1 2014-03-18 Richard Biener <rguenther@suse.de>
3 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
4 by using bitmap_and_compl instead of bitmap_and_compl_into.
5 (df_rd_transfer_function): Likewise.
7 2014-03-18 Richard Biener <rguenther@suse.de>
9 * doc/lto.texi (fresolution): Fix typo.
11 2014-03-18 Richard Biener <rguenther@suse.de>
13 * doc/invoke.texi (flto): Update for changes in 4.9.
15 2014-03-18 Richard Biener <rguenther@suse.de>
17 * doc/loop.texi: Remove section on the removed lambda framework.
18 Update loop docs with recent changes in preserving loop structure.
20 2014-03-18 Richard Biener <rguenther@suse.de>
22 * doc/lto.texi (-fresolution): Document.
24 2014-03-18 Richard Biener <rguenther@suse.de>
26 * doc/contrib.texi: Adjust my name.
28 2014-03-18 Jakub Jelinek <jakub@redhat.com>
31 * internal-fn.c: Include diagnostic-core.h.
32 (expand_BUILTIN_EXPECT): New function.
33 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
34 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
35 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
36 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
38 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
39 Revert 3 argument __builtin_expect code.
40 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
41 * gimple-fold.c (gimple_fold_call): Likewise.
42 * tree.h (fold_builtin_expect): New prototype.
43 * builtins.c (build_builtin_expect_predicate): Add predictor
44 argument, if non-NULL, create 3 argument __builtin_expect.
45 (fold_builtin_expect): No longer static. Add ARG2 argument,
46 pass it through to build_builtin_expect_predicate.
47 (fold_builtin_2): Adjust caller.
48 (fold_builtin_3): Handle BUILT_IN_EXPECT.
49 * internal-fn.def (BUILTIN_EXPECT): New.
51 2014-03-18 Tobias Burnus <burnus@net-b.de>
54 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
55 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
56 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
58 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
61 * predict.c (combine_predictions_for_bb): Fix up formatting.
62 (expr_expected_value_1, expr_expected_value): Add predictor argument,
63 fill what it points to if non-NULL.
64 (tree_predict_by_opcode): Adjust caller, use the predictor.
65 * predict.def (PRED_COMPARE_AND_SWAP): Add.
67 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
69 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
70 proper constant for the store mode.
72 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
74 * symtab.c (change_decl_assembler_name): Fix transparent alias
77 2014-03-16 Renlin Li <Renlin.Li@arm.com>
79 * config/aarch64/aarch64.c: Correct the comments about the
82 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
84 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
85 check for GF_OMP_FOR_KIND_FOR.
87 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
89 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
90 ymm and zmm register names.
92 2014-03-17 Jakub Jelinek <jakub@redhat.com>
95 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
96 note creation for the 2010-08-31 changes.
98 2014-03-17 Marek Polacek <polacek@redhat.com>
101 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
102 as -fno-tree-loop-vectorize.
103 (expand_omp_simd): Likewise.
105 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
107 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
108 (eligible_for_call_delay): New prototype.
109 * config/sparc/sparc.c (tls_call_delay): Rename into...
110 (eligible_for_call_delay): ...this. Return false if the instruction
111 cannot be put in the delay slot of a branch.
112 (eligible_for_restore_insn): Simplify.
113 (eligible_for_return_delay): Return false if the instruction cannot be
114 put in the delay slot of a branch and simplify.
115 (eligible_for_sibcall_delay): Return false if the instruction cannot be
116 put in the delay slot of a branch.
117 * config/sparc/sparc.md (fix_ut699): New attribute.
118 (tls_call_delay): Delete.
119 (in_call_delay): Reimplement.
120 (eligible_for_sibcall_delay): Rename into...
121 (in_sibcall_delay): ...this.
122 (eligible_for_return_delay): Rename into...
123 (in_return_delay): ...this.
124 (in_branch_delay): Reimplement.
125 (in_uncond_branch_delay): Delete.
126 (in_annul_branch_delay): Delete.
128 2014-03-14 Richard Henderson <rth@redhat.com>
131 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
132 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
133 (*floathi<X87MODEF>2_i387_with_temp): Remove.
134 (floathi splitters): Remove.
135 (float<SWI48x>xf2): New pattern.
136 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
137 code that tried to handle DImode for 32-bit, but which was excluded
138 by the pattern's condition. Drop allocation of stack temporary.
139 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
140 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
141 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
142 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
143 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
144 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
145 (*float<SWI48><MODEF>2_sse_interunit): Remove.
146 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
147 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
148 (*float<SWI48x><X87MODEF>2_i387): Remove.
149 (all float _with_temp splitters): Remove.
150 (*float<SWI48x><MODEF>2_i387): New pattern.
151 (*float<SWI48><MODEF>2_sse): New pattern.
152 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
153 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
155 2014-03-14 Jakub Jelinek <jakub@redhat.com>
156 Marek Polacek <polacek@redhat.com>
159 * common.opt (dump_base_name_prefixed): New Variable.
160 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
161 if x_dump_base_name_prefixed is already set, set it at the end.
163 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
165 PR rtl-optimization/60508
166 * lra-constraints.c (get_reload_reg): Add new parameter
168 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
169 Pass the new parameter values.
171 2014-03-14 Richard Biener <rguenther@suse.de>
173 * common.opt: Revert unintented changes from r205065.
176 2014-03-14 Richard Biener <rguenther@suse.de>
179 * cfghooks.c (split_block): Properly adjust all loops the
180 block was a latch of.
182 2014-03-14 Martin Jambor <mjambor@suse.cz>
185 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
188 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
191 * config/avr/avr.c (avr_set_current_function): Pass function name
192 through default_strip_name_encoding before sanity checking instead
193 of skipping the first char of the assembler name.
195 2014-03-13 Richard Henderson <rth@redhat.com>
198 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
199 (ix86_force_to_memory, ix86_free_from_memory): Remove.
200 * config/i386/i386-protos.h: Likewise.
201 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
202 in the expander instead of a splitter.
203 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
204 any possibility of requiring a memory.
205 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
206 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
207 (fp branch splitters): Update for ix86_split_fp_branch.
208 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
209 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
210 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
211 (*fop_<MODEF>_2_i387): Remove f/r alternative.
212 (*fop_<MODEF>_3_i387): Likewise.
213 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
214 (splitters for the fop_* register patterns): Remove.
215 (fscalexf4_i387): Rename from *fscalexf4_i387.
216 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
218 2014-03-13 Jakub Jelinek <jakub@redhat.com>
220 PR tree-optimization/59779
221 * tree-dfa.c (get_ref_base_and_extent): Use double_int
222 type for bitsize and maxsize instead of HOST_WIDE_INT.
224 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
226 PR rtl-optimization/57320
227 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
228 the CFG after thread_prologue_and_epilogue_insns.
230 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
232 PR rtl-optimization/57189
233 * lra-constraints.c (process_alt_operands): Disfavor spilling
236 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
238 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
241 2014-03-13 Jakub Jelinek <jakub@redhat.com>
243 PR tree-optimization/59025
245 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
246 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
248 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
251 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
252 calls of avr_out_plus_1.
254 2014-03-13 Bin Cheng <bin.cheng@arm.com>
256 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
257 BB's single pred and update the father loop's latch info later.
259 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
261 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
265 (VEC_base): Likewise.
266 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
267 registers, we need to swap double words in little endian mode.
269 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
270 to be a container mode for 128-bit integer operations added in ISA
271 2.07. Unlike TImode and PTImode, the preferred register set is
272 the Altivec/VMX registers for the 128-bit operations.
274 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
276 (rs6000_split_128bit_ok_p): Likewise.
278 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
279 macros for creating ISA 2.07 normal and overloaded builtin
280 functions with 3 arguments.
281 (BU_P8V_OVERLOAD_3): Likewise.
282 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
283 for use as overloaded functions.
284 (VPERM_1TI_UNS): Likewise.
285 (VSEL_1TI): Likewise.
286 (VSEL_1TI_UNS): Likewise.
287 (ST_INTERNAL_1ti): Likewise.
288 (LD_INTERNAL_1ti): Likewise.
289 (XXSEL_1TI): Likewise.
290 (XXSEL_1TI_UNS): Likewise.
291 (VPERM_1TI): Likewise.
292 (VPERM_1TI_UNS): Likewise.
293 (XXPERMDI_1TI): Likewise.
295 (LXVD2X_V1TI): Likewise.
296 (STXVD2X_V1TI): Likewise.
297 (VEC_INIT_V1TI): Likewise.
298 (VEC_SET_V1TI): Likewise.
299 (VEC_EXT_V1TI): Likewise.
300 (EQV_V1TI): Likewise.
301 (NAND_V1TI): Likewise.
302 (ORC_V1TI): Likewise.
303 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
304 added in ISA 2.07. Add both normal 'altivec' builtins, and the
308 (VADDEUQM): Likewise.
309 (VADDECUQ): Likewise.
310 (VSUBEUQM): Likewise.
311 (VSUBECUQ): Likewise.
313 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
314 __int128_t and __uint128_t types.
315 (__uint128_type): Likewise.
316 (altivec_categorize_keyword): Add support for vector __int128_t,
317 vector __uint128_t, vector __int128, and vector unsigned __int128
318 as a container type for TImode operations that need to be done in
319 VSX/Altivec registers.
320 (rs6000_macro_to_expand): Likewise.
321 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
322 to support 128-bit integer instructions vaddcuq, vadduqm,
323 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
324 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
326 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
327 for V1TImode, and set up preferences to use VSX/Altivec registers.
328 Setup VSX reload handlers.
329 (rs6000_debug_reg_global): Likewise.
330 (rs6000_init_hard_regno_mode_ok): Likewise.
331 (rs6000_preferred_simd_mode): Likewise.
332 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
333 (easy_altivec_constant): Likewise.
334 (output_vec_const_move): Likewise.
335 (rs6000_expand_vector_set): Convert V1TImode set and extract to
337 (rs6000_expand_vector_extract): Likewise.
338 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
340 (rs6000_const_vec): Add support for V1TImode.
341 (rs6000_emit_le_vsx_load): Swap double words when loading or
342 storing TImode/V1TImode.
343 (rs6000_emit_le_vsx_store): Likewise.
344 (rs6000_emit_le_vsx_move): Likewise.
345 (rs6000_emit_move): Add support for V1TImode.
346 (altivec_expand_ld_builtin): Likewise.
347 (altivec_expand_st_builtin): Likewise.
348 (altivec_expand_vec_init_builtin): Likewise.
349 (altivec_expand_builtin): Likewise.
350 (rs6000_init_builtins): Add support for V1TImode type. Add
351 support for ISA 2.07 128-bit integer builtins. Define type names
352 for the VSX/Altivec vector types.
353 (altivec_init_builtins): Add support for overloaded vector
354 functions with V1TImode type.
355 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
356 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
358 (rs6000_split_128bit_ok_p): Likewise.
359 (rs6000_handle_altivec_attribute): Create V1TImode from vector
360 __int128_t and vector __uint128_t.
362 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
370 (VS_scalar): Likewise.
371 (VS_double): Likewise.
372 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
374 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
375 we support the ISA 2.07 128-bit integer arithmetic instructions.
376 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
377 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
378 and TImode types for use with the builtin functions.
379 (V1TI_type_node): Likewise.
380 (unsigned_V1TI_type_node): Likewise.
381 (intTI_type_internal_node): Likewise.
382 (uintTI_type_internal_node): Likewise.
384 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
385 128-bit builtin functions.
386 (UNSPEC_VADDEUQM): Likewise.
387 (UNSPEC_VADDECUQ): Likewise.
388 (UNSPEC_VSUBCUQ): Likewise.
389 (UNSPEC_VSUBEUQM): Likewise.
390 (UNSPEC_VSUBECUQ): Likewise.
391 (VM): Add V1TImode to vector mode iterators.
394 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
395 (altivec_vaddcuq): Likewise.
396 (altivec_vsubuqm): Likewise.
397 (altivec_vsubcuq): Likewise.
398 (altivec_vaddeuqm): Likewise.
399 (altivec_vaddecuq): Likewise.
400 (altivec_vsubeuqm): Likewise.
401 (altivec_vsubecuq): Likewise.
403 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
405 (BOOL_128): Likewise.
406 (BOOL_REGS_OUTPUT): Likewise.
407 (BOOL_REGS_OP1): Likewise.
408 (BOOL_REGS_OP2): Likewise.
409 (BOOL_REGS_UNARY): Likewise.
410 (BOOL_REGS_AND_CR0): Likewise.
412 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
413 128-bit integer builtin support.
414 (vec_vadduqm): Likewise.
415 (vec_vaddecuq): Likewise.
416 (vec_vaddeuqm): Likewise.
417 (vec_vsubecuq): Likewise.
418 (vec_vsubeuqm): Likewise.
419 (vec_vsubcuq): Likewise.
420 (vec_vsubuqm): Likewise.
422 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
423 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
424 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
425 128-bit integer add/subtract to ISA 2.07.
427 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
429 * config/arc/arc.c (arc_predicate_delay_insns):
430 Fix third argument passed to conditionalize_nonjump.
432 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
434 * config/aarch64/aarch64-builtins.c
435 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
436 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
437 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
438 instead of __builtin_lfloor.
439 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
441 2014-03-12 Jakub Jelinek <jakub@redhat.com>
443 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
444 (tree_ssa_ifcombine_bb_1): New function.
445 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
446 is an empty forwarder block to then_bb or vice versa and then_bb
447 and else_bb are effectively swapped.
449 2014-03-12 Christian Bruel <christian.bruel@st.com>
452 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
453 REG_CFA_DEF_CFA note.
454 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
455 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
457 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
459 PR tree-optimization/60454
460 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
462 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
464 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
465 Do not define target_cpu_default2 to generic.
466 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
467 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
468 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
470 2014-03-12 Jakub Jelinek <jakub@redhat.com>
471 Marc Glisse <marc.glisse@inria.fr>
473 PR tree-optimization/60502
474 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
475 instead of build_low_bits_mask.
477 2014-03-12 Jakub Jelinek <jakub@redhat.com>
480 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
481 if there are multiple uses, but op doesn't live on E edge.
482 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
483 clobber stmts before __builtin_unreachable.
485 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
487 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
488 hard_frame_pointer_rtx.
489 * cse.c (cse_insn): Remove volatile check.
490 * cselib.c (cselib_process_insn): Likewise.
491 * dse.c (scan_insn): Likewise.
493 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
495 * config/arc/arc.c (conditionalize_nonjump): New function,
497 (arc_ifcvt): ... this.
498 (arc_predicate_delay_insns): Use it.
500 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
502 * config/arc/predicates.md (extend_operand): During/after reload,
503 allow const_int_operand.
504 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
505 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
506 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
508 (umulsi3_highpart_i): Likewise.
510 2014-03-11 Richard Biener <rguenther@suse.de>
512 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
513 Add asserts to guard possible wrong-code bugs.
515 2014-03-11 Richard Biener <rguenther@suse.de>
517 PR tree-optimization/60429
518 PR tree-optimization/60485
519 * tree-ssa-structalias.c (set_union_with_increment): Properly
520 take into account all fields that overlap the shifted vars.
521 (do_sd_constraint): Likewise.
522 (do_ds_constraint): Likewise.
523 (get_constraint_for_ptr_offset): Likewise.
525 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
527 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
528 (nios2_compute_frame_layout):
529 Add calculation of cfun->machine->fp_save_offset.
530 (nios2_expand_prologue): Correct setting of frame pointer register
532 (nios2_expand_epilogue): Update recovery of stack pointer from
533 frame pointer accordingly.
534 (nios2_initial_elimination_offset): Update calculation of offset
535 for eliminating to HARD_FRAME_POINTER_REGNUM.
537 2014-03-10 Jakub Jelinek <jakub@redhat.com>
540 * ipa.c (symtab_remove_unreachable_nodes): Don't call
541 cgraph_get_create_node on VAR_DECLs.
543 2014-03-10 Richard Biener <rguenther@suse.de>
546 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
548 2014-03-08 Douglas B Rupp <rupp@gnat.com>
550 * config/vms/vms.opt (vms_float_format): New variable.
552 2014-03-08 Tobias Burnus <burnus@net-b.de>
554 * doc/invoke.texi (-fcilkplus): Update implementation status.
556 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
557 Richard Biener <rguenther@suse.de>
559 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
560 consistently accross all TUs.
561 (run_gcc): Enable -fshort-double automatically at link at link-time
562 and disallow override.
564 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
567 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
568 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
569 if they can't be used.
571 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
573 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
574 for Solaris 11/x86 ld.
575 * configure: Regenerate.
577 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
579 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
580 (LIB_TLS_SPEC): Save as ld_tls_libs.
581 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
582 (HAVE_AS_IX86_TLSLDM): New test.
583 * configure, config.in: Regenerate.
584 * config/i386/i386.c (legitimize_tls_address): Fall back to
585 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
586 cannot support TLS_MODEL_LOCAL_DYNAMIC.
587 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
588 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
590 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
592 * common.opt (fira-loop-pressure): Mark as optimization.
594 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
596 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
597 an OpenMP mappable type.
599 2014-03-06 Matthias Klose <doko@ubuntu.com>
601 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
602 MULTILIB_OSDIRNAMES is not defined.
604 2014-03-06 Jakub Jelinek <jakub@redhat.com>
605 Meador Inge <meadori@codesourcery.com>
608 * config/arm/arm.c (arm_tls_symbol_p): Remove.
609 (arm_legitimize_address): Call legitimize_tls_address for any
610 arm_tls_referenced_p expression, handle constant addend. Call it
611 before testing for !TARGET_ARM.
612 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
614 2014-03-06 Richard Biener <rguenther@suse.de>
620 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
622 * tree-streamer.c (record_common_node): Assert we don't record
623 nodes with type double.
624 (preload_common_node): Skip type double, complex double and double
625 pointer since it is now frontend dependent due to fshort-double option.
627 2014-03-06 Richard Biener <rguenther@suse.de>
629 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
630 or -fno-lto is specified and the linker has full plugin support.
631 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
632 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
633 * lto-wrapper.c (merge_and_complain): Merge compile-time
635 (run_gcc): And pass it through to the link options.
637 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
641 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
643 * cselib.c (remove_useless_values): Skip to avoid quadratic
644 behavior if the condition moved from...
645 (cselib_process_insn): ... here holds.
647 2014-03-05 Jakub Jelinek <jakub@redhat.com>
650 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
651 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
654 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
655 (TM_H): Add x86-tune.def.
657 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
659 * config/aarch64/aarch64.c (generic_tunings):
660 Use cortexa57_extra_costs.
662 2014-03-05 Jakub Jelinek <jakub@redhat.com>
665 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
666 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
667 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
670 2014-03-04 Heiher <r@hev.cc>
672 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
673 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
675 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
677 * config/i386/predicates.md (const2356_operand): Change to ...
678 (const2367_operand): ... this.
679 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
681 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
682 (*avx512pf_scatterpf<mode>sf): Ditto.
683 (avx512pf_scatterpf<mode>df): Ditto.
684 (*avx512pf_scatterpf<mode>df_mask): Ditto.
685 (*avx512pf_scatterpf<mode>df): Ditto.
686 * config/i386/i386.c (ix86_expand_builtin): Update
687 incorrect hint operand error message.
689 2014-03-04 Richard Biener <rguenther@suse.de>
691 * lto-section-in.c (lto_get_section_data): Fix const cast.
693 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
695 * tree-streamer.c (record_common_node): Assert we don't record
696 nodes with type double.
697 (preload_common_node): Skip type double, complex double and double
698 pointer since it is now frontend dependent due to fshort-double option.
700 2014-03-04 Richard Biener <rguenther@suse.de>
703 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
704 (lto_input_toplevel_asms): Likewise.
705 * lto-section-in.c (lto_get_section_data): Instead do it here
708 2014-03-04 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/60382
711 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
712 dead PHIs a reduction.
714 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
716 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
718 (_mm_prefetch): Move out of GCC target("sse") pragma.
719 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
720 GCC target("prfchw") pragma.
721 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
723 * config/i386/i386.c (ix86_option_override_internal): Enable
724 -mprfchw with -mprefetchwt1.
726 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
728 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
731 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
733 * opts.h (CL_PCH_IGNORE): Define.
734 * targhooks.c (option_affects_pch_p):
735 Return false for options that have CL_PCH_IGNORE set.
736 * opt-functions.awk: Process PchIgnore.
737 * doc/options.texi: Document PchIgnore.
739 * config/arc/arc.opt (misize): Add PchIgnore property.
741 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
743 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
744 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
745 constraint on constants to permit them being loaded into
746 GENERAL_REGS or BASE_REGS.
748 2014-03-03 Nick Clifton <nickc@redhat.com>
750 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
751 anti-cacnonical alternatives.
752 (negandhi3_real): New pattern.
753 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
755 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
757 * config/avr/avr-mcus.def: Remove atxmega16x1.
758 * config/avr/avr-tables.opt: Regenerate.
759 * config/avr/t-multilib: Regenerate.
760 * doc/avr-mmcu.texi: Regenerate.
762 2014-03-03 Tobias Grosser <tobias@grosser.es>
763 Mircea Namolaru <mircea.namolaru@inria.fr>
765 PR tree-optimization/58028
766 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
769 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
771 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
772 not handled by recognizers.
774 2014-03-03 Jakub Jelinek <jakub@redhat.com>
777 * function.c (expand_function_end): Don't emit
778 clobber_return_register sequence if clobber_after is a BARRIER.
779 * cfgexpand.c (construct_exit_block): Append instructions before
780 return_label to prev_bb.
782 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
784 * config/rs6000/constraints.md: Document reserved use of "wc".
786 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
789 * ipa.c (function_and_variable_visibility): When dissolving comdat
790 group, also set all symbols to local.
792 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
797 2013-12-14 Jan Hubicka <jh@suse.cz>
799 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
801 2014-03-02 Jon Beniston <jon@beniston.com>
807 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
808 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
809 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
810 (simple_return, *simple_return): New patterns
811 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
812 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
814 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
816 * dwarf2out.c (gen_subprogram_die): Tidy.
818 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
821 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
822 (*mov_t_msb_neg_negc): ... this new insn.
824 2014-02-28 Jason Merrill <jason@redhat.com>
827 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
830 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
833 * dwarf2out.c (decltype_auto_die): New static.
834 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
835 (gen_type_die_with_usage): Handle 'decltype(auto)'.
836 (is_cxx_auto): Likewise.
838 2014-02-28 Ian Bolton <ian.bolton@arm.com>
840 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
841 we are not using general regs only.
843 2014-02-28 Richard Biener <rguenther@suse.de>
846 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
847 previous fix and only allow to remove trivial pre-headers
848 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
849 (remove_forwarder_block): Properly update the latch of a loop.
851 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
854 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
855 (cselib_preserved_hash_table): New.
856 (preserve_constants_and_equivs): Move preserved vals to it.
857 (cselib_find_slot): Look it up first.
858 (cselib_init): Initialize it.
859 (cselib_finish): Release it.
860 (dump_cselib_table): Dump it.
862 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
865 * cselib.c (remove_useless_values): Skip to avoid quadratic
866 behavior if the condition moved from...
867 (cselib_process_insn): ... here holds.
869 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
872 * var-tracking.c (vt_initialize): Apply the same condition to
873 preserve the CFA base value.
875 2014-02-28 Joey Ye <joey.ye@arm.com>
878 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
879 if reload in progress or completed.
881 2014-02-28 Tobias Burnus <burnus@net-b.de>
884 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
887 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
889 * doc/tm.texi.in (Condition Code Status): Update documention for
890 relative locations of cc0-setter and cc0-user.
892 2014-02-27 Jeff Law <law@redhat.com>
894 PR rtl-optimization/52714
895 * combine.c (try_combine): When splitting an unrecognized PARALLEL
896 into two independent simple sets, if I3 is a jump, ensure the
897 pattern we place into I3 is a (set (pc) ...).
899 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
900 Jeff Law <law@redhat.com>
902 PR rtl-optimization/49847
903 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
904 are in different blocks.
905 * doc/tm.texi (Condition Code Status): Update documention for
906 relative locations of cc0-setter and cc0-user.
908 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
911 * lra.c (lra_emit_add): Check SUBREG too.
913 2014-02-27 Andreas Schwab <schwab@suse.de>
915 * config/m68k/m68k.c (m68k_option_override): Disable
916 -flive-range-shrinkage for classic m68k.
917 (m68k_override_options_after_change): Likewise.
919 2014-02-27 Marek Polacek <polacek@redhat.com>
922 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
923 -Wmaybe-uninitialized.
925 2014-02-27 Alan Modra <amodra@gmail.com>
928 * reload1.c (emit_input_reload_insns): When reload_override_in,
929 set old to rl->in_reg when rl->in_reg is a subreg.
931 2014-02-26 Richard Biener <rguenther@suse.de>
934 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
936 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
938 * common/config/i386/predicates.md (const1256_operand): Remove.
939 (const2356_operand): New.
940 (const_1_to_2_operand): Remove.
941 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
942 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
943 (*avx512pf_gatherpf<mode>sf): Ditto.
944 (avx512pf_gatherpf<mode>df): Ditto.
945 (*avx512pf_gatherpf<mode>df_mask): Ditto.
946 (*avx512pf_gatherpf<mode>df): Ditto.
947 (avx512pf_scatterpf<mode>sf): Ditto.
948 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
949 (*avx512pf_scatterpf<mode>sf): Ditto.
950 (avx512pf_scatterpf<mode>df): Ditto.
951 (*avx512pf_scatterpf<mode>df_mask): Ditto.
952 (*avx512pf_scatterpf<mode>df): Ditto.
953 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
955 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
957 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
958 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
959 (_mm512_mask_testn_epi64_mask): Move to ...
960 * config/i386/avx512cdintrin.h: Here.
961 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
962 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
963 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
964 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
965 TARGET_AVX512F from TARGET_AVX512CD.
967 2014-02-26 Richard Biener <rguenther@suse.de>
970 * ipa.c (walk_polymorphic_call_targets): Properly guard
971 call to inline_update_overall_summary.
973 2014-02-26 Bin Cheng <bin.cheng@arm.com>
976 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
977 and latches only if requested. Fix latch if it is removed.
978 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
979 LOOPS_HAVE_PREHEADERS.
981 2014-02-25 Andrew Pinski <apinski@cavium.com>
983 * builtins.c (expand_builtin_thread_pointer): Create a new target
984 when the target is NULL.
986 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
988 PR rtl-optimization/60317
989 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
990 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
991 * lra-assigns.c: Include params.h.
992 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
993 other reload pseudos considerations.
995 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
997 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
998 to use canonical form for nor<mode>3.
1000 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1003 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1006 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1008 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1009 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1010 (ix86_handle_option): Handle OPT_mprefetchwt1.
1011 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1012 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1014 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1015 OPTION_MASK_ISA_PREFETCHWT1.
1016 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1017 (PTA_PREFETCHWT1): New.
1018 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1019 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1020 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1021 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1022 (*prefetch_avx512pf_<mode>_: Change into ...
1023 (*prefetch_prefetchwt1_<mode>: This.
1024 * config/i386/i386.opt (mprefetchwt1): New.
1025 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1026 (_mm_prefetch): Handle intent to write.
1027 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1029 2014-02-25 Richard Biener <rguenther@suse.de>
1032 * emit-rtl.c (mem_attrs_htab): Remove.
1033 (mem_attrs_htab_hash): Likewise.
1034 (mem_attrs_htab_eq): Likewise.
1035 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1036 (init_emit_once): Do not allocate mem_attrs_htab.
1038 2014-02-25 Richard Biener <rguenther@suse.de>
1041 * lto-opts.c (lto_write_options): Output non-explicit conservative
1042 -fwrapv, -fno-trapv and -fno-strict-overflow.
1043 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1044 (run_gcc): And pass them through.
1046 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1048 * sel-sched.c (calculate_new_fences): New parameter ptime.
1049 Calculate it as a maximum over all fence cycles.
1050 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1051 Print the final schedule timing when sched_verbose.
1053 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1055 PR rtl-optimization/60292
1056 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1057 bit fot the fence instruction.
1059 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1061 * calls.h: Fix typo in comment.
1063 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1065 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1066 adjusting offsetable addresses.
1068 2014-02-24 Guozhi Wei <carrot@google.com>
1070 * sparseset.h (sparseset_pop): Fix the wrong index.
1072 2014-02-24 Walter Lee <walt@tilera.com>
1074 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1075 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1077 * common/config/tilegx/tilegx-common.c
1078 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1079 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1081 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1082 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1083 (tilegx_gimplify_va_arg_expr): Handle big endian.
1084 (tilegx_expand_unaligned_load): Ditto.
1085 (tilegx_expand_unaligned_store): Ditto.
1086 (TARGET_RETURN_IN_MSB): New.
1087 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1088 (TARGET_ENDIAN_DEFAULT): New.
1089 (TARGET_BIG_ENDIAN): Handle big endian.
1090 (BYTES_BIG_ENDIAN): Ditto.
1091 (WORDS_BIG_ENDIAN): Ditto.
1092 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1095 * config/tilegx/tilegx.md (extv): Handle big endian.
1097 (insn_st<n>): Ditto.
1098 (insn_st<n>_add<bitsuffix>): Ditto.
1099 (insn_stnt<n>): Ditto.
1100 (insn_stnt<n>_add<bitsuffix>):Ditto.
1101 (vec_interleave_highv8qi): Handle big endian.
1102 (vec_interleave_highv8qi_be): New.
1103 (vec_interleave_highv8qi_le): New.
1104 (insn_v1int_h): Handle big endian.
1105 (vec_interleave_lowv8qi): Handle big endian.
1106 (vec_interleave_lowv8qi_be): New.
1107 (vec_interleave_lowv8qi_le): New.
1108 (insn_v1int_l): Handle big endian.
1109 (vec_interleave_highv4hi): Handle big endian.
1110 (vec_interleave_highv4hi_be): New.
1111 (vec_interleave_highv4hi_le): New.
1112 (insn_v2int_h): Handle big endian.
1113 (vec_interleave_lowv4hi): Handle big endian.
1114 (vec_interleave_lowv4hi_be): New.
1115 (vec_interleave_lowv4hi_le): New.
1116 (insn_v2int_l): Handle big endian.
1117 (vec_interleave_highv2si): Handle big endian.
1118 (vec_interleave_highv2si_be): New.
1119 (vec_interleave_highv2si_le): New.
1120 (insn_v4int_h): Handle big endian.
1121 (vec_interleave_lowv2si): Handle big endian.
1122 (vec_interleave_lowv2si_be): New.
1123 (vec_interleave_lowv2si_le): New.
1124 (insn_v4int_l): Handle big endian.
1125 * config/tilegx/tilegx.opt (mbig-endian): New option.
1126 (mlittle-endian): New option.
1127 * doc/install.texi: Document tilegxbe-linux.
1128 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1130 2014-02-24 Martin Jambor <mjambor@suse.cz>
1133 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1134 there are no parameter descriptors.
1136 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1138 PR rtl-optimization/60268
1139 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1140 initialization to ...
1141 (sched_rgn_init): ... here.
1142 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1144 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1146 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1149 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1151 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1154 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1156 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1157 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1159 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1161 * config/microblaze/predicates.md: Add cmp_op predicate.
1162 * config/microblaze/microblaze.md: Add branch_compare instruction
1163 which uses cmp_op predicate and emits cmp insn before branch.
1164 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1165 to microblaze_expand_conditional_branch and consolidate logic.
1166 (microblaze_expand_conditional_branch): emit branch_compare
1167 insn instead of handling cmp op separate from branch insn.
1169 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1171 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1174 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1176 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1177 define_insn with define_expand and new define_insn
1178 *altivec_lve<VI_char>x_internal.
1179 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1180 and new define_insn *altivec_stve<VI_char>x_internal.
1181 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1183 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1185 (altivec_expand_stvex_be): New function.
1187 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1189 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1190 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1191 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1192 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1194 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1197 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1198 instead of emit_move_insn.
1200 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1202 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1204 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1205 gen_altivec_vsumsws.
1207 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1209 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1210 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1211 (altivec_lvxl_<mode>): New define_expand incorporating
1212 -maltivec=be semantics where needed.
1213 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1214 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1215 semantics where needed.
1216 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1217 (altivec_stvx_<mode>): New define_expand incorporating
1218 -maltivec=be semantics where needed.
1219 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1220 VM2 iterator instead of V4SI.
1221 (altivec_stvxl_<mode>): New define_expand incorporating
1222 -maltivec=be semantics where needed.
1223 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1224 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1225 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1226 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1227 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1228 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1229 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1230 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1231 ALTIVEC_BUILTIN_STVXL.
1232 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1233 (altivec_expand_stvx_be): Likewise.
1234 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1235 (altivec_expand_lvx_be): Likewise.
1236 (altivec_expand_stvx_be): Likewise.
1237 (altivec_expand_builtin): Add cases for
1238 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1239 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1240 (altivec_init_builtins): Add definitions for
1241 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1242 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1244 2014-02-21 Catherine Moore <clm@codesourcery.com>
1246 * doc/invoke.texi (mvirt, mno-virt): Document.
1247 * config/mips/mips.opt (mvirt): New option.
1248 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1250 2014-02-21 Richard Biener <rguenther@suse.de>
1252 PR tree-optimization/60276
1253 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1254 (STMT_VINFO_MIN_NEG_DIST): New macro.
1255 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1256 STMT_VINFO_MIN_NEG_DIST.
1257 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1258 made for negative dependence distances still hold.
1260 2014-02-21 Richard Biener <rguenther@suse.de>
1263 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1264 DECL_INITIAL for globals not in the current function context.
1266 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1268 PR tree-optimization/56490
1269 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1270 * tree-ssa-uninit.c: Include params.h.
1271 (compute_control_dep_chain): Add num_calls argument, return false
1272 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1273 num_calls to recursive call.
1274 (find_predicates): Change dep_chain into normal array,
1275 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1276 variable and adjust compute_control_dep_chain caller.
1277 (find_def_preds): Likewise.
1279 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1281 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1282 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1284 2014-02-21 Nick Clifton <nickc@redhat.com>
1286 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1287 (pushhi1): Likewise.
1288 (popqi1): Add mode to pre_dec.
1291 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1293 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1294 mode for mask of V8SFmode permutation.
1296 2014-02-20 Richard Henderson <rth@redhat.com>
1299 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1300 a new pseudo for OLDVAL.
1302 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1305 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1306 gen_reg_rtx if d->testing_p.
1307 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1308 if d->testing_p and we will certainly return true.
1309 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1312 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1314 * emit-rtl.c (gen_reg_rtx): Assert that
1315 crtl->emit.regno_pointer_align_length is non-zero.
1317 2014-02-20 Richard Henderson <rth@redhat.com>
1320 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1321 on failure the store back into EXPECT.
1323 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1324 Sandra Loosemore <sandra@codesourcery.com>
1326 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1327 * config/nios2/nios2.c (nios2_function_profiler): Add
1328 -fPIC (flag_pic == 2) support.
1329 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1330 (nios2_large_offset_p): New function.
1331 (nios2_unspec_reloc_p): Move up position, update to use
1332 nios2_large_offset_p.
1333 (nios2_unspec_address): Remove function.
1334 (nios2_unspec_offset): New function.
1335 (nios2_large_got_address): New function.
1336 (nios2_got_address): Add large offset support.
1337 (nios2_legitimize_tls_address): Update usage of removed and new
1339 (nios2_symbol_binds_local_p): New function.
1340 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1341 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1342 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1343 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1344 processing for (const (unspec ...)).
1345 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1347 2014-02-20 Richard Biener <rguenther@suse.de>
1349 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1350 doing the substitution.
1351 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1353 2014-02-20 Martin Jambor <mjambor@suse.cz>
1356 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1357 info when checking whether lattices are bottom.
1359 2014-02-20 Richard Biener <rguenther@suse.de>
1362 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1365 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1368 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1369 parameter specifying the scaling.
1370 (inline_call): Update.
1371 (want_inline_recursively): Guard division by zero.
1372 (recursive_inlining): Update.
1373 * ipa-inline.h (clone_inlined_nodes): Update.
1375 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1378 * config/i386/i386.c (classify_argument): Pass structures of size
1379 64 bytes or less in register.
1381 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1382 Kirill Yukhin <kirill.yukhin@intel.com>
1384 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1385 (_mm_rcp28_round_ss): Ditto.
1386 (_mm_rsqrt28_round_sd): Ditto.
1387 (_mm_rsqrt28_round_ss): Ditto.
1388 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1389 (_mm_rcp14_round_ss): Ditto.
1390 (_mm_rsqrt14_round_sd): Ditto.
1391 (_mm_rsqrt14_round_ss): Ditto.
1392 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1393 the first input operand, get rid of match_dup.
1394 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1396 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1398 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1399 operand as the first input operand, set type attribute.
1400 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1402 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1403 operand as the first input operand, set type attribute.
1405 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1407 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1410 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1413 * config/i386/i386.c (construct_container): Remove TFmode check
1414 for X86_64_INTEGER_CLASS.
1416 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1419 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1420 only when -Wpsabi is enabled.
1422 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1425 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1426 passing arrays in registers are the same as for structs, so remove the
1427 special case for them.
1429 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1431 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1432 destination type, extract only the valid bits if the source type is not
1433 integral and has a different mode.
1435 2014-02-19 Richard Biener <rguenther@suse.de>
1438 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1441 2014-02-19 Richard Biener <rguenther@suse.de>
1444 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1445 (ipa_modify_call_arguments): Emit an argument load explicitely and
1446 preserve virtual SSA form there and for the replacement call.
1447 Do not update SSA form nor free dominance info.
1449 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1451 * ipa.c (function_and_variable_visibility): Also clear WEAK
1452 flag when disolving COMDAT_GROUP.
1454 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1456 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1457 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1459 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1460 do more sanity checks.
1461 (detect_type_change): Return true when giving up early.
1462 (compute_complex_assign_jump_func): Fix type parameter of
1463 ipa_set_ancestor_jf.
1464 (compute_complex_ancestor_jump_func): Likewise.
1465 (update_jump_functions_after_inlining): Fix updating of
1467 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1469 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1471 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1472 inline clones when edge disappears.
1474 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1477 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1478 Split 64-bit moves into 2 patterns. Do not allow the use of
1479 direct move for TDmode in little endian, since the decimal value
1480 has little endian bytes within a word, but the 64-bit pieces are
1481 ordered in a big endian fashion, and normal subreg's of TDmode are
1483 (mov<mode>_64bit_dm): Likewise.
1484 (movtd_64bit_nodm): Likewise.
1486 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1488 PR tree-optimization/60174
1489 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1490 statement of an SSA_NAME that occurs in an abnormal PHI node.
1492 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1495 * final.c (SEEN_BB): Remove.
1496 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1497 (final_scan_insn): Don't force_source_line on second
1498 NOTE_INSN_BASIC_BLOCK.
1500 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1503 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1504 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1505 (type_natural_mode): Warn ABI change when %zmm register is not
1506 available for AVX512F vector value passing.
1508 2014-02-18 Kai Tietz <ktietz@redhat.com>
1511 * config/i386/i386.c (ix86_expand_prologue): Use value in
1512 rax register as displacement when restoring %r10 or %rax.
1513 Fix wrong offset when restoring both registers.
1515 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1517 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1518 assertion with conditional return.
1520 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1521 Uros Bizjak <ubizjak@gmail.com>
1524 * config/i386/driver-i386.c (host_detect_local_cpu): If
1525 YMM state is not saved by the OS, also clear has_f16c. Move
1526 CPUID 0x80000001 handling before YMM state saving checking.
1528 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1530 PR rtl-optimization/58960
1531 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1532 factored out from ...
1533 (sched_init): ... here.
1534 (free_global_sched_pressure_data): New, factored out from ...
1535 (sched_finish): ... here.
1536 * sched-int.h (free_global_sched_pressure_data): Declare.
1537 * sched-rgn.c (nr_regions_initial): New static global.
1538 (haifa_find_rgns): Initialize it.
1539 (schedule_region): Disable sched-pressure for the newly
1542 2014-02-17 Richard Biener <rguenther@suse.de>
1544 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1545 release SSA defs of pattern stmts.
1547 2014-02-17 Richard Biener <rguenther@suse.de>
1549 * tree-inline.c (expand_call_inline): Release the virtual
1550 operand defined by the call we are about to inline.
1552 2014-02-17 Richard Biener <rguenther@suse.de>
1554 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1556 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1557 Ilya Tocar <ilya.tocar@intel.com>
1559 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1560 arguments order in builtin.
1561 (_mm512_permutexvar_epi64): Ditto.
1562 (_mm512_mask_permutexvar_epi64): Ditto
1563 (_mm512_maskz_permutexvar_epi32): Ditto
1564 (_mm512_permutexvar_epi32): Ditto
1565 (_mm512_mask_permutexvar_epi32): Ditto
1567 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1569 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1570 (p8_vmrgow): Likewise.
1572 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1574 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1577 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1580 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1581 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1582 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1583 using direct move instructions on ISA 2.07. Also adjust
1584 instruction length for 64-bit.
1585 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1586 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1588 2014-02-15 Alan Modra <amodra@gmail.com>
1592 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1593 find_replacement on parts of insn rtl that might be reloaded.
1595 2014-02-15 Richard Biener <rguenther@suse.de>
1597 PR tree-optimization/60183
1598 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1599 (tree_ssa_phiprop): Calculate and free post-dominators.
1601 2014-02-14 Jeff Law <law@redhat.com>
1603 PR rtl-optimization/60131
1604 * ree.c (get_extended_src_reg): New function.
1605 (combine_reaching_defs): Use it rather than assuming location of REG.
1606 (find_and_remove_re): Verify first operand of extension is
1607 a REG before adding the insns to the copy list.
1609 2014-02-14 Roland McGrath <mcgrathr@google.com>
1611 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1612 * configure: Regenerated.
1613 * config.in: Regenerated.
1614 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1615 instead of ASM_SHORT.
1617 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1618 Richard Earnshaw <rearnsha@arm.com>
1620 PR rtl-optimization/59535
1621 * lra-constraints.c (process_alt_operands): Encourage alternative
1622 when unassigned pseudo class is superset of the alternative class.
1623 (inherit_reload_reg): Don't inherit when optimizing for code size.
1624 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1625 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1626 modes not less than 4 for Thumb1.
1628 2014-02-14 Kyle McMartin <kyle@redhat.com>
1631 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1633 2014-02-14 Richard Biener <rguenther@suse.de>
1635 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1636 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1637 Do not strip INDIRECT_REFs.
1639 2014-02-14 Richard Biener <rguenther@suse.de>
1642 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1643 DECL_FUNCTION_SPECIFIC_TARGET.
1644 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1645 * tree-streamer-out.c (pack_ts_target_option): Remove.
1646 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1647 (write_ts_function_decl_tree_pointers): Do not stream
1648 DECL_FUNCTION_SPECIFIC_TARGET.
1649 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1650 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1651 (lto_input_ts_function_decl_tree_pointers): Do not stream
1652 DECL_FUNCTION_SPECIFIC_TARGET.
1654 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1656 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1657 (get_initial_def_for_induction, vectorizable_induction): Ignore
1658 debug stmts when looking for exit_phi.
1659 (vectorizable_live_operation): Fix up condition.
1661 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1663 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1664 nreverse() because it changes the content of original tree list.
1666 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1668 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1669 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1671 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1673 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1674 GNU coding standards.
1676 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1679 * dwarf2out.c (gen_subprogram_die): Don't call
1680 add_calling_convention_attribute if subr_die is old_die.
1682 2014-02-13 Sharad Singhai <singhai@google.com>
1684 * doc/optinfo.texi: Fix order of nodes.
1686 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1688 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1689 operands[2], not operands[3].
1691 2014-02-13 Richard Biener <rguenther@suse.de>
1694 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1695 mention the possibility of an in-tree build.
1696 (CLooG): Update recommended version to 0.18.1, mention the
1697 possibility of an in-tree build and clarify that the ISL
1698 bundled with CLooG does not work.
1700 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1703 * expr.c (compress_float_constant): If x is a hard register,
1704 extend into a pseudo and then move to x.
1706 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1708 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1709 caused by bad second argument to warning_at() with -mhotpatch and
1710 nested functions (e.g. with gfortran).
1712 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1714 * opts.c (option_name): Remove "enabled by default" rider.
1716 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1718 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1720 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1721 Uros Bizjak <ubizjak@gmail.com>
1724 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1725 * configure: Regenerated.
1727 2014-02-12 Richard Biener <rguenther@suse.de>
1729 * vec.c (vec_prefix::calculate_allocation): Move as
1730 inline variant to vec.h.
1731 (vec_prefix::calculate_allocation_1): New out-of-line version.
1732 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1733 (vec_prefix::m_has_auto_buf): Rename to ...
1734 (vec_prefix::m_using_auto_storage): ... this.
1735 (vec_prefix::calculate_allocation): Inline the easy cases
1736 and dispatch to calculate_allocation_1 which doesn't need the
1738 (va_heap::reserve): Use gcc_checking_assert.
1739 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1740 m_using_auto_storage.
1741 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1743 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1744 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1745 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1747 2014-02-12 Richard Biener <rguenther@suse.de>
1749 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1750 when we found a dependence.
1752 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1754 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1756 (maybe_fold_stmt): ... into this new function.
1757 * omp-low.c (lower_omp): Update comment.
1759 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1762 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1765 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1767 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1768 identifiers in comments.
1769 (cortexa53_extra_costs): Likewise.
1770 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1771 (cortexa7_extra_costs): Likewise.
1772 (cortexa12_extra_costs): Likewise.
1773 (cortexa15_extra_costs): Likewise.
1774 (v7m_extra_costs): Likewise.
1776 2014-02-12 Richard Biener <rguenther@suse.de>
1779 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1780 of posix_memalign being successful.
1781 (lower_stmt): Restrict lowering of posix_memalign to when
1782 -ftree-bit-ccp is enabled.
1784 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1786 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1788 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1790 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1792 PR rtl-optimization/60116
1793 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1794 other_insn once the combination has been validated.
1796 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1799 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1801 * ipa-devirt.c: Include demangle.h
1802 (odr_violation_reported): New static variable.
1803 (add_type_duplicate): Update odr_violations.
1804 (maybe_record_node): Add completep parameter; update it.
1805 (record_target_from_binfo): Add COMPLETEP parameter;
1806 update it as needed.
1807 (possible_polymorphic_call_targets_1): Likewise.
1808 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1809 rename FINAL to COMPLETE.
1810 (record_targets_from_bases): Sanity check we found the binfo;
1811 fix COMPLETEP updating.
1812 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1813 parameter, fix computing of COMPLETEP.
1814 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1815 at LTO time do demangling.
1816 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1817 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1819 (gimple_get_virt_method_for_binfo): Likewise.
1820 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1821 gimple_get_virt_method_for_vtable): Update prototypes.
1823 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1826 * genautomata.c (add_presence_absence): Fix typo with
1827 {final_}presence_list.
1829 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1832 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1833 for VSX/Altivec vectors that land in GPR registers.
1835 2014-02-11 Richard Henderson <rth@redhat.com>
1836 Jakub Jelinek <jakub@redhat.com>
1839 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1840 around drhs if type conversion to lacc->type is not useless.
1842 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1844 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1846 (cortex-a57.cortex-a53): Likewise.
1847 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1849 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1851 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1854 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1856 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1857 add_options_for_arm_vfp3.
1859 2014-02-11 Jeff Law <law@redhat.com>
1862 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1863 object with an undesirable mode.
1865 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1868 * config/i386/sol2-9.h: New file.
1869 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1870 *-*-solaris2.9*): Use it.
1872 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1874 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1875 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1877 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1879 * config/microblaze/microblaze.c: Extend mcpu version format
1881 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1883 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1885 2014-02-10 Richard Henderson <rth@redhat.com>
1888 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1889 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1890 ms-abi vs -mno-accumulate-outgoing-args.
1891 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1892 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1895 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1898 * cfgexpand.c (expand_asm_operands): Attach source location to
1899 ASM_INPUT rtx objects.
1900 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1902 2014-02-10 Nick Clifton <nickc@redhat.com>
1904 * config/mn10300/mn10300.c (popcount): New function.
1905 (mn10300_expand_prologue): Include saved registers in stack usage
1908 2014-02-10 Jeff Law <law@redhat.com>
1911 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1912 when changing the SET_DEST of a prior insn to avoid an input reload.
1914 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1916 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1917 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1918 -mcall-openbsd, or -mcall-linux.
1919 (CC1_ENDIAN_BIG_SPEC): Remove.
1920 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1921 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1922 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1923 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1924 and %cc1_endian_default.
1925 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1927 2014-02-10 Richard Biener <rguenther@suse.de>
1929 PR tree-optimization/60115
1930 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1931 MEM_REF handling. Properly verify that the accesses are not
1932 out of the objects bound.
1934 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1936 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1939 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1941 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1942 proper constants and fix formatting.
1943 (possible_polymorphic_call_targets): Fix formatting.
1945 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1946 Ilya Tocar <ilya.tocar@intel.com>
1948 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1949 (_mm512_loadu_epi32): Renamed into...
1950 (_mm512_loadu_si512): This.
1951 (_mm512_storeu_epi32): Renamed into...
1952 (_mm512_storeu_si512): This.
1953 (_mm512_maskz_ceil_ps): Removed.
1954 (_mm512_maskz_ceil_pd): Ditto.
1955 (_mm512_maskz_floor_ps): Ditto.
1956 (_mm512_maskz_floor_pd): Ditto.
1957 (_mm512_floor_round_ps): Ditto.
1958 (_mm512_floor_round_pd): Ditto.
1959 (_mm512_ceil_round_ps): Ditto.
1960 (_mm512_ceil_round_pd): Ditto.
1961 (_mm512_mask_floor_round_ps): Ditto.
1962 (_mm512_mask_floor_round_pd): Ditto.
1963 (_mm512_mask_ceil_round_ps): Ditto.
1964 (_mm512_mask_ceil_round_pd): Ditto.
1965 (_mm512_maskz_floor_round_ps): Ditto.
1966 (_mm512_maskz_floor_round_pd): Ditto.
1967 (_mm512_maskz_ceil_round_ps): Ditto.
1968 (_mm512_maskz_ceil_round_pd): Ditto.
1969 (_mm512_expand_pd): Ditto.
1970 (_mm512_expand_ps): Ditto.
1971 * config/i386/i386.c (ix86_builtins): Remove
1972 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1973 (bdesc_args): Ditto.
1974 * config/i386/predicates.md (const1256_operand): New.
1975 (const_1_to_2_operand): Ditto.
1976 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1977 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1978 (*avx512pf_gatherpf<mode>sf): Ditto.
1979 (avx512pf_gatherpf<mode>df): Ditto.
1980 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1981 (*avx512pf_gatherpf<mode>df): Ditto.
1982 (avx512pf_scatterpf<mode>sf): Ditto.
1983 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1984 (*avx512pf_scatterpf<mode>sf): Ditto.
1985 (avx512pf_scatterpf<mode>df): Ditto.
1986 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1987 (*avx512pf_scatterpf<mode>df): Ditto.
1988 (avx512f_expand<mode>): Removed.
1989 (<shift_insn><mode>3<mask_name>): Change predicate type.
1991 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1993 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1994 not at the end of datarefs vector use ordered_remove to avoid
1995 reordering datarefs vector.
1998 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1999 mark local addressable non-static vars as GOVD_PRIVATE
2000 instead of GOVD_LOCAL.
2001 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2002 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2006 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2007 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2008 assume_aligned or alloc_align attributes.
2009 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2010 arguments. Handle also assume_aligned and alloc_align attributes.
2011 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2012 calls to functions with assume_aligned or alloc_align attributes.
2013 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2015 2014-02-08 Terry Guo <terry.guo@arm.com>
2017 * doc/invoke.texi: Document ARM -march=armv7e-m.
2019 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2021 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2022 flag on __cilkrts_rethrow builtin.
2025 * ipa-cp.c (determine_versionability): Fail at -O0
2026 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2027 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2030 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2033 * tree-inline.c (copy_forbidden): Fail for
2034 __attribute__((optimize (0))) functions.
2036 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2038 * varpool.c: Include pointer-set.h.
2039 (varpool_remove_unreferenced_decls): Variables in other partitions
2040 will not be output; be however careful to not lose information
2043 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2045 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2046 lookup in the vtable constructor.
2048 2014-02-07 Jeff Law <law@redhat.com>
2051 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2052 define_insn_and_split.
2054 * ipa-inline.c (inline_small_functions): Fix typos.
2056 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2058 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2059 (s390_can_use_return_insn): Declare.
2060 * config/s390/s390.h (EPILOGUE_USES): Define.
2061 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2063 (s390_chunkify_start): Handle return JUMP_LABELs.
2064 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2065 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2066 (s390_can_use_return_insn): New functions.
2067 (s390_fix_long_loop_prediction): Handle conditional returns.
2068 (TARGET_SET_UP_BY_PROLOGUE): Define.
2069 * config/s390/s390.md (ANY_RETURN): New code iterator.
2070 (*creturn, *csimple_return, return, simple_return): New patterns.
2072 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2074 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2075 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2076 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2077 REG_CFA_RESTORE list when deciding not to restore a register.
2079 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2081 * config/s390/s390.c: Include tree-pass.h and context.h.
2082 (s390_early_mach): New function, split out from...
2083 (s390_emit_prologue): ...here.
2084 (pass_data_s390_early_mach): New pass structure.
2085 (pass_s390_early_mach): New class.
2086 (s390_option_override): Create and register early_mach pass.
2087 Move to end of file.
2089 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2091 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2092 to match for the exit block.
2094 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2096 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2097 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2098 Reject misaligned operands.
2100 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2102 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2104 2014-02-07 Richard Biener <rguenther@suse.de>
2107 * gimple-low.c (lower_builtin_posix_memalign): New function.
2108 (lower_stmt): Call it to lower posix_memalign in a way
2109 to make alignment info accessible.
2111 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2114 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2115 __builtin_setjmp_receiver.
2117 2014-02-07 Richard Biener <rguenther@suse.de>
2120 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2121 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2122 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2123 Handle BUILT_IN_POSIX_MEMALIGN.
2124 (find_func_clobbers): Likewise.
2125 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2126 (call_may_clobber_ref_p_1): Likewise.
2128 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2131 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2134 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2137 * lto-cgraph.c (lto_output_node): Use
2138 symtab_get_symbol_partitioning_class.
2139 (lto_output_varpool_node): likewise.
2140 (symtab_get_symbol_partitioning_class): Move here from
2142 * cgraph.h (symbol_partitioning_class): Likewise.
2143 (symtab_get_symbol_partitioning_class): Declare.
2145 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2147 * ggc.h (ggc_internal_cleared_alloc): New macro.
2148 * vec.h (vec_safe_copy): Handle memory stats.
2149 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2150 * target-globals.c (save_target_globals): Likewise.
2152 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2155 * expr.c (emit_move_resolve_push): Export; be bit more selective
2156 on when to clear alias set.
2157 * expr.h (emit_move_resolve_push): Declare.
2158 * function.h (struct function): Add tail_call_marked.
2159 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2160 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2161 * config/i386/i386.md (TImode move expander): De not call
2163 (FP push expanders): Preserve memory attributes.
2164 * config/i386/sse.md (push<mode>1): Remove.
2165 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2166 (ix86_expand_push): Remove.
2167 * config/i386/mmx.md (push<mode>1): Remove.
2169 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2171 PR rtl-optimization/60030
2172 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2173 lopart with paradoxical subreg before shifting it up by hprec.
2175 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2177 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2178 Remove extra newline at end of file.
2179 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2180 (arm_issue_rate): Handle cortexa57.
2181 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2182 (cortex-a57.cortex-a53): Likewise.
2184 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2187 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2188 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2190 (arm_expand_prologue): Adjust all callers.
2191 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2192 info, registers also at the lowest numbered registers side. Use
2193 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2197 * var-tracking.c (adjust_mems): Before adding a SET to
2198 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2200 2014-02-06 Alan Modra <amodra@gmail.com>
2203 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2204 change SDmode to DDmode when lra_in_progress.
2206 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2209 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2210 free_data_ref on the dr first, and before goto again also set dr
2211 to the next dr. For simd_lane_access, free old datarefs[i] before
2212 overwriting it. For get_vectype_for_scalar_type failure, don't
2213 free_data_ref if simd_lane_access.
2215 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2218 * tree.h (opts_for_fn): New inline function.
2219 (opt_for_fn): Define.
2220 * config/i386/i386.c (ix86_function_regparm): Use
2221 opt_for_fn (decl, optimize) instead of optimize.
2223 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2225 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2226 for SYMBOL_REF in large memory model.
2228 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2230 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2232 (cortex-a57): Likewise.
2233 (cortex-a57.cortex-a53): Likewise.
2235 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2236 Kugan Vivekanandarajah <kuganv@linaro.org>
2238 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2240 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2242 2014-02-06 Richard Biener <rguenther@suse.de>
2244 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2245 set_loop_copy and initialize_original_copy_tables.
2247 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2249 * config/aarch64/aarch64-simd.md
2250 (aarch64_ashr_simddi): Change QI to SI.
2252 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2253 Jakub Jelinek <jakub@redhat.com>
2256 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2259 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2261 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2262 CODE_FOR_altivec_vpku[hw]um to
2263 CODE_FOR_altivec_vpku[hw]um_direct.
2264 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2265 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2266 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2267 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2269 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2271 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2272 generation for -maltivec=be.
2273 (altivec_vsumsws): Simplify redundant test.
2275 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2277 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2278 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2279 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2280 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2281 gen_altivec_vpkuwum.
2282 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2284 (altivec_vpks<VI_char>ss): Likewise.
2285 (altivec_vpks<VI_char>us): Likewise.
2286 (altivec_vpku<VI_char>us): Likewise.
2287 (altivec_vpku<VI_char>um): Likewise.
2288 (altivec_vpku<VI_char>um_direct): New (copy of
2289 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2291 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2292 target is little endian and -maltivec=be is not specified.
2293 (*altivec_vupkhs<VU_char>_direct): New (copy of
2294 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2295 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2296 target is little endian and -maltivec=be is not specified.
2297 (*altivec_vupkls<VU_char>_direct): New (copy of
2298 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2299 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2300 little endian and -maltivec=be is not specified.
2301 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2302 little endian and -maltivec=be is not specified.
2304 2014-02-05 Richard Henderson <rth@redhat.com>
2307 * combine-stack-adj.c: Revert r206943.
2308 * sched-int.h (struct deps_desc): Add last_args_size.
2309 * sched-deps.c (init_deps): Initialize it.
2310 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2311 contain REG_ARGS_SIZE notes.
2313 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2315 * lto-cgraph.c (asm_nodes_output): Make global.
2316 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2317 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2318 (driver_handle_option): Handle OPT_fwpa.
2320 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2323 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2324 a comment typo and formatting issue. If odr_hash hasn't been
2325 created, return vNULL and set *completep to false.
2328 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2329 bb with no successors.
2331 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2334 * doc/invoke.texi (-march): Clarify documentation for ARM.
2338 2014-02-05 Richard Biener <rguenther@suse.de>
2340 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2341 when not vectorizing because of too many alias checks.
2342 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2343 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2345 2014-02-05 Nick Clifton <nickc@redhat.com>
2347 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2348 accept extended registers in any mode when compiling for the MN10300.
2350 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2352 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2353 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2354 sanitization attributes.
2355 (can_inline_edge_p): Likewise.
2356 (sanitize_attrs_match_for_inline_p): New function.
2358 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2360 * ipa-prop.c (detect_type_change): Shor circuit testing of
2361 type changes on THIS pointer.
2363 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2366 * config/pa/pa.c (legitimize_tls_address): Return original address
2367 if not passed a SYMBOL_REF rtx.
2368 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2370 (pa_emit_move_sequence): Simplify TLS source operands.
2371 (pa_legitimate_constant_p): Reject all TLS constants.
2372 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2373 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2375 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2377 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2378 groups when we know they are controlled by LTO.
2379 * varasm.c (default_binds_local_p_1): If object is in other partition,
2380 it will be resolved locally.
2382 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2384 * config/host-linux.c (linux_gt_pch_use_address): Don't
2385 use SSIZE_MAX because it is not always defined.
2387 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2390 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2391 threshold for pseudo splitting.
2392 (update_ebb_live_info): Process call argument hard registers and
2393 hard registers from insn definition too.
2394 (max_small_class_regs_num): New constant.
2395 (inherit_in_ebb): Update live hard regs through EBBs. Update
2396 reloads_num only for small register classes. Don't split for
2399 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2402 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2405 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2407 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2410 2014-02-04 Marek Polacek <polacek@redhat.com>
2412 * gdbinit.in (pel): Define.
2414 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2416 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2419 2014-02-04 Richard Biener <rguenther@suse.de>
2422 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2423 in function context local.
2424 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2425 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2426 similar to LTO_imported_decl_ref.
2428 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2430 PR tree-optimization/60002
2431 * cgraphclones.c (build_function_decl_skip_args): Clear
2434 PR tree-optimization/60023
2435 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2436 false to gsi_replace.
2437 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2438 has been in some EH region and vec_stmt could throw, add
2439 vec_stmt into the same EH region.
2440 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2441 has no lhs, ignore it.
2442 * internal-fn.c (expand_MASK_LOAD): Likewise.
2445 * tree-inline.c (copy_forbidden): Fail for
2446 __attribute__((optimize (0))) functions.
2449 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2450 is set, copy one less argument.
2451 (expand_simd_clones): Don't subtract clone_info->inbranch
2452 from simd_clone_struct_alloc argument.
2454 PR rtl-optimization/57915
2455 * recog.c (simplify_while_replacing): If all unary/binary/relational
2456 operation arguments are constant, attempt to simplify those.
2459 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2460 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2462 2014-02-04 Richard Biener <rguenther@suse.de>
2464 PR tree-optimization/60012
2465 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2466 TBAA disambiguation to all DDRs.
2468 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2471 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2472 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2474 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2477 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2479 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2481 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2482 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2484 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2487 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2488 to figure out targets of polymorphic calls with known decl.
2489 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2490 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2491 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2492 (get_polymorphic_call_info): ... here.
2493 (get_polymorphic_call_info_from_invariant): New function.
2495 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2497 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2498 lookup via vtable pointer; check for type consistency
2499 and turn inconsitent facts into UNREACHABLE.
2500 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2501 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2502 type inconsistent querries; return UNREACHABLE instead.
2504 2014-02-03 Richard Henderson <rth@twiddle.net>
2507 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2508 already processed this node.
2509 (normalize_one_pred_1): Pass along mark_set.
2510 (normalize_one_pred): Create and destroy a pointer_set_t.
2511 (normalize_one_pred_chain): Likewise.
2513 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2515 PR gcov-profile/58602
2516 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2518 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2521 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2522 -fno-devirtualize; try to devirtualize by the knowledge of
2523 virtual table pointer given by aggregate propagation.
2524 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2525 (ipa_print_node_jump_functions): Dump also offset that
2526 is relevant for polymorphic calls.
2527 (determine_known_aggregate_parts): Add arg_type parameter; use it
2528 instead of determining the type from pointer type.
2529 (ipa_compute_jump_functions_for_edge): Update call of
2530 determine_known_aggregate_parts.
2531 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2532 (gimple_get_virt_method_for_binfo): ... here; simplify using
2533 vtable_pointer_value_to_vtable.
2534 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2535 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2536 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2537 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2539 (vtable_pointer_value_to_binfo): ... here.
2540 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2542 2014-02-03 Teresa Johnson <tejohnson@google.com>
2544 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2545 redef of outer loop index variable.
2547 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2551 * doc/extend.texi (Function Attributes): Typo.
2553 2014-02-03 Cong Hou <congh@google.com>
2555 PR tree-optimization/60000
2556 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2557 if the vectorized statement is a store. A store statement can only
2558 appear at the end of pattern statements.
2560 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2562 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2563 (ix86_option_override_internal): Default long double to 64-bit for
2564 32-bit Bionic and to 128-bit for 64-bit Bionic.
2566 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2567 TARGET_LONG_DOUBLE_128 is true.
2568 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2570 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2571 (mlong-double-64): Negate -mlong-double-128.
2572 (mlong-double-128): New option.
2574 * config/i386/i386-c.c (ix86_target_macros): Define
2575 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2577 * doc/invoke.texi: Document -mlong-double-128.
2579 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2581 PR rtl-optimization/60024
2582 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2584 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2586 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2588 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2590 PR rtl-optimization/57662
2591 * sel-sched.c (code_motion_path_driver): Do not mark already not
2592 existing blocks in the visiting bitmap.
2594 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2596 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2597 on the insn being emitted.
2599 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2600 Will Deacon <will.deacon@arm.com>
2602 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2604 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2606 * config/arm/arm-tables.opt: Regenerate.
2608 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2610 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2611 for vector types other than V16QImode.
2612 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2613 define_expand, and call altivec_expand_vec_perm_le when producing
2614 code with little endian element order.
2615 (*altivec_vperm_<mode>_internal): New insn having previous
2616 behavior of altivec_vperm_<mode>.
2617 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2618 altivec_expand_vec_perm_le when producing code with little endian
2620 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2621 behavior of altivec_vperm_<mode>_uns.
2623 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2625 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2626 (altivec_vsumsws): Add handling for -maltivec=be with a little
2628 (altivec_vsumsws_direct): New.
2629 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2630 gen_altivec_vsumsws.
2632 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2634 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2635 vtable_pointer_value_to_binfo): New functions.
2636 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2637 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2639 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2641 * config/nios2/nios2.md (load_got_register): Initialize GOT
2642 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2643 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2645 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2647 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2648 preserverd by passthrough, do not propagate the type.
2650 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2652 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2653 (mips_atomic_assign_expand_fenv): New function.
2654 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2656 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2658 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2659 (__builtin_mips_set_fcsr): Likewise.
2660 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2661 MIPS_USI_FTYPE_VOID.
2662 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2663 (mips16_expand_set_fcsr): Likewise.
2664 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2665 (mips16_set_fcsr_stub): Likewise.
2666 (mips16_get_fcsr_one_only_stub): New class.
2667 (mips16_set_fcsr_one_only_stub): Likewise.
2668 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2669 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2670 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2671 (hard_float): New availability predicate.
2672 (mips_builtins): Add get_fcsr and set_fcsr.
2673 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2674 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2675 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2676 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2677 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2680 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2682 * config/mips/mips.c (mips_one_only_stub): New class.
2683 (mips_need_mips16_rdhwr_p): Replace with...
2684 (mips16_rdhwr_stub): ...this new variable.
2685 (mips16_stub_call_address): New function.
2686 (mips16_rdhwr_one_only_stub): New class.
2687 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2688 (mips_output_mips16_rdhwr): Delete.
2689 (mips_finish_stub): New function.
2690 (mips_code_end): Use it to handle rdhwr stubs.
2692 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2695 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2696 when calculating size of integer atomic types.
2698 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2700 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2702 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2704 PR tree-optimization/60003
2705 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2706 * profile.c (branch_prob): Use gimple_call_builtin_p
2707 to check for BUILT_IN_SETJMP_RECEIVER.
2708 * tree-inline.c (copy_bb): Call notice_special_calls.
2710 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2713 * lra-constraints.c (process_alt_operands): Update reload_sum only
2716 2014-01-31 Richard Henderson <rth@redhat.com>
2719 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2720 until after else_eh is processed.
2722 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2724 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2725 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2726 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2727 in smmintrin.h, remove them.
2728 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2729 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2730 * config/i386/i386.md (ROUND_SAE): Fix value.
2731 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2732 (const48_operand): New.
2733 * config/i386/subst.md (round), (round_expand): Use
2734 const_4_or_8_to_11_operand.
2735 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2737 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2739 * config/i386/constraints.md (Yk): Swap meaning with k.
2740 * config/i386/i386.md (movhi_internal): Change Yk to k.
2741 (movqi_internal): Ditto.
2742 (*k<logic><mode>): Ditto.
2745 (kandn<mode>): Ditto.
2746 (*<code>hi_1): Ditto.
2747 (*<code>qi_1): Ditto.
2748 (kxnor<mode>): Ditto.
2749 (kortestzhi): Ditto.
2750 (kortestchi): Ditto.
2752 (*one_cmplhi2_1): Ditto.
2753 (*one_cmplqi2_1): Ditto.
2754 * config/i386/sse.md (): Change k to Yk.
2755 (avx512f_load<mode>_mask): Ditto.
2756 (avx512f_blendm<mode>): Ditto.
2757 (avx512f_store<mode>_mask): Ditto.
2758 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2759 (avx512f_storedqu<mode>_mask): Ditto.
2760 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2762 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2763 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2764 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2765 (avx512f_maskcmp<mode>3): Ditto.
2766 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2767 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2768 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2769 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2770 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2771 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2772 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2773 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2774 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2775 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2776 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2777 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2778 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2779 (vec_extract_lo_<mode>_maskm): Ditto.
2780 (vec_extract_hi_<mode>_maskm): Ditto.
2781 (avx512f_vternlog<mode>_mask): Ditto.
2782 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2783 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2784 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2785 (avx512f_<code>v8div16qi2_mask): Ditto.
2786 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2787 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2788 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2789 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2790 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2791 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2792 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2793 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2794 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2795 (avx512cd_maskb_vec_dupv8di): Ditto.
2796 (avx512cd_maskw_vec_dupv16si): Ditto.
2797 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2798 (avx512f_vpermi2var<mode>3_mask): Ditto.
2799 (avx512f_vpermi2var<mode>3_mask): Ditto.
2800 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2801 (*avx512f_gathersi<mode>): Ditto.
2802 (*avx512f_gathersi<mode>_2): Ditto.
2803 (*avx512f_gatherdi<mode>): Ditto.
2804 (*avx512f_gatherdi<mode>_2): Ditto.
2805 (*avx512f_scattersi<mode>): Ditto.
2806 (*avx512f_scatterdi<mode>): Ditto.
2807 (avx512f_compress<mode>_mask): Ditto.
2808 (avx512f_compressstore<mode>_mask): Ditto.
2809 (avx512f_expand<mode>_mask): Ditto.
2810 * config/i386/subst.md (mask): Change k to Yk.
2811 (mask_scalar_merge): Ditto.
2814 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2816 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2818 2014-01-31 Richard Biener <rguenther@suse.de>
2821 * builtins.c (fold_builtin_memory_op): Make sure to not
2822 use a floating-point mode or a boolean or enumeral type for
2825 2014-01-30 DJ Delorie <dj@redhat.com>
2827 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2828 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2829 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2830 whenever main() has an epilogue.
2832 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2834 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2835 unused variable "field".
2836 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2837 (vsx_mergeh_<mode>): Likewise.
2838 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2839 (altivec_vmrghh): Likewise.
2840 (altivec_vmrghw): Likewise.
2841 (altivec_vmrglb): Likewise.
2842 (altivec_vmrglh): Likewise.
2843 (altivec_vmrglw): Likewise.
2844 (altivec_vspltb): Add missing uses.
2845 (altivec_vsplth): Likewise.
2846 (altivec_vspltw): Likewise.
2847 (altivec_vspltsf): Likewise.
2849 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2852 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2853 frame related instructions.
2855 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2857 PR rtl-optimization/59959
2858 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2859 any reload of register whose subreg is invalid.
2861 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2863 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2864 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2865 Add missing return type - void.
2867 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2869 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2870 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2871 remove element index adjustment for endian (now handled in vsx.md
2873 (altivec_expand_vec_perm_const): Use
2874 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2875 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2876 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2877 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2878 define_expand and a new define_insn *altivec_vspltb_internal;
2879 adjust for -maltivec=be on a little endian target.
2880 (altivec_vspltb_direct): New.
2881 (altivec_vsplth): Divide into a define_expand and a new
2882 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2883 little endian target.
2884 (altivec_vsplth_direct): New.
2885 (altivec_vspltw): Divide into a define_expand and a new
2886 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2887 little endian target.
2888 (altivec_vspltw_direct): New.
2889 (altivec_vspltsf): Divide into a define_expand and a new
2890 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2891 a little endian target.
2893 2014-01-30 Richard Biener <rguenther@suse.de>
2895 PR tree-optimization/59993
2896 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2897 can propagate form the earlier stmt and avoid the transform
2898 when the intermediate result is needed.
2900 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2902 * README.Portability: Fix typo.
2904 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2906 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2907 comparison_operator with ordered_comparison_operator.
2909 2014-01-30 Nick Clifton <nickc@redhat.com>
2911 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2912 Rename to mn10300_store_multiple_regs.
2913 * config/mn10300/mn10300.c: Likewise.
2914 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2915 store_multiple_regs.
2916 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2917 Call mn10300_store_multiple_regs.
2919 2014-01-30 Nick Clifton <nickc@redhat.com>
2920 DJ Delorie <dj@redhat.com>
2922 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2923 %fp 2 to keep registers after it properly word-aligned.
2924 (rl78_alloc_physical_registers_umul): Handle the case where both
2925 input operands are the same.
2927 2014-01-30 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/59903
2930 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2933 2014-01-30 Jason Merrill <jason@redhat.com>
2936 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2939 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2941 2014-01-30 Richard Biener <rguenther@suse.de>
2943 PR tree-optimization/59951
2944 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2946 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2949 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2950 SFmode to DFmode case.
2952 2014-01-29 DJ Delorie <dj@redhat.com>
2954 * config/msp430/msp430.opt (-minrt): New.
2955 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2957 (ENDFILE_SPEC): Likewise.
2959 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2961 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2962 (estimate_function_body_sizes): Use it.
2964 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2967 * dwarf2out.c (is_cxx_auto): New.
2968 (is_base_type): Use it.
2969 (gen_type_die_with_usage): Likewise.
2971 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2973 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2974 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2975 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2976 -maltivec=be with LE targets.
2977 (vsx_mergeh_<mode>): Likewise.
2978 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2979 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2980 (altivec_vmrghb): Replace with define_expand and new
2981 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2982 (altivec_vmrghb_direct): New define_insn.
2983 (altivec_vmrghh): Replace with define_expand and new
2984 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2985 (altivec_vmrghh_direct): New define_insn.
2986 (altivec_vmrghw): Replace with define_expand and new
2987 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2988 (altivec_vmrghw_direct): New define_insn.
2989 (*altivec_vmrghsf): Adjust for endianness.
2990 (altivec_vmrglb): Replace with define_expand and new
2991 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2992 (altivec_vmrglb_direct): New define_insn.
2993 (altivec_vmrglh): Replace with define_expand and new
2994 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2995 (altivec_vmrglh_direct): New define_insn.
2996 (altivec_vmrglw): Replace with define_expand and new
2997 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2998 (altivec_vmrglw_direct): New define_insn.
2999 (*altivec_vmrglsf): Adjust for endianness.
3000 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3001 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3002 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3003 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3004 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3005 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3006 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3007 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3009 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3011 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3012 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3015 2014-01-29 Richard Biener <rguenther@suse.de>
3017 PR tree-optimization/58742
3018 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3019 associate_pointerplus_align.
3020 (associate_pointerplus_diff): New function.
3021 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3022 and associate_pointerplus_diff.
3024 2014-01-29 Richard Biener <rguenther@suse.de>
3026 * lto-streamer.h (LTO_major_version): Bump to 3.
3027 (LTO_minor_version): Reset to 0.
3029 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3031 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3032 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3033 (arm_file_start): Generate correct asm header for armv7ve.
3034 * config/arm/bpabi.h: Add multilib support for armv7ve.
3035 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3036 and cortex-a15 to armv7ve.
3037 * config/arm/t-aprofile: Add multilib support for armv7ve.
3038 * doc/invoke.texi: Document -march=armv7ve.
3040 2014-01-29 Richard Biener <rguenther@suse.de>
3042 PR tree-optimization/58742
3043 * tree-ssa-forwprop.c (associate_plusminus): Return true
3044 if we changed sth, defer EH cleanup to ...
3045 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3046 (simplify_mult): New function.
3048 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3051 PR tree-optimization/59920
3052 * tree.c (build_common_builtin_nodes): Remove
3053 __builtin_setjmp_dispatcher initialization.
3054 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3055 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3056 instead of gsi_after_labels + manually skipping debug stmts.
3057 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3058 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3059 * tree-inline.c (copy_edges_for_bb): Remove
3060 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3061 argument. Ignore computed_goto_p stmts. Don't call
3062 make_abnormal_goto_edges. If a call might need abnormal edges
3063 for non-local gotos, see if it already has an edge to
3064 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3065 with true argument, don't do anything then, otherwise add
3066 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3067 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3069 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3070 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3071 (lower_stmt): Don't set data->calls_builtin_setjmp.
3072 (lower_builtin_setjmp): Adjust comment.
3073 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3074 * tree-cfg.c (found_computed_goto): Remove.
3075 (factor_computed_gotos): Remove.
3076 (make_goto_expr_edges): Return bool, true for computed gotos.
3077 Don't call make_abnormal_goto_edges.
3078 (build_gimple_cfg): Don't set found_computed_goto, don't call
3079 factor_computed_gotos.
3080 (computed_goto_p): No longer static.
3081 (make_blocks): Don't set found_computed_goto.
3082 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3083 (make_edges): If make_goto_expr_edges returns true, push bb
3084 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3085 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3086 vector. Record mapping between bbs and OpenMP regions if there
3087 are any, adjust make_gimple_omp_edges caller. Call
3088 handle_abnormal_edges.
3089 (make_abnormal_goto_edges): Remove.
3090 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3091 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3092 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3093 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3094 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3095 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3096 filling *region also set *region_idx to (*region)->entry->index.
3099 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3100 For REGs set ORIGINAL_REGNO.
3102 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3104 * doc/md.texi: Mention that a target shouldn't implement
3105 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3108 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3110 PR tree-optimization/59594
3111 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3112 a copy of the datarefs vector rather than the vector itself.
3114 2014-01-28 Jason Merrill <jason@redhat.com>
3117 * dwarf2out.c (auto_die): New static.
3118 (gen_type_die_with_usage): Handle C++1y 'auto'.
3119 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3122 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3125 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3126 (SPEC_X32): Likewise.
3127 (SPEC_64): Likewise.
3128 * config/i386/i386.c (ix86_option_override_internal): Turn off
3129 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3131 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3132 * config/i386/i386.h (TARGET_16BIT): New macro.
3133 (TARGET_16BIT_P): Likewise.
3134 * config/i386/i386.opt: Add m16.
3135 * doc/invoke.texi: Document -m16.
3137 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3139 PR preprocessor/59935
3140 * input.c (location_get_source_line): Bail out on when line number
3141 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3143 2014-01-28 Richard Biener <rguenther@suse.de>
3145 PR tree-optimization/58742
3146 * tree-ssa-forwprop.c (associate_plusminus): Handle
3147 pointer subtraction of the form (T)(P + A) - (T)P.
3149 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3151 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3154 2014-01-28 Richard Biener <rguenther@suse.de>
3157 2014-01-28 Richard Biener <rguenther@suse.de>
3159 PR rtl-optimization/45364
3160 PR rtl-optimization/59890
3161 * var-tracking.c (local_get_addr_clear_given_value): Handle
3162 already cleared slot.
3163 (val_reset): Handle not allocated local_get_addr_cache.
3164 (vt_find_locations): Use post-order on the inverted CFG.
3166 2014-01-28 Richard Biener <rguenther@suse.de>
3168 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3170 2014-01-28 Richard Biener <rguenther@suse.de>
3172 PR rtl-optimization/45364
3173 PR rtl-optimization/59890
3174 * var-tracking.c (local_get_addr_clear_given_value): Handle
3175 already cleared slot.
3176 (val_reset): Handle not allocated local_get_addr_cache.
3177 (vt_find_locations): Use post-order on the inverted CFG.
3179 2014-01-28 Alan Modra <amodra@gmail.com>
3181 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3182 * configure.ac <recursive call for build != host>: Define
3183 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3184 and LD_FOR_BUILD too.
3185 * configure: Regenerate.
3187 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3189 * config/i386/i386.c (get_builtin_code_for_version): Separate
3190 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3191 Broadwell from Haswell.
3193 2014-01-27 Steve Ellcey <sellcey@mips.com>
3195 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3196 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3197 * config/mips/mips.c (mips_option_override): Change setting
3199 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3200 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3201 Change from Mask to Var.
3203 2014-01-27 Jeff Law <law@redhat.com>
3205 * ipa-inline.c (inline_small_functions): Fix typo.
3207 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3209 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3210 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3211 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3212 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3213 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3214 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3215 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3216 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3217 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3218 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3219 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3220 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3221 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3222 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3223 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3224 (_mm512_storeu_epi64): Ditto.
3225 (_mm512_cmpge_epi32_mask): Ditto.
3226 (_mm512_cmpge_epu32_mask): Ditto.
3227 (_mm512_cmpge_epi64_mask): Ditto.
3228 (_mm512_cmpge_epu64_mask): Ditto.
3229 (_mm512_cmple_epi32_mask): Ditto.
3230 (_mm512_cmple_epu32_mask): Ditto.
3231 (_mm512_cmple_epi64_mask): Ditto.
3232 (_mm512_cmple_epu64_mask): Ditto.
3233 (_mm512_cmplt_epi32_mask): Ditto.
3234 (_mm512_cmplt_epu32_mask): Ditto.
3235 (_mm512_cmplt_epi64_mask): Ditto.
3236 (_mm512_cmplt_epu64_mask): Ditto.
3237 (_mm512_cmpneq_epi32_mask): Ditto.
3238 (_mm512_cmpneq_epu32_mask): Ditto.
3239 (_mm512_cmpneq_epi64_mask): Ditto.
3240 (_mm512_cmpneq_epu64_mask): Ditto.
3241 (_mm512_expand_pd): Ditto.
3242 (_mm512_expand_ps): Ditto.
3243 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3244 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3245 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3246 * config/i386/i386.c (ix86_builtins): Add
3247 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3248 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3249 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3250 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3251 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3252 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3253 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3254 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3255 IX86_BUILTIN_PMOVUSQW512_MEM.
3256 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3257 __builtin_ia32_pmovsqd512mem_mask,
3258 __builtin_ia32_pmovqd512mem_mask,
3259 __builtin_ia32_pmovusqw512mem_mask,
3260 __builtin_ia32_pmovsqw512mem_mask,
3261 __builtin_ia32_pmovqw512mem_mask,
3262 __builtin_ia32_pmovusdw512mem_mask,
3263 __builtin_ia32_pmovsdw512mem_mask,
3264 __builtin_ia32_pmovdw512mem_mask,
3265 __builtin_ia32_pmovqb512mem_mask,
3266 __builtin_ia32_pmovusqb512mem_mask,
3267 __builtin_ia32_pmovsqb512mem_mask,
3268 __builtin_ia32_pmovusdb512mem_mask,
3269 __builtin_ia32_pmovsdb512mem_mask,
3270 __builtin_ia32_pmovdb512mem_mask.
3271 (bdesc_args): Add __builtin_ia32_expanddf512,
3272 __builtin_ia32_expandsf512.
3273 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3274 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3275 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3276 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3277 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3278 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3279 (avx512f_<code>v8div16qi2_mask_store): This.
3280 (avx512f_expand<mode>): New.
3282 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3284 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3286 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3287 (_mm512_prefetch_i32scatter_pd): Ditto.
3288 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3289 (_mm512_prefetch_i64scatter_pd): Ditto.
3290 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3291 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3292 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3293 (_mm512_prefetch_i32scatter_ps): Ditto.
3294 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3295 (_mm512_prefetch_i64scatter_ps): Ditto.
3296 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3297 * config/i386/i386-builtin-types.def: Define
3298 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3299 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3300 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3301 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3302 IX86_BUILTIN_SCATTERPFQPD.
3303 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3304 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3305 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3306 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3307 __builtin_ia32_scatterpfqps.
3308 (ix86_expand_builtin): Expand new built-ins.
3309 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3310 fix memory access data type.
3311 (*avx512pf_gatherpf<mode>_mask): Ditto.
3312 (*avx512pf_gatherpf<mode>): Ditto.
3313 (avx512pf_scatterpf<mode>): Ditto.
3314 (*avx512pf_scatterpf<mode>_mask): Ditto.
3315 (*avx512pf_scatterpf<mode>): Ditto.
3316 (GATHER_SCATTER_SF_MEM_MODE): New.
3317 (avx512pf_gatherpf<mode>df): Ditto.
3318 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3319 (*avx512pf_scatterpf<mode>df): Ditto.
3321 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3324 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3325 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3328 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3330 * common/config/arm/arm-common.c
3331 (arm_rewrite_mcpu): Handle multiple names.
3333 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3335 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3337 * gimple-builder.h (create_gimple_tmp): Delete.
3339 2014-01-27 Christian Bruel <christian.bruel@st.com>
3341 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3344 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3346 * config/pa/pa.md (call): Generate indirect long calls to non-local
3347 functions when outputing 32-bit code.
3348 (call_value): Likewise except for special call to buggy powf function.
3350 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3351 portable runtime and PIC indirect calls.
3352 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3353 and PIC call sequences. Use ldo instead of blr to set return register
3354 in PIC call sequence.
3356 2014-01-25 Walter Lee <walt@tilera.com>
3358 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3359 avoid clobbering a live register.
3361 2014-01-25 Walter Lee <walt@tilera.com>
3363 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3364 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3365 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3366 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3368 2014-01-25 Walter Lee <walt@tilera.com>
3370 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3371 arguments on even registers.
3372 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3374 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3375 (BIGGEST_ALIGNMENT): Ditto.
3376 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3378 2014-01-25 Walter Lee <walt@tilera.com>
3380 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3381 insns before bundling.
3382 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3384 2014-01-25 Walter Lee <walt@tilera.com>
3386 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3387 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3388 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3390 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3392 * config/mips/constraints.md (kl): Delete.
3393 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3394 define expands, using...
3395 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3396 instructions for MIPS16.
3397 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3398 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3400 2014-01-25 Walter Lee <walt@tilera.com>
3402 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3406 2014-01-25 Walter Lee <walt@tilera.com>
3408 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3409 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3411 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3413 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3416 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3418 * print-rtl.c (in_call_function_usage): New var.
3419 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3420 EXPR_LIST mode as mode and not as reg note name.
3423 * cfgloopmanip.c (copy_loop_info): If
3424 loop->warned_aggressive_loop_optimizations, make sure
3425 the flag is set in target loop too.
3427 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3429 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3431 * builtins.def: Likewise.
3432 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3433 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3434 * ira.c (ira_setup_eliminable_regset): Likewise.
3435 * omp-low.c (gate_expand_omp): Likewise.
3436 (execute_lower_omp): Likewise.
3437 (diagnose_sb_0): Likewise.
3438 (gate_diagnose_omp_blocks): Likewise.
3439 (simd_clone_clauses_extract): Likewise.
3442 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3444 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3445 correction for little endian...
3446 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3449 2014-01-24 Jeff Law <law@redhat.com>
3451 PR tree-optimization/59919
3452 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3453 for non-returning calls.
3455 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3457 * common/config/aarch64/aarch64-common.c
3458 (aarch64_rewrite_mcpu): Handle multiple names.
3459 * config/aarch64/aarch64.h
3460 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3462 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3464 * input.c (add_file_to_cache_tab): Handle the case where fopen
3467 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3470 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3471 from push operand if code of push isn't PRE_DEC.
3473 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3476 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3477 -mquad-memory-atomic. Update -mquad-memory documentation to say
3478 it is only used for non-atomic loads/stores.
3480 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3481 -mquad-memory or -mquad-memory-atomic switches.
3483 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3484 -mquad-memory-atomic to ISA 2.07 support.
3486 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3487 to separate support of normal quad word memory operations (ldq, stq)
3488 from the atomic quad word memory operations.
3490 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3491 support to separate non-atomic quad word operations from atomic
3492 quad word operations. Disable non-atomic quad word operations in
3493 little endian mode so that we don't have to swap words after the
3494 load and before the store.
3495 (quad_load_store_p): Add comment about atomic quad word support.
3496 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3497 options printed with -mdebug=reg.
3499 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3500 -mquad-memory-atomic as the test for whether we have quad word
3501 atomic instructions.
3502 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3503 or -mp8-vector are used, allow byte/half-word atomic operations.
3505 * config/rs6000/sync.md (load_lockedti): Insure that the address
3506 is a proper indexed or indirect address for the lqarx instruction.
3507 On little endian systems, swap the hi/lo registers after the lqarx
3509 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3510 insure the address is valid for the lqarx instruction.
3511 (store_conditionalti): Insure that the address is a proper indexed
3512 or indirect address for the stqcrx. instruction. On little endian
3513 systems, swap the hi/lo registers before doing the stqcrx.
3515 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3516 insure the address is valid for the stqcrx. instruction.
3518 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3519 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3520 type of quad memory support is available.
3522 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3525 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3526 there is a danger of looping.
3528 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3530 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3531 force flag_ira_loop_pressure if set via command line.
3533 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3535 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3536 (ashr_simd): New builtin handling DI mode.
3537 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3538 (aarch64_sshr_simddi): New match pattern.
3539 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3540 (vshrd_n_s64): Likewise.
3541 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3543 2014-01-23 Nick Clifton <nickc@redhat.com>
3545 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3546 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3547 favour of mcu specific scripts.
3548 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3551 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3552 Alex Velenko <Alex.Velenko@arm.com>
3554 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3555 (vaddv_s16): Likewise.
3556 (vaddv_s32): Likewise.
3557 (vaddv_u8): Likewise.
3558 (vaddv_u16): Likewise.
3559 (vaddv_u32): Likewise.
3560 (vaddvq_s8): Likewise.
3561 (vaddvq_s16): Likewise.
3562 (vaddvq_s32): Likewise.
3563 (vaddvq_s64): Likewise.
3564 (vaddvq_u8): Likewise.
3565 (vaddvq_u16): Likewise.
3566 (vaddvq_u32): Likewise.
3567 (vaddvq_u64): Likewise.
3568 (vaddv_f32): Likewise.
3569 (vaddvq_f32): Likewise.
3570 (vaddvq_f64): Likewise.
3571 (vmaxv_f32): Likewise.
3572 (vmaxv_s8): Likewise.
3573 (vmaxv_s16): Likewise.
3574 (vmaxv_s32): Likewise.
3575 (vmaxv_u8): Likewise.
3576 (vmaxv_u16): Likewise.
3577 (vmaxv_u32): Likewise.
3578 (vmaxvq_f32): Likewise.
3579 (vmaxvq_f64): Likewise.
3580 (vmaxvq_s8): Likewise.
3581 (vmaxvq_s16): Likewise.
3582 (vmaxvq_s32): Likewise.
3583 (vmaxvq_u8): Likewise.
3584 (vmaxvq_u16): Likewise.
3585 (vmaxvq_u32): Likewise.
3586 (vmaxnmv_f32): Likewise.
3587 (vmaxnmvq_f32): Likewise.
3588 (vmaxnmvq_f64): Likewise.
3589 (vminv_f32): Likewise.
3590 (vminv_s8): Likewise.
3591 (vminv_s16): Likewise.
3592 (vminv_s32): Likewise.
3593 (vminv_u8): Likewise.
3594 (vminv_u16): Likewise.
3595 (vminv_u32): Likewise.
3596 (vminvq_f32): Likewise.
3597 (vminvq_f64): Likewise.
3598 (vminvq_s8): Likewise.
3599 (vminvq_s16): Likewise.
3600 (vminvq_s32): Likewise.
3601 (vminvq_u8): Likewise.
3602 (vminvq_u16): Likewise.
3603 (vminvq_u32): Likewise.
3604 (vminnmv_f32): Likewise.
3605 (vminnmvq_f32): Likewise.
3606 (vminnmvq_f64): Likewise.
3608 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3610 * config/aarch64/aarch64-simd.md
3611 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3612 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3613 (*aarch64_mul3_elt<mode>): Likewise.
3614 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3615 (*aarch64_mul3_elt_to_64v2df): Likewise.
3616 (*aarch64_mla_elt<mode>): Likewise.
3617 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3618 (*aarch64_mls_elt<mode>): Likewise.
3619 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3620 (*aarch64_fma4_elt<mode>): Likewise.
3621 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3622 (*aarch64_fma4_elt_to_64v2df): Likewise.
3623 (*aarch64_fnma4_elt<mode>): Likewise.
3624 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3625 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3626 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3627 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3628 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3629 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3630 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3631 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3632 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3634 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3636 * config/aarch64/aarch64-simd.md
3637 (aarch64_be_checked_get_lane<mode>): New define_expand.
3638 * config/aarch64/aarch64-simd-builtins.def
3639 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3640 New builtin definition.
3641 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3642 Use new safe be builtin.
3644 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3646 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3648 (aarch64_be_st1<mode>): Likewise.
3649 (aarch_ld1<VALL:mode>): Define_expand modified.
3650 (aarch_st1<VALL:mode>): Likewise.
3651 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3652 (UNSPEC_ST1): Likewise.
3654 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3656 * config/microblaze/microblaze.md: Add trap insn and attribute
3658 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3660 PR preprocessor/58580
3661 * input.h (location_get_source_line): Take an additional line_size
3663 (void diagnostics_file_cache_fini): Declare new function.
3664 * input.c (struct fcache): New type.
3665 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3666 New static constants.
3667 (diagnostic_file_cache_init, total_lines_num)
3668 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3669 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3670 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3671 (get_next_line, read_next_line, goto_next_line, read_line_num):
3672 New static function definitions.
3673 (diagnostic_file_cache_fini): New function.
3674 (location_get_source_line): Take an additional output line_len
3675 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3677 * diagnostic.c (diagnostic_finish): Call
3678 diagnostic_file_cache_fini.
3679 (adjust_line): Take an additional input parameter for the length
3680 of the line, rather than calculating it with strlen.
3681 (diagnostic_show_locus): Adjust the use of
3682 location_get_source_line and adjust_line with respect to their new
3683 signature. While displaying a line now, do not stop at the first
3684 null byte. Rather, display the zero byte as a space and keep
3685 going until we reach the size of the line.
3686 * Makefile.in: Add vec.o to OBJS-libcommon
3688 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3689 Ilya Tocar <ilya.tocar@intel.com>
3691 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3692 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3693 (__builtin_ia32_kmov16): Ditto.
3694 * config/i386/i386.md (UNSPEC_KMOV): New.
3697 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3699 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3700 (_mm512_storeu_si512): Ditto.
3702 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3705 * rtl.h (get_referenced_operands): Declare.
3706 * recog.c (get_referenced_operands): New function.
3707 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3708 operands have been referenced when recording LO_SUM references.
3710 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3712 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3714 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3716 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3717 Enable for generic and recent AMD targets.
3719 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3721 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3722 ARG_SIZE note when adjustment was eliminated.
3724 2014-01-22 Jeff Law <law@redhat.com>
3726 PR tree-optimization/59597
3727 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3728 in file. Accept new argument REGISTERING and use it to modify
3729 dump output appropriately.
3730 (register_jump_thread): Corresponding changes.
3731 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3732 thread paths involving joiner blocks. Add code to dump cancelled
3733 jump threading paths.
3735 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3737 PR rtl-optimization/59477
3738 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3739 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3741 2014-01-22 Tom Tromey <tromey@redhat.com>
3743 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3745 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3747 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3749 PR rtl-optimization/59896
3750 * lra-constraints.c (process_alt_operands): Check unused note for
3751 matched operands of insn with no output reloads.
3753 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3755 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3756 (mips_move_from_gpr_cost): Likewise.
3758 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3760 PR rtl-optimization/59858
3761 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3762 ira_class_hard_regs_num.
3763 (process_alt_operands): Increase reject for dying matched operand.
3765 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3768 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3769 smaller than size, perform several stores or loads and stores
3770 at dst + count - size to store or copy all of size bytes, rather
3771 than just last modesize bytes.
3773 2014-01-20 DJ Delorie <dj@redhat.com>
3775 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3776 that CLOBBERs are REGs before propogating their values.
3778 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3781 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3782 (cgraph_inline_failed_type): New function.
3783 * cgraph.h (DEFCIFCODE): Add type.
3784 (cgraph_inline_failed_type_t): New enum.
3785 (cgraph_inline_failed_type): New prototype.
3786 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3787 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3788 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3789 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3790 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3791 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3792 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3793 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3794 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3795 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3796 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3797 OPTIMIZATION_MISMATCH.
3798 * tree-inline.c (expand_call_inline): Emit errors during
3799 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3801 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3804 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3805 mode attribute in insn output.
3807 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3809 * output.h (output_constant): Delete.
3810 * varasm.c (output_constant): Make private.
3812 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3814 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3816 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3819 * tree.h (fold_builtin_strcat): New prototype.
3820 * builtins.c (fold_builtin_strcat): No longer static. Add len
3821 argument, if non-NULL, don't call c_strlen. Optimize
3822 directly into __builtin_memcpy instead of __builtin_strcpy.
3823 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3824 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3826 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3828 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3829 for SImode_address_operand operands, having only a REG argument.
3831 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3833 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3834 loader name using mbig-endian.
3835 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3837 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3839 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3843 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3845 * config/aarch64/aarch64-protos.h
3846 (aarch64_cannot_change_mode_class_ptr): Declare.
3847 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3848 aarch64_cannot_change_mode_class_ptr): New.
3849 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3850 backend hook aarch64_cannot_change_mode_class.
3852 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3854 * common/config/aarch64/aarch64-common.c
3855 (aarch64_handle_option): Don't handle any option order logic here.
3856 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3857 selected_cpu, warn on architecture version mismatch.
3858 (aarch64_override_options): Fix parsing order for option strings.
3860 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3861 Iain Sandoe <iain@codesourcery.com>
3864 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3865 warning. Amend comment to reflect current functionality.
3867 2014-01-20 Richard Biener <rguenther@suse.de>
3870 * builtins.c (fold_builtin_strcat): Remove case better handled
3871 by tree-ssa-strlen.c.
3873 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3875 * config/aarch64/aarch64.opt
3876 (mcpu, march, mtune): Make case-insensitive.
3878 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3881 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3882 if operands[1] is a REG or ZERO_EXTEND of a REG.
3884 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3886 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3888 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3890 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3891 long non-pic millicode calls.
3893 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3895 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3897 2014-01-19 Kito Cheng <kito@0xlab.org>
3899 * builtins.c (expand_movstr): Check movstr expand done or fail.
3901 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3902 H.J. Lu <hongjiu.lu@intel.com>
3905 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3906 to DImode for zero-extended addresses.
3908 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3910 PR rtl-optimization/57763
3911 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3912 on the new indirect jump_insn and increment LABEL_NUSES (label).
3914 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3918 * config.gcc (x86_archs): New variable.
3919 (x86_64_archs): Likewise.
3920 (x86_cpus): Likewise.
3921 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3922 --with-arch/--with-cpu= options.
3923 Support --with-arch=/--with-cpu={nehalem,westmere,
3924 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3926 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3928 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3929 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3931 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3933 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3935 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3938 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3939 clear cpp_get_options (parse_in)->warn_unused_macros for
3940 ix86_target_macros_internal with cpp_define.
3942 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3944 * jump.c (delete_related_insns): Keep (use (insn))s.
3945 * reorg.c (redundant_insn): Check for barriers too.
3947 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3949 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3951 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3953 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3954 call to $$dyncall when TARGET_LONG_CALLS is true.
3956 2014-01-17 Jeff Law <law@redhat.com>
3958 * ree.c (combine_set_extension): Temporarily disable test for
3959 changing number of hard registers.
3961 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3964 * ipa-inline-analysis.c (inline_free_summary):
3965 Do not free summary of aliases.
3967 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3970 * gimplify.c (gimplify_expr): Use create_tmp_var
3971 instead of create_tmp_var_raw. If cond doesn't have
3972 integral type, don't add the IFN_ANNOTATE builtin at all.
3974 2014-01-17 Martin Jambor <mjambor@suse.cz>
3977 * ipa-cp.c (prev_edge_clone): New variable.
3978 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3979 Also resize prev_edge_clone vector.
3980 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3981 (ipcp_edge_removal_hook): New function.
3982 (ipcp_driver): Register ipcp_edge_removal_hook.
3984 2014-01-17 Andrew Pinski <apinski@cavium.com>
3985 Steve Ellcey <sellcey@mips.com>
3988 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3991 2014-01-17 Jeff Law <law@redhat.com>
3994 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3995 so that pass_ccp runs first.
3997 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3999 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4000 (ix86_adjust_cost): Use !TARGET_XXX.
4001 (do_reorder_for_imul): Likewise.
4002 (swap_top_of_ready_list): Likewise.
4003 (ix86_sched_reorder): Likewise.
4005 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4007 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4008 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4009 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4010 (intel_memset): New. Duplicate slm_memset.
4011 (intel_cost): New. Duplicate slm_cost.
4012 (m_INTEL): New macro.
4013 (processor_target_table): Add "intel".
4014 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4015 with PROCESSOR_INTEL for "intel".
4016 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4017 PROCESSOR_SILVERMONT.
4018 (ix86_issue_rate): Likewise.
4019 (ix86_adjust_cost): Likewise.
4020 (ia32_multipass_dfa_lookahead): Likewise.
4021 (swap_top_of_ready_list): Likewise.
4022 (ix86_sched_reorder): Likewise.
4023 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4024 instead of TARGET_OPT_AGU.
4025 * config/i386/i386.h (TARGET_INTEL): New.
4026 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4027 (processor_type): Add PROCESSOR_INTEL.
4028 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4029 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4031 2014-01-17 Marek Polacek <polacek@redhat.com>
4034 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4037 2014-01-17 Richard Biener <rguenther@suse.de>
4039 PR tree-optimization/46590
4040 * opts.c (default_options_table): Add entries for
4041 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4042 all enabled at -O1 but not for -Og.
4043 * common.opt (fbranch-count-reg): Remove Init(1).
4044 (fmove-loop-invariants): Likewise.
4045 (ftree-pta): Likewise.
4047 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4049 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4050 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4051 decls to at least the GCC 4.8 used alignments.
4054 * tree-nested.c (convert_nonlocal_reference_stmt,
4055 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4056 of GIMPLE_BIND stmts, adjust associated decls.
4058 2014-01-17 Richard Biener <rguenther@suse.de>
4060 PR tree-optimization/46590
4061 * vec.h (vec<>::bseach): New member function implementing
4062 binary search according to C89 bsearch.
4063 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4064 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4065 bitmap pointer again. Make accesses_in_loop a flat array.
4066 (mem_ref_obstack): New global.
4067 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4068 (mark_ref_stored): Likewise.
4069 (ref_indep_loop_p_2): Likewise.
4070 (set_ref_stored_in_loop): New helper function.
4071 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4072 (memref_free): Adjust.
4073 (record_mem_ref_loc): Simplify.
4074 (gather_mem_refs_stmt): Adjust.
4075 (sort_locs_in_loop_postorder_cmp): New function.
4076 (analyze_memory_references): Sort accesses_in_loop after
4077 loop postorder number.
4078 (find_ref_loc_in_loop_cmp): New function.
4079 (for_all_locs_in_loop): Find relevant cluster of locs in
4080 accesses_in_loop and iterate without recursion.
4081 (execute_sm): Avoid uninit warning.
4082 (struct ref_always_accessed): Simplify.
4083 (ref_always_accessed::operator ()): Likewise.
4084 (ref_always_accessed_p): Likewise.
4085 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4086 loop postorder numbers here.
4087 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4090 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4093 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4094 on decls for which assemble_alias has been called.
4096 2014-01-17 Nick Clifton <nickc@redhat.com>
4098 * config/msp430/msp430.opt: (mcpu): New option.
4099 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4100 (msp430_option_override): Parse target_cpu. If the MCU name
4101 matches a generic string, clear target_mcu.
4102 (msp430_attr): Allow numeric interrupt values up to 63.
4103 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4104 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4106 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4108 * config/msp430/msp430.md (popm): Use %J rather than %I.
4109 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4110 (addhi_cy_i): Use immediate_operand for operand 2.
4111 * doc/invoke.texi: Document -mcpu option.
4113 2014-01-17 Richard Biener <rguenther@suse.de>
4115 PR rtl-optimization/38518
4116 * df.h (df_analyze_loop): Declare.
4117 * df-core.c: Include cfgloop.h.
4118 (df_analyze_1): Split out main part of df_analyze.
4119 (df_analyze): Adjust.
4120 (loop_inverted_post_order_compute): New function.
4121 (loop_post_order_compute): Likewise.
4122 (df_analyze_loop): New function avoiding whole-function
4124 * loop-invariant.c (find_defs): Use df_analyze_loop.
4125 (find_invariants): Adjust.
4126 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4128 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4130 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4131 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4133 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4135 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4136 traversal when removing references.
4138 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4141 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4143 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4146 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4147 pushing a reload for an autoinc when we had previously reloaded an
4148 inner part of the address.
4150 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4152 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4154 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4155 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4156 when not giving up or versioning for alias only because of
4158 (vect_analyze_data_ref_dependences): Set to true.
4159 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4161 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4162 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4166 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4169 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4170 operand 0 predicate for gathers, use a new pseudo as subtarget.
4172 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4175 * lra-constraints.c (process_alt_operands): Add printing debug info.
4176 Check absence of input/output reloads for matched operands too.
4178 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4180 PR rtl-optimization/59835
4181 * ira.c (ira_init_register_move_cost): Increase cost for
4184 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4186 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4188 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4191 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4192 non-register objects. Use gen_(high/low)part more consistently.
4195 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4198 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4199 endian support, remove tests for WORDS_BIG_ENDIAN.
4200 (p8_mfvsrd_3_<mode>): Likewise.
4201 (reload_gpr_from_vsx<mode>): Likewise.
4202 (reload_gpr_from_vsxsf): Likewise.
4203 (p8_mfvsrd_4_disf): Likewise.
4205 2014-01-16 Richard Biener <rguenther@suse.de>
4207 PR rtl-optimization/46590
4208 * lcm.c (compute_antinout_edge): Use postorder iteration.
4209 (compute_laterin): Use inverted postorder iteration.
4211 2014-01-16 Nick Clifton <nickc@redhat.com>
4214 * varasm.c (output_constant): Return the number of bytes actually
4216 (output_constructor_array_range): Update the field size with the
4217 number of bytes emitted by output_constant.
4218 (output_constructor_regular_field): Likewise. Also do not
4219 complain if the total number of bytes emitted is now greater
4220 than the expected fieldpos.
4221 * output.h (output_constant): Update prototype and descriptive comment.
4223 2014-01-16 Marek Polacek <polacek@redhat.com>
4226 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4227 it is error_mark_node.
4229 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4231 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4232 VALID_AVX256_REG_OR_OI_MODE.
4234 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4236 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4237 current procedure should be profiled.
4239 2014-01-15 Andrew Pinski <apinski@cavium.com>
4241 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4242 of moving from/to the STACK_REG register class.
4244 2014-01-15 Richard Henderson <rth@redhat.com>
4247 * reginfo.c (global_regs_decl): Globalize.
4248 * rtl.h (global_regs_decl): Declare.
4249 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4250 reserved via global_regs.
4252 2014-01-15 Teresa Johnson <tejohnson@google.com>
4254 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4256 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4258 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4259 and vmulosh rather than call gen_vec_widen_smult_*.
4260 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4261 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4262 (vec_widen_smult_even_v16qi): Likewise.
4263 (vec_widen_umult_even_v8hi): Likewise.
4264 (vec_widen_smult_even_v8hi): Likewise.
4265 (vec_widen_umult_odd_v16qi): Likewise.
4266 (vec_widen_smult_odd_v16qi): Likewise.
4267 (vec_widen_umult_odd_v8hi): Likewise.
4268 (vec_widen_smult_odd_v8hi): Likewise.
4269 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4270 vmuloub rather than call gen_vec_widen_umult_*.
4271 (vec_widen_umult_lo_v16qi): Likewise.
4272 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4273 vmulosb rather than call gen_vec_widen_smult_*.
4274 (vec_widen_smult_lo_v16qi): Likewise.
4275 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4276 rather than call gen_vec_widen_umult_*.
4277 (vec_widen_umult_lo_v8hi): Likewise.
4278 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4279 rather than call gen_vec_widen_smult_*.
4280 (vec_widen_smult_lo_v8hi): Likewise.
4282 2014-01-15 Jeff Law <law@redhat.com>
4284 PR tree-optimization/59747
4285 * ree.c (find_and_remove_re): Properly handle case where a second
4286 eliminated extension requires widening a copy created for elimination
4287 of a prior extension.
4288 (combine_set_extension): Ensure that the number of hard regs needed
4289 for a destination register does not change when we widen it.
4291 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4293 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4294 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4295 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4296 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4297 (avr-*-rtems*): Likewise.
4298 (bfin*-rtems*): Likewise.
4299 (moxie-*-rtems*): Likewise.
4300 (h8300-*-rtems*): Likewise.
4301 (i[34567]86-*-rtems*): Likewise.
4302 (lm32-*-rtems*): Likewise.
4303 (m32r-*-rtems*): Likewise.
4304 (m68k-*-rtems*): Likewise.
4305 (microblaze*-*-rtems*): Likewise.
4306 (mips*-*-rtems*): Likewise.
4307 (powerpc-*-rtems*): Likewise.
4308 (sh-*-rtems*): Likewise.
4309 (sparc-*-rtems*): Likewise.
4310 (sparc64-*-rtems*): Likewise.
4311 (v850-*-rtems*): Likewise.
4312 (m32c-*-rtems*): Likewise.
4314 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4316 PR rtl-optimization/59511
4317 * ira.c (ira_init_register_move_cost): Use memory costs for some
4318 cases of register move cost calculations.
4319 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4320 instead of BB frequency.
4321 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4322 * lra-assigns.c (find_hard_regno_for): Ditto.
4324 2014-01-15 Richard Biener <rguenther@suse.de>
4326 PR tree-optimization/59822
4327 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4328 (vectorizable_load): Use it to hoist defs of uses of invariant
4329 loads out of the loop.
4331 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4332 Kugan Vivekanandarajah <kuganv@linaro.org>
4335 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4338 2014-01-15 Richard Biener <rguenther@suse.de>
4340 PR rtl-optimization/59802
4341 * lcm.c (compute_available): Use inverted postorder to seed
4342 the initial worklist.
4344 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4347 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4348 ADDR_REGS for invalid symrefs in non-PIC code.
4350 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4353 * builtins.c (determine_block_size): Initialize *probable_max_size
4354 even if len_rtx is CONST_INT.
4356 2014-01-14 Andrew Pinski <apinski@cavium.com>
4358 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4359 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4360 (cortexa53_tunings): Likewise.
4361 (aarch64_sched_issue_rate): New function.
4362 (TARGET_SCHED_ISSUE_RATE): Define.
4364 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4366 * ira-costs.c (find_costs_and_classes): Add missed
4367 ira_init_register_move_cost_if_necessary.
4369 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4372 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4374 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4377 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4378 to indicate if type is used for function return value. Warn ABI
4379 change if the vector mode isn't available for function return value.
4380 (ix86_function_arg_advance): Pass false to type_natural_mode.
4381 (ix86_function_arg): Likewise.
4382 (ix86_gimplify_va_arg): Likewise.
4383 (function_arg_32): Don't warn ABI change.
4384 (ix86_function_value): Pass true to type_natural_mode.
4385 (ix86_return_in_memory): Likewise.
4386 (ix86_struct_value_rtx): Removed.
4387 (TARGET_STRUCT_VALUE_RTX): Likewise.
4389 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4391 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4392 converting a conditional jump into a conditional return.
4394 2014-01-14 Richard Biener <rguenther@suse.de>
4396 PR tree-optimization/58921
4397 PR tree-optimization/59006
4398 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4399 hoisting invariant stmts.
4400 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4401 invariant loads on the preheader edge if possible.
4403 2014-01-14 Joey Ye <joey.ye@arm.com>
4405 * doc/plugin.texi (Building GCC plugins): Update to C++.
4407 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4409 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4410 (_mm_rcp28_round_ss): Ditto.
4411 (_mm_rsqrt28_round_sd): Ditto.
4412 (_mm_rsqrt28_round_ss): Ditto.
4413 (_mm_rcp28_sd): Ditto.
4414 (_mm_rcp28_ss): Ditto.
4415 (_mm_rsqrt28_sd): Ditto.
4416 (_mm_rsqrt28_ss): Ditto.
4417 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4418 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4419 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4420 (IX86_BUILTIN_RCP28SD): Ditto.
4421 (IX86_BUILTIN_RCP28SS): Ditto.
4422 (IX86_BUILTIN_RSQRT28SD): Ditto.
4423 (IX86_BUILTIN_RSQRT28SS): Ditto.
4424 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4425 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4426 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4427 (ix86_expand_special_args_builtin): Expand new FTYPE.
4428 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4429 (srcp14<mode>): Make insn unary.
4430 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4431 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4432 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4433 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4434 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4435 Fix rounding: make it SAE only.
4436 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4438 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4440 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4441 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4442 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4443 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4444 (round_saeonly_mask_scalar_operand4): Ditto.
4445 (round_saeonly_mask_scalar_op3): Ditto.
4446 (round_saeonly_mask_scalar_op4): Ditto.
4448 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4450 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4451 Implement -maltivec=be for vec_insert and vec_extract.
4453 2014-01-10 DJ Delorie <dj@redhat.com>
4455 * config/msp430/msp430.md (call_internal): Don't allow memory
4456 references with SP as the base register.
4457 (call_value_internal): Likewise.
4458 * config/msp430/constraints.md (Yc): New. For memory references
4459 that don't use SP as a base register.
4461 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4462 "an integer without a # prefix"
4463 * config/msp430/msp430.md (epilogue_helper): Use it.
4465 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4468 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4469 AVX512F gather builtins.
4470 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4471 on gather decls with INTEGER_TYPE masktype.
4472 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4473 directly into the builtin rather than hoisting it before loop.
4475 PR tree-optimization/59387
4476 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4477 (scev_const_prop): If folded_casts and type has undefined overflow,
4478 use force_gimple_operand instead of force_gimple_operand_gsi and
4479 for each added stmt if it is assign with
4480 arith_code_with_undefined_signed_overflow, call
4481 rewrite_to_defined_overflow.
4482 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4483 gimple-fold.h instead.
4484 (arith_code_with_undefined_signed_overflow,
4485 rewrite_to_defined_overflow): Moved to ...
4486 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4487 rewrite_to_defined_overflow): ... here. No longer static.
4488 Include gimplify-me.h.
4489 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4490 rewrite_to_defined_overflow): New prototypes.
4492 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4494 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4496 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4498 * builtins.c (get_object_alignment_2): Minor tweak.
4499 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4501 2014-01-13 Christian Bruel <christian.bruel@st.com>
4503 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4504 optimized non constant lengths.
4506 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4509 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4510 load as __atomic_load_N if possible.
4512 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4514 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4516 (rs6000_expand_builtin): Adjust call.
4518 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4521 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4522 * config/rs6000/rs6000.c: Include target-globals.h.
4523 (rs6000_set_current_function): Instead of doing target_reinit
4524 unconditionally, use save_target_globals_default_opts and
4525 restore_target_globals.
4527 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4529 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4530 (rs6000_expand_builtin): Handle mffs and mtfsf.
4531 (rs6000_init_builtins): Define mffs and mtfsf.
4532 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4533 (rs6000_mffs): New pattern.
4534 (rs6000_mtfsf): New pattern.
4536 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4538 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4539 Start narrowing with START. Apply candidate-use pair
4540 and check overall cost in narrowing.
4541 (iv_ca_prune): Pass new argument.
4543 2014-01-10 Jeff Law <law@redhat.com>
4546 * ree.c (combine_reaching_defs): Ensure the defining statement
4547 occurs before the extension when optimizing extensions with
4548 different source and destination hard registers.
4550 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4553 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4554 vtables into the type inheritance graph.
4556 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4558 PR rtl-optimization/59754
4559 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4560 modes in the REGNO != REGNO case.
4562 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4564 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4566 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4568 PR tree-optimization/59745
4569 * tree-predcom.c (tree_predictive_commoning_loop): Call
4570 free_affine_expand_cache if giving up because components is NULL.
4572 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4573 GC in payload of target_globals struct instead of allocating them on
4574 the heap and the larger structs separately using GC.
4575 * target-globals.h (struct target_globals): Make regs, hard_regs,
4576 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4577 of GTY((skip)) and change type to void *.
4578 (reset_target_globals): Cast loads from those fields to corresponding
4581 2014-01-10 Steve Ellcey <sellcey@mips.com>
4584 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4585 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4586 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4588 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4591 * aarch64-modes.def (CC_Zmode): New flags mode.
4592 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4593 represents an equality.
4594 (aarch64_get_condition_code): Handle CC_Zmode.
4595 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4597 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4599 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4600 extraction in good case.
4602 2014-01-10 Richard Biener <rguenther@suse.de>
4604 PR tree-optimization/59374
4605 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4606 checking after SLP discovery. Mark stmts not participating
4607 in any SLP instance properly.
4609 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4611 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4612 when handling a SET rtx.
4614 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4616 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4617 (cortex-a57): Likewise.
4618 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4620 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4622 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4623 non-iwmmxt builtins.
4625 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4629 * ipa-devirt.c record_target_from_binfo): Take as argument
4630 stack of binfos and lookup matching one for virtual inheritance.
4631 (possible_polymorphic_call_targets_1): Update.
4633 2014-01-10 Huacai Chen <chenhc@lemote.com>
4635 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4636 kernel strings for Loongson-2E/2F/3A.
4638 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4641 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4642 is_gimple_call before calling gimple_call_internal_p.
4644 2014-01-09 Steve Ellcey <sellcey@mips.com>
4646 * Makefile.in (TREE_FLOW_H): Remove.
4647 (TREE_SSA_H): Add file names from tree-flow.h.
4648 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4649 * tree.h: Remove tree-flow.h reference.
4650 * hash-table.h: Remove tree-flow.h reference.
4651 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4652 reference with tree-ssa-loop.h.
4654 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4656 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4657 default element-order behavior for -maltivec.
4658 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4660 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4661 when targeting big endian, at least for now.
4662 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4664 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4667 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4668 var satisfies use_register_for_decl, just take into account type
4669 alignment, rather than decl alignment.
4671 PR tree-optimization/59622
4672 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4673 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4674 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4675 Don't devirtualize for inplace at all. For targets.length () == 1,
4676 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4678 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4680 * config/i386/i386.md (cpu): Remove the unused btver1.
4682 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4684 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4686 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4689 * tree-core.h (struct target_globals): New forward declaration.
4690 (struct tree_target_option): Add globals field.
4691 * tree.h (TREE_TARGET_GLOBALS): Define.
4692 (prepare_target_option_nodes_for_pch): New prototype.
4693 * target-globals.h (struct target_globals): Define even if
4695 * tree.c (prepare_target_option_node_for_pch,
4696 prepare_target_option_nodes_for_pch): New functions.
4697 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4698 * config/i386/i386.c: Include target-globals.h.
4699 (ix86_set_current_function): Instead of doing target_reinit
4700 unconditionally, use save_target_globals_default_opts and
4701 restore_target_globals.
4703 2014-01-09 Richard Biener <rguenther@suse.de>
4705 PR tree-optimization/59715
4706 * tree-cfg.h (split_critical_edges): Declare.
4707 * tree-cfg.c (split_critical_edges): Export.
4708 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4710 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4712 * cfgexpand.c (expand_stack_vars): Optionally disable
4713 asan stack protection.
4714 (expand_used_vars): Likewise.
4715 (partition_stack_vars): Likewise.
4716 * asan.c (asan_emit_stack_protection): Optionally disable
4717 after return stack usage.
4718 (instrument_derefs): Optionally disable memory access instrumentation.
4719 (instrument_builtin_call): Likewise.
4720 (instrument_strlen_call): Likewise.
4721 (asan_protect_global): Optionally disable global variables protection.
4722 * doc/invoke.texi: Added doc for new options.
4723 * params.def: Added new options.
4724 * params.h: Likewise.
4726 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4728 PR rtl-optimization/59724
4729 * ifcvt.c (cond_exec_process_if_block): Don't call
4730 flow_find_head_matching_sequence with 0 longest_match.
4731 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4732 non-active insns if !stop_after.
4733 (try_head_merge_bb): Revert 2014-01-07 changes.
4735 2014-01-08 Jeff Law <law@redhat.com>
4737 * ree.c (get_sub_rtx): New function, extracted from...
4738 (merge_def_and_ext): Here.
4739 (combine_reaching_defs): Use get_sub_rtx.
4741 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4743 * cgraph.h (varpool_variable_node): Do not choke on null node.
4745 2014-01-08 Catherine Moore <clm@codesourcery.com>
4747 * config/mips/mips.md (simple_return): Attempt to use JRC
4749 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4751 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4753 PR rtl-optimization/59137
4754 * reorg.c (steal_delay_list_from_target): Call update_block for
4756 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4758 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4761 two duplicate entries.
4763 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4766 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4768 * config/mips/mips.c (mips_truncated_op_cost): New function.
4769 (mips_rtx_costs): Adjust test for BADDU.
4770 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4772 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4774 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4775 (*baddu_si): ...this new pattern.
4777 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4780 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4782 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4785 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4787 (expand_expr, expand_normal): Adjust.
4788 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4789 inner_reference_p. Use inner_reference_p to expand inner references.
4790 (store_expr): Adjust.
4791 * cfgexpand.c (expand_call_stmt): Adjust.
4793 2014-01-08 Rong Xu <xur@google.com>
4795 * gcov-io.c (gcov_var): Move from gcov-io.h.
4796 (gcov_position): Ditto.
4797 (gcov_is_error): Ditto.
4798 (gcov_rewrite): Ditto.
4799 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4800 only part to libgcc/libgcov.h.
4802 2014-01-08 Marek Polacek <polacek@redhat.com>
4805 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4807 2014-01-08 Marek Polacek <polacek@redhat.com>
4810 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4812 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4814 PR rtl-optimization/59649
4815 * stor-layout.c (get_mode_bounds): For BImode return
4816 0 and STORE_FLAG_VALUE.
4818 2014-01-08 Richard Biener <rguenther@suse.de>
4821 * gimple.h (is_gimple_builtin_call): Remove.
4822 (gimple_builtin_call_types_compatible_p): New.
4823 (gimple_call_builtin_p): New overload.
4824 * gimple.c (is_gimple_builtin_call): Remove.
4825 (validate_call): Rename to ...
4826 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4828 (validate_type): New static function.
4829 (gimple_call_builtin_p): New overload and adjust.
4830 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4831 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4832 (gimple_fold_stmt_to_constant_1): Likewise.
4833 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4835 2014-01-08 Richard Biener <rguenther@suse.de>
4838 * gimplify.c (gimplify_expr): Gimplify register-register type
4839 VIEW_CONVERT_EXPRs to separate stmts.
4841 2014-01-07 Jeff Law <law@redhat.com>
4844 * ree.c (combine_set_extension): Handle case where source
4845 and destination registers in an extension insn are different.
4846 (combine_reaching_defs): Allow source and destination registers
4847 in extension to be different under limited circumstances.
4848 (add_removable_extension): Remove restriction that the
4849 source and destination registers in the extension are the same.
4850 (find_and_remove_re): Emit a copy from the extension's
4851 destination to its source after the defining insn if
4852 the source and destination registers are different.
4855 * ifcvt.c (merge_if_block): If we are merging a block with more than
4856 one successor with a block with no successors, remove any BARRIER
4857 after the second block.
4859 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4861 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4863 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4866 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4867 for 14-bit register offsets when INT14_OK_STRICT is false.
4869 2014-01-07 Roland Stigge <stigge@antcom.de>
4870 Michael Meissner <meissner@linux.vnet.ibm.com>
4873 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4874 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4876 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4878 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4881 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4883 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4884 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4885 rtx is const0_rtx or not.
4887 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4890 * target-globals.c (save_target_globals): Remove this_fn_optab
4892 * toplev.c: Include optabs.h.
4893 (target_reinit): Temporarily restore the global options if another
4894 set of options are in force.
4896 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4898 PR rtl-optimization/58668
4899 * cfgcleanup.c (flow_find_cross_jump): Don't count
4900 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4901 to determine what is counted.
4902 (flow_find_head_matching_sequence): Use active_insn_p to determine
4904 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4906 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4907 determine what is counted.
4909 PR tree-optimization/59643
4910 * tree-predcom.c (split_data_refs_to_components): If one dr is
4911 read and one write, determine_offset fails and the write isn't
4912 in the bad component, just put the read into the bad component.
4914 2014-01-07 Mike Stump <mikestump@comcast.net>
4915 Jakub Jelinek <jakub@redhat.com>
4918 * tree-core.h (struct tree_optimization_option): Change optabs
4919 type from unsigned char * to void *.
4920 * optabs.c (init_tree_optimization_optabs): Adjust
4921 TREE_OPTIMIZATION_OPTABS initialization.
4923 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4926 * config/i386/i386.h (struct machine_function): Add
4927 no_drap_save_restore field.
4928 * config/i386/i386.c (ix86_save_reg): Use
4929 !cfun->machine->no_drap_save_restore instead of
4930 crtl->stack_realign_needed.
4931 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4932 this function clears frame_pointer_needed. Set
4933 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4934 and DRAP reg is needed.
4936 2014-01-06 Marek Polacek <polacek@redhat.com>
4939 * doc/implement-c.texi: Mention that other integer types are
4940 permitted as bit-field types in strictly conforming mode.
4942 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4944 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4947 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4949 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4951 2014-01-06 Martin Jambor <mjambor@suse.cz>
4954 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4956 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4958 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4962 * var-tracking.c (add_stores): Preserve the value of the source even if
4963 we don't record the store.
4965 2014-01-06 Terry Guo <terry.guo@arm.com>
4967 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4969 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4972 * config/darwin.c (darwin_function_section): Adjust return values to
4973 correspond to optimisation changes made in r206070.
4975 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4977 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4978 from prefetch_block tune setting.
4979 (nocona_cost): Correct size of prefetch block to 64.
4981 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4983 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4984 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4985 used to save the static chain register in the computation of the offset
4986 from which the FP registers need to be restored.
4988 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4990 PR tree-optimization/59519
4991 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4992 ICE if get_current_def (current_new_name) is already non-NULL, as long
4993 as it is a phi result of some other phi in *new_exit_bb that has
4996 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4997 or vmovdqu* for misaligned_operand.
4998 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4999 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5000 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5001 aligned_mem for AVX512F masked aligned load and store builtins and for
5004 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5006 PR tree-optimization/59651
5007 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5008 Address range for negative step should be added by TYPE_SIZE_UNIT.
5010 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5012 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5013 overlapping registers also for registers other than the stack pointer.
5015 2014-01-03 Marek Polacek <polacek@redhat.com>
5018 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5021 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5024 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5027 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5028 (push splitter): Use <P:MODE_SIZE> instead of
5029 GET_MODE_SIZE (<P:MODE>mode).
5030 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5031 (mov -1, reg peephole2): Likewise.
5032 * config/i386/sse.md (*mov<mode>_internal,
5033 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5034 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5035 *<code><mode>3, *andnot<mode>3<mask_name>,
5036 <mask_codefor><code><mode>3<mask_name>): Likewise.
5037 * config/i386/subst.md (mask_mode512bit_condition,
5038 sd_mask_mode512bit_condition): Likewise.
5040 2014-01-02 Xinliang David Li <davidxl@google.com>
5042 PR tree-optimization/59303
5043 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5044 (dump_predicates): Better output format.
5045 (pred_equal_p): New function.
5046 (is_neq_relop_p): Ditto.
5047 (is_neq_zero_form_p): Ditto.
5048 (pred_expr_equal_p): Ditto.
5049 (pred_neg_p): Ditto.
5050 (simplify_pred): Ditto.
5051 (simplify_preds_2): Ditto.
5052 (simplify_preds_3): Ditto.
5053 (simplify_preds_4): Ditto.
5054 (simplify_preds): Ditto.
5056 (push_to_worklist): Ditto.
5057 (get_pred_info_from_cmp): Ditto.
5058 (is_degenerated_phi): Ditto.
5059 (normalize_one_pred_1): Ditto.
5060 (normalize_one_pred): Ditto.
5061 (normalize_one_pred_chain): Ditto.
5062 (normalize_preds): Ditto.
5063 (normalize_cond_1): Remove function.
5064 (normalize_cond): Ditto.
5065 (is_gcond_subset_of): Ditto.
5066 (is_subset_of_any): Ditto.
5067 (is_or_set_subset_of): Ditto.
5068 (is_and_set_subset_of): Ditto.
5069 (is_norm_cond_subset_of): Ditto.
5070 (pred_chain_length_cmp): Ditto.
5071 (convert_control_dep_chain_into_preds): Type change.
5072 (find_predicates): Ditto.
5073 (find_def_preds): Ditto.
5074 (destroy_predicates_vecs): Ditto.
5075 (find_matching_predicates_in_rest_chains): Ditto.
5076 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5077 (is_pred_expr_subset): Ditto.
5078 (is_pred_chain_subset_of): Ditto.
5079 (is_included_in): Ditto.
5080 (is_superset_of): Ditto.
5082 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5084 Update copyright years.
5086 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5088 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5089 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5090 config/arc/arc.md, config/arc/arc.opt,
5091 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5092 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5093 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5094 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5095 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5096 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5097 vtable-verify.c, vtable-verify.h: Use the standard form for the
5100 2014-01-02 Tobias Burnus <burnus@net-b.de>
5102 * gcc.c (process_command): Update copyright notice dates.
5103 * gcov-dump.c: Ditto.
5105 * doc/cpp.texi: Bump @copying's copyright year.
5106 * doc/cppinternals.texi: Ditto.
5107 * doc/gcc.texi: Ditto.
5108 * doc/gccint.texi: Ditto.
5109 * doc/gcov.texi: Ditto.
5110 * doc/install.texi: Ditto.
5111 * doc/invoke.texi: Ditto.
5113 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5115 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5117 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5119 * config/i386/sse.md (*mov<mode>_internal): Guard
5120 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5122 PR rtl-optimization/59647
5123 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5124 new_rtx into UNSIGNED_FLOAT rtxes.
5126 Copyright (C) 2014 Free Software Foundation, Inc.
5128 Copying and distribution of this file, with or without modification,
5129 are permitted in any medium without royalty provided the copyright
5130 notice and this notice are preserved.