that land in GPR registers.
1821 2014-02-11 Richard Henderson <rth@redhat.com>
1822 Jakub Jelinek <jakub@redhat.com>
1825 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1826 around drhs if type conversion to lacc->type is not useless.
1828 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1830 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1832 (cortex-a57.cortex-a53): Likewise.
1833 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1835 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1837 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1840 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1842 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1843 add_options_for_arm_vfp3.
1845 2014-02-11 Jeff Law <law@redhat.com>
1848 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1849 object with an undesirable mode.
1851 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1854 * config/i386/sol2-9.h: New file.
1855 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1856 *-*-solaris2.9*): Use it.
1858 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1860 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1861 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1863 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1865 * config/microblaze/microblaze.c: Extend mcpu version format
1867 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1869 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1871 2014-02-10 Richard Henderson <rth@redhat.com>
1874 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1875 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1876 ms-abi vs -mno-accumulate-outgoing-args.
1877 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1878 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1881 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1884 * cfgexpand.c (expand_asm_operands): Attach source location to
1885 ASM_INPUT rtx objects.
1886 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1888 2014-02-10 Nick Clifton <nickc@redhat.com>
1890 * config/mn10300/mn10300.c (popcount): New function.
1891 (mn10300_expand_prologue): Include saved registers in stack usage
1894 2014-02-10 Jeff Law <law@redhat.com>
1897 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1898 when changing the SET_DEST of a prior insn to avoid an input reload.
1900 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1902 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1903 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1904 -mcall-openbsd, or -mcall-linux.
1905 (CC1_ENDIAN_BIG_SPEC): Remove.
1906 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1907 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1908 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1909 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1910 and %cc1_endian_default.
1911 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1913 2014-02-10 Richard Biener <rguenther@suse.de>
1915 PR tree-optimization/60115
1916 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1917 MEM_REF handling. Properly verify that the accesses are not
1918 out of the objects bound.
1920 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1922 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1925 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1927 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1928 proper constants and fix formatting.
1929 (possible_polymorphic_call_targets): Fix formatting.
1931 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1932 Ilya Tocar <ilya.tocar@intel.com>
1934 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1935 (_mm512_loadu_epi32): Renamed into...
1936 (_mm512_loadu_si512): This.
1937 (_mm512_storeu_epi32): Renamed into...
1938 (_mm512_storeu_si512): This.
1939 (_mm512_maskz_ceil_ps): Removed.
1940 (_mm512_maskz_ceil_pd): Ditto.
1941 (_mm512_maskz_floor_ps): Ditto.
1942 (_mm512_maskz_floor_pd): Ditto.
1943 (_mm512_floor_round_ps): Ditto.
1944 (_mm512_floor_round_pd): Ditto.
1945 (_mm512_ceil_round_ps): Ditto.
1946 (_mm512_ceil_round_pd): Ditto.
1947 (_mm512_mask_floor_round_ps): Ditto.
1948 (_mm512_mask_floor_round_pd): Ditto.
1949 (_mm512_mask_ceil_round_ps): Ditto.
1950 (_mm512_mask_ceil_round_pd): Ditto.
1951 (_mm512_maskz_floor_round_ps): Ditto.
1952 (_mm512_maskz_floor_round_pd): Ditto.
1953 (_mm512_maskz_ceil_round_ps): Ditto.
1954 (_mm512_maskz_ceil_round_pd): Ditto.
1955 (_mm512_expand_pd): Ditto.
1956 (_mm512_expand_ps): Ditto.
1957 * config/i386/i386.c (ix86_builtins): Remove
1958 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1959 (bdesc_args): Ditto.
1960 * config/i386/predicates.md (const1256_operand): New.
1961 (const_1_to_2_operand): Ditto.
1962 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1963 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1964 (*avx512pf_gatherpf<mode>sf): Ditto.
1965 (avx512pf_gatherpf<mode>df): Ditto.
1966 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1967 (*avx512pf_gatherpf<mode>df): Ditto.
1968 (avx512pf_scatterpf<mode>sf): Ditto.
1969 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1970 (*avx512pf_scatterpf<mode>sf): Ditto.
1971 (avx512pf_scatterpf<mode>df): Ditto.
1972 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1973 (*avx512pf_scatterpf<mode>df): Ditto.
1974 (avx512f_expand<mode>): Removed.
1975 (<shift_insn><mode>3<mask_name>): Change predicate type.
1977 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1979 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1980 not at the end of datarefs vector use ordered_remove to avoid
1981 reordering datarefs vector.
1984 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1985 mark local addressable non-static vars as GOVD_PRIVATE
1986 instead of GOVD_LOCAL.
1987 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1988 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1992 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1993 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1994 assume_aligned or alloc_align attributes.
1995 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1996 arguments. Handle also assume_aligned and alloc_align attributes.
1997 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1998 calls to functions with assume_aligned or alloc_align attributes.
1999 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2001 2014-02-08 Terry Guo <terry.guo@arm.com>
2003 * doc/invoke.texi: Document ARM -march=armv7e-m.
2005 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2007 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2008 flag on __cilkrts_rethrow builtin.
2011 * ipa-cp.c (determine_versionability): Fail at -O0
2012 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2013 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2016 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2019 * tree-inline.c (copy_forbidden): Fail for
2020 __attribute__((optimize (0))) functions.
2022 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2024 * varpool.c: Include pointer-set.h.
2025 (varpool_remove_unreferenced_decls): Variables in other partitions
2026 will not be output; be however careful to not lose information
2029 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2031 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2032 lookup in the vtable constructor.
2034 2014-02-07 Jeff Law <law@redhat.com>
2037 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2038 define_insn_and_split.
2040 * ipa-inline.c (inline_small_functions): Fix typos.
2042 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2044 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2045 (s390_can_use_return_insn): Declare.
2046 * config/s390/s390.h (EPILOGUE_USES): Define.
2047 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2049 (s390_chunkify_start): Handle return JUMP_LABELs.
2050 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2051 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2052 (s390_can_use_return_insn): New functions.
2053 (s390_fix_long_loop_prediction): Handle conditional returns.
2054 (TARGET_SET_UP_BY_PROLOGUE): Define.
2055 * config/s390/s390.md (ANY_RETURN): New code iterator.
2056 (*creturn, *csimple_return, return, simple_return): New patterns.
2058 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2060 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2061 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2062 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2063 REG_CFA_RESTORE list when deciding not to restore a register.
2065 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2067 * config/s390/s390.c: Include tree-pass.h and context.h.
2068 (s390_early_mach): New function, split out from...
2069 (s390_emit_prologue): ...here.
2070 (pass_data_s390_early_mach): New pass structure.
2071 (pass_s390_early_mach): New class.
2072 (s390_option_override): Create and register early_mach pass.
2073 Move to end of file.
2075 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2077 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2078 to match for the exit block.
2080 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2082 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2083 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2084 Reject misaligned operands.
2086 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2088 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2090 2014-02-07 Richard Biener <rguenther@suse.de>
2093 * gimple-low.c (lower_builtin_posix_memalign): New function.
2094 (lower_stmt): Call it to lower posix_memalign in a way
2095 to make alignment info accessible.
2097 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2100 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2101 __builtin_setjmp_receiver.
2103 2014-02-07 Richard Biener <rguenther@suse.de>
2106 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2107 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2108 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2109 Handle BUILT_IN_POSIX_MEMALIGN.
2110 (find_func_clobbers): Likewise.
2111 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2112 (call_may_clobber_ref_p_1): Likewise.
2114 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2117 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2120 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2123 * lto-cgraph.c (lto_output_node): Use
2124 symtab_get_symbol_partitioning_class.
2125 (lto_output_varpool_node): likewise.
2126 (symtab_get_symbol_partitioning_class): Move here from
2128 * cgraph.h (symbol_partitioning_class): Likewise.
2129 (symtab_get_symbol_partitioning_class): Declare.
2131 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2133 * ggc.h (ggc_internal_cleared_alloc): New macro.
2134 * vec.h (vec_safe_copy): Handle memory stats.
2135 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2136 * target-globals.c (save_target_globals): Likewise.
2138 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2141 * expr.c (emit_move_resolve_push): Export; be bit more selective
2142 on when to clear alias set.
2143 * expr.h (emit_move_resolve_push): Declare.
2144 * function.h (struct function): Add tail_call_marked.
2145 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2146 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2147 * config/i386/i386.md (TImode move expander): De not call
2149 (FP push expanders): Preserve memory attributes.
2150 * config/i386/sse.md (push<mode>1): Remove.
2151 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2152 (ix86_expand_push): Remove.
2153 * config/i386/mmx.md (push<mode>1): Remove.
2155 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2157 PR rtl-optimization/60030
2158 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2159 lopart with paradoxical subreg before shifting it up by hprec.
2161 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2163 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2164 Remove extra newline at end of file.
2165 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2166 (arm_issue_rate): Handle cortexa57.
2167 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2168 (cortex-a57.cortex-a53): Likewise.
2170 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2173 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2174 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2176 (arm_expand_prologue): Adjust all callers.
2177 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2178 info, registers also at the lowest numbered registers side. Use
2179 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2183 * var-tracking.c (adjust_mems): Before adding a SET to
2184 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2186 2014-02-06 Alan Modra <amodra@gmail.com>
2189 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2190 change SDmode to DDmode when lra_in_progress.
2192 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2195 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2196 free_data_ref on the dr first, and before goto again also set dr
2197 to the next dr. For simd_lane_access, free old datarefs[i] before
2198 overwriting it. For get_vectype_for_scalar_type failure, don't
2199 free_data_ref if simd_lane_access.
2201 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2204 * tree.h (opts_for_fn): New inline function.
2205 (opt_for_fn): Define.
2206 * config/i386/i386.c (ix86_function_regparm): Use
2207 opt_for_fn (decl, optimize) instead of optimize.
2209 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2211 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2212 for SYMBOL_REF in large memory model.
2214 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2216 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2218 (cortex-a57): Likewise.
2219 (cortex-a57.cortex-a53): Likewise.
2221 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2222 Kugan Vivekanandarajah <kuganv@linaro.org>
2224 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2226 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2228 2014-02-06 Richard Biener <rguenther@suse.de>
2230 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2231 set_loop_copy and initialize_original_copy_tables.
2233 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2235 * config/aarch64/aarch64-simd.md
2236 (aarch64_ashr_simddi): Change QI to SI.
2238 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2239 Jakub Jelinek <jakub@redhat.com>
2242 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2245 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2247 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2248 CODE_FOR_altivec_vpku[hw]um to
2249 CODE_FOR_altivec_vpku[hw]um_direct.
2250 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2251 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2252 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2253 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2255 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2257 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2258 generation for -maltivec=be.
2259 (altivec_vsumsws): Simplify redundant test.
2261 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2263 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2264 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2265 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2266 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2267 gen_altivec_vpkuwum.
2268 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2270 (altivec_vpks<VI_char>ss): Likewise.
2271 (altivec_vpks<VI_char>us): Likewise.
2272 (altivec_vpku<VI_char>us): Likewise.
2273 (altivec_vpku<VI_char>um): Likewise.
2274 (altivec_vpku<VI_char>um_direct): New (copy of
2275 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2277 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2278 target is little endian and -maltivec=be is not specified.
2279 (*altivec_vupkhs<VU_char>_direct): New (copy of
2280 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2281 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2282 target is little endian and -maltivec=be is not specified.
2283 (*altivec_vupkls<VU_char>_direct): New (copy of
2284 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2285 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2286 little endian and -maltivec=be is not specified.
2287 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2288 little endian and -maltivec=be is not specified.
2290 2014-02-05 Richard Henderson <rth@redhat.com>
2293 * combine-stack-adj.c: Revert r206943.
2294 * sched-int.h (struct deps_desc): Add last_args_size.
2295 * sched-deps.c (init_deps): Initialize it.
2296 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2297 contain REG_ARGS_SIZE notes.
2299 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2301 * lto-cgraph.c (asm_nodes_output): Make global.
2302 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2303 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2304 (driver_handle_option): Handle OPT_fwpa.
2306 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2309 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2310 a comment typo and formatting issue. If odr_hash hasn't been
2311 created, return vNULL and set *completep to false.
2314 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2315 bb with no successors.
2317 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2320 * doc/invoke.texi (-march): Clarify documentation for ARM.
2324 2014-02-05 Richard Biener <rguenther@suse.de>
2326 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2327 when not vectorizing because of too many alias checks.
2328 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2329 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2331 2014-02-05 Nick Clifton <nickc@redhat.com>
2333 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2334 accept extended registers in any mode when compiling for the MN10300.
2336 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2338 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2339 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2340 sanitization attributes.
2341 (can_inline_edge_p): Likewise.
2342 (sanitize_attrs_match_for_inline_p): New function.
2344 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2346 * ipa-prop.c (detect_type_change): Shor circuit testing of
2347 type changes on THIS pointer.
2349 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2352 * config/pa/pa.c (legitimize_tls_address): Return original address
2353 if not passed a SYMBOL_REF rtx.
2354 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2356 (pa_emit_move_sequence): Simplify TLS source operands.
2357 (pa_legitimate_constant_p): Reject all TLS constants.
2358 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2359 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2361 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2363 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2364 groups when we know they are controlled by LTO.
2365 * varasm.c (default_binds_local_p_1): If object is in other partition,
2366 it will be resolved locally.
2368 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2370 * config/host-linux.c (linux_gt_pch_use_address): Don't
2371 use SSIZE_MAX because it is not always defined.
2373 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2376 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2377 threshold for pseudo splitting.
2378 (update_ebb_live_info): Process call argument hard registers and
2379 hard registers from insn definition too.
2380 (max_small_class_regs_num): New constant.
2381 (inherit_in_ebb): Update live hard regs through EBBs. Update
2382 reloads_num only for small register classes. Don't split for
2385 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2388 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2391 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2393 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2396 2014-02-04 Marek Polacek <polacek@redhat.com>
2398 * gdbinit.in (pel): Define.
2400 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2402 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2405 2014-02-04 Richard Biener <rguenther@suse.de>
2408 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2409 in function context local.
2410 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2411 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2412 similar to LTO_imported_decl_ref.
2414 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2416 PR tree-optimization/60002
2417 * cgraphclones.c (build_function_decl_skip_args): Clear
2420 PR tree-optimization/60023
2421 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2422 false to gsi_replace.
2423 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2424 has been in some EH region and vec_stmt could throw, add
2425 vec_stmt into the same EH region.
2426 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2427 has no lhs, ignore it.
2428 * internal-fn.c (expand_MASK_LOAD): Likewise.
2431 * tree-inline.c (copy_forbidden): Fail for
2432 __attribute__((optimize (0))) functions.
2435 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2436 is set, copy one less argument.
2437 (expand_simd_clones): Don't subtract clone_info->inbranch
2438 from simd_clone_struct_alloc argument.
2440 PR rtl-optimization/57915
2441 * recog.c (simplify_while_replacing): If all unary/binary/relational
2442 operation arguments are constant, attempt to simplify those.
2445 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2446 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2448 2014-02-04 Richard Biener <rguenther@suse.de>
2450 PR tree-optimization/60012
2451 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2452 TBAA disambiguation to all DDRs.
2454 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2457 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2458 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2460 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2463 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2465 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2467 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2468 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2470 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2473 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2474 to figure out targets of polymorphic calls with known decl.
2475 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2476 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2477 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2478 (get_polymorphic_call_info): ... here.
2479 (get_polymorphic_call_info_from_invariant): New function.
2481 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2483 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2484 lookup via vtable pointer; check for type consistency
2485 and turn inconsitent facts into UNREACHABLE.
2486 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2487 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2488 type inconsistent querries; return UNREACHABLE instead.
2490 2014-02-03 Richard Henderson <rth@twiddle.net>
2493 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2494 already processed this node.
2495 (normalize_one_pred_1): Pass along mark_set.
2496 (normalize_one_pred): Create and destroy a pointer_set_t.
2497 (normalize_one_pred_chain): Likewise.
2499 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2501 PR gcov-profile/58602
2502 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2504 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2507 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2508 -fno-devirtualize; try to devirtualize by the knowledge of
2509 virtual table pointer given by aggregate propagation.
2510 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2511 (ipa_print_node_jump_functions): Dump also offset that
2512 is relevant for polymorphic calls.
2513 (determine_known_aggregate_parts): Add arg_type parameter; use it
2514 instead of determining the type from pointer type.
2515 (ipa_compute_jump_functions_for_edge): Update call of
2516 determine_known_aggregate_parts.
2517 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2518 (gimple_get_virt_method_for_binfo): ... here; simplify using
2519 vtable_pointer_value_to_vtable.
2520 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2521 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2522 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2523 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2525 (vtable_pointer_value_to_binfo): ... here.
2526 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2528 2014-02-03 Teresa Johnson <tejohnson@google.com>
2530 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2531 redef of outer loop index variable.
2533 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2537 * doc/extend.texi (Function Attributes): Typo.
2539 2014-02-03 Cong Hou <congh@google.com>
2541 PR tree-optimization/60000
2542 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2543 if the vectorized statement is a store. A store statement can only
2544 appear at the end of pattern statements.
2546 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2548 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2549 (ix86_option_override_internal): Default long double to 64-bit for
2550 32-bit Bionic and to 128-bit for 64-bit Bionic.
2552 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2553 TARGET_LONG_DOUBLE_128 is true.
2554 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2556 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2557 (mlong-double-64): Negate -mlong-double-128.
2558 (mlong-double-128): New option.
2560 * config/i386/i386-c.c (ix86_target_macros): Define
2561 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2563 * doc/invoke.texi: Document -mlong-double-128.
2565 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2567 PR rtl-optimization/60024
2568 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2570 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2572 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2574 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2576 PR rtl-optimization/57662
2577 * sel-sched.c (code_motion_path_driver): Do not mark already not
2578 existing blocks in the visiting bitmap.
2580 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2582 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2583 on the insn being emitted.
2585 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2586 Will Deacon <will.deacon@arm.com>
2588 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2590 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2592 * config/arm/arm-tables.opt: Regenerate.
2594 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2596 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2597 for vector types other than V16QImode.
2598 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2599 define_expand, and call altivec_expand_vec_perm_le when producing
2600 code with little endian element order.
2601 (*altivec_vperm_<mode>_internal): New insn having previous
2602 behavior of altivec_vperm_<mode>.
2603 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2604 altivec_expand_vec_perm_le when producing code with little endian
2606 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2607 behavior of altivec_vperm_<mode>_uns.
2609 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2611 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2612 (altivec_vsumsws): Add handling for -maltivec=be with a little
2614 (altivec_vsumsws_direct): New.
2615 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2616 gen_altivec_vsumsws.
2618 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2620 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2621 vtable_pointer_value_to_binfo): New functions.
2622 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2623 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2625 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2627 * config/nios2/nios2.md (load_got_register): Initialize GOT
2628 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2629 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2631 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2633 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2634 preserverd by passthrough, do not propagate the type.
2636 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2638 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2639 (mips_atomic_assign_expand_fenv): New function.
2640 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2642 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2644 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2645 (__builtin_mips_set_fcsr): Likewise.
2646 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2647 MIPS_USI_FTYPE_VOID.
2648 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2649 (mips16_expand_set_fcsr): Likewise.
2650 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2651 (mips16_set_fcsr_stub): Likewise.
2652 (mips16_get_fcsr_one_only_stub): New class.
2653 (mips16_set_fcsr_one_only_stub): Likewise.
2654 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2655 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2656 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2657 (hard_float): New availability predicate.
2658 (mips_builtins): Add get_fcsr and set_fcsr.
2659 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2660 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2661 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2662 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2663 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2666 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2668 * config/mips/mips.c (mips_one_only_stub): New class.
2669 (mips_need_mips16_rdhwr_p): Replace with...
2670 (mips16_rdhwr_stub): ...this new variable.
2671 (mips16_stub_call_address): New function.
2672 (mips16_rdhwr_one_only_stub): New class.
2673 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2674 (mips_output_mips16_rdhwr): Delete.
2675 (mips_finish_stub): New function.
2676 (mips_code_end): Use it to handle rdhwr stubs.
2678 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2681 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2682 when calculating size of integer atomic types.
2684 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2686 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2688 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2690 PR tree-optimization/60003
2691 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2692 * profile.c (branch_prob): Use gimple_call_builtin_p
2693 to check for BUILT_IN_SETJMP_RECEIVER.
2694 * tree-inline.c (copy_bb): Call notice_special_calls.
2696 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2699 * lra-constraints.c (process_alt_operands): Update reload_sum only
2702 2014-01-31 Richard Henderson <rth@redhat.com>
2705 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2706 until after else_eh is processed.
2708 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2710 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2711 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2712 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2713 in smmintrin.h, remove them.
2714 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2715 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2716 * config/i386/i386.md (ROUND_SAE): Fix value.
2717 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2718 (const48_operand): New.
2719 * config/i386/subst.md (round), (round_expand): Use
2720 const_4_or_8_to_11_operand.
2721 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2723 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2725 * config/i386/constraints.md (Yk): Swap meaning with k.
2726 * config/i386/i386.md (movhi_internal): Change Yk to k.
2727 (movqi_internal): Ditto.
2728 (*k<logic><mode>): Ditto.
2731 (kandn<mode>): Ditto.
2732 (*<code>hi_1): Ditto.
2733 (*<code>qi_1): Ditto.
2734 (kxnor<mode>): Ditto.
2735 (kortestzhi): Ditto.
2736 (kortestchi): Ditto.
2738 (*one_cmplhi2_1): Ditto.
2739 (*one_cmplqi2_1): Ditto.
2740 * config/i386/sse.md (): Change k to Yk.
2741 (avx512f_load<mode>_mask): Ditto.
2742 (avx512f_blendm<mode>): Ditto.
2743 (avx512f_store<mode>_mask): Ditto.
2744 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2745 (avx512f_storedqu<mode>_mask): Ditto.
2746 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2748 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2749 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2750 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2751 (avx512f_maskcmp<mode>3): Ditto.
2752 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2753 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2754 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2755 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2756 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2757 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2758 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2759 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2760 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2761 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2762 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2763 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2764 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2765 (vec_extract_lo_<mode>_maskm): Ditto.
2766 (vec_extract_hi_<mode>_maskm): Ditto.
2767 (avx512f_vternlog<mode>_mask): Ditto.
2768 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2769 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2770 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2771 (avx512f_<code>v8div16qi2_mask): Ditto.
2772 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2773 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2774 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2775 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2776 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2777 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2778 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2779 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2780 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2781 (avx512cd_maskb_vec_dupv8di): Ditto.
2782 (avx512cd_maskw_vec_dupv16si): Ditto.
2783 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2784 (avx512f_vpermi2var<mode>3_mask): Ditto.
2785 (avx512f_vpermi2var<mode>3_mask): Ditto.
2786 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2787 (*avx512f_gathersi<mode>): Ditto.
2788 (*avx512f_gathersi<mode>_2): Ditto.
2789 (*avx512f_gatherdi<mode>): Ditto.
2790 (*avx512f_gatherdi<mode>_2): Ditto.
2791 (*avx512f_scattersi<mode>): Ditto.
2792 (*avx512f_scatterdi<mode>): Ditto.
2793 (avx512f_compress<mode>_mask): Ditto.
2794 (avx512f_compressstore<mode>_mask): Ditto.
2795 (avx512f_expand<mode>_mask): Ditto.
2796 * config/i386/subst.md (mask): Change k to Yk.
2797 (mask_scalar_merge): Ditto.
2800 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2802 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2804 2014-01-31 Richard Biener <rguenther@suse.de>
2807 * builtins.c (fold_builtin_memory_op): Make sure to not
2808 use a floating-point mode or a boolean or enumeral type for
2811 2014-01-30 DJ Delorie <dj@redhat.com>
2813 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2814 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2815 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2816 whenever main() has an epilogue.
2818 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2820 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2821 unused variable "field".
2822 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2823 (vsx_mergeh_<mode>): Likewise.
2824 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2825 (altivec_vmrghh): Likewise.
2826 (altivec_vmrghw): Likewise.
2827 (altivec_vmrglb): Likewise.
2828 (altivec_vmrglh): Likewise.
2829 (altivec_vmrglw): Likewise.
2830 (altivec_vspltb): Add missing uses.
2831 (altivec_vsplth): Likewise.
2832 (altivec_vspltw): Likewise.
2833 (altivec_vspltsf): Likewise.
2835 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2838 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2839 frame related instructions.
2841 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2843 PR rtl-optimization/59959
2844 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2845 any reload of register whose subreg is invalid.
2847 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2849 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2850 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2851 Add missing return type - void.
2853 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2855 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2856 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2857 remove element index adjustment for endian (now handled in vsx.md
2859 (altivec_expand_vec_perm_const): Use
2860 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2861 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2862 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2863 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2864 define_expand and a new define_insn *altivec_vspltb_internal;
2865 adjust for -maltivec=be on a little endian target.
2866 (altivec_vspltb_direct): New.
2867 (altivec_vsplth): Divide into a define_expand and a new
2868 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2869 little endian target.
2870 (altivec_vsplth_direct): New.
2871 (altivec_vspltw): Divide into a define_expand and a new
2872 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2873 little endian target.
2874 (altivec_vspltw_direct): New.
2875 (altivec_vspltsf): Divide into a define_expand and a new
2876 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2877 a little endian target.
2879 2014-01-30 Richard Biener <rguenther@suse.de>
2881 PR tree-optimization/59993
2882 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2883 can propagate form the earlier stmt and avoid the transform
2884 when the intermediate result is needed.
2886 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2888 * README.Portability: Fix typo.
2890 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2892 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2893 comparison_operator with ordered_comparison_operator.
2895 2014-01-30 Nick Clifton <nickc@redhat.com>
2897 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2898 Rename to mn10300_store_multiple_regs.
2899 * config/mn10300/mn10300.c: Likewise.
2900 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2901 store_multiple_regs.
2902 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2903 Call mn10300_store_multiple_regs.
2905 2014-01-30 Nick Clifton <nickc@redhat.com>
2906 DJ Delorie <dj@redhat.com>
2908 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2909 %fp 2 to keep registers after it properly word-aligned.
2910 (rl78_alloc_physical_registers_umul): Handle the case where both
2911 input operands are the same.
2913 2014-01-30 Richard Biener <rguenther@suse.de>
2915 PR tree-optimization/59903
2916 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2919 2014-01-30 Jason Merrill <jason@redhat.com>
2922 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2925 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2927 2014-01-30 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/59951
2930 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2932 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2935 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2936 SFmode to DFmode case.
2938 2014-01-29 DJ Delorie <dj@redhat.com>
2940 * config/msp430/msp430.opt (-minrt): New.
2941 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2943 (ENDFILE_SPEC): Likewise.
2945 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2947 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2948 (estimate_function_body_sizes): Use it.
2950 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2953 * dwarf2out.c (is_cxx_auto): New.
2954 (is_base_type): Use it.
2955 (gen_type_die_with_usage): Likewise.
2957 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2959 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2960 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2961 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2962 -maltivec=be with LE targets.
2963 (vsx_mergeh_<mode>): Likewise.
2964 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2965 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2966 (altivec_vmrghb): Replace with define_expand and new
2967 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2968 (altivec_vmrghb_direct): New define_insn.
2969 (altivec_vmrghh): Replace with define_expand and new
2970 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2971 (altivec_vmrghh_direct): New define_insn.
2972 (altivec_vmrghw): Replace with define_expand and new
2973 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2974 (altivec_vmrghw_direct): New define_insn.
2975 (*altivec_vmrghsf): Adjust for endianness.
2976 (altivec_vmrglb): Replace with define_expand and new
2977 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2978 (altivec_vmrglb_direct): New define_insn.
2979 (altivec_vmrglh): Replace with define_expand and new
2980 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2981 (altivec_vmrglh_direct): New define_insn.
2982 (altivec_vmrglw): Replace with define_expand and new
2983 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2984 (altivec_vmrglw_direct): New define_insn.
2985 (*altivec_vmrglsf): Adjust for endianness.
2986 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2987 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2988 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2989 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2990 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2991 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2992 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2993 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2995 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2997 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2998 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3001 2014-01-29 Richard Biener <rguenther@suse.de>
3003 PR tree-optimization/58742
3004 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3005 associate_pointerplus_align.
3006 (associate_pointerplus_diff): New function.
3007 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3008 and associate_pointerplus_diff.
3010 2014-01-29 Richard Biener <rguenther@suse.de>
3012 * lto-streamer.h (LTO_major_version): Bump to 3.
3013 (LTO_minor_version): Reset to 0.
3015 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3017 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3018 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3019 (arm_file_start): Generate correct asm header for armv7ve.
3020 * config/arm/bpabi.h: Add multilib support for armv7ve.
3021 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3022 and cortex-a15 to armv7ve.
3023 * config/arm/t-aprofile: Add multilib support for armv7ve.
3024 * doc/invoke.texi: Document -march=armv7ve.
3026 2014-01-29 Richard Biener <rguenther@suse.de>
3028 PR tree-optimization/58742
3029 * tree-ssa-forwprop.c (associate_plusminus): Return true
3030 if we changed sth, defer EH cleanup to ...
3031 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3032 (simplify_mult): New function.
3034 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3037 PR tree-optimization/59920
3038 * tree.c (build_common_builtin_nodes): Remove
3039 __builtin_setjmp_dispatcher initialization.
3040 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3041 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3042 instead of gsi_after_labels + manually skipping debug stmts.
3043 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3044 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3045 * tree-inline.c (copy_edges_for_bb): Remove
3046 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3047 argument. Ignore computed_goto_p stmts. Don't call
3048 make_abnormal_goto_edges. If a call might need abnormal edges
3049 for non-local gotos, see if it already has an edge to
3050 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3051 with true argument, don't do anything then, otherwise add
3052 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3053 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3055 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3056 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3057 (lower_stmt): Don't set data->calls_builtin_setjmp.
3058 (lower_builtin_setjmp): Adjust comment.
3059 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3060 * tree-cfg.c (found_computed_goto): Remove.
3061 (factor_computed_gotos): Remove.
3062 (make_goto_expr_edges): Return bool, true for computed gotos.
3063 Don't call make_abnormal_goto_edges.
3064 (build_gimple_cfg): Don't set found_computed_goto, don't call
3065 factor_computed_gotos.
3066 (computed_goto_p): No longer static.
3067 (make_blocks): Don't set found_computed_goto.
3068 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3069 (make_edges): If make_goto_expr_edges returns true, push bb
3070 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3071 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3072 vector. Record mapping between bbs and OpenMP regions if there
3073 are any, adjust make_gimple_omp_edges caller. Call
3074 handle_abnormal_edges.
3075 (make_abnormal_goto_edges): Remove.
3076 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3077 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3078 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3079 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3080 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3081 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3082 filling *region also set *region_idx to (*region)->entry->index.
3085 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3086 For REGs set ORIGINAL_REGNO.
3088 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3090 * doc/md.texi: Mention that a target shouldn't implement
3091 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3094 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3096 PR tree-optimization/59594
3097 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3098 a copy of the datarefs vector rather than the vector itself.
3100 2014-01-28 Jason Merrill <jason@redhat.com>
3103 * dwarf2out.c (auto_die): New static.
3104 (gen_type_die_with_usage): Handle C++1y 'auto'.
3105 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3108 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3111 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3112 (SPEC_X32): Likewise.
3113 (SPEC_64): Likewise.
3114 * config/i386/i386.c (ix86_option_override_internal): Turn off
3115 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3117 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3118 * config/i386/i386.h (TARGET_16BIT): New macro.
3119 (TARGET_16BIT_P): Likewise.
3120 * config/i386/i386.opt: Add m16.
3121 * doc/invoke.texi: Document -m16.
3123 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3125 PR preprocessor/59935
3126 * input.c (location_get_source_line): Bail out on when line number
3127 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3129 2014-01-28 Richard Biener <rguenther@suse.de>
3131 PR tree-optimization/58742
3132 * tree-ssa-forwprop.c (associate_plusminus): Handle
3133 pointer subtraction of the form (T)(P + A) - (T)P.
3135 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3137 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3140 2014-01-28 Richard Biener <rguenther@suse.de>
3143 2014-01-28 Richard Biener <rguenther@suse.de>
3145 PR rtl-optimization/45364
3146 PR rtl-optimization/59890
3147 * var-tracking.c (local_get_addr_clear_given_value): Handle
3148 already cleared slot.
3149 (val_reset): Handle not allocated local_get_addr_cache.
3150 (vt_find_locations): Use post-order on the inverted CFG.
3152 2014-01-28 Richard Biener <rguenther@suse.de>
3154 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3156 2014-01-28 Richard Biener <rguenther@suse.de>
3158 PR rtl-optimization/45364
3159 PR rtl-optimization/59890
3160 * var-tracking.c (local_get_addr_clear_given_value): Handle
3161 already cleared slot.
3162 (val_reset): Handle not allocated local_get_addr_cache.
3163 (vt_find_locations): Use post-order on the inverted CFG.
3165 2014-01-28 Alan Modra <amodra@gmail.com>
3167 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3168 * configure.ac <recursive call for build != host>: Define
3169 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3170 and LD_FOR_BUILD too.
3171 * configure: Regenerate.
3173 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3175 * config/i386/i386.c (get_builtin_code_for_version): Separate
3176 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3177 Broadwell from Haswell.
3179 2014-01-27 Steve Ellcey <sellcey@mips.com>
3181 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3182 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3183 * config/mips/mips.c (mips_option_override): Change setting
3185 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3186 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3187 Change from Mask to Var.
3189 2014-01-27 Jeff Law <law@redhat.com>
3191 * ipa-inline.c (inline_small_functions): Fix typo.
3193 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3195 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3196 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3197 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3198 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3199 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3200 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3201 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3202 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3203 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3204 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3205 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3206 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3207 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3208 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3209 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3210 (_mm512_storeu_epi64): Ditto.
3211 (_mm512_cmpge_epi32_mask): Ditto.
3212 (_mm512_cmpge_epu32_mask): Ditto.
3213 (_mm512_cmpge_epi64_mask): Ditto.
3214 (_mm512_cmpge_epu64_mask): Ditto.
3215 (_mm512_cmple_epi32_mask): Ditto.
3216 (_mm512_cmple_epu32_mask): Ditto.
3217 (_mm512_cmple_epi64_mask): Ditto.
3218 (_mm512_cmple_epu64_mask): Ditto.
3219 (_mm512_cmplt_epi32_mask): Ditto.
3220 (_mm512_cmplt_epu32_mask): Ditto.
3221 (_mm512_cmplt_epi64_mask): Ditto.
3222 (_mm512_cmplt_epu64_mask): Ditto.
3223 (_mm512_cmpneq_epi32_mask): Ditto.
3224 (_mm512_cmpneq_epu32_mask): Ditto.
3225 (_mm512_cmpneq_epi64_mask): Ditto.
3226 (_mm512_cmpneq_epu64_mask): Ditto.
3227 (_mm512_expand_pd): Ditto.
3228 (_mm512_expand_ps): Ditto.
3229 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3230 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3231 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3232 * config/i386/i386.c (ix86_builtins): Add
3233 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3234 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3235 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3236 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3237 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3238 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3239 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3240 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3241 IX86_BUILTIN_PMOVUSQW512_MEM.
3242 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3243 __builtin_ia32_pmovsqd512mem_mask,
3244 __builtin_ia32_pmovqd512mem_mask,
3245 __builtin_ia32_pmovusqw512mem_mask,
3246 __builtin_ia32_pmovsqw512mem_mask,
3247 __builtin_ia32_pmovqw512mem_mask,
3248 __builtin_ia32_pmovusdw512mem_mask,
3249 __builtin_ia32_pmovsdw512mem_mask,
3250 __builtin_ia32_pmovdw512mem_mask,
3251 __builtin_ia32_pmovqb512mem_mask,
3252 __builtin_ia32_pmovusqb512mem_mask,
3253 __builtin_ia32_pmovsqb512mem_mask,
3254 __builtin_ia32_pmovusdb512mem_mask,
3255 __builtin_ia32_pmovsdb512mem_mask,
3256 __builtin_ia32_pmovdb512mem_mask.
3257 (bdesc_args): Add __builtin_ia32_expanddf512,
3258 __builtin_ia32_expandsf512.
3259 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3260 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3261 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3262 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3263 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3264 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3265 (avx512f_<code>v8div16qi2_mask_store): This.
3266 (avx512f_expand<mode>): New.
3268 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3270 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3272 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3273 (_mm512_prefetch_i32scatter_pd): Ditto.
3274 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3275 (_mm512_prefetch_i64scatter_pd): Ditto.
3276 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3277 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3278 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3279 (_mm512_prefetch_i32scatter_ps): Ditto.
3280 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3281 (_mm512_prefetch_i64scatter_ps): Ditto.
3282 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3283 * config/i386/i386-builtin-types.def: Define
3284 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3285 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3286 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3287 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3288 IX86_BUILTIN_SCATTERPFQPD.
3289 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3290 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3291 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3292 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3293 __builtin_ia32_scatterpfqps.
3294 (ix86_expand_builtin): Expand new built-ins.
3295 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3296 fix memory access data type.
3297 (*avx512pf_gatherpf<mode>_mask): Ditto.
3298 (*avx512pf_gatherpf<mode>): Ditto.
3299 (avx512pf_scatterpf<mode>): Ditto.
3300 (*avx512pf_scatterpf<mode>_mask): Ditto.
3301 (*avx512pf_scatterpf<mode>): Ditto.
3302 (GATHER_SCATTER_SF_MEM_MODE): New.
3303 (avx512pf_gatherpf<mode>df): Ditto.
3304 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3305 (*avx512pf_scatterpf<mode>df): Ditto.
3307 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3310 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3311 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3314 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3316 * common/config/arm/arm-common.c
3317 (arm_rewrite_mcpu): Handle multiple names.
3319 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3321 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3323 * gimple-builder.h (create_gimple_tmp): Delete.
3325 2014-01-27 Christian Bruel <christian.bruel@st.com>
3327 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3330 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3332 * config/pa/pa.md (call): Generate indirect long calls to non-local
3333 functions when outputing 32-bit code.
3334 (call_value): Likewise except for special call to buggy powf function.
3336 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3337 portable runtime and PIC indirect calls.
3338 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3339 and PIC call sequences. Use ldo instead of blr to set return register
3340 in PIC call sequence.
3342 2014-01-25 Walter Lee <walt@tilera.com>
3344 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3345 avoid clobbering a live register.
3347 2014-01-25 Walter Lee <walt@tilera.com>
3349 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3350 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3351 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3352 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3354 2014-01-25 Walter Lee <walt@tilera.com>
3356 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3357 arguments on even registers.
3358 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3360 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3361 (BIGGEST_ALIGNMENT): Ditto.
3362 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3364 2014-01-25 Walter Lee <walt@tilera.com>
3366 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3367 insns before bundling.
3368 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3370 2014-01-25 Walter Lee <walt@tilera.com>
3372 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3373 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3374 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3376 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3378 * config/mips/constraints.md (kl): Delete.
3379 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3380 define expands, using...
3381 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3382 instructions for MIPS16.
3383 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3384 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3386 2014-01-25 Walter Lee <walt@tilera.com>
3388 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3392 2014-01-25 Walter Lee <walt@tilera.com>
3394 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3395 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3397 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3399 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3402 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3404 * print-rtl.c (in_call_function_usage): New var.
3405 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3406 EXPR_LIST mode as mode and not as reg note name.
3409 * cfgloopmanip.c (copy_loop_info): If
3410 loop->warned_aggressive_loop_optimizations, make sure
3411 the flag is set in target loop too.
3413 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3415 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3417 * builtins.def: Likewise.
3418 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3419 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3420 * ira.c (ira_setup_eliminable_regset): Likewise.
3421 * omp-low.c (gate_expand_omp): Likewise.
3422 (execute_lower_omp): Likewise.
3423 (diagnose_sb_0): Likewise.
3424 (gate_diagnose_omp_blocks): Likewise.
3425 (simd_clone_clauses_extract): Likewise.
3428 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3430 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3431 correction for little endian...
3432 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3435 2014-01-24 Jeff Law <law@redhat.com>
3437 PR tree-optimization/59919
3438 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3439 for non-returning calls.
3441 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3443 * common/config/aarch64/aarch64-common.c
3444 (aarch64_rewrite_mcpu): Handle multiple names.
3445 * config/aarch64/aarch64.h
3446 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3448 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3450 * input.c (add_file_to_cache_tab): Handle the case where fopen
3453 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3456 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3457 from push operand if code of push isn't PRE_DEC.
3459 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3462 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3463 -mquad-memory-atomic. Update -mquad-memory documentation to say
3464 it is only used for non-atomic loads/stores.
3466 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3467 -mquad-memory or -mquad-memory-atomic switches.
3469 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3470 -mquad-memory-atomic to ISA 2.07 support.
3472 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3473 to separate support of normal quad word memory operations (ldq, stq)
3474 from the atomic quad word memory operations.
3476 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3477 support to separate non-atomic quad word operations from atomic
3478 quad word operations. Disable non-atomic quad word operations in
3479 little endian mode so that we don't have to swap words after the
3480 load and before the store.
3481 (quad_load_store_p): Add comment about atomic quad word support.
3482 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3483 options printed with -mdebug=reg.
3485 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3486 -mquad-memory-atomic as the test for whether we have quad word
3487 atomic instructions.
3488 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3489 or -mp8-vector are used, allow byte/half-word atomic operations.
3491 * config/rs6000/sync.md (load_lockedti): Insure that the address
3492 is a proper indexed or indirect address for the lqarx instruction.
3493 On little endian systems, swap the hi/lo registers after the lqarx
3495 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3496 insure the address is valid for the lqarx instruction.
3497 (store_conditionalti): Insure that the address is a proper indexed
3498 or indirect address for the stqcrx. instruction. On little endian
3499 systems, swap the hi/lo registers before doing the stqcrx.
3501 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3502 insure the address is valid for the stqcrx. instruction.
3504 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3505 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3506 type of quad memory support is available.
3508 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3511 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3512 there is a danger of looping.
3514 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3516 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3517 force flag_ira_loop_pressure if set via command line.
3519 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3521 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3522 (ashr_simd): New builtin handling DI mode.
3523 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3524 (aarch64_sshr_simddi): New match pattern.
3525 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3526 (vshrd_n_s64): Likewise.
3527 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3529 2014-01-23 Nick Clifton <nickc@redhat.com>
3531 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3532 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3533 favour of mcu specific scripts.
3534 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3537 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3538 Alex Velenko <Alex.Velenko@arm.com>
3540 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3541 (vaddv_s16): Likewise.
3542 (vaddv_s32): Likewise.
3543 (vaddv_u8): Likewise.
3544 (vaddv_u16): Likewise.
3545 (vaddv_u32): Likewise.
3546 (vaddvq_s8): Likewise.
3547 (vaddvq_s16): Likewise.
3548 (vaddvq_s32): Likewise.
3549 (vaddvq_s64): Likewise.
3550 (vaddvq_u8): Likewise.
3551 (vaddvq_u16): Likewise.
3552 (vaddvq_u32): Likewise.
3553 (vaddvq_u64): Likewise.
3554 (vaddv_f32): Likewise.
3555 (vaddvq_f32): Likewise.
3556 (vaddvq_f64): Likewise.
3557 (vmaxv_f32): Likewise.
3558 (vmaxv_s8): Likewise.
3559 (vmaxv_s16): Likewise.
3560 (vmaxv_s32): Likewise.
3561 (vmaxv_u8): Likewise.
3562 (vmaxv_u16): Likewise.
3563 (vmaxv_u32): Likewise.
3564 (vmaxvq_f32): Likewise.
3565 (vmaxvq_f64): Likewise.
3566 (vmaxvq_s8): Likewise.
3567 (vmaxvq_s16): Likewise.
3568 (vmaxvq_s32): Likewise.
3569 (vmaxvq_u8): Likewise.
3570 (vmaxvq_u16): Likewise.
3571 (vmaxvq_u32): Likewise.
3572 (vmaxnmv_f32): Likewise.
3573 (vmaxnmvq_f32): Likewise.
3574 (vmaxnmvq_f64): Likewise.
3575 (vminv_f32): Likewise.
3576 (vminv_s8): Likewise.
3577 (vminv_s16): Likewise.
3578 (vminv_s32): Likewise.
3579 (vminv_u8): Likewise.
3580 (vminv_u16): Likewise.
3581 (vminv_u32): Likewise.
3582 (vminvq_f32): Likewise.
3583 (vminvq_f64): Likewise.
3584 (vminvq_s8): Likewise.
3585 (vminvq_s16): Likewise.
3586 (vminvq_s32): Likewise.
3587 (vminvq_u8): Likewise.
3588 (vminvq_u16): Likewise.
3589 (vminvq_u32): Likewise.
3590 (vminnmv_f32): Likewise.
3591 (vminnmvq_f32): Likewise.
3592 (vminnmvq_f64): Likewise.
3594 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3596 * config/aarch64/aarch64-simd.md
3597 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3598 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3599 (*aarch64_mul3_elt<mode>): Likewise.
3600 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3601 (*aarch64_mul3_elt_to_64v2df): Likewise.
3602 (*aarch64_mla_elt<mode>): Likewise.
3603 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3604 (*aarch64_mls_elt<mode>): Likewise.
3605 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3606 (*aarch64_fma4_elt<mode>): Likewise.
3607 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3608 (*aarch64_fma4_elt_to_64v2df): Likewise.
3609 (*aarch64_fnma4_elt<mode>): Likewise.
3610 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3611 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3612 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3613 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3614 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3615 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3616 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3617 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3618 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3620 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3622 * config/aarch64/aarch64-simd.md
3623 (aarch64_be_checked_get_lane<mode>): New define_expand.
3624 * config/aarch64/aarch64-simd-builtins.def
3625 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3626 New builtin definition.
3627 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3628 Use new safe be builtin.
3630 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3632 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3634 (aarch64_be_st1<mode>): Likewise.
3635 (aarch_ld1<VALL:mode>): Define_expand modified.
3636 (aarch_st1<VALL:mode>): Likewise.
3637 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3638 (UNSPEC_ST1): Likewise.
3640 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3642 * config/microblaze/microblaze.md: Add trap insn and attribute
3644 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3646 PR preprocessor/58580
3647 * input.h (location_get_source_line): Take an additional line_size
3649 (void diagnostics_file_cache_fini): Declare new function.
3650 * input.c (struct fcache): New type.
3651 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3652 New static constants.
3653 (diagnostic_file_cache_init, total_lines_num)
3654 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3655 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3656 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3657 (get_next_line, read_next_line, goto_next_line, read_line_num):
3658 New static function definitions.
3659 (diagnostic_file_cache_fini): New function.
3660 (location_get_source_line): Take an additional output line_len
3661 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3663 * diagnostic.c (diagnostic_finish): Call
3664 diagnostic_file_cache_fini.
3665 (adjust_line): Take an additional input parameter for the length
3666 of the line, rather than calculating it with strlen.
3667 (diagnostic_show_locus): Adjust the use of
3668 location_get_source_line and adjust_line with respect to their new
3669 signature. While displaying a line now, do not stop at the first
3670 null byte. Rather, display the zero byte as a space and keep
3671 going until we reach the size of the line.
3672 * Makefile.in: Add vec.o to OBJS-libcommon
3674 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3675 Ilya Tocar <ilya.tocar@intel.com>
3677 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3678 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3679 (__builtin_ia32_kmov16): Ditto.
3680 * config/i386/i386.md (UNSPEC_KMOV): New.
3683 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3685 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3686 (_mm512_storeu_si512): Ditto.
3688 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3691 * rtl.h (get_referenced_operands): Declare.
3692 * recog.c (get_referenced_operands): New function.
3693 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3694 operands have been referenced when recording LO_SUM references.
3696 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3698 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3700 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3702 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3703 Enable for generic and recent AMD targets.
3705 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3707 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3708 ARG_SIZE note when adjustment was eliminated.
3710 2014-01-22 Jeff Law <law@redhat.com>
3712 PR tree-optimization/59597
3713 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3714 in file. Accept new argument REGISTERING and use it to modify
3715 dump output appropriately.
3716 (register_jump_thread): Corresponding changes.
3717 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3718 thread paths involving joiner blocks. Add code to dump cancelled
3719 jump threading paths.
3721 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3723 PR rtl-optimization/59477
3724 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3725 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3727 2014-01-22 Tom Tromey <tromey@redhat.com>
3729 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3731 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3733 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3735 PR rtl-optimization/59896
3736 * lra-constraints.c (process_alt_operands): Check unused note for
3737 matched operands of insn with no output reloads.
3739 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3741 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3742 (mips_move_from_gpr_cost): Likewise.
3744 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3746 PR rtl-optimization/59858
3747 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3748 ira_class_hard_regs_num.
3749 (process_alt_operands): Increase reject for dying matched operand.
3751 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3754 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3755 smaller than size, perform several stores or loads and stores
3756 at dst + count - size to store or copy all of size bytes, rather
3757 than just last modesize bytes.
3759 2014-01-20 DJ Delorie <dj@redhat.com>
3761 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3762 that CLOBBERs are REGs before propogating their values.
3764 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3767 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3768 (cgraph_inline_failed_type): New function.
3769 * cgraph.h (DEFCIFCODE): Add type.
3770 (cgraph_inline_failed_type_t): New enum.
3771 (cgraph_inline_failed_type): New prototype.
3772 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3773 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3774 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3775 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3776 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3777 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3778 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3779 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3780 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3781 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3782 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3783 OPTIMIZATION_MISMATCH.
3784 * tree-inline.c (expand_call_inline): Emit errors during
3785 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3787 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3790 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3791 mode attribute in insn output.
3793 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3795 * output.h (output_constant): Delete.
3796 * varasm.c (output_constant): Make private.
3798 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3800 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3802 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3805 * tree.h (fold_builtin_strcat): New prototype.
3806 * builtins.c (fold_builtin_strcat): No longer static. Add len
3807 argument, if non-NULL, don't call c_strlen. Optimize
3808 directly into __builtin_memcpy instead of __builtin_strcpy.
3809 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3810 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3812 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3814 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3815 for SImode_address_operand operands, having only a REG argument.
3817 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3819 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3820 loader name using mbig-endian.
3821 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3823 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3825 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3829 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3831 * config/aarch64/aarch64-protos.h
3832 (aarch64_cannot_change_mode_class_ptr): Declare.
3833 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3834 aarch64_cannot_change_mode_class_ptr): New.
3835 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3836 backend hook aarch64_cannot_change_mode_class.
3838 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3840 * common/config/aarch64/aarch64-common.c
3841 (aarch64_handle_option): Don't handle any option order logic here.
3842 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3843 selected_cpu, warn on architecture version mismatch.
3844 (aarch64_override_options): Fix parsing order for option strings.
3846 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3847 Iain Sandoe <iain@codesourcery.com>
3850 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3851 warning. Amend comment to reflect current functionality.
3853 2014-01-20 Richard Biener <rguenther@suse.de>
3856 * builtins.c (fold_builtin_strcat): Remove case better handled
3857 by tree-ssa-strlen.c.
3859 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3861 * config/aarch64/aarch64.opt
3862 (mcpu, march, mtune): Make case-insensitive.
3864 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3867 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3868 if operands[1] is a REG or ZERO_EXTEND of a REG.
3870 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3872 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3874 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3876 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3877 long non-pic millicode calls.
3879 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3881 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3883 2014-01-19 Kito Cheng <kito@0xlab.org>
3885 * builtins.c (expand_movstr): Check movstr expand done or fail.
3887 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3888 H.J. Lu <hongjiu.lu@intel.com>
3891 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3892 to DImode for zero-extended addresses.
3894 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3896 PR rtl-optimization/57763
3897 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3898 on the new indirect jump_insn and increment LABEL_NUSES (label).
3900 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3904 * config.gcc (x86_archs): New variable.
3905 (x86_64_archs): Likewise.
3906 (x86_cpus): Likewise.
3907 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3908 --with-arch/--with-cpu= options.
3909 Support --with-arch=/--with-cpu={nehalem,westmere,
3910 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3912 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3914 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3915 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3917 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3919 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3921 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3924 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3925 clear cpp_get_options (parse_in)->warn_unused_macros for
3926 ix86_target_macros_internal with cpp_define.
3928 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3930 * jump.c (delete_related_insns): Keep (use (insn))s.
3931 * reorg.c (redundant_insn): Check for barriers too.
3933 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3935 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3937 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3939 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3940 call to $$dyncall when TARGET_LONG_CALLS is true.
3942 2014-01-17 Jeff Law <law@redhat.com>
3944 * ree.c (combine_set_extension): Temporarily disable test for
3945 changing number of hard registers.
3947 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3950 * ipa-inline-analysis.c (inline_free_summary):
3951 Do not free summary of aliases.
3953 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3956 * gimplify.c (gimplify_expr): Use create_tmp_var
3957 instead of create_tmp_var_raw. If cond doesn't have
3958 integral type, don't add the IFN_ANNOTATE builtin at all.
3960 2014-01-17 Martin Jambor <mjambor@suse.cz>
3963 * ipa-cp.c (prev_edge_clone): New variable.
3964 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3965 Also resize prev_edge_clone vector.
3966 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3967 (ipcp_edge_removal_hook): New function.
3968 (ipcp_driver): Register ipcp_edge_removal_hook.
3970 2014-01-17 Andrew Pinski <apinski@cavium.com>
3971 Steve Ellcey <sellcey@mips.com>
3974 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3977 2014-01-17 Jeff Law <law@redhat.com>
3980 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3981 so that pass_ccp runs first.
3983 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3985 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3986 (ix86_adjust_cost): Use !TARGET_XXX.
3987 (do_reorder_for_imul): Likewise.
3988 (swap_top_of_ready_list): Likewise.
3989 (ix86_sched_reorder): Likewise.
3991 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3993 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3994 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3995 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3996 (intel_memset): New. Duplicate slm_memset.
3997 (intel_cost): New. Duplicate slm_cost.
3998 (m_INTEL): New macro.
3999 (processor_target_table): Add "intel".
4000 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4001 with PROCESSOR_INTEL for "intel".
4002 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4003 PROCESSOR_SILVERMONT.
4004 (ix86_issue_rate): Likewise.
4005 (ix86_adjust_cost): Likewise.
4006 (ia32_multipass_dfa_lookahead): Likewise.
4007 (swap_top_of_ready_list): Likewise.
4008 (ix86_sched_reorder): Likewise.
4009 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4010 instead of TARGET_OPT_AGU.
4011 * config/i386/i386.h (TARGET_INTEL): New.
4012 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4013 (processor_type): Add PROCESSOR_INTEL.
4014 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4015 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4017 2014-01-17 Marek Polacek <polacek@redhat.com>
4020 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4023 2014-01-17 Richard Biener <rguenther@suse.de>
4025 PR tree-optimization/46590
4026 * opts.c (default_options_table): Add entries for
4027 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4028 all enabled at -O1 but not for -Og.
4029 * common.opt (fbranch-count-reg): Remove Init(1).
4030 (fmove-loop-invariants): Likewise.
4031 (ftree-pta): Likewise.
4033 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4035 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4036 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4037 decls to at least the GCC 4.8 used alignments.
4040 * tree-nested.c (convert_nonlocal_reference_stmt,
4041 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4042 of GIMPLE_BIND stmts, adjust associated decls.
4044 2014-01-17 Richard Biener <rguenther@suse.de>
4046 PR tree-optimization/46590
4047 * vec.h (vec<>::bseach): New member function implementing
4048 binary search according to C89 bsearch.
4049 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4050 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4051 bitmap pointer again. Make accesses_in_loop a flat array.
4052 (mem_ref_obstack): New global.
4053 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4054 (mark_ref_stored): Likewise.
4055 (ref_indep_loop_p_2): Likewise.
4056 (set_ref_stored_in_loop): New helper function.
4057 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4058 (memref_free): Adjust.
4059 (record_mem_ref_loc): Simplify.
4060 (gather_mem_refs_stmt): Adjust.
4061 (sort_locs_in_loop_postorder_cmp): New function.
4062 (analyze_memory_references): Sort accesses_in_loop after
4063 loop postorder number.
4064 (find_ref_loc_in_loop_cmp): New function.
4065 (for_all_locs_in_loop): Find relevant cluster of locs in
4066 accesses_in_loop and iterate without recursion.
4067 (execute_sm): Avoid uninit warning.
4068 (struct ref_always_accessed): Simplify.
4069 (ref_always_accessed::operator ()): Likewise.
4070 (ref_always_accessed_p): Likewise.
4071 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4072 loop postorder numbers here.
4073 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4076 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4079 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4080 on decls for which assemble_alias has been called.
4082 2014-01-17 Nick Clifton <nickc@redhat.com>
4084 * config/msp430/msp430.opt: (mcpu): New option.
4085 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4086 (msp430_option_override): Parse target_cpu. If the MCU name
4087 matches a generic string, clear target_mcu.
4088 (msp430_attr): Allow numeric interrupt values up to 63.
4089 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4090 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4092 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4094 * config/msp430/msp430.md (popm): Use %J rather than %I.
4095 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4096 (addhi_cy_i): Use immediate_operand for operand 2.
4097 * doc/invoke.texi: Document -mcpu option.
4099 2014-01-17 Richard Biener <rguenther@suse.de>
4101 PR rtl-optimization/38518
4102 * df.h (df_analyze_loop): Declare.
4103 * df-core.c: Include cfgloop.h.
4104 (df_analyze_1): Split out main part of df_analyze.
4105 (df_analyze): Adjust.
4106 (loop_inverted_post_order_compute): New function.
4107 (loop_post_order_compute): Likewise.
4108 (df_analyze_loop): New function avoiding whole-function
4110 * loop-invariant.c (find_defs): Use df_analyze_loop.
4111 (find_invariants): Adjust.
4112 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4114 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4116 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4117 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4119 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4121 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4122 traversal when removing references.
4124 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4127 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4129 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4132 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4133 pushing a reload for an autoinc when we had previously reloaded an
4134 inner part of the address.
4136 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4138 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4140 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4141 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4142 when not giving up or versioning for alias only because of
4144 (vect_analyze_data_ref_dependences): Set to true.
4145 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4147 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4148 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4152 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4155 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4156 operand 0 predicate for gathers, use a new pseudo as subtarget.
4158 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4161 * lra-constraints.c (process_alt_operands): Add printing debug info.
4162 Check absence of input/output reloads for matched operands too.
4164 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4166 PR rtl-optimization/59835
4167 * ira.c (ira_init_register_move_cost): Increase cost for
4170 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4172 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4174 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4177 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4178 non-register objects. Use gen_(high/low)part more consistently.
4181 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4184 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4185 endian support, remove tests for WORDS_BIG_ENDIAN.
4186 (p8_mfvsrd_3_<mode>): Likewise.
4187 (reload_gpr_from_vsx<mode>): Likewise.
4188 (reload_gpr_from_vsxsf): Likewise.
4189 (p8_mfvsrd_4_disf): Likewise.
4191 2014-01-16 Richard Biener <rguenther@suse.de>
4193 PR rtl-optimization/46590
4194 * lcm.c (compute_antinout_edge): Use postorder iteration.
4195 (compute_laterin): Use inverted postorder iteration.
4197 2014-01-16 Nick Clifton <nickc@redhat.com>
4200 * varasm.c (output_constant): Return the number of bytes actually
4202 (output_constructor_array_range): Update the field size with the
4203 number of bytes emitted by output_constant.
4204 (output_constructor_regular_field): Likewise. Also do not
4205 complain if the total number of bytes emitted is now greater
4206 than the expected fieldpos.
4207 * output.h (output_constant): Update prototype and descriptive comment.
4209 2014-01-16 Marek Polacek <polacek@redhat.com>
4212 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4213 it is error_mark_node.
4215 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4217 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4218 VALID_AVX256_REG_OR_OI_MODE.
4220 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4222 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4223 current procedure should be profiled.
4225 2014-01-15 Andrew Pinski <apinski@cavium.com>
4227 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4228 of moving from/to the STACK_REG register class.
4230 2014-01-15 Richard Henderson <rth@redhat.com>
4233 * reginfo.c (global_regs_decl): Globalize.
4234 * rtl.h (global_regs_decl): Declare.
4235 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4236 reserved via global_regs.
4238 2014-01-15 Teresa Johnson <tejohnson@google.com>
4240 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4242 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4244 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4245 and vmulosh rather than call gen_vec_widen_smult_*.
4246 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4247 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4248 (vec_widen_smult_even_v16qi): Likewise.
4249 (vec_widen_umult_even_v8hi): Likewise.
4250 (vec_widen_smult_even_v8hi): Likewise.
4251 (vec_widen_umult_odd_v16qi): Likewise.
4252 (vec_widen_smult_odd_v16qi): Likewise.
4253 (vec_widen_umult_odd_v8hi): Likewise.
4254 (vec_widen_smult_odd_v8hi): Likewise.
4255 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4256 vmuloub rather than call gen_vec_widen_umult_*.
4257 (vec_widen_umult_lo_v16qi): Likewise.
4258 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4259 vmulosb rather than call gen_vec_widen_smult_*.
4260 (vec_widen_smult_lo_v16qi): Likewise.
4261 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4262 rather than call gen_vec_widen_umult_*.
4263 (vec_widen_umult_lo_v8hi): Likewise.
4264 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4265 rather than call gen_vec_widen_smult_*.
4266 (vec_widen_smult_lo_v8hi): Likewise.
4268 2014-01-15 Jeff Law <law@redhat.com>
4270 PR tree-optimization/59747
4271 * ree.c (find_and_remove_re): Properly handle case where a second
4272 eliminated extension requires widening a copy created for elimination
4273 of a prior extension.
4274 (combine_set_extension): Ensure that the number of hard regs needed
4275 for a destination register does not change when we widen it.
4277 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4279 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4280 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4281 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4282 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4283 (avr-*-rtems*): Likewise.
4284 (bfin*-rtems*): Likewise.
4285 (moxie-*-rtems*): Likewise.
4286 (h8300-*-rtems*): Likewise.
4287 (i[34567]86-*-rtems*): Likewise.
4288 (lm32-*-rtems*): Likewise.
4289 (m32r-*-rtems*): Likewise.
4290 (m68k-*-rtems*): Likewise.
4291 (microblaze*-*-rtems*): Likewise.
4292 (mips*-*-rtems*): Likewise.
4293 (powerpc-*-rtems*): Likewise.
4294 (sh-*-rtems*): Likewise.
4295 (sparc-*-rtems*): Likewise.
4296 (sparc64-*-rtems*): Likewise.
4297 (v850-*-rtems*): Likewise.
4298 (m32c-*-rtems*): Likewise.
4300 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4302 PR rtl-optimization/59511
4303 * ira.c (ira_init_register_move_cost): Use memory costs for some
4304 cases of register move cost calculations.
4305 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4306 instead of BB frequency.
4307 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4308 * lra-assigns.c (find_hard_regno_for): Ditto.
4310 2014-01-15 Richard Biener <rguenther@suse.de>
4312 PR tree-optimization/59822
4313 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4314 (vectorizable_load): Use it to hoist defs of uses of invariant
4315 loads out of the loop.
4317 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4318 Kugan Vivekanandarajah <kuganv@linaro.org>
4321 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4324 2014-01-15 Richard Biener <rguenther@suse.de>
4326 PR rtl-optimization/59802
4327 * lcm.c (compute_available): Use inverted postorder to seed
4328 the initial worklist.
4330 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4333 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4334 ADDR_REGS for invalid symrefs in non-PIC code.
4336 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4339 * builtins.c (determine_block_size): Initialize *probable_max_size
4340 even if len_rtx is CONST_INT.
4342 2014-01-14 Andrew Pinski <apinski@cavium.com>
4344 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4345 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4346 (cortexa53_tunings): Likewise.
4347 (aarch64_sched_issue_rate): New function.
4348 (TARGET_SCHED_ISSUE_RATE): Define.
4350 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4352 * ira-costs.c (find_costs_and_classes): Add missed
4353 ira_init_register_move_cost_if_necessary.
4355 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4358 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4360 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4363 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4364 to indicate if type is used for function return value. Warn ABI
4365 change if the vector mode isn't available for function return value.
4366 (ix86_function_arg_advance): Pass false to type_natural_mode.
4367 (ix86_function_arg): Likewise.
4368 (ix86_gimplify_va_arg): Likewise.
4369 (function_arg_32): Don't warn ABI change.
4370 (ix86_function_value): Pass true to type_natural_mode.
4371 (ix86_return_in_memory): Likewise.
4372 (ix86_struct_value_rtx): Removed.
4373 (TARGET_STRUCT_VALUE_RTX): Likewise.
4375 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4377 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4378 converting a conditional jump into a conditional return.
4380 2014-01-14 Richard Biener <rguenther@suse.de>
4382 PR tree-optimization/58921
4383 PR tree-optimization/59006
4384 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4385 hoisting invariant stmts.
4386 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4387 invariant loads on the preheader edge if possible.
4389 2014-01-14 Joey Ye <joey.ye@arm.com>
4391 * doc/plugin.texi (Building GCC plugins): Update to C++.
4393 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4395 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4396 (_mm_rcp28_round_ss): Ditto.
4397 (_mm_rsqrt28_round_sd): Ditto.
4398 (_mm_rsqrt28_round_ss): Ditto.
4399 (_mm_rcp28_sd): Ditto.
4400 (_mm_rcp28_ss): Ditto.
4401 (_mm_rsqrt28_sd): Ditto.
4402 (_mm_rsqrt28_ss): Ditto.
4403 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4404 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4405 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4406 (IX86_BUILTIN_RCP28SD): Ditto.
4407 (IX86_BUILTIN_RCP28SS): Ditto.
4408 (IX86_BUILTIN_RSQRT28SD): Ditto.
4409 (IX86_BUILTIN_RSQRT28SS): Ditto.
4410 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4411 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4412 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4413 (ix86_expand_special_args_builtin): Expand new FTYPE.
4414 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4415 (srcp14<mode>): Make insn unary.
4416 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4417 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4418 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4419 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4420 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4421 Fix rounding: make it SAE only.
4422 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4424 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4426 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4427 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4428 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4429 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4430 (round_saeonly_mask_scalar_operand4): Ditto.
4431 (round_saeonly_mask_scalar_op3): Ditto.
4432 (round_saeonly_mask_scalar_op4): Ditto.
4434 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4436 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4437 Implement -maltivec=be for vec_insert and vec_extract.
4439 2014-01-10 DJ Delorie <dj@redhat.com>
4441 * config/msp430/msp430.md (call_internal): Don't allow memory
4442 references with SP as the base register.
4443 (call_value_internal): Likewise.
4444 * config/msp430/constraints.md (Yc): New. For memory references
4445 that don't use SP as a base register.
4447 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4448 "an integer without a # prefix"
4449 * config/msp430/msp430.md (epilogue_helper): Use it.
4451 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4454 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4455 AVX512F gather builtins.
4456 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4457 on gather decls with INTEGER_TYPE masktype.
4458 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4459 directly into the builtin rather than hoisting it before loop.
4461 PR tree-optimization/59387
4462 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4463 (scev_const_prop): If folded_casts and type has undefined overflow,
4464 use force_gimple_operand instead of force_gimple_operand_gsi and
4465 for each added stmt if it is assign with
4466 arith_code_with_undefined_signed_overflow, call
4467 rewrite_to_defined_overflow.
4468 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4469 gimple-fold.h instead.
4470 (arith_code_with_undefined_signed_overflow,
4471 rewrite_to_defined_overflow): Moved to ...
4472 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4473 rewrite_to_defined_overflow): ... here. No longer static.
4474 Include gimplify-me.h.
4475 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4476 rewrite_to_defined_overflow): New prototypes.
4478 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4480 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4482 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4484 * builtins.c (get_object_alignment_2): Minor tweak.
4485 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4487 2014-01-13 Christian Bruel <christian.bruel@st.com>
4489 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4490 optimized non constant lengths.
4492 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4495 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4496 load as __atomic_load_N if possible.
4498 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4500 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4502 (rs6000_expand_builtin): Adjust call.
4504 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4507 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4508 * config/rs6000/rs6000.c: Include target-globals.h.
4509 (rs6000_set_current_function): Instead of doing target_reinit
4510 unconditionally, use save_target_globals_default_opts and
4511 restore_target_globals.
4513 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4515 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4516 (rs6000_expand_builtin): Handle mffs and mtfsf.
4517 (rs6000_init_builtins): Define mffs and mtfsf.
4518 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4519 (rs6000_mffs): New pattern.
4520 (rs6000_mtfsf): New pattern.
4522 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4524 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4525 Start narrowing with START. Apply candidate-use pair
4526 and check overall cost in narrowing.
4527 (iv_ca_prune): Pass new argument.
4529 2014-01-10 Jeff Law <law@redhat.com>
4532 * ree.c (combine_reaching_defs): Ensure the defining statement
4533 occurs before the extension when optimizing extensions with
4534 different source and destination hard registers.
4536 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4539 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4540 vtables into the type inheritance graph.
4542 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4544 PR rtl-optimization/59754
4545 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4546 modes in the REGNO != REGNO case.
4548 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4550 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4552 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4554 PR tree-optimization/59745
4555 * tree-predcom.c (tree_predictive_commoning_loop): Call
4556 free_affine_expand_cache if giving up because components is NULL.
4558 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4559 GC in payload of target_globals struct instead of allocating them on
4560 the heap and the larger structs separately using GC.
4561 * target-globals.h (struct target_globals): Make regs, hard_regs,
4562 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4563 of GTY((skip)) and change type to void *.
4564 (reset_target_globals): Cast loads from those fields to corresponding
4567 2014-01-10 Steve Ellcey <sellcey@mips.com>
4570 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4571 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4572 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4574 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4577 * aarch64-modes.def (CC_Zmode): New flags mode.
4578 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4579 represents an equality.
4580 (aarch64_get_condition_code): Handle CC_Zmode.
4581 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4583 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4585 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4586 extraction in good case.
4588 2014-01-10 Richard Biener <rguenther@suse.de>
4590 PR tree-optimization/59374
4591 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4592 checking after SLP discovery. Mark stmts not participating
4593 in any SLP instance properly.
4595 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4597 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4598 when handling a SET rtx.
4600 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4602 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4603 (cortex-a57): Likewise.
4604 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4606 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4608 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4609 non-iwmmxt builtins.
4611 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4615 * ipa-devirt.c record_target_from_binfo): Take as argument
4616 stack of binfos and lookup matching one for virtual inheritance.
4617 (possible_polymorphic_call_targets_1): Update.
4619 2014-01-10 Huacai Chen <chenhc@lemote.com>
4621 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4622 kernel strings for Loongson-2E/2F/3A.
4624 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4627 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4628 is_gimple_call before calling gimple_call_internal_p.
4630 2014-01-09 Steve Ellcey <sellcey@mips.com>
4632 * Makefile.in (TREE_FLOW_H): Remove.
4633 (TREE_SSA_H): Add file names from tree-flow.h.
4634 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4635 * tree.h: Remove tree-flow.h reference.
4636 * hash-table.h: Remove tree-flow.h reference.
4637 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4638 reference with tree-ssa-loop.h.
4640 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4642 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4643 default element-order behavior for -maltivec.
4644 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4645 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4646 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4647 when targeting big endian, at least for now.
4648 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4650 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4653 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4654 var satisfies use_register_for_decl, just take into account type
4655 alignment, rather than decl alignment.
4657 PR tree-optimization/59622
4658 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4659 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4660 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4661 Don't devirtualize for inplace at all. For targets.length () == 1,
4662 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4664 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4666 * config/i386/i386.md (cpu): Remove the unused btver1.
4668 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4670 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4672 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4675 * tree-core.h (struct target_globals): New forward declaration.
4676 (struct tree_target_option): Add globals field.
4677 * tree.h (TREE_TARGET_GLOBALS): Define.
4678 (prepare_target_option_nodes_for_pch): New prototype.
4679 * target-globals.h (struct target_globals): Define even if
4681 * tree.c (prepare_target_option_node_for_pch,
4682 prepare_target_option_nodes_for_pch): New functions.
4683 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4684 * config/i386/i386.c: Include target-globals.h.
4685 (ix86_set_current_function): Instead of doing target_reinit
4686 unconditionally, use save_target_globals_default_opts and
4687 restore_target_globals.
4689 2014-01-09 Richard Biener <rguenther@suse.de>
4691 PR tree-optimization/59715
4692 * tree-cfg.h (split_critical_edges): Declare.
4693 * tree-cfg.c (split_critical_edges): Export.
4694 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4696 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4698 * cfgexpand.c (expand_stack_vars): Optionally disable
4699 asan stack protection.
4700 (expand_used_vars): Likewise.
4701 (partition_stack_vars): Likewise.
4702 * asan.c (asan_emit_stack_protection): Optionally disable
4703 after return stack usage.
4704 (instrument_derefs): Optionally disable memory access instrumentation.
4705 (instrument_builtin_call): Likewise.
4706 (instrument_strlen_call): Likewise.
4707 (asan_protect_global): Optionally disable global variables protection.
4708 * doc/invoke.texi: Added doc for new options.
4709 * params.def: Added new options.
4710 * params.h: Likewise.
4712 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4714 PR rtl-optimization/59724
4715 * ifcvt.c (cond_exec_process_if_block): Don't call
4716 flow_find_head_matching_sequence with 0 longest_match.
4717 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4718 non-active insns if !stop_after.
4719 (try_head_merge_bb): Revert 2014-01-07 changes.
4721 2014-01-08 Jeff Law <law@redhat.com>
4723 * ree.c (get_sub_rtx): New function, extracted from...
4724 (merge_def_and_ext): Here.
4725 (combine_reaching_defs): Use get_sub_rtx.
4727 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4729 * cgraph.h (varpool_variable_node): Do not choke on null node.
4731 2014-01-08 Catherine Moore <clm@codesourcery.com>
4733 * config/mips/mips.md (simple_return): Attempt to use JRC
4735 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4737 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4739 PR rtl-optimization/59137
4740 * reorg.c (steal_delay_list_from_target): Call update_block for
4742 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4744 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4746 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4747 two duplicate entries.
4749 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4752 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4754 * config/mips/mips.c (mips_truncated_op_cost): New function.
4755 (mips_rtx_costs): Adjust test for BADDU.
4756 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4758 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4760 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4761 (*baddu_si): ...this new pattern.
4763 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4766 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4768 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4771 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4773 (expand_expr, expand_normal): Adjust.
4774 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4775 inner_reference_p. Use inner_reference_p to expand inner references.
4776 (store_expr): Adjust.
4777 * cfgexpand.c (expand_call_stmt): Adjust.
4779 2014-01-08 Rong Xu <xur@google.com>
4781 * gcov-io.c (gcov_var): Move from gcov-io.h.
4782 (gcov_position): Ditto.
4783 (gcov_is_error): Ditto.
4784 (gcov_rewrite): Ditto.
4785 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4786 only part to libgcc/libgcov.h.
4788 2014-01-08 Marek Polacek <polacek@redhat.com>
4791 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4793 2014-01-08 Marek Polacek <polacek@redhat.com>
4796 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4798 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4800 PR rtl-optimization/59649
4801 * stor-layout.c (get_mode_bounds): For BImode return
4802 0 and STORE_FLAG_VALUE.
4804 2014-01-08 Richard Biener <rguenther@suse.de>
4807 * gimple.h (is_gimple_builtin_call): Remove.
4808 (gimple_builtin_call_types_compatible_p): New.
4809 (gimple_call_builtin_p): New overload.
4810 * gimple.c (is_gimple_builtin_call): Remove.
4811 (validate_call): Rename to ...
4812 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4814 (validate_type): New static function.
4815 (gimple_call_builtin_p): New overload and adjust.
4816 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4817 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4818 (gimple_fold_stmt_to_constant_1): Likewise.
4819 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4821 2014-01-08 Richard Biener <rguenther@suse.de>
4824 * gimplify.c (gimplify_expr): Gimplify register-register type
4825 VIEW_CONVERT_EXPRs to separate stmts.
4827 2014-01-07 Jeff Law <law@redhat.com>
4830 * ree.c (combine_set_extension): Handle case where source
4831 and destination registers in an extension insn are different.
4832 (combine_reaching_defs): Allow source and destination registers
4833 in extension to be different under limited circumstances.
4834 (add_removable_extension): Remove restriction that the
4835 source and destination registers in the extension are the same.
4836 (find_and_remove_re): Emit a copy from the extension's
4837 destination to its source after the defining insn if
4838 the source and destination registers are different.
4841 * ifcvt.c (merge_if_block): If we are merging a block with more than
4842 one successor with a block with no successors, remove any BARRIER
4843 after the second block.
4845 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4847 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4849 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4852 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4853 for 14-bit register offsets when INT14_OK_STRICT is false.
4855 2014-01-07 Roland Stigge <stigge@antcom.de>
4856 Michael Meissner <meissner@linux.vnet.ibm.com>
4859 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4860 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4862 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4864 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4867 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4869 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4870 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4871 rtx is const0_rtx or not.
4873 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4876 * target-globals.c (save_target_globals): Remove this_fn_optab
4878 * toplev.c: Include optabs.h.
4879 (target_reinit): Temporarily restore the global options if another
4880 set of options are in force.
4882 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4884 PR rtl-optimization/58668
4885 * cfgcleanup.c (flow_find_cross_jump): Don't count
4886 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4887 to determine what is counted.
4888 (flow_find_head_matching_sequence): Use active_insn_p to determine
4890 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4892 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4893 determine what is counted.
4895 PR tree-optimization/59643
4896 * tree-predcom.c (split_data_refs_to_components): If one dr is
4897 read and one write, determine_offset fails and the write isn't
4898 in the bad component, just put the read into the bad component.
4900 2014-01-07 Mike Stump <mikestump@comcast.net>
4901 Jakub Jelinek <jakub@redhat.com>
4904 * tree-core.h (struct tree_optimization_option): Change optabs
4905 type from unsigned char * to void *.
4906 * optabs.c (init_tree_optimization_optabs): Adjust
4907 TREE_OPTIMIZATION_OPTABS initialization.
4909 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4912 * config/i386/i386.h (struct machine_function): Add
4913 no_drap_save_restore field.
4914 * config/i386/i386.c (ix86_save_reg): Use
4915 !cfun->machine->no_drap_save_restore instead of
4916 crtl->stack_realign_needed.
4917 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4918 this function clears frame_pointer_needed. Set
4919 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4920 and DRAP reg is needed.
4922 2014-01-06 Marek Polacek <polacek@redhat.com>
4925 * doc/implement-c.texi: Mention that other integer types are
4926 permitted as bit-field types in strictly conforming mode.
4928 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4930 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4933 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4935 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4937 2014-01-06 Martin Jambor <mjambor@suse.cz>
4940 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4942 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4944 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4948 * var-tracking.c (add_stores): Preserve the value of the source even if
4949 we don't record the store.
4951 2014-01-06 Terry Guo <terry.guo@arm.com>
4953 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4955 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4958 * config/darwin.c (darwin_function_section): Adjust return values to
4959 correspond to optimisation changes made in r206070.
4961 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4963 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4964 from prefetch_block tune setting.
4965 (nocona_cost): Correct size of prefetch block to 64.
4967 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4969 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4970 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4971 used to save the static chain register in the computation of the offset
4972 from which the FP registers need to be restored.
4974 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4976 PR tree-optimization/59519
4977 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4978 ICE if get_current_def (current_new_name) is already non-NULL, as long
4979 as it is a phi result of some other phi in *new_exit_bb that has
4982 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4983 or vmovdqu* for misaligned_operand.
4984 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4985 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4986 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4987 aligned_mem for AVX512F masked aligned load and store builtins and for
4990 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4992 PR tree-optimization/59651
4993 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4994 Address range for negative step should be added by TYPE_SIZE_UNIT.
4996 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4998 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4999 overlapping registers also for registers other than the stack pointer.
5001 2014-01-03 Marek Polacek <polacek@redhat.com>
5004 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5007 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5010 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5013 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5014 (push splitter): Use <P:MODE_SIZE> instead of
5015 GET_MODE_SIZE (<P:MODE>mode).
5016 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5017 (mov -1, reg peephole2): Likewise.
5018 * config/i386/sse.md (*mov<mode>_internal,
5019 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5020 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5021 *<code><mode>3, *andnot<mode>3<mask_name>,
5022 <mask_codefor><code><mode>3<mask_name>): Likewise.
5023 * config/i386/subst.md (mask_mode512bit_condition,
5024 sd_mask_mode512bit_condition): Likewise.
5026 2014-01-02 Xinliang David Li <davidxl@google.com>
5028 PR tree-optimization/59303
5029 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5030 (dump_predicates): Better output format.
5031 (pred_equal_p): New function.
5032 (is_neq_relop_p): Ditto.
5033 (is_neq_zero_form_p): Ditto.
5034 (pred_expr_equal_p): Ditto.
5035 (pred_neg_p): Ditto.
5036 (simplify_pred): Ditto.
5037 (simplify_preds_2): Ditto.
5038 (simplify_preds_3): Ditto.
5039 (simplify_preds_4): Ditto.
5040 (simplify_preds): Ditto.
5042 (push_to_worklist): Ditto.
5043 (get_pred_info_from_cmp): Ditto.
5044 (is_degenerated_phi): Ditto.
5045 (normalize_one_pred_1): Ditto.
5046 (normalize_one_pred): Ditto.
5047 (normalize_one_pred_chain): Ditto.
5048 (normalize_preds): Ditto.
5049 (normalize_cond_1): Remove function.
5050 (normalize_cond): Ditto.
5051 (is_gcond_subset_of): Ditto.
5052 (is_subset_of_any): Ditto.
5053 (is_or_set_subset_of): Ditto.
5054 (is_and_set_subset_of): Ditto.
5055 (is_norm_cond_subset_of): Ditto.
5056 (pred_chain_length_cmp): Ditto.
5057 (convert_control_dep_chain_into_preds): Type change.
5058 (find_predicates): Ditto.
5059 (find_def_preds): Ditto.
5060 (destroy_predicates_vecs): Ditto.
5061 (find_matching_predicates_in_rest_chains): Ditto.
5062 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5063 (is_pred_expr_subset): Ditto.
5064 (is_pred_chain_subset_of): Ditto.
5065 (is_included_in): Ditto.
5066 (is_superset_of): Ditto.
5068 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5070 Update copyright years.
5072 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5074 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5075 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5076 config/arc/arc.md, config/arc/arc.opt,
5077 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5078 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5079 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5080 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5081 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5082 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5083 vtable-verify.c, vtable-verify.h: Use the standard form for the
5086 2014-01-02 Tobias Burnus <burnus@net-b.de>
5088 * gcc.c (process_command): Update copyright notice dates.
5089 * gcov-dump.c: Ditto.
5091 * doc/cpp.texi: Bump @copying's copyright year.
5092 * doc/cppinternals.texi: Ditto.
5093 * doc/gcc.texi: Ditto.
5094 * doc/gccint.texi: Ditto.
5095 * doc/gcov.texi: Ditto.
5096 * doc/install.texi: Ditto.
5097 * doc/invoke.texi: Ditto.
5099 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5101 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5103 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5105 * config/i386/sse.md (*mov<mode>_internal): Guard
5106 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5108 PR rtl-optimization/59647
5109 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5110 new_rtx into UNSIGNED_FLOAT rtxes.
5112 Copyright (C) 2014 Free Software Foundation, Inc.
5114 Copying and distribution of this file, with or without modification,
5115 are permitted in any medium without royalty provided the copyright
5116 notice and this notice are preserved.