1 2014-03-18 Richard Biener <rguenther@suse.de>
3 * doc/invoke.texi (flto): Update for changes in 4.9.
5 2014-03-18 Richard Biener <rguenther@suse.de>
7 * doc/loop.texi: Remove section on the removed lambda framework.
8 Update loop docs with recent changes in preserving loop structure.
10 2014-03-18 Richard Biener <rguenther@suse.de>
12 * doc/lto.texi (-fresolution): Document.
14 2014-03-18 Richard Biener <rguenther@suse.de>
16 * doc/contrib.texi: Adjust my name.
18 2014-03-18 Jakub Jelinek <jakub@redhat.com>
21 * internal-fn.c: Include diagnostic-core.h.
22 (expand_BUILTIN_EXPECT): New function.
23 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
24 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
25 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
26 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
28 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
29 Revert 3 argument __builtin_expect code.
30 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
31 * gimple-fold.c (gimple_fold_call): Likewise.
32 * tree.h (fold_builtin_expect): New prototype.
33 * builtins.c (build_builtin_expect_predicate): Add predictor
34 argument, if non-NULL, create 3 argument __builtin_expect.
35 (fold_builtin_expect): No longer static. Add ARG2 argument,
36 pass it through to build_builtin_expect_predicate.
37 (fold_builtin_2): Adjust caller.
38 (fold_builtin_3): Handle BUILT_IN_EXPECT.
39 * internal-fn.def (BUILTIN_EXPECT): New.
41 2014-03-18 Tobias Burnus <burnus@net-b.de>
44 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
45 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
46 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
48 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
51 * predict.c (combine_predictions_for_bb): Fix up formatting.
52 (expr_expected_value_1, expr_expected_value): Add predictor argument,
53 fill what it points to if non-NULL.
54 (tree_predict_by_opcode): Adjust caller, use the predictor.
55 * predict.def (PRED_COMPARE_AND_SWAP): Add.
57 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
59 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
60 proper constant for the store mode.
62 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
64 * symtab.c (change_decl_assembler_name): Fix transparent alias
67 2014-03-16 Renlin Li <Renlin.Li@arm.com>
69 * config/aarch64/aarch64.c: Correct the comments about the
72 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
74 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
75 check for GF_OMP_FOR_KIND_FOR.
77 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
79 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
80 ymm and zmm register names.
82 2014-03-17 Jakub Jelinek <jakub@redhat.com>
85 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
86 note creation for the 2010-08-31 changes.
88 2014-03-17 Marek Polacek <polacek@redhat.com>
91 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
92 as -fno-tree-loop-vectorize.
93 (expand_omp_simd): Likewise.
95 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
97 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
98 (eligible_for_call_delay): New prototype.
99 * config/sparc/sparc.c (tls_call_delay): Rename into...
100 (eligible_for_call_delay): ...this. Return false if the instruction
101 cannot be put in the delay slot of a branch.
102 (eligible_for_restore_insn): Simplify.
103 (eligible_for_return_delay): Return false if the instruction cannot be
104 put in the delay slot of a branch and simplify.
105 (eligible_for_sibcall_delay): Return false if the instruction cannot be
106 put in the delay slot of a branch.
107 * config/sparc/sparc.md (fix_ut699): New attribute.
108 (tls_call_delay): Delete.
109 (in_call_delay): Reimplement.
110 (eligible_for_sibcall_delay): Rename into...
111 (in_sibcall_delay): ...this.
112 (eligible_for_return_delay): Rename into...
113 (in_return_delay): ...this.
114 (in_branch_delay): Reimplement.
115 (in_uncond_branch_delay): Delete.
116 (in_annul_branch_delay): Delete.
118 2014-03-14 Richard Henderson <rth@redhat.com>
121 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
122 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
123 (*floathi<X87MODEF>2_i387_with_temp): Remove.
124 (floathi splitters): Remove.
125 (float<SWI48x>xf2): New pattern.
126 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
127 code that tried to handle DImode for 32-bit, but which was excluded
128 by the pattern's condition. Drop allocation of stack temporary.
129 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
130 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
131 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
132 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
133 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
134 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
135 (*float<SWI48><MODEF>2_sse_interunit): Remove.
136 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
137 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
138 (*float<SWI48x><X87MODEF>2_i387): Remove.
139 (all float _with_temp splitters): Remove.
140 (*float<SWI48x><MODEF>2_i387): New pattern.
141 (*float<SWI48><MODEF>2_sse): New pattern.
142 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
143 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
145 2014-03-14 Jakub Jelinek <jakub@redhat.com>
146 Marek Polacek <polacek@redhat.com>
149 * common.opt (dump_base_name_prefixed): New Variable.
150 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
151 if x_dump_base_name_prefixed is already set, set it at the end.
153 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
155 PR rtl-optimization/60508
156 * lra-constraints.c (get_reload_reg): Add new parameter
158 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
159 Pass the new parameter values.
161 2014-03-14 Richard Biener <rguenther@suse.de>
163 * common.opt: Revert unintented changes from r205065.
166 2014-03-14 Richard Biener <rguenther@suse.de>
169 * cfghooks.c (split_block): Properly adjust all loops the
170 block was a latch of.
172 2014-03-14 Martin Jambor <mjambor@suse.cz>
175 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
178 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
181 * config/avr/avr.c (avr_set_current_function): Pass function name
182 through default_strip_name_encoding before sanity checking instead
183 of skipping the first char of the assembler name.
185 2014-03-13 Richard Henderson <rth@redhat.com>
188 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
189 (ix86_force_to_memory, ix86_free_from_memory): Remove.
190 * config/i386/i386-protos.h: Likewise.
191 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
192 in the expander instead of a splitter.
193 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
194 any possibility of requiring a memory.
195 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
196 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
197 (fp branch splitters): Update for ix86_split_fp_branch.
198 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
199 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
200 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
201 (*fop_<MODEF>_2_i387): Remove f/r alternative.
202 (*fop_<MODEF>_3_i387): Likewise.
203 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
204 (splitters for the fop_* register patterns): Remove.
205 (fscalexf4_i387): Rename from *fscalexf4_i387.
206 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
208 2014-03-13 Jakub Jelinek <jakub@redhat.com>
210 PR tree-optimization/59779
211 * tree-dfa.c (get_ref_base_and_extent): Use double_int
212 type for bitsize and maxsize instead of HOST_WIDE_INT.
214 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
216 PR rtl-optimization/57320
217 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
218 the CFG after thread_prologue_and_epilogue_insns.
220 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
222 PR rtl-optimization/57189
223 * lra-constraints.c (process_alt_operands): Disfavor spilling
226 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
228 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
231 2014-03-13 Jakub Jelinek <jakub@redhat.com>
233 PR tree-optimization/59025
235 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
236 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
238 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
241 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
242 calls of avr_out_plus_1.
244 2014-03-13 Bin Cheng <bin.cheng@arm.com>
246 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
247 BB's single pred and update the father loop's latch info later.
249 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
251 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
255 (VEC_base): Likewise.
256 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
257 registers, we need to swap double words in little endian mode.
259 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
260 to be a container mode for 128-bit integer operations added in ISA
261 2.07. Unlike TImode and PTImode, the preferred register set is
262 the Altivec/VMX registers for the 128-bit operations.
264 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
266 (rs6000_split_128bit_ok_p): Likewise.
268 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
269 macros for creating ISA 2.07 normal and overloaded builtin
270 functions with 3 arguments.
271 (BU_P8V_OVERLOAD_3): Likewise.
272 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
273 for use as overloaded functions.
274 (VPERM_1TI_UNS): Likewise.
275 (VSEL_1TI): Likewise.
276 (VSEL_1TI_UNS): Likewise.
277 (ST_INTERNAL_1ti): Likewise.
278 (LD_INTERNAL_1ti): Likewise.
279 (XXSEL_1TI): Likewise.
280 (XXSEL_1TI_UNS): Likewise.
281 (VPERM_1TI): Likewise.
282 (VPERM_1TI_UNS): Likewise.
283 (XXPERMDI_1TI): Likewise.
285 (LXVD2X_V1TI): Likewise.
286 (STXVD2X_V1TI): Likewise.
287 (VEC_INIT_V1TI): Likewise.
288 (VEC_SET_V1TI): Likewise.
289 (VEC_EXT_V1TI): Likewise.
290 (EQV_V1TI): Likewise.
291 (NAND_V1TI): Likewise.
292 (ORC_V1TI): Likewise.
293 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
294 added in ISA 2.07. Add both normal 'altivec' builtins, and the
298 (VADDEUQM): Likewise.
299 (VADDECUQ): Likewise.
300 (VSUBEUQM): Likewise.
301 (VSUBECUQ): Likewise.
303 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
304 __int128_t and __uint128_t types.
305 (__uint128_type): Likewise.
306 (altivec_categorize_keyword): Add support for vector __int128_t,
307 vector __uint128_t, vector __int128, and vector unsigned __int128
308 as a container type for TImode operations that need to be done in
309 VSX/Altivec registers.
310 (rs6000_macro_to_expand): Likewise.
311 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
312 to support 128-bit integer instructions vaddcuq, vadduqm,
313 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
314 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
316 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
317 for V1TImode, and set up preferences to use VSX/Altivec registers.
318 Setup VSX reload handlers.
319 (rs6000_debug_reg_global): Likewise.
320 (rs6000_init_hard_regno_mode_ok): Likewise.
321 (rs6000_preferred_simd_mode): Likewise.
322 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
323 (easy_altivec_constant): Likewise.
324 (output_vec_const_move): Likewise.
325 (rs6000_expand_vector_set): Convert V1TImode set and extract to
327 (rs6000_expand_vector_extract): Likewise.
328 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
330 (rs6000_const_vec): Add support for V1TImode.
331 (rs6000_emit_le_vsx_load): Swap double words when loading or
332 storing TImode/V1TImode.
333 (rs6000_emit_le_vsx_store): Likewise.
334 (rs6000_emit_le_vsx_move): Likewise.
335 (rs6000_emit_move): Add support for V1TImode.
336 (altivec_expand_ld_builtin): Likewise.
337 (altivec_expand_st_builtin): Likewise.
338 (altivec_expand_vec_init_builtin): Likewise.
339 (altivec_expand_builtin): Likewise.
340 (rs6000_init_builtins): Add support for V1TImode type. Add
341 support for ISA 2.07 128-bit integer builtins. Define type names
342 for the VSX/Altivec vector types.
343 (altivec_init_builtins): Add support for overloaded vector
344 functions with V1TImode type.
345 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
346 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
348 (rs6000_split_128bit_ok_p): Likewise.
349 (rs6000_handle_altivec_attribute): Create V1TImode from vector
350 __int128_t and vector __uint128_t.
352 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
360 (VS_scalar): Likewise.
361 (VS_double): Likewise.
362 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
364 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
365 we support the ISA 2.07 128-bit integer arithmetic instructions.
366 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
367 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
368 and TImode types for use with the builtin functions.
369 (V1TI_type_node): Likewise.
370 (unsigned_V1TI_type_node): Likewise.
371 (intTI_type_internal_node): Likewise.
372 (uintTI_type_internal_node): Likewise.
374 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
375 128-bit builtin functions.
376 (UNSPEC_VADDEUQM): Likewise.
377 (UNSPEC_VADDECUQ): Likewise.
378 (UNSPEC_VSUBCUQ): Likewise.
379 (UNSPEC_VSUBEUQM): Likewise.
380 (UNSPEC_VSUBECUQ): Likewise.
381 (VM): Add V1TImode to vector mode iterators.
384 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
385 (altivec_vaddcuq): Likewise.
386 (altivec_vsubuqm): Likewise.
387 (altivec_vsubcuq): Likewise.
388 (altivec_vaddeuqm): Likewise.
389 (altivec_vaddecuq): Likewise.
390 (altivec_vsubeuqm): Likewise.
391 (altivec_vsubecuq): Likewise.
393 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
395 (BOOL_128): Likewise.
396 (BOOL_REGS_OUTPUT): Likewise.
397 (BOOL_REGS_OP1): Likewise.
398 (BOOL_REGS_OP2): Likewise.
399 (BOOL_REGS_UNARY): Likewise.
400 (BOOL_REGS_AND_CR0): Likewise.
402 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
403 128-bit integer builtin support.
404 (vec_vadduqm): Likewise.
405 (vec_vaddecuq): Likewise.
406 (vec_vaddeuqm): Likewise.
407 (vec_vsubecuq): Likewise.
408 (vec_vsubeuqm): Likewise.
409 (vec_vsubcuq): Likewise.
410 (vec_vsubuqm): Likewise.
412 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
413 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
414 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
415 128-bit integer add/subtract to ISA 2.07.
417 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
419 * config/arc/arc.c (arc_predicate_delay_insns):
420 Fix third argument passed to conditionalize_nonjump.
422 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
424 * config/aarch64/aarch64-builtins.c
425 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
426 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
427 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
428 instead of __builtin_lfloor.
429 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
431 2014-03-12 Jakub Jelinek <jakub@redhat.com>
433 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
434 (tree_ssa_ifcombine_bb_1): New function.
435 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
436 is an empty forwarder block to then_bb or vice versa and then_bb
437 and else_bb are effectively swapped.
439 2014-03-12 Christian Bruel <christian.bruel@st.com>
442 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
443 REG_CFA_DEF_CFA note.
444 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
445 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
447 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
449 PR tree-optimization/60454
450 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
452 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
454 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
455 Do not define target_cpu_default2 to generic.
456 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
457 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
458 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
460 2014-03-12 Jakub Jelinek <jakub@redhat.com>
461 Marc Glisse <marc.glisse@inria.fr>
463 PR tree-optimization/60502
464 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
465 instead of build_low_bits_mask.
467 2014-03-12 Jakub Jelinek <jakub@redhat.com>
470 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
471 if there are multiple uses, but op doesn't live on E edge.
472 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
473 clobber stmts before __builtin_unreachable.
475 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
477 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
478 hard_frame_pointer_rtx.
479 * cse.c (cse_insn): Remove volatile check.
480 * cselib.c (cselib_process_insn): Likewise.
481 * dse.c (scan_insn): Likewise.
483 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
485 * config/arc/arc.c (conditionalize_nonjump): New function,
487 (arc_ifcvt): ... this.
488 (arc_predicate_delay_insns): Use it.
490 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
492 * config/arc/predicates.md (extend_operand): During/after reload,
493 allow const_int_operand.
494 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
495 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
496 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
498 (umulsi3_highpart_i): Likewise.
500 2014-03-11 Richard Biener <rguenther@suse.de>
502 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
503 Add asserts to guard possible wrong-code bugs.
505 2014-03-11 Richard Biener <rguenther@suse.de>
507 PR tree-optimization/60429
508 PR tree-optimization/60485
509 * tree-ssa-structalias.c (set_union_with_increment): Properly
510 take into account all fields that overlap the shifted vars.
511 (do_sd_constraint): Likewise.
512 (do_ds_constraint): Likewise.
513 (get_constraint_for_ptr_offset): Likewise.
515 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
517 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
518 (nios2_compute_frame_layout):
519 Add calculation of cfun->machine->fp_save_offset.
520 (nios2_expand_prologue): Correct setting of frame pointer register
522 (nios2_expand_epilogue): Update recovery of stack pointer from
523 frame pointer accordingly.
524 (nios2_initial_elimination_offset): Update calculation of offset
525 for eliminating to HARD_FRAME_POINTER_REGNUM.
527 2014-03-10 Jakub Jelinek <jakub@redhat.com>
530 * ipa.c (symtab_remove_unreachable_nodes): Don't call
531 cgraph_get_create_node on VAR_DECLs.
533 2014-03-10 Richard Biener <rguenther@suse.de>
536 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
538 2014-03-08 Douglas B Rupp <rupp@gnat.com>
540 * config/vms/vms.opt (vms_float_format): New variable.
542 2014-03-08 Tobias Burnus <burnus@net-b.de>
544 * doc/invoke.texi (-fcilkplus): Update implementation status.
546 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
547 Richard Biener <rguenther@suse.de>
549 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
550 consistently accross all TUs.
551 (run_gcc): Enable -fshort-double automatically at link at link-time
552 and disallow override.
554 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
557 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
558 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
559 if they can't be used.
561 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
563 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
564 for Solaris 11/x86 ld.
565 * configure: Regenerate.
567 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
569 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
570 (LIB_TLS_SPEC): Save as ld_tls_libs.
571 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
572 (HAVE_AS_IX86_TLSLDM): New test.
573 * configure, config.in: Regenerate.
574 * config/i386/i386.c (legitimize_tls_address): Fall back to
575 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
576 cannot support TLS_MODEL_LOCAL_DYNAMIC.
577 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
578 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
580 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
582 * common.opt (fira-loop-pressure): Mark as optimization.
584 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
586 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
587 an OpenMP mappable type.
589 2014-03-06 Matthias Klose <doko@ubuntu.com>
591 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
592 MULTILIB_OSDIRNAMES is not defined.
594 2014-03-06 Jakub Jelinek <jakub@redhat.com>
595 Meador Inge <meadori@codesourcery.com>
598 * config/arm/arm.c (arm_tls_symbol_p): Remove.
599 (arm_legitimize_address): Call legitimize_tls_address for any
600 arm_tls_referenced_p expression, handle constant addend. Call it
601 before testing for !TARGET_ARM.
602 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
604 2014-03-06 Richard Biener <rguenther@suse.de>
610 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
612 * tree-streamer.c (record_common_node): Assert we don't record
613 nodes with type double.
614 (preload_common_node): Skip type double, complex double and double
615 pointer since it is now frontend dependent due to fshort-double option.
617 2014-03-06 Richard Biener <rguenther@suse.de>
619 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
620 or -fno-lto is specified and the linker has full plugin support.
621 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
622 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
623 * lto-wrapper.c (merge_and_complain): Merge compile-time
625 (run_gcc): And pass it through to the link options.
627 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
631 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
633 * cselib.c (remove_useless_values): Skip to avoid quadratic
634 behavior if the condition moved from...
635 (cselib_process_insn): ... here holds.
637 2014-03-05 Jakub Jelinek <jakub@redhat.com>
640 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
641 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
644 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
645 (TM_H): Add x86-tune.def.
647 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
649 * config/aarch64/aarch64.c (generic_tunings):
650 Use cortexa57_extra_costs.
652 2014-03-05 Jakub Jelinek <jakub@redhat.com>
655 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
656 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
657 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
660 2014-03-04 Heiher <r@hev.cc>
662 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
663 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
665 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
667 * config/i386/predicates.md (const2356_operand): Change to ...
668 (const2367_operand): ... this.
669 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
671 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
672 (*avx512pf_scatterpf<mode>sf): Ditto.
673 (avx512pf_scatterpf<mode>df): Ditto.
674 (*avx512pf_scatterpf<mode>df_mask): Ditto.
675 (*avx512pf_scatterpf<mode>df): Ditto.
676 * config/i386/i386.c (ix86_expand_builtin): Update
677 incorrect hint operand error message.
679 2014-03-04 Richard Biener <rguenther@suse.de>
681 * lto-section-in.c (lto_get_section_data): Fix const cast.
683 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
685 * tree-streamer.c (record_common_node): Assert we don't record
686 nodes with type double.
687 (preload_common_node): Skip type double, complex double and double
688 pointer since it is now frontend dependent due to fshort-double option.
690 2014-03-04 Richard Biener <rguenther@suse.de>
693 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
694 (lto_input_toplevel_asms): Likewise.
695 * lto-section-in.c (lto_get_section_data): Instead do it here
698 2014-03-04 Richard Biener <rguenther@suse.de>
700 PR tree-optimization/60382
701 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
702 dead PHIs a reduction.
704 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
706 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
708 (_mm_prefetch): Move out of GCC target("sse") pragma.
709 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
710 GCC target("prfchw") pragma.
711 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
713 * config/i386/i386.c (ix86_option_override_internal): Enable
714 -mprfchw with -mprefetchwt1.
716 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
718 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
721 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
723 * opts.h (CL_PCH_IGNORE): Define.
724 * targhooks.c (option_affects_pch_p):
725 Return false for options that have CL_PCH_IGNORE set.
726 * opt-functions.awk: Process PchIgnore.
727 * doc/options.texi: Document PchIgnore.
729 * config/arc/arc.opt (misize): Add PchIgnore property.
731 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
733 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
734 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
735 constraint on constants to permit them being loaded into
736 GENERAL_REGS or BASE_REGS.
738 2014-03-03 Nick Clifton <nickc@redhat.com>
740 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
741 anti-cacnonical alternatives.
742 (negandhi3_real): New pattern.
743 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
745 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
747 * config/avr/avr-mcus.def: Remove atxmega16x1.
748 * config/avr/avr-tables.opt: Regenerate.
749 * config/avr/t-multilib: Regenerate.
750 * doc/avr-mmcu.texi: Regenerate.
752 2014-03-03 Tobias Grosser <tobias@grosser.es>
753 Mircea Namolaru <mircea.namolaru@inria.fr>
755 PR tree-optimization/58028
756 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
759 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
761 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
762 not handled by recognizers.
764 2014-03-03 Jakub Jelinek <jakub@redhat.com>
767 * function.c (expand_function_end): Don't emit
768 clobber_return_register sequence if clobber_after is a BARRIER.
769 * cfgexpand.c (construct_exit_block): Append instructions before
770 return_label to prev_bb.
772 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
774 * config/rs6000/constraints.md: Document reserved use of "wc".
776 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
779 * ipa.c (function_and_variable_visibility): When dissolving comdat
780 group, also set all symbols to local.
782 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
787 2013-12-14 Jan Hubicka <jh@suse.cz>
789 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
791 2014-03-02 Jon Beniston <jon@beniston.com>
797 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
798 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
799 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
800 (simple_return, *simple_return): New patterns
801 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
802 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
804 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
806 * dwarf2out.c (gen_subprogram_die): Tidy.
808 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
811 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
812 (*mov_t_msb_neg_negc): ... this new insn.
814 2014-02-28 Jason Merrill <jason@redhat.com>
817 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
820 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
823 * dwarf2out.c (decltype_auto_die): New static.
824 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
825 (gen_type_die_with_usage): Handle 'decltype(auto)'.
826 (is_cxx_auto): Likewise.
828 2014-02-28 Ian Bolton <ian.bolton@arm.com>
830 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
831 we are not using general regs only.
833 2014-02-28 Richard Biener <rguenther@suse.de>
836 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
837 previous fix and only allow to remove trivial pre-headers
838 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
839 (remove_forwarder_block): Properly update the latch of a loop.
841 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
844 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
845 (cselib_preserved_hash_table): New.
846 (preserve_constants_and_equivs): Move preserved vals to it.
847 (cselib_find_slot): Look it up first.
848 (cselib_init): Initialize it.
849 (cselib_finish): Release it.
850 (dump_cselib_table): Dump it.
852 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
855 * cselib.c (remove_useless_values): Skip to avoid quadratic
856 behavior if the condition moved from...
857 (cselib_process_insn): ... here holds.
859 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
862 * var-tracking.c (vt_initialize): Apply the same condition to
863 preserve the CFA base value.
865 2014-02-28 Joey Ye <joey.ye@arm.com>
868 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
869 if reload in progress or completed.
871 2014-02-28 Tobias Burnus <burnus@net-b.de>
874 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
877 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
879 * doc/tm.texi.in (Condition Code Status): Update documention for
880 relative locations of cc0-setter and cc0-user.
882 2014-02-27 Jeff Law <law@redhat.com>
884 PR rtl-optimization/52714
885 * combine.c (try_combine): When splitting an unrecognized PARALLEL
886 into two independent simple sets, if I3 is a jump, ensure the
887 pattern we place into I3 is a (set (pc) ...).
889 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
890 Jeff Law <law@redhat.com>
892 PR rtl-optimization/49847
893 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
894 are in different blocks.
895 * doc/tm.texi (Condition Code Status): Update documention for
896 relative locations of cc0-setter and cc0-user.
898 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
901 * lra.c (lra_emit_add): Check SUBREG too.
903 2014-02-27 Andreas Schwab <schwab@suse.de>
905 * config/m68k/m68k.c (m68k_option_override): Disable
906 -flive-range-shrinkage for classic m68k.
907 (m68k_override_options_after_change): Likewise.
909 2014-02-27 Marek Polacek <polacek@redhat.com>
912 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
913 -Wmaybe-uninitialized.
915 2014-02-27 Alan Modra <amodra@gmail.com>
918 * reload1.c (emit_input_reload_insns): When reload_override_in,
919 set old to rl->in_reg when rl->in_reg is a subreg.
921 2014-02-26 Richard Biener <rguenther@suse.de>
924 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
926 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
928 * common/config/i386/predicates.md (const1256_operand): Remove.
929 (const2356_operand): New.
930 (const_1_to_2_operand): Remove.
931 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
932 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
933 (*avx512pf_gatherpf<mode>sf): Ditto.
934 (avx512pf_gatherpf<mode>df): Ditto.
935 (*avx512pf_gatherpf<mode>df_mask): Ditto.
936 (*avx512pf_gatherpf<mode>df): Ditto.
937 (avx512pf_scatterpf<mode>sf): Ditto.
938 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
939 (*avx512pf_scatterpf<mode>sf): Ditto.
940 (avx512pf_scatterpf<mode>df): Ditto.
941 (*avx512pf_scatterpf<mode>df_mask): Ditto.
942 (*avx512pf_scatterpf<mode>df): Ditto.
943 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
945 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
947 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
948 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
949 (_mm512_mask_testn_epi64_mask): Move to ...
950 * config/i386/avx512cdintrin.h: Here.
951 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
952 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
953 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
954 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
955 TARGET_AVX512F from TARGET_AVX512CD.
957 2014-02-26 Richard Biener <rguenther@suse.de>
960 * ipa.c (walk_polymorphic_call_targets): Properly guard
961 call to inline_update_overall_summary.
963 2014-02-26 Bin Cheng <bin.cheng@arm.com>
966 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
967 and latches only if requested. Fix latch if it is removed.
968 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
969 LOOPS_HAVE_PREHEADERS.
971 2014-02-25 Andrew Pinski <apinski@cavium.com>
973 * builtins.c (expand_builtin_thread_pointer): Create a new target
974 when the target is NULL.
976 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
978 PR rtl-optimization/60317
979 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
980 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
981 * lra-assigns.c: Include params.h.
982 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
983 other reload pseudos considerations.
985 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
987 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
988 to use canonical form for nor<mode>3.
990 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
993 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
996 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
998 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
999 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1000 (ix86_handle_option): Handle OPT_mprefetchwt1.
1001 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1002 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1004 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1005 OPTION_MASK_ISA_PREFETCHWT1.
1006 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1007 (PTA_PREFETCHWT1): New.
1008 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1009 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1010 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1011 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1012 (*prefetch_avx512pf_<mode>_: Change into ...
1013 (*prefetch_prefetchwt1_<mode>: This.
1014 * config/i386/i386.opt (mprefetchwt1): New.
1015 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1016 (_mm_prefetch): Handle intent to write.
1017 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1019 2014-02-25 Richard Biener <rguenther@suse.de>
1022 * emit-rtl.c (mem_attrs_htab): Remove.
1023 (mem_attrs_htab_hash): Likewise.
1024 (mem_attrs_htab_eq): Likewise.
1025 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1026 (init_emit_once): Do not allocate mem_attrs_htab.
1028 2014-02-25 Richard Biener <rguenther@suse.de>
1031 * lto-opts.c (lto_write_options): Output non-explicit conservative
1032 -fwrapv, -fno-trapv and -fno-strict-overflow.
1033 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1034 (run_gcc): And pass them through.
1036 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1038 * sel-sched.c (calculate_new_fences): New parameter ptime.
1039 Calculate it as a maximum over all fence cycles.
1040 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1041 Print the final schedule timing when sched_verbose.
1043 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1045 PR rtl-optimization/60292
1046 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1047 bit fot the fence instruction.
1049 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1051 * calls.h: Fix typo in comment.
1053 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1055 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1056 adjusting offsetable addresses.
1058 2014-02-24 Guozhi Wei <carrot@google.com>
1060 * sparseset.h (sparseset_pop): Fix the wrong index.
1062 2014-02-24 Walter Lee <walt@tilera.com>
1064 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1065 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1067 * common/config/tilegx/tilegx-common.c
1068 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1069 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1071 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1072 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1073 (tilegx_gimplify_va_arg_expr): Handle big endian.
1074 (tilegx_expand_unaligned_load): Ditto.
1075 (tilegx_expand_unaligned_store): Ditto.
1076 (TARGET_RETURN_IN_MSB): New.
1077 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1078 (TARGET_ENDIAN_DEFAULT): New.
1079 (TARGET_BIG_ENDIAN): Handle big endian.
1080 (BYTES_BIG_ENDIAN): Ditto.
1081 (WORDS_BIG_ENDIAN): Ditto.
1082 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1085 * config/tilegx/tilegx.md (extv): Handle big endian.
1087 (insn_st<n>): Ditto.
1088 (insn_st<n>_add<bitsuffix>): Ditto.
1089 (insn_stnt<n>): Ditto.
1090 (insn_stnt<n>_add<bitsuffix>):Ditto.
1091 (vec_interleave_highv8qi): Handle big endian.
1092 (vec_interleave_highv8qi_be): New.
1093 (vec_interleave_highv8qi_le): New.
1094 (insn_v1int_h): Handle big endian.
1095 (vec_interleave_lowv8qi): Handle big endian.
1096 (vec_interleave_lowv8qi_be): New.
1097 (vec_interleave_lowv8qi_le): New.
1098 (insn_v1int_l): Handle big endian.
1099 (vec_interleave_highv4hi): Handle big endian.
1100 (vec_interleave_highv4hi_be): New.
1101 (vec_interleave_highv4hi_le): New.
1102 (insn_v2int_h): Handle big endian.
1103 (vec_interleave_lowv4hi): Handle big endian.
1104 (vec_interleave_lowv4hi_be): New.
1105 (vec_interleave_lowv4hi_le): New.
1106 (insn_v2int_l): Handle big endian.
1107 (vec_interleave_highv2si): Handle big endian.
1108 (vec_interleave_highv2si_be): New.
1109 (vec_interleave_highv2si_le): New.
1110 (insn_v4int_h): Handle big endian.
1111 (vec_interleave_lowv2si): Handle big endian.
1112 (vec_interleave_lowv2si_be): New.
1113 (vec_interleave_lowv2si_le): New.
1114 (insn_v4int_l): Handle big endian.
1115 * config/tilegx/tilegx.opt (mbig-endian): New option.
1116 (mlittle-endian): New option.
1117 * doc/install.texi: Document tilegxbe-linux.
1118 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1120 2014-02-24 Martin Jambor <mjambor@suse.cz>
1123 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1124 there are no parameter descriptors.
1126 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1128 PR rtl-optimization/60268
1129 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1130 initialization to ...
1131 (sched_rgn_init): ... here.
1132 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1134 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1136 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1139 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1141 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1144 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1146 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1147 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1149 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1151 * config/microblaze/predicates.md: Add cmp_op predicate.
1152 * config/microblaze/microblaze.md: Add branch_compare instruction
1153 which uses cmp_op predicate and emits cmp insn before branch.
1154 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1155 to microblaze_expand_conditional_branch and consolidate logic.
1156 (microblaze_expand_conditional_branch): emit branch_compare
1157 insn instead of handling cmp op separate from branch insn.
1159 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1161 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1164 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1166 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1167 define_insn with define_expand and new define_insn
1168 *altivec_lve<VI_char>x_internal.
1169 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1170 and new define_insn *altivec_stve<VI_char>x_internal.
1171 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1173 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1175 (altivec_expand_stvex_be): New function.
1177 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1179 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1180 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1181 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1182 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1184 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1187 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1188 instead of emit_move_insn.
1190 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1192 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1194 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1195 gen_altivec_vsumsws.
1197 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1199 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1200 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1201 (altivec_lvxl_<mode>): New define_expand incorporating
1202 -maltivec=be semantics where needed.
1203 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1204 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1205 semantics where needed.
1206 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1207 (altivec_stvx_<mode>): New define_expand incorporating
1208 -maltivec=be semantics where needed.
1209 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1210 VM2 iterator instead of V4SI.
1211 (altivec_stvxl_<mode>): New define_expand incorporating
1212 -maltivec=be semantics where needed.
1213 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1214 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1215 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1216 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1217 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1218 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1219 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1220 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1221 ALTIVEC_BUILTIN_STVXL.
1222 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1223 (altivec_expand_stvx_be): Likewise.
1224 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1225 (altivec_expand_lvx_be): Likewise.
1226 (altivec_expand_stvx_be): Likewise.
1227 (altivec_expand_builtin): Add cases for
1228 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1229 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1230 (altivec_init_builtins): Add definitions for
1231 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1232 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1234 2014-02-21 Catherine Moore <clm@codesourcery.com>
1236 * doc/invoke.texi (mvirt, mno-virt): Document.
1237 * config/mips/mips.opt (mvirt): New option.
1238 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1240 2014-02-21 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/60276
1243 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1244 (STMT_VINFO_MIN_NEG_DIST): New macro.
1245 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1246 STMT_VINFO_MIN_NEG_DIST.
1247 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1248 made for negative dependence distances still hold.
1250 2014-02-21 Richard Biener <rguenther@suse.de>
1253 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1254 DECL_INITIAL for globals not in the current function context.
1256 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1258 PR tree-optimization/56490
1259 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1260 * tree-ssa-uninit.c: Include params.h.
1261 (compute_control_dep_chain): Add num_calls argument, return false
1262 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1263 num_calls to recursive call.
1264 (find_predicates): Change dep_chain into normal array,
1265 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1266 variable and adjust compute_control_dep_chain caller.
1267 (find_def_preds): Likewise.
1269 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1271 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1272 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1274 2014-02-21 Nick Clifton <nickc@redhat.com>
1276 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1277 (pushhi1): Likewise.
1278 (popqi1): Add mode to pre_dec.
1281 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1283 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1284 mode for mask of V8SFmode permutation.
1286 2014-02-20 Richard Henderson <rth@redhat.com>
1289 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1290 a new pseudo for OLDVAL.
1292 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1295 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1296 gen_reg_rtx if d->testing_p.
1297 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1298 if d->testing_p and we will certainly return true.
1299 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1302 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1304 * emit-rtl.c (gen_reg_rtx): Assert that
1305 crtl->emit.regno_pointer_align_length is non-zero.
1307 2014-02-20 Richard Henderson <rth@redhat.com>
1310 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1311 on failure the store back into EXPECT.
1313 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1314 Sandra Loosemore <sandra@codesourcery.com>
1316 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1317 * config/nios2/nios2.c (nios2_function_profiler): Add
1318 -fPIC (flag_pic == 2) support.
1319 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1320 (nios2_large_offset_p): New function.
1321 (nios2_unspec_reloc_p): Move up position, update to use
1322 nios2_large_offset_p.
1323 (nios2_unspec_address): Remove function.
1324 (nios2_unspec_offset): New function.
1325 (nios2_large_got_address): New function.
1326 (nios2_got_address): Add large offset support.
1327 (nios2_legitimize_tls_address): Update usage of removed and new
1329 (nios2_symbol_binds_local_p): New function.
1330 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1331 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1332 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1333 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1334 processing for (const (unspec ...)).
1335 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1337 2014-02-20 Richard Biener <rguenther@suse.de>
1339 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1340 doing the substitution.
1341 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1343 2014-02-20 Martin Jambor <mjambor@suse.cz>
1346 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1347 info when checking whether lattices are bottom.
1349 2014-02-20 Richard Biener <rguenther@suse.de>
1352 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1355 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1358 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1359 parameter specifying the scaling.
1360 (inline_call): Update.
1361 (want_inline_recursively): Guard division by zero.
1362 (recursive_inlining): Update.
1363 * ipa-inline.h (clone_inlined_nodes): Update.
1365 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1368 * config/i386/i386.c (classify_argument): Pass structures of size
1369 64 bytes or less in register.
1371 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1372 Kirill Yukhin <kirill.yukhin@intel.com>
1374 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1375 (_mm_rcp28_round_ss): Ditto.
1376 (_mm_rsqrt28_round_sd): Ditto.
1377 (_mm_rsqrt28_round_ss): Ditto.
1378 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1379 (_mm_rcp14_round_ss): Ditto.
1380 (_mm_rsqrt14_round_sd): Ditto.
1381 (_mm_rsqrt14_round_ss): Ditto.
1382 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1383 the first input operand, get rid of match_dup.
1384 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1386 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1388 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1389 operand as the first input operand, set type attribute.
1390 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1392 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1393 operand as the first input operand, set type attribute.
1395 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1397 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1400 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1403 * config/i386/i386.c (construct_container): Remove TFmode check
1404 for X86_64_INTEGER_CLASS.
1406 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1409 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1410 only when -Wpsabi is enabled.
1412 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1415 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1416 passing arrays in registers are the same as for structs, so remove the
1417 special case for them.
1419 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1421 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1422 destination type, extract only the valid bits if the source type is not
1423 integral and has a different mode.
1425 2014-02-19 Richard Biener <rguenther@suse.de>
1428 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1431 2014-02-19 Richard Biener <rguenther@suse.de>
1434 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1435 (ipa_modify_call_arguments): Emit an argument load explicitely and
1436 preserve virtual SSA form there and for the replacement call.
1437 Do not update SSA form nor free dominance info.
1439 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1441 * ipa.c (function_and_variable_visibility): Also clear WEAK
1442 flag when disolving COMDAT_GROUP.
1444 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1446 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1447 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1449 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1450 do more sanity checks.
1451 (detect_type_change): Return true when giving up early.
1452 (compute_complex_assign_jump_func): Fix type parameter of
1453 ipa_set_ancestor_jf.
1454 (compute_complex_ancestor_jump_func): Likewise.
1455 (update_jump_functions_after_inlining): Fix updating of
1457 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1459 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1461 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1462 inline clones when edge disappears.
1464 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1467 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1468 Split 64-bit moves into 2 patterns. Do not allow the use of
1469 direct move for TDmode in little endian, since the decimal value
1470 has little endian bytes within a word, but the 64-bit pieces are
1471 ordered in a big endian fashion, and normal subreg's of TDmode are
1473 (mov<mode>_64bit_dm): Likewise.
1474 (movtd_64bit_nodm): Likewise.
1476 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1478 PR tree-optimization/60174
1479 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1480 statement of an SSA_NAME that occurs in an abnormal PHI node.
1482 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1485 * final.c (SEEN_BB): Remove.
1486 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1487 (final_scan_insn): Don't force_source_line on second
1488 NOTE_INSN_BASIC_BLOCK.
1490 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1493 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1494 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1495 (type_natural_mode): Warn ABI change when %zmm register is not
1496 available for AVX512F vector value passing.
1498 2014-02-18 Kai Tietz <ktietz@redhat.com>
1501 * config/i386/i386.c (ix86_expand_prologue): Use value in
1502 rax register as displacement when restoring %r10 or %rax.
1503 Fix wrong offset when restoring both registers.
1505 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1507 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1508 assertion with conditional return.
1510 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1511 Uros Bizjak <ubizjak@gmail.com>
1514 * config/i386/driver-i386.c (host_detect_local_cpu): If
1515 YMM state is not saved by the OS, also clear has_f16c. Move
1516 CPUID 0x80000001 handling before YMM state saving checking.
1518 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1520 PR rtl-optimization/58960
1521 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1522 factored out from ...
1523 (sched_init): ... here.
1524 (free_global_sched_pressure_data): New, factored out from ...
1525 (sched_finish): ... here.
1526 * sched-int.h (free_global_sched_pressure_data): Declare.
1527 * sched-rgn.c (nr_regions_initial): New static global.
1528 (haifa_find_rgns): Initialize it.
1529 (schedule_region): Disable sched-pressure for the newly
1532 2014-02-17 Richard Biener <rguenther@suse.de>
1534 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1535 release SSA defs of pattern stmts.
1537 2014-02-17 Richard Biener <rguenther@suse.de>
1539 * tree-inline.c (expand_call_inline): Release the virtual
1540 operand defined by the call we are about to inline.
1542 2014-02-17 Richard Biener <rguenther@suse.de>
1544 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1546 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1547 Ilya Tocar <ilya.tocar@intel.com>
1549 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1550 arguments order in builtin.
1551 (_mm512_permutexvar_epi64): Ditto.
1552 (_mm512_mask_permutexvar_epi64): Ditto
1553 (_mm512_maskz_permutexvar_epi32): Ditto
1554 (_mm512_permutexvar_epi32): Ditto
1555 (_mm512_mask_permutexvar_epi32): Ditto
1557 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1559 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1560 (p8_vmrgow): Likewise.
1562 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1564 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1567 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1570 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1571 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1572 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1573 using direct move instructions on ISA 2.07. Also adjust
1574 instruction length for 64-bit.
1575 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1576 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1578 2014-02-15 Alan Modra <amodra@gmail.com>
1582 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1583 find_replacement on parts of insn rtl that might be reloaded.
1585 2014-02-15 Richard Biener <rguenther@suse.de>
1587 PR tree-optimization/60183
1588 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1589 (tree_ssa_phiprop): Calculate and free post-dominators.
1591 2014-02-14 Jeff Law <law@redhat.com>
1593 PR rtl-optimization/60131
1594 * ree.c (get_extended_src_reg): New function.
1595 (combine_reaching_defs): Use it rather than assuming location of REG.
1596 (find_and_remove_re): Verify first operand of extension is
1597 a REG before adding the insns to the copy list.
1599 2014-02-14 Roland McGrath <mcgrathr@google.com>
1601 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1602 * configure: Regenerated.
1603 * config.in: Regenerated.
1604 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1605 instead of ASM_SHORT.
1607 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1608 Richard Earnshaw <rearnsha@arm.com>
1610 PR rtl-optimization/59535
1611 * lra-constraints.c (process_alt_operands): Encourage alternative
1612 when unassigned pseudo class is superset of the alternative class.
1613 (inherit_reload_reg): Don't inherit when optimizing for code size.
1614 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1615 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1616 modes not less than 4 for Thumb1.
1618 2014-02-14 Kyle McMartin <kyle@redhat.com>
1621 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1623 2014-02-14 Richard Biener <rguenther@suse.de>
1625 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1626 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1627 Do not strip INDIRECT_REFs.
1629 2014-02-14 Richard Biener <rguenther@suse.de>
1632 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1633 DECL_FUNCTION_SPECIFIC_TARGET.
1634 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1635 * tree-streamer-out.c (pack_ts_target_option): Remove.
1636 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1637 (write_ts_function_decl_tree_pointers): Do not stream
1638 DECL_FUNCTION_SPECIFIC_TARGET.
1639 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1640 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1641 (lto_input_ts_function_decl_tree_pointers): Do not stream
1642 DECL_FUNCTION_SPECIFIC_TARGET.
1644 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1646 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1647 (get_initial_def_for_induction, vectorizable_induction): Ignore
1648 debug stmts when looking for exit_phi.
1649 (vectorizable_live_operation): Fix up condition.
1651 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1653 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1654 nreverse() because it changes the content of original tree list.
1656 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1658 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1659 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1661 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1663 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1664 GNU coding standards.
1666 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1669 * dwarf2out.c (gen_subprogram_die): Don't call
1670 add_calling_convention_attribute if subr_die is old_die.
1672 2014-02-13 Sharad Singhai <singhai@google.com>
1674 * doc/optinfo.texi: Fix order of nodes.
1676 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1678 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1679 operands[2], not operands[3].
1681 2014-02-13 Richard Biener <rguenther@suse.de>
1684 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1685 mention the possibility of an in-tree build.
1686 (CLooG): Update recommended version to 0.18.1, mention the
1687 possibility of an in-tree build and clarify that the ISL
1688 bundled with CLooG does not work.
1690 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1693 * expr.c (compress_float_constant): If x is a hard register,
1694 extend into a pseudo and then move to x.
1696 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1698 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1699 caused by bad second argument to warning_at() with -mhotpatch and
1700 nested functions (e.g. with gfortran).
1702 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1704 * opts.c (option_name): Remove "enabled by default" rider.
1706 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1708 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1710 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1711 Uros Bizjak <ubizjak@gmail.com>
1714 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1715 * configure: Regenerated.
1717 2014-02-12 Richard Biener <rguenther@suse.de>
1719 * vec.c (vec_prefix::calculate_allocation): Move as
1720 inline variant to vec.h.
1721 (vec_prefix::calculate_allocation_1): New out-of-line version.
1722 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1723 (vec_prefix::m_has_auto_buf): Rename to ...
1724 (vec_prefix::m_using_auto_storage): ... this.
1725 (vec_prefix::calculate_allocation): Inline the easy cases
1726 and dispatch to calculate_allocation_1 which doesn't need the
1728 (va_heap::reserve): Use gcc_checking_assert.
1729 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1730 m_using_auto_storage.
1731 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1733 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1734 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1735 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1737 2014-02-12 Richard Biener <rguenther@suse.de>
1739 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1740 when we found a dependence.
1742 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1744 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1746 (maybe_fold_stmt): ... into this new function.
1747 * omp-low.c (lower_omp): Update comment.
1749 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1752 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1755 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1757 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1758 identifiers in comments.
1759 (cortexa53_extra_costs): Likewise.
1760 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1761 (cortexa7_extra_costs): Likewise.
1762 (cortexa12_extra_costs): Likewise.
1763 (cortexa15_extra_costs): Likewise.
1764 (v7m_extra_costs): Likewise.
1766 2014-02-12 Richard Biener <rguenther@suse.de>
1769 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1770 of posix_memalign being successful.
1771 (lower_stmt): Restrict lowering of posix_memalign to when
1772 -ftree-bit-ccp is enabled.
1774 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1776 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1778 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1780 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1782 PR rtl-optimization/60116
1783 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1784 other_insn once the combination has been validated.
1786 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1789 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1791 * ipa-devirt.c: Include demangle.h
1792 (odr_violation_reported): New static variable.
1793 (add_type_duplicate): Update odr_violations.
1794 (maybe_record_node): Add completep parameter; update it.
1795 (record_target_from_binfo): Add COMPLETEP parameter;
1796 update it as needed.
1797 (possible_polymorphic_call_targets_1): Likewise.
1798 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1799 rename FINAL to COMPLETE.
1800 (record_targets_from_bases): Sanity check we found the binfo;
1801 fix COMPLETEP updating.
1802 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1803 parameter, fix computing of COMPLETEP.
1804 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1805 at LTO time do demangling.
1806 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1807 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1809 (gimple_get_virt_method_for_binfo): Likewise.
1810 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1811 gimple_get_virt_method_for_vtable): Update prototypes.
1813 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1816 * genautomata.c (add_presence_absence): Fix typo with
1817 {final_}presence_list.
1819 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1822 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1823 for VSX/Altivec vectors that land in GPR registers.
1825 2014-02-11 Richard Henderson <rth@redhat.com>
1826 Jakub Jelinek <jakub@redhat.com>
1829 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1830 around drhs if type conversion to lacc->type is not useless.
1832 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1834 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1836 (cortex-a57.cortex-a53): Likewise.
1837 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1839 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1841 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1844 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1846 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1847 add_options_for_arm_vfp3.
1849 2014-02-11 Jeff Law <law@redhat.com>
1852 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1853 object with an undesirable mode.
1855 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1858 * config/i386/sol2-9.h: New file.
1859 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1860 *-*-solaris2.9*): Use it.
1862 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1864 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1865 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1867 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1869 * config/microblaze/microblaze.c: Extend mcpu version format
1871 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1873 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1875 2014-02-10 Richard Henderson <rth@redhat.com>
1878 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1879 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1880 ms-abi vs -mno-accumulate-outgoing-args.
1881 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1882 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1885 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1888 * cfgexpand.c (expand_asm_operands): Attach source location to
1889 ASM_INPUT rtx objects.
1890 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1892 2014-02-10 Nick Clifton <nickc@redhat.com>
1894 * config/mn10300/mn10300.c (popcount): New function.
1895 (mn10300_expand_prologue): Include saved registers in stack usage
1898 2014-02-10 Jeff Law <law@redhat.com>
1901 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1902 when changing the SET_DEST of a prior insn to avoid an input reload.
1904 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1906 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1907 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1908 -mcall-openbsd, or -mcall-linux.
1909 (CC1_ENDIAN_BIG_SPEC): Remove.
1910 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1911 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1912 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1913 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1914 and %cc1_endian_default.
1915 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1917 2014-02-10 Richard Biener <rguenther@suse.de>
1919 PR tree-optimization/60115
1920 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1921 MEM_REF handling. Properly verify that the accesses are not
1922 out of the objects bound.
1924 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1926 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1929 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1931 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1932 proper constants and fix formatting.
1933 (possible_polymorphic_call_targets): Fix formatting.
1935 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1936 Ilya Tocar <ilya.tocar@intel.com>
1938 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1939 (_mm512_loadu_epi32): Renamed into...
1940 (_mm512_loadu_si512): This.
1941 (_mm512_storeu_epi32): Renamed into...
1942 (_mm512_storeu_si512): This.
1943 (_mm512_maskz_ceil_ps): Removed.
1944 (_mm512_maskz_ceil_pd): Ditto.
1945 (_mm512_maskz_floor_ps): Ditto.
1946 (_mm512_maskz_floor_pd): Ditto.
1947 (_mm512_floor_round_ps): Ditto.
1948 (_mm512_floor_round_pd): Ditto.
1949 (_mm512_ceil_round_ps): Ditto.
1950 (_mm512_ceil_round_pd): Ditto.
1951 (_mm512_mask_floor_round_ps): Ditto.
1952 (_mm512_mask_floor_round_pd): Ditto.
1953 (_mm512_mask_ceil_round_ps): Ditto.
1954 (_mm512_mask_ceil_round_pd): Ditto.
1955 (_mm512_maskz_floor_round_ps): Ditto.
1956 (_mm512_maskz_floor_round_pd): Ditto.
1957 (_mm512_maskz_ceil_round_ps): Ditto.
1958 (_mm512_maskz_ceil_round_pd): Ditto.
1959 (_mm512_expand_pd): Ditto.
1960 (_mm512_expand_ps): Ditto.
1961 * config/i386/i386.c (ix86_builtins): Remove
1962 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1963 (bdesc_args): Ditto.
1964 * config/i386/predicates.md (const1256_operand): New.
1965 (const_1_to_2_operand): Ditto.
1966 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1967 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1968 (*avx512pf_gatherpf<mode>sf): Ditto.
1969 (avx512pf_gatherpf<mode>df): Ditto.
1970 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1971 (*avx512pf_gatherpf<mode>df): Ditto.
1972 (avx512pf_scatterpf<mode>sf): Ditto.
1973 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1974 (*avx512pf_scatterpf<mode>sf): Ditto.
1975 (avx512pf_scatterpf<mode>df): Ditto.
1976 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1977 (*avx512pf_scatterpf<mode>df): Ditto.
1978 (avx512f_expand<mode>): Removed.
1979 (<shift_insn><mode>3<mask_name>): Change predicate type.
1981 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1983 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1984 not at the end of datarefs vector use ordered_remove to avoid
1985 reordering datarefs vector.
1988 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1989 mark local addressable non-static vars as GOVD_PRIVATE
1990 instead of GOVD_LOCAL.
1991 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1992 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1996 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1997 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1998 assume_aligned or alloc_align attributes.
1999 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2000 arguments. Handle also assume_aligned and alloc_align attributes.
2001 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2002 calls to functions with assume_aligned or alloc_align attributes.
2003 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2005 2014-02-08 Terry Guo <terry.guo@arm.com>
2007 * doc/invoke.texi: Document ARM -march=armv7e-m.
2009 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2011 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2012 flag on __cilkrts_rethrow builtin.
2015 * ipa-cp.c (determine_versionability): Fail at -O0
2016 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2017 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2020 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2023 * tree-inline.c (copy_forbidden): Fail for
2024 __attribute__((optimize (0))) functions.
2026 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2028 * varpool.c: Include pointer-set.h.
2029 (varpool_remove_unreferenced_decls): Variables in other partitions
2030 will not be output; be however careful to not lose information
2033 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2035 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2036 lookup in the vtable constructor.
2038 2014-02-07 Jeff Law <law@redhat.com>
2041 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2042 define_insn_and_split.
2044 * ipa-inline.c (inline_small_functions): Fix typos.
2046 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2048 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2049 (s390_can_use_return_insn): Declare.
2050 * config/s390/s390.h (EPILOGUE_USES): Define.
2051 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2053 (s390_chunkify_start): Handle return JUMP_LABELs.
2054 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2055 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2056 (s390_can_use_return_insn): New functions.
2057 (s390_fix_long_loop_prediction): Handle conditional returns.
2058 (TARGET_SET_UP_BY_PROLOGUE): Define.
2059 * config/s390/s390.md (ANY_RETURN): New code iterator.
2060 (*creturn, *csimple_return, return, simple_return): New patterns.
2062 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2064 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2065 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2066 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2067 REG_CFA_RESTORE list when deciding not to restore a register.
2069 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2071 * config/s390/s390.c: Include tree-pass.h and context.h.
2072 (s390_early_mach): New function, split out from...
2073 (s390_emit_prologue): ...here.
2074 (pass_data_s390_early_mach): New pass structure.
2075 (pass_s390_early_mach): New class.
2076 (s390_option_override): Create and register early_mach pass.
2077 Move to end of file.
2079 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2081 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2082 to match for the exit block.
2084 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2086 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2087 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2088 Reject misaligned operands.
2090 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2092 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2094 2014-02-07 Richard Biener <rguenther@suse.de>
2097 * gimple-low.c (lower_builtin_posix_memalign): New function.
2098 (lower_stmt): Call it to lower posix_memalign in a way
2099 to make alignment info accessible.
2101 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2104 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2105 __builtin_setjmp_receiver.
2107 2014-02-07 Richard Biener <rguenther@suse.de>
2110 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2111 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2112 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2113 Handle BUILT_IN_POSIX_MEMALIGN.
2114 (find_func_clobbers): Likewise.
2115 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2116 (call_may_clobber_ref_p_1): Likewise.
2118 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2121 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2124 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2127 * lto-cgraph.c (lto_output_node): Use
2128 symtab_get_symbol_partitioning_class.
2129 (lto_output_varpool_node): likewise.
2130 (symtab_get_symbol_partitioning_class): Move here from
2132 * cgraph.h (symbol_partitioning_class): Likewise.
2133 (symtab_get_symbol_partitioning_class): Declare.
2135 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2137 * ggc.h (ggc_internal_cleared_alloc): New macro.
2138 * vec.h (vec_safe_copy): Handle memory stats.
2139 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2140 * target-globals.c (save_target_globals): Likewise.
2142 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2145 * expr.c (emit_move_resolve_push): Export; be bit more selective
2146 on when to clear alias set.
2147 * expr.h (emit_move_resolve_push): Declare.
2148 * function.h (struct function): Add tail_call_marked.
2149 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2150 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2151 * config/i386/i386.md (TImode move expander): De not call
2153 (FP push expanders): Preserve memory attributes.
2154 * config/i386/sse.md (push<mode>1): Remove.
2155 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2156 (ix86_expand_push): Remove.
2157 * config/i386/mmx.md (push<mode>1): Remove.
2159 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2161 PR rtl-optimization/60030
2162 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2163 lopart with paradoxical subreg before shifting it up by hprec.
2165 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2167 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2168 Remove extra newline at end of file.
2169 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2170 (arm_issue_rate): Handle cortexa57.
2171 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2172 (cortex-a57.cortex-a53): Likewise.
2174 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2177 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2178 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2180 (arm_expand_prologue): Adjust all callers.
2181 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2182 info, registers also at the lowest numbered registers side. Use
2183 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2187 * var-tracking.c (adjust_mems): Before adding a SET to
2188 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2190 2014-02-06 Alan Modra <amodra@gmail.com>
2193 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2194 change SDmode to DDmode when lra_in_progress.
2196 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2199 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2200 free_data_ref on the dr first, and before goto again also set dr
2201 to the next dr. For simd_lane_access, free old datarefs[i] before
2202 overwriting it. For get_vectype_for_scalar_type failure, don't
2203 free_data_ref if simd_lane_access.
2205 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2208 * tree.h (opts_for_fn): New inline function.
2209 (opt_for_fn): Define.
2210 * config/i386/i386.c (ix86_function_regparm): Use
2211 opt_for_fn (decl, optimize) instead of optimize.
2213 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2215 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2216 for SYMBOL_REF in large memory model.
2218 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2220 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2222 (cortex-a57): Likewise.
2223 (cortex-a57.cortex-a53): Likewise.
2225 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2226 Kugan Vivekanandarajah <kuganv@linaro.org>
2228 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2230 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2232 2014-02-06 Richard Biener <rguenther@suse.de>
2234 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2235 set_loop_copy and initialize_original_copy_tables.
2237 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2239 * config/aarch64/aarch64-simd.md
2240 (aarch64_ashr_simddi): Change QI to SI.
2242 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2243 Jakub Jelinek <jakub@redhat.com>
2246 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2249 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2251 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2252 CODE_FOR_altivec_vpku[hw]um to
2253 CODE_FOR_altivec_vpku[hw]um_direct.
2254 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2255 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2256 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2257 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2259 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2261 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2262 generation for -maltivec=be.
2263 (altivec_vsumsws): Simplify redundant test.
2265 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2267 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2268 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2269 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2270 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2271 gen_altivec_vpkuwum.
2272 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2274 (altivec_vpks<VI_char>ss): Likewise.
2275 (altivec_vpks<VI_char>us): Likewise.
2276 (altivec_vpku<VI_char>us): Likewise.
2277 (altivec_vpku<VI_char>um): Likewise.
2278 (altivec_vpku<VI_char>um_direct): New (copy of
2279 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2281 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2282 target is little endian and -maltivec=be is not specified.
2283 (*altivec_vupkhs<VU_char>_direct): New (copy of
2284 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2285 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2286 target is little endian and -maltivec=be is not specified.
2287 (*altivec_vupkls<VU_char>_direct): New (copy of
2288 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2289 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2290 little endian and -maltivec=be is not specified.
2291 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2292 little endian and -maltivec=be is not specified.
2294 2014-02-05 Richard Henderson <rth@redhat.com>
2297 * combine-stack-adj.c: Revert r206943.
2298 * sched-int.h (struct deps_desc): Add last_args_size.
2299 * sched-deps.c (init_deps): Initialize it.
2300 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2301 contain REG_ARGS_SIZE notes.
2303 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2305 * lto-cgraph.c (asm_nodes_output): Make global.
2306 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2307 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2308 (driver_handle_option): Handle OPT_fwpa.
2310 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2313 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2314 a comment typo and formatting issue. If odr_hash hasn't been
2315 created, return vNULL and set *completep to false.
2318 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2319 bb with no successors.
2321 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2324 * doc/invoke.texi (-march): Clarify documentation for ARM.
2328 2014-02-05 Richard Biener <rguenther@suse.de>
2330 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2331 when not vectorizing because of too many alias checks.
2332 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2333 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2335 2014-02-05 Nick Clifton <nickc@redhat.com>
2337 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2338 accept extended registers in any mode when compiling for the MN10300.
2340 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2342 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2343 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2344 sanitization attributes.
2345 (can_inline_edge_p): Likewise.
2346 (sanitize_attrs_match_for_inline_p): New function.
2348 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2350 * ipa-prop.c (detect_type_change): Shor circuit testing of
2351 type changes on THIS pointer.
2353 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2356 * config/pa/pa.c (legitimize_tls_address): Return original address
2357 if not passed a SYMBOL_REF rtx.
2358 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2360 (pa_emit_move_sequence): Simplify TLS source operands.
2361 (pa_legitimate_constant_p): Reject all TLS constants.
2362 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2363 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2365 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2367 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2368 groups when we know they are controlled by LTO.
2369 * varasm.c (default_binds_local_p_1): If object is in other partition,
2370 it will be resolved locally.
2372 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2374 * config/host-linux.c (linux_gt_pch_use_address): Don't
2375 use SSIZE_MAX because it is not always defined.
2377 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2380 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2381 threshold for pseudo splitting.
2382 (update_ebb_live_info): Process call argument hard registers and
2383 hard registers from insn definition too.
2384 (max_small_class_regs_num): New constant.
2385 (inherit_in_ebb): Update live hard regs through EBBs. Update
2386 reloads_num only for small register classes. Don't split for
2389 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2392 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2395 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2397 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2400 2014-02-04 Marek Polacek <polacek@redhat.com>
2402 * gdbinit.in (pel): Define.
2404 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2406 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2409 2014-02-04 Richard Biener <rguenther@suse.de>
2412 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2413 in function context local.
2414 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2415 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2416 similar to LTO_imported_decl_ref.
2418 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2420 PR tree-optimization/60002
2421 * cgraphclones.c (build_function_decl_skip_args): Clear
2424 PR tree-optimization/60023
2425 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2426 false to gsi_replace.
2427 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2428 has been in some EH region and vec_stmt could throw, add
2429 vec_stmt into the same EH region.
2430 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2431 has no lhs, ignore it.
2432 * internal-fn.c (expand_MASK_LOAD): Likewise.
2435 * tree-inline.c (copy_forbidden): Fail for
2436 __attribute__((optimize (0))) functions.
2439 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2440 is set, copy one less argument.
2441 (expand_simd_clones): Don't subtract clone_info->inbranch
2442 from simd_clone_struct_alloc argument.
2444 PR rtl-optimization/57915
2445 * recog.c (simplify_while_replacing): If all unary/binary/relational
2446 operation arguments are constant, attempt to simplify those.
2449 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2450 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2452 2014-02-04 Richard Biener <rguenther@suse.de>
2454 PR tree-optimization/60012
2455 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2456 TBAA disambiguation to all DDRs.
2458 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2461 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2462 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2464 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2467 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2469 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2471 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2472 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2474 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2477 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2478 to figure out targets of polymorphic calls with known decl.
2479 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2480 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2481 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2482 (get_polymorphic_call_info): ... here.
2483 (get_polymorphic_call_info_from_invariant): New function.
2485 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2487 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2488 lookup via vtable pointer; check for type consistency
2489 and turn inconsitent facts into UNREACHABLE.
2490 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2491 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2492 type inconsistent querries; return UNREACHABLE instead.
2494 2014-02-03 Richard Henderson <rth@twiddle.net>
2497 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2498 already processed this node.
2499 (normalize_one_pred_1): Pass along mark_set.
2500 (normalize_one_pred): Create and destroy a pointer_set_t.
2501 (normalize_one_pred_chain): Likewise.
2503 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2505 PR gcov-profile/58602
2506 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2508 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2511 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2512 -fno-devirtualize; try to devirtualize by the knowledge of
2513 virtual table pointer given by aggregate propagation.
2514 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2515 (ipa_print_node_jump_functions): Dump also offset that
2516 is relevant for polymorphic calls.
2517 (determine_known_aggregate_parts): Add arg_type parameter; use it
2518 instead of determining the type from pointer type.
2519 (ipa_compute_jump_functions_for_edge): Update call of
2520 determine_known_aggregate_parts.
2521 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2522 (gimple_get_virt_method_for_binfo): ... here; simplify using
2523 vtable_pointer_value_to_vtable.
2524 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2525 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2526 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2527 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2529 (vtable_pointer_value_to_binfo): ... here.
2530 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2532 2014-02-03 Teresa Johnson <tejohnson@google.com>
2534 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2535 redef of outer loop index variable.
2537 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2541 * doc/extend.texi (Function Attributes): Typo.
2543 2014-02-03 Cong Hou <congh@google.com>
2545 PR tree-optimization/60000
2546 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2547 if the vectorized statement is a store. A store statement can only
2548 appear at the end of pattern statements.
2550 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2552 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2553 (ix86_option_override_internal): Default long double to 64-bit for
2554 32-bit Bionic and to 128-bit for 64-bit Bionic.
2556 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2557 TARGET_LONG_DOUBLE_128 is true.
2558 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2560 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2561 (mlong-double-64): Negate -mlong-double-128.
2562 (mlong-double-128): New option.
2564 * config/i386/i386-c.c (ix86_target_macros): Define
2565 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2567 * doc/invoke.texi: Document -mlong-double-128.
2569 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2571 PR rtl-optimization/60024
2572 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2574 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2576 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2578 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2580 PR rtl-optimization/57662
2581 * sel-sched.c (code_motion_path_driver): Do not mark already not
2582 existing blocks in the visiting bitmap.
2584 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2586 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2587 on the insn being emitted.
2589 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2590 Will Deacon <will.deacon@arm.com>
2592 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2594 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2596 * config/arm/arm-tables.opt: Regenerate.
2598 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2600 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2601 for vector types other than V16QImode.
2602 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2603 define_expand, and call altivec_expand_vec_perm_le when producing
2604 code with little endian element order.
2605 (*altivec_vperm_<mode>_internal): New insn having previous
2606 behavior of altivec_vperm_<mode>.
2607 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2608 altivec_expand_vec_perm_le when producing code with little endian
2610 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2611 behavior of altivec_vperm_<mode>_uns.
2613 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2615 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2616 (altivec_vsumsws): Add handling for -maltivec=be with a little
2618 (altivec_vsumsws_direct): New.
2619 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2620 gen_altivec_vsumsws.
2622 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2624 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2625 vtable_pointer_value_to_binfo): New functions.
2626 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2627 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2629 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2631 * config/nios2/nios2.md (load_got_register): Initialize GOT
2632 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2633 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2635 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2637 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2638 preserverd by passthrough, do not propagate the type.
2640 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2642 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2643 (mips_atomic_assign_expand_fenv): New function.
2644 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2646 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2648 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2649 (__builtin_mips_set_fcsr): Likewise.
2650 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2651 MIPS_USI_FTYPE_VOID.
2652 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2653 (mips16_expand_set_fcsr): Likewise.
2654 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2655 (mips16_set_fcsr_stub): Likewise.
2656 (mips16_get_fcsr_one_only_stub): New class.
2657 (mips16_set_fcsr_one_only_stub): Likewise.
2658 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2659 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2660 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2661 (hard_float): New availability predicate.
2662 (mips_builtins): Add get_fcsr and set_fcsr.
2663 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2664 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2665 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2666 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2667 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2670 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2672 * config/mips/mips.c (mips_one_only_stub): New class.
2673 (mips_need_mips16_rdhwr_p): Replace with...
2674 (mips16_rdhwr_stub): ...this new variable.
2675 (mips16_stub_call_address): New function.
2676 (mips16_rdhwr_one_only_stub): New class.
2677 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2678 (mips_output_mips16_rdhwr): Delete.
2679 (mips_finish_stub): New function.
2680 (mips_code_end): Use it to handle rdhwr stubs.
2682 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2685 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2686 when calculating size of integer atomic types.
2688 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2690 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2692 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2694 PR tree-optimization/60003
2695 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2696 * profile.c (branch_prob): Use gimple_call_builtin_p
2697 to check for BUILT_IN_SETJMP_RECEIVER.
2698 * tree-inline.c (copy_bb): Call notice_special_calls.
2700 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2703 * lra-constraints.c (process_alt_operands): Update reload_sum only
2706 2014-01-31 Richard Henderson <rth@redhat.com>
2709 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2710 until after else_eh is processed.
2712 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2714 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2715 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2716 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2717 in smmintrin.h, remove them.
2718 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2719 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2720 * config/i386/i386.md (ROUND_SAE): Fix value.
2721 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2722 (const48_operand): New.
2723 * config/i386/subst.md (round), (round_expand): Use
2724 const_4_or_8_to_11_operand.
2725 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2727 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2729 * config/i386/constraints.md (Yk): Swap meaning with k.
2730 * config/i386/i386.md (movhi_internal): Change Yk to k.
2731 (movqi_internal): Ditto.
2732 (*k<logic><mode>): Ditto.
2735 (kandn<mode>): Ditto.
2736 (*<code>hi_1): Ditto.
2737 (*<code>qi_1): Ditto.
2738 (kxnor<mode>): Ditto.
2739 (kortestzhi): Ditto.
2740 (kortestchi): Ditto.
2742 (*one_cmplhi2_1): Ditto.
2743 (*one_cmplqi2_1): Ditto.
2744 * config/i386/sse.md (): Change k to Yk.
2745 (avx512f_load<mode>_mask): Ditto.
2746 (avx512f_blendm<mode>): Ditto.
2747 (avx512f_store<mode>_mask): Ditto.
2748 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2749 (avx512f_storedqu<mode>_mask): Ditto.
2750 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2752 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2753 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2754 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2755 (avx512f_maskcmp<mode>3): Ditto.
2756 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2757 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2758 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2759 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2760 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2761 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2762 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2763 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2764 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2765 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2766 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2767 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2768 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2769 (vec_extract_lo_<mode>_maskm): Ditto.
2770 (vec_extract_hi_<mode>_maskm): Ditto.
2771 (avx512f_vternlog<mode>_mask): Ditto.
2772 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2773 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2774 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2775 (avx512f_<code>v8div16qi2_mask): Ditto.
2776 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2777 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2778 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2779 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2780 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2781 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2782 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2783 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2784 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2785 (avx512cd_maskb_vec_dupv8di): Ditto.
2786 (avx512cd_maskw_vec_dupv16si): Ditto.
2787 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2788 (avx512f_vpermi2var<mode>3_mask): Ditto.
2789 (avx512f_vpermi2var<mode>3_mask): Ditto.
2790 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2791 (*avx512f_gathersi<mode>): Ditto.
2792 (*avx512f_gathersi<mode>_2): Ditto.
2793 (*avx512f_gatherdi<mode>): Ditto.
2794 (*avx512f_gatherdi<mode>_2): Ditto.
2795 (*avx512f_scattersi<mode>): Ditto.
2796 (*avx512f_scatterdi<mode>): Ditto.
2797 (avx512f_compress<mode>_mask): Ditto.
2798 (avx512f_compressstore<mode>_mask): Ditto.
2799 (avx512f_expand<mode>_mask): Ditto.
2800 * config/i386/subst.md (mask): Change k to Yk.
2801 (mask_scalar_merge): Ditto.
2804 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2806 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2808 2014-01-31 Richard Biener <rguenther@suse.de>
2811 * builtins.c (fold_builtin_memory_op): Make sure to not
2812 use a floating-point mode or a boolean or enumeral type for
2815 2014-01-30 DJ Delorie <dj@redhat.com>
2817 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2818 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2819 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2820 whenever main() has an epilogue.
2822 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2824 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2825 unused variable "field".
2826 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2827 (vsx_mergeh_<mode>): Likewise.
2828 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2829 (altivec_vmrghh): Likewise.
2830 (altivec_vmrghw): Likewise.
2831 (altivec_vmrglb): Likewise.
2832 (altivec_vmrglh): Likewise.
2833 (altivec_vmrglw): Likewise.
2834 (altivec_vspltb): Add missing uses.
2835 (altivec_vsplth): Likewise.
2836 (altivec_vspltw): Likewise.
2837 (altivec_vspltsf): Likewise.
2839 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2842 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2843 frame related instructions.
2845 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2847 PR rtl-optimization/59959
2848 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2849 any reload of register whose subreg is invalid.
2851 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2853 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2854 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2855 Add missing return type - void.
2857 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2859 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2860 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2861 remove element index adjustment for endian (now handled in vsx.md
2863 (altivec_expand_vec_perm_const): Use
2864 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2865 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2866 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2867 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2868 define_expand and a new define_insn *altivec_vspltb_internal;
2869 adjust for -maltivec=be on a little endian target.
2870 (altivec_vspltb_direct): New.
2871 (altivec_vsplth): Divide into a define_expand and a new
2872 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2873 little endian target.
2874 (altivec_vsplth_direct): New.
2875 (altivec_vspltw): Divide into a define_expand and a new
2876 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2877 little endian target.
2878 (altivec_vspltw_direct): New.
2879 (altivec_vspltsf): Divide into a define_expand and a new
2880 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2881 a little endian target.
2883 2014-01-30 Richard Biener <rguenther@suse.de>
2885 PR tree-optimization/59993
2886 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2887 can propagate form the earlier stmt and avoid the transform
2888 when the intermediate result is needed.
2890 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2892 * README.Portability: Fix typo.
2894 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2896 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2897 comparison_operator with ordered_comparison_operator.
2899 2014-01-30 Nick Clifton <nickc@redhat.com>
2901 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2902 Rename to mn10300_store_multiple_regs.
2903 * config/mn10300/mn10300.c: Likewise.
2904 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2905 store_multiple_regs.
2906 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2907 Call mn10300_store_multiple_regs.
2909 2014-01-30 Nick Clifton <nickc@redhat.com>
2910 DJ Delorie <dj@redhat.com>
2912 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2913 %fp 2 to keep registers after it properly word-aligned.
2914 (rl78_alloc_physical_registers_umul): Handle the case where both
2915 input operands are the same.
2917 2014-01-30 Richard Biener <rguenther@suse.de>
2919 PR tree-optimization/59903
2920 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2923 2014-01-30 Jason Merrill <jason@redhat.com>
2926 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2929 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2931 2014-01-30 Richard Biener <rguenther@suse.de>
2933 PR tree-optimization/59951
2934 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2936 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2939 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2940 SFmode to DFmode case.
2942 2014-01-29 DJ Delorie <dj@redhat.com>
2944 * config/msp430/msp430.opt (-minrt): New.
2945 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2947 (ENDFILE_SPEC): Likewise.
2949 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2951 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2952 (estimate_function_body_sizes): Use it.
2954 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2957 * dwarf2out.c (is_cxx_auto): New.
2958 (is_base_type): Use it.
2959 (gen_type_die_with_usage): Likewise.
2961 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2963 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2964 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2965 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2966 -maltivec=be with LE targets.
2967 (vsx_mergeh_<mode>): Likewise.
2968 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2969 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2970 (altivec_vmrghb): Replace with define_expand and new
2971 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2972 (altivec_vmrghb_direct): New define_insn.
2973 (altivec_vmrghh): Replace with define_expand and new
2974 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2975 (altivec_vmrghh_direct): New define_insn.
2976 (altivec_vmrghw): Replace with define_expand and new
2977 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2978 (altivec_vmrghw_direct): New define_insn.
2979 (*altivec_vmrghsf): Adjust for endianness.
2980 (altivec_vmrglb): Replace with define_expand and new
2981 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2982 (altivec_vmrglb_direct): New define_insn.
2983 (altivec_vmrglh): Replace with define_expand and new
2984 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2985 (altivec_vmrglh_direct): New define_insn.
2986 (altivec_vmrglw): Replace with define_expand and new
2987 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2988 (altivec_vmrglw_direct): New define_insn.
2989 (*altivec_vmrglsf): Adjust for endianness.
2990 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2991 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2992 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2993 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2994 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2995 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2996 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2997 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2999 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3001 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3002 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3005 2014-01-29 Richard Biener <rguenther@suse.de>
3007 PR tree-optimization/58742
3008 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3009 associate_pointerplus_align.
3010 (associate_pointerplus_diff): New function.
3011 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3012 and associate_pointerplus_diff.
3014 2014-01-29 Richard Biener <rguenther@suse.de>
3016 * lto-streamer.h (LTO_major_version): Bump to 3.
3017 (LTO_minor_version): Reset to 0.
3019 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3021 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3022 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3023 (arm_file_start): Generate correct asm header for armv7ve.
3024 * config/arm/bpabi.h: Add multilib support for armv7ve.
3025 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3026 and cortex-a15 to armv7ve.
3027 * config/arm/t-aprofile: Add multilib support for armv7ve.
3028 * doc/invoke.texi: Document -march=armv7ve.
3030 2014-01-29 Richard Biener <rguenther@suse.de>
3032 PR tree-optimization/58742
3033 * tree-ssa-forwprop.c (associate_plusminus): Return true
3034 if we changed sth, defer EH cleanup to ...
3035 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3036 (simplify_mult): New function.
3038 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3041 PR tree-optimization/59920
3042 * tree.c (build_common_builtin_nodes): Remove
3043 __builtin_setjmp_dispatcher initialization.
3044 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3045 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3046 instead of gsi_after_labels + manually skipping debug stmts.
3047 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3048 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3049 * tree-inline.c (copy_edges_for_bb): Remove
3050 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3051 argument. Ignore computed_goto_p stmts. Don't call
3052 make_abnormal_goto_edges. If a call might need abnormal edges
3053 for non-local gotos, see if it already has an edge to
3054 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3055 with true argument, don't do anything then, otherwise add
3056 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3057 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3059 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3060 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3061 (lower_stmt): Don't set data->calls_builtin_setjmp.
3062 (lower_builtin_setjmp): Adjust comment.
3063 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3064 * tree-cfg.c (found_computed_goto): Remove.
3065 (factor_computed_gotos): Remove.
3066 (make_goto_expr_edges): Return bool, true for computed gotos.
3067 Don't call make_abnormal_goto_edges.
3068 (build_gimple_cfg): Don't set found_computed_goto, don't call
3069 factor_computed_gotos.
3070 (computed_goto_p): No longer static.
3071 (make_blocks): Don't set found_computed_goto.
3072 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3073 (make_edges): If make_goto_expr_edges returns true, push bb
3074 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3075 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3076 vector. Record mapping between bbs and OpenMP regions if there
3077 are any, adjust make_gimple_omp_edges caller. Call
3078 handle_abnormal_edges.
3079 (make_abnormal_goto_edges): Remove.
3080 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3081 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3082 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3083 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3084 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3085 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3086 filling *region also set *region_idx to (*region)->entry->index.
3089 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3090 For REGs set ORIGINAL_REGNO.
3092 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3094 * doc/md.texi: Mention that a target shouldn't implement
3095 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3098 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3100 PR tree-optimization/59594
3101 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3102 a copy of the datarefs vector rather than the vector itself.
3104 2014-01-28 Jason Merrill <jason@redhat.com>
3107 * dwarf2out.c (auto_die): New static.
3108 (gen_type_die_with_usage): Handle C++1y 'auto'.
3109 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3112 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3115 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3116 (SPEC_X32): Likewise.
3117 (SPEC_64): Likewise.
3118 * config/i386/i386.c (ix86_option_override_internal): Turn off
3119 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3121 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3122 * config/i386/i386.h (TARGET_16BIT): New macro.
3123 (TARGET_16BIT_P): Likewise.
3124 * config/i386/i386.opt: Add m16.
3125 * doc/invoke.texi: Document -m16.
3127 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3129 PR preprocessor/59935
3130 * input.c (location_get_source_line): Bail out on when line number
3131 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3133 2014-01-28 Richard Biener <rguenther@suse.de>
3135 PR tree-optimization/58742
3136 * tree-ssa-forwprop.c (associate_plusminus): Handle
3137 pointer subtraction of the form (T)(P + A) - (T)P.
3139 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3141 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3144 2014-01-28 Richard Biener <rguenther@suse.de>
3147 2014-01-28 Richard Biener <rguenther@suse.de>
3149 PR rtl-optimization/45364
3150 PR rtl-optimization/59890
3151 * var-tracking.c (local_get_addr_clear_given_value): Handle
3152 already cleared slot.
3153 (val_reset): Handle not allocated local_get_addr_cache.
3154 (vt_find_locations): Use post-order on the inverted CFG.
3156 2014-01-28 Richard Biener <rguenther@suse.de>
3158 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3160 2014-01-28 Richard Biener <rguenther@suse.de>
3162 PR rtl-optimization/45364
3163 PR rtl-optimization/59890
3164 * var-tracking.c (local_get_addr_clear_given_value): Handle
3165 already cleared slot.
3166 (val_reset): Handle not allocated local_get_addr_cache.
3167 (vt_find_locations): Use post-order on the inverted CFG.
3169 2014-01-28 Alan Modra <amodra@gmail.com>
3171 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3172 * configure.ac <recursive call for build != host>: Define
3173 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3174 and LD_FOR_BUILD too.
3175 * configure: Regenerate.
3177 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3179 * config/i386/i386.c (get_builtin_code_for_version): Separate
3180 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3181 Broadwell from Haswell.
3183 2014-01-27 Steve Ellcey <sellcey@mips.com>
3185 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3186 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3187 * config/mips/mips.c (mips_option_override): Change setting
3189 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3190 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3191 Change from Mask to Var.
3193 2014-01-27 Jeff Law <law@redhat.com>
3195 * ipa-inline.c (inline_small_functions): Fix typo.
3197 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3199 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3200 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3201 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3202 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3203 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3204 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3205 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3206 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3207 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3208 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3209 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3210 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3211 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3212 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3213 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3214 (_mm512_storeu_epi64): Ditto.
3215 (_mm512_cmpge_epi32_mask): Ditto.
3216 (_mm512_cmpge_epu32_mask): Ditto.
3217 (_mm512_cmpge_epi64_mask): Ditto.
3218 (_mm512_cmpge_epu64_mask): Ditto.
3219 (_mm512_cmple_epi32_mask): Ditto.
3220 (_mm512_cmple_epu32_mask): Ditto.
3221 (_mm512_cmple_epi64_mask): Ditto.
3222 (_mm512_cmple_epu64_mask): Ditto.
3223 (_mm512_cmplt_epi32_mask): Ditto.
3224 (_mm512_cmplt_epu32_mask): Ditto.
3225 (_mm512_cmplt_epi64_mask): Ditto.
3226 (_mm512_cmplt_epu64_mask): Ditto.
3227 (_mm512_cmpneq_epi32_mask): Ditto.
3228 (_mm512_cmpneq_epu32_mask): Ditto.
3229 (_mm512_cmpneq_epi64_mask): Ditto.
3230 (_mm512_cmpneq_epu64_mask): Ditto.
3231 (_mm512_expand_pd): Ditto.
3232 (_mm512_expand_ps): Ditto.
3233 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3234 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3235 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3236 * config/i386/i386.c (ix86_builtins): Add
3237 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3238 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3239 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3240 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3241 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3242 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3243 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3244 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3245 IX86_BUILTIN_PMOVUSQW512_MEM.
3246 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3247 __builtin_ia32_pmovsqd512mem_mask,
3248 __builtin_ia32_pmovqd512mem_mask,
3249 __builtin_ia32_pmovusqw512mem_mask,
3250 __builtin_ia32_pmovsqw512mem_mask,
3251 __builtin_ia32_pmovqw512mem_mask,
3252 __builtin_ia32_pmovusdw512mem_mask,
3253 __builtin_ia32_pmovsdw512mem_mask,
3254 __builtin_ia32_pmovdw512mem_mask,
3255 __builtin_ia32_pmovqb512mem_mask,
3256 __builtin_ia32_pmovusqb512mem_mask,
3257 __builtin_ia32_pmovsqb512mem_mask,
3258 __builtin_ia32_pmovusdb512mem_mask,
3259 __builtin_ia32_pmovsdb512mem_mask,
3260 __builtin_ia32_pmovdb512mem_mask.
3261 (bdesc_args): Add __builtin_ia32_expanddf512,
3262 __builtin_ia32_expandsf512.
3263 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3264 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3265 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3266 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3267 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3268 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3269 (avx512f_<code>v8div16qi2_mask_store): This.
3270 (avx512f_expand<mode>): New.
3272 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3274 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3276 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3277 (_mm512_prefetch_i32scatter_pd): Ditto.
3278 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3279 (_mm512_prefetch_i64scatter_pd): Ditto.
3280 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3281 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3282 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3283 (_mm512_prefetch_i32scatter_ps): Ditto.
3284 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3285 (_mm512_prefetch_i64scatter_ps): Ditto.
3286 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3287 * config/i386/i386-builtin-types.def: Define
3288 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3289 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3290 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3291 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3292 IX86_BUILTIN_SCATTERPFQPD.
3293 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3294 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3295 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3296 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3297 __builtin_ia32_scatterpfqps.
3298 (ix86_expand_builtin): Expand new built-ins.
3299 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3300 fix memory access data type.
3301 (*avx512pf_gatherpf<mode>_mask): Ditto.
3302 (*avx512pf_gatherpf<mode>): Ditto.
3303 (avx512pf_scatterpf<mode>): Ditto.
3304 (*avx512pf_scatterpf<mode>_mask): Ditto.
3305 (*avx512pf_scatterpf<mode>): Ditto.
3306 (GATHER_SCATTER_SF_MEM_MODE): New.
3307 (avx512pf_gatherpf<mode>df): Ditto.
3308 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3309 (*avx512pf_scatterpf<mode>df): Ditto.
3311 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3314 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3315 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3318 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3320 * common/config/arm/arm-common.c
3321 (arm_rewrite_mcpu): Handle multiple names.
3323 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3325 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3327 * gimple-builder.h (create_gimple_tmp): Delete.
3329 2014-01-27 Christian Bruel <christian.bruel@st.com>
3331 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3334 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3336 * config/pa/pa.md (call): Generate indirect long calls to non-local
3337 functions when outputing 32-bit code.
3338 (call_value): Likewise except for special call to buggy powf function.
3340 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3341 portable runtime and PIC indirect calls.
3342 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3343 and PIC call sequences. Use ldo instead of blr to set return register
3344 in PIC call sequence.
3346 2014-01-25 Walter Lee <walt@tilera.com>
3348 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3349 avoid clobbering a live register.
3351 2014-01-25 Walter Lee <walt@tilera.com>
3353 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3354 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3355 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3356 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3358 2014-01-25 Walter Lee <walt@tilera.com>
3360 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3361 arguments on even registers.
3362 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3364 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3365 (BIGGEST_ALIGNMENT): Ditto.
3366 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3368 2014-01-25 Walter Lee <walt@tilera.com>
3370 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3371 insns before bundling.
3372 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3374 2014-01-25 Walter Lee <walt@tilera.com>
3376 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3377 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3378 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3380 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3382 * config/mips/constraints.md (kl): Delete.
3383 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3384 define expands, using...
3385 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3386 instructions for MIPS16.
3387 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3388 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3390 2014-01-25 Walter Lee <walt@tilera.com>
3392 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3396 2014-01-25 Walter Lee <walt@tilera.com>
3398 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3399 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3401 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3403 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3406 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3408 * print-rtl.c (in_call_function_usage): New var.
3409 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3410 EXPR_LIST mode as mode and not as reg note name.
3413 * cfgloopmanip.c (copy_loop_info): If
3414 loop->warned_aggressive_loop_optimizations, make sure
3415 the flag is set in target loop too.
3417 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3419 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3421 * builtins.def: Likewise.
3422 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3423 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3424 * ira.c (ira_setup_eliminable_regset): Likewise.
3425 * omp-low.c (gate_expand_omp): Likewise.
3426 (execute_lower_omp): Likewise.
3427 (diagnose_sb_0): Likewise.
3428 (gate_diagnose_omp_blocks): Likewise.
3429 (simd_clone_clauses_extract): Likewise.
3432 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3434 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3435 correction for little endian...
3436 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3439 2014-01-24 Jeff Law <law@redhat.com>
3441 PR tree-optimization/59919
3442 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3443 for non-returning calls.
3445 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3447 * common/config/aarch64/aarch64-common.c
3448 (aarch64_rewrite_mcpu): Handle multiple names.
3449 * config/aarch64/aarch64.h
3450 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3452 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3454 * input.c (add_file_to_cache_tab): Handle the case where fopen
3457 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3460 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3461 from push operand if code of push isn't PRE_DEC.
3463 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3466 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3467 -mquad-memory-atomic. Update -mquad-memory documentation to say
3468 it is only used for non-atomic loads/stores.
3470 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3471 -mquad-memory or -mquad-memory-atomic switches.
3473 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3474 -mquad-memory-atomic to ISA 2.07 support.
3476 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3477 to separate support of normal quad word memory operations (ldq, stq)
3478 from the atomic quad word memory operations.
3480 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3481 support to separate non-atomic quad word operations from atomic
3482 quad word operations. Disable non-atomic quad word operations in
3483 little endian mode so that we don't have to swap words after the
3484 load and before the store.
3485 (quad_load_store_p): Add comment about atomic quad word support.
3486 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3487 options printed with -mdebug=reg.
3489 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3490 -mquad-memory-atomic as the test for whether we have quad word
3491 atomic instructions.
3492 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3493 or -mp8-vector are used, allow byte/half-word atomic operations.
3495 * config/rs6000/sync.md (load_lockedti): Insure that the address
3496 is a proper indexed or indirect address for the lqarx instruction.
3497 On little endian systems, swap the hi/lo registers after the lqarx
3499 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3500 insure the address is valid for the lqarx instruction.
3501 (store_conditionalti): Insure that the address is a proper indexed
3502 or indirect address for the stqcrx. instruction. On little endian
3503 systems, swap the hi/lo registers before doing the stqcrx.
3505 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3506 insure the address is valid for the stqcrx. instruction.
3508 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3509 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3510 type of quad memory support is available.
3512 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3515 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3516 there is a danger of looping.
3518 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3520 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3521 force flag_ira_loop_pressure if set via command line.
3523 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3525 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3526 (ashr_simd): New builtin handling DI mode.
3527 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3528 (aarch64_sshr_simddi): New match pattern.
3529 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3530 (vshrd_n_s64): Likewise.
3531 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3533 2014-01-23 Nick Clifton <nickc@redhat.com>
3535 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3536 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3537 favour of mcu specific scripts.
3538 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3541 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3542 Alex Velenko <Alex.Velenko@arm.com>
3544 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3545 (vaddv_s16): Likewise.
3546 (vaddv_s32): Likewise.
3547 (vaddv_u8): Likewise.
3548 (vaddv_u16): Likewise.
3549 (vaddv_u32): Likewise.
3550 (vaddvq_s8): Likewise.
3551 (vaddvq_s16): Likewise.
3552 (vaddvq_s32): Likewise.
3553 (vaddvq_s64): Likewise.
3554 (vaddvq_u8): Likewise.
3555 (vaddvq_u16): Likewise.
3556 (vaddvq_u32): Likewise.
3557 (vaddvq_u64): Likewise.
3558 (vaddv_f32): Likewise.
3559 (vaddvq_f32): Likewise.
3560 (vaddvq_f64): Likewise.
3561 (vmaxv_f32): Likewise.
3562 (vmaxv_s8): Likewise.
3563 (vmaxv_s16): Likewise.
3564 (vmaxv_s32): Likewise.
3565 (vmaxv_u8): Likewise.
3566 (vmaxv_u16): Likewise.
3567 (vmaxv_u32): Likewise.
3568 (vmaxvq_f32): Likewise.
3569 (vmaxvq_f64): Likewise.
3570 (vmaxvq_s8): Likewise.
3571 (vmaxvq_s16): Likewise.
3572 (vmaxvq_s32): Likewise.
3573 (vmaxvq_u8): Likewise.
3574 (vmaxvq_u16): Likewise.
3575 (vmaxvq_u32): Likewise.
3576 (vmaxnmv_f32): Likewise.
3577 (vmaxnmvq_f32): Likewise.
3578 (vmaxnmvq_f64): Likewise.
3579 (vminv_f32): Likewise.
3580 (vminv_s8): Likewise.
3581 (vminv_s16): Likewise.
3582 (vminv_s32): Likewise.
3583 (vminv_u8): Likewise.
3584 (vminv_u16): Likewise.
3585 (vminv_u32): Likewise.
3586 (vminvq_f32): Likewise.
3587 (vminvq_f64): Likewise.
3588 (vminvq_s8): Likewise.
3589 (vminvq_s16): Likewise.
3590 (vminvq_s32): Likewise.
3591 (vminvq_u8): Likewise.
3592 (vminvq_u16): Likewise.
3593 (vminvq_u32): Likewise.
3594 (vminnmv_f32): Likewise.
3595 (vminnmvq_f32): Likewise.
3596 (vminnmvq_f64): Likewise.
3598 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3600 * config/aarch64/aarch64-simd.md
3601 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3602 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3603 (*aarch64_mul3_elt<mode>): Likewise.
3604 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3605 (*aarch64_mul3_elt_to_64v2df): Likewise.
3606 (*aarch64_mla_elt<mode>): Likewise.
3607 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3608 (*aarch64_mls_elt<mode>): Likewise.
3609 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3610 (*aarch64_fma4_elt<mode>): Likewise.
3611 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3612 (*aarch64_fma4_elt_to_64v2df): Likewise.
3613 (*aarch64_fnma4_elt<mode>): Likewise.
3614 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3615 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3616 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3617 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3618 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3619 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3620 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3621 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3622 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3624 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3626 * config/aarch64/aarch64-simd.md
3627 (aarch64_be_checked_get_lane<mode>): New define_expand.
3628 * config/aarch64/aarch64-simd-builtins.def
3629 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3630 New builtin definition.
3631 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3632 Use new safe be builtin.
3634 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3636 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3638 (aarch64_be_st1<mode>): Likewise.
3639 (aarch_ld1<VALL:mode>): Define_expand modified.
3640 (aarch_st1<VALL:mode>): Likewise.
3641 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3642 (UNSPEC_ST1): Likewise.
3644 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3646 * config/microblaze/microblaze.md: Add trap insn and attribute
3648 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3650 PR preprocessor/58580
3651 * input.h (location_get_source_line): Take an additional line_size
3653 (void diagnostics_file_cache_fini): Declare new function.
3654 * input.c (struct fcache): New type.
3655 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3656 New static constants.
3657 (diagnostic_file_cache_init, total_lines_num)
3658 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3659 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3660 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3661 (get_next_line, read_next_line, goto_next_line, read_line_num):
3662 New static function definitions.
3663 (diagnostic_file_cache_fini): New function.
3664 (location_get_source_line): Take an additional output line_len
3665 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3667 * diagnostic.c (diagnostic_finish): Call
3668 diagnostic_file_cache_fini.
3669 (adjust_line): Take an additional input parameter for the length
3670 of the line, rather than calculating it with strlen.
3671 (diagnostic_show_locus): Adjust the use of
3672 location_get_source_line and adjust_line with respect to their new
3673 signature. While displaying a line now, do not stop at the first
3674 null byte. Rather, display the zero byte as a space and keep
3675 going until we reach the size of the line.
3676 * Makefile.in: Add vec.o to OBJS-libcommon
3678 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3679 Ilya Tocar <ilya.tocar@intel.com>
3681 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3682 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3683 (__builtin_ia32_kmov16): Ditto.
3684 * config/i386/i386.md (UNSPEC_KMOV): New.
3687 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3689 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3690 (_mm512_storeu_si512): Ditto.
3692 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3695 * rtl.h (get_referenced_operands): Declare.
3696 * recog.c (get_referenced_operands): New function.
3697 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3698 operands have been referenced when recording LO_SUM references.
3700 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3702 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3704 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3706 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3707 Enable for generic and recent AMD targets.
3709 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3711 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3712 ARG_SIZE note when adjustment was eliminated.
3714 2014-01-22 Jeff Law <law@redhat.com>
3716 PR tree-optimization/59597
3717 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3718 in file. Accept new argument REGISTERING and use it to modify
3719 dump output appropriately.
3720 (register_jump_thread): Corresponding changes.
3721 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3722 thread paths involving joiner blocks. Add code to dump cancelled
3723 jump threading paths.
3725 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3727 PR rtl-optimization/59477
3728 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3729 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3731 2014-01-22 Tom Tromey <tromey@redhat.com>
3733 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3735 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3737 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3739 PR rtl-optimization/59896
3740 * lra-constraints.c (process_alt_operands): Check unused note for
3741 matched operands of insn with no output reloads.
3743 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3745 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3746 (mips_move_from_gpr_cost): Likewise.
3748 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3750 PR rtl-optimization/59858
3751 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3752 ira_class_hard_regs_num.
3753 (process_alt_operands): Increase reject for dying matched operand.
3755 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3758 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3759 smaller than size, perform several stores or loads and stores
3760 at dst + count - size to store or copy all of size bytes, rather
3761 than just last modesize bytes.
3763 2014-01-20 DJ Delorie <dj@redhat.com>
3765 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3766 that CLOBBERs are REGs before propogating their values.
3768 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3771 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3772 (cgraph_inline_failed_type): New function.
3773 * cgraph.h (DEFCIFCODE): Add type.
3774 (cgraph_inline_failed_type_t): New enum.
3775 (cgraph_inline_failed_type): New prototype.
3776 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3777 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3778 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3779 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3780 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3781 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3782 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3783 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3784 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3785 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3786 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3787 OPTIMIZATION_MISMATCH.
3788 * tree-inline.c (expand_call_inline): Emit errors during
3789 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3791 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3794 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3795 mode attribute in insn output.
3797 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3799 * output.h (output_constant): Delete.
3800 * varasm.c (output_constant): Make private.
3802 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3804 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3806 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3809 * tree.h (fold_builtin_strcat): New prototype.
3810 * builtins.c (fold_builtin_strcat): No longer static. Add len
3811 argument, if non-NULL, don't call c_strlen. Optimize
3812 directly into __builtin_memcpy instead of __builtin_strcpy.
3813 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3814 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3816 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3818 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3819 for SImode_address_operand operands, having only a REG argument.
3821 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3823 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3824 loader name using mbig-endian.
3825 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3827 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3829 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3833 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3835 * config/aarch64/aarch64-protos.h
3836 (aarch64_cannot_change_mode_class_ptr): Declare.
3837 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3838 aarch64_cannot_change_mode_class_ptr): New.
3839 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3840 backend hook aarch64_cannot_change_mode_class.
3842 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3844 * common/config/aarch64/aarch64-common.c
3845 (aarch64_handle_option): Don't handle any option order logic here.
3846 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3847 selected_cpu, warn on architecture version mismatch.
3848 (aarch64_override_options): Fix parsing order for option strings.
3850 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3851 Iain Sandoe <iain@codesourcery.com>
3854 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3855 warning. Amend comment to reflect current functionality.
3857 2014-01-20 Richard Biener <rguenther@suse.de>
3860 * builtins.c (fold_builtin_strcat): Remove case better handled
3861 by tree-ssa-strlen.c.
3863 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3865 * config/aarch64/aarch64.opt
3866 (mcpu, march, mtune): Make case-insensitive.
3868 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3871 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3872 if operands[1] is a REG or ZERO_EXTEND of a REG.
3874 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3876 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3878 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3880 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3881 long non-pic millicode calls.
3883 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3885 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3887 2014-01-19 Kito Cheng <kito@0xlab.org>
3889 * builtins.c (expand_movstr): Check movstr expand done or fail.
3891 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3892 H.J. Lu <hongjiu.lu@intel.com>
3895 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3896 to DImode for zero-extended addresses.
3898 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3900 PR rtl-optimization/57763
3901 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3902 on the new indirect jump_insn and increment LABEL_NUSES (label).
3904 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3908 * config.gcc (x86_archs): New variable.
3909 (x86_64_archs): Likewise.
3910 (x86_cpus): Likewise.
3911 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3912 --with-arch/--with-cpu= options.
3913 Support --with-arch=/--with-cpu={nehalem,westmere,
3914 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3916 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3918 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3919 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3921 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3923 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3925 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3928 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3929 clear cpp_get_options (parse_in)->warn_unused_macros for
3930 ix86_target_macros_internal with cpp_define.
3932 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3934 * jump.c (delete_related_insns): Keep (use (insn))s.
3935 * reorg.c (redundant_insn): Check for barriers too.
3937 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3939 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3941 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3943 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3944 call to $$dyncall when TARGET_LONG_CALLS is true.
3946 2014-01-17 Jeff Law <law@redhat.com>
3948 * ree.c (combine_set_extension): Temporarily disable test for
3949 changing number of hard registers.
3951 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3954 * ipa-inline-analysis.c (inline_free_summary):
3955 Do not free summary of aliases.
3957 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3960 * gimplify.c (gimplify_expr): Use create_tmp_var
3961 instead of create_tmp_var_raw. If cond doesn't have
3962 integral type, don't add the IFN_ANNOTATE builtin at all.
3964 2014-01-17 Martin Jambor <mjambor@suse.cz>
3967 * ipa-cp.c (prev_edge_clone): New variable.
3968 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3969 Also resize prev_edge_clone vector.
3970 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3971 (ipcp_edge_removal_hook): New function.
3972 (ipcp_driver): Register ipcp_edge_removal_hook.
3974 2014-01-17 Andrew Pinski <apinski@cavium.com>
3975 Steve Ellcey <sellcey@mips.com>
3978 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3981 2014-01-17 Jeff Law <law@redhat.com>
3984 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3985 so that pass_ccp runs first.
3987 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3989 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3990 (ix86_adjust_cost): Use !TARGET_XXX.
3991 (do_reorder_for_imul): Likewise.
3992 (swap_top_of_ready_list): Likewise.
3993 (ix86_sched_reorder): Likewise.
3995 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3997 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3998 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3999 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4000 (intel_memset): New. Duplicate slm_memset.
4001 (intel_cost): New. Duplicate slm_cost.
4002 (m_INTEL): New macro.
4003 (processor_target_table): Add "intel".
4004 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4005 with PROCESSOR_INTEL for "intel".
4006 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4007 PROCESSOR_SILVERMONT.
4008 (ix86_issue_rate): Likewise.
4009 (ix86_adjust_cost): Likewise.
4010 (ia32_multipass_dfa_lookahead): Likewise.
4011 (swap_top_of_ready_list): Likewise.
4012 (ix86_sched_reorder): Likewise.
4013 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4014 instead of TARGET_OPT_AGU.
4015 * config/i386/i386.h (TARGET_INTEL): New.
4016 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4017 (processor_type): Add PROCESSOR_INTEL.
4018 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4019 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4021 2014-01-17 Marek Polacek <polacek@redhat.com>
4024 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4027 2014-01-17 Richard Biener <rguenther@suse.de>
4029 PR tree-optimization/46590
4030 * opts.c (default_options_table): Add entries for
4031 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4032 all enabled at -O1 but not for -Og.
4033 * common.opt (fbranch-count-reg): Remove Init(1).
4034 (fmove-loop-invariants): Likewise.
4035 (ftree-pta): Likewise.
4037 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4039 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4040 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4041 decls to at least the GCC 4.8 used alignments.
4044 * tree-nested.c (convert_nonlocal_reference_stmt,
4045 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4046 of GIMPLE_BIND stmts, adjust associated decls.
4048 2014-01-17 Richard Biener <rguenther@suse.de>
4050 PR tree-optimization/46590
4051 * vec.h (vec<>::bseach): New member function implementing
4052 binary search according to C89 bsearch.
4053 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4054 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4055 bitmap pointer again. Make accesses_in_loop a flat array.
4056 (mem_ref_obstack): New global.
4057 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4058 (mark_ref_stored): Likewise.
4059 (ref_indep_loop_p_2): Likewise.
4060 (set_ref_stored_in_loop): New helper function.
4061 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4062 (memref_free): Adjust.
4063 (record_mem_ref_loc): Simplify.
4064 (gather_mem_refs_stmt): Adjust.
4065 (sort_locs_in_loop_postorder_cmp): New function.
4066 (analyze_memory_references): Sort accesses_in_loop after
4067 loop postorder number.
4068 (find_ref_loc_in_loop_cmp): New function.
4069 (for_all_locs_in_loop): Find relevant cluster of locs in
4070 accesses_in_loop and iterate without recursion.
4071 (execute_sm): Avoid uninit warning.
4072 (struct ref_always_accessed): Simplify.
4073 (ref_always_accessed::operator ()): Likewise.
4074 (ref_always_accessed_p): Likewise.
4075 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4076 loop postorder numbers here.
4077 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4080 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4083 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4084 on decls for which assemble_alias has been called.
4086 2014-01-17 Nick Clifton <nickc@redhat.com>
4088 * config/msp430/msp430.opt: (mcpu): New option.
4089 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4090 (msp430_option_override): Parse target_cpu. If the MCU name
4091 matches a generic string, clear target_mcu.
4092 (msp430_attr): Allow numeric interrupt values up to 63.
4093 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4094 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4096 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4098 * config/msp430/msp430.md (popm): Use %J rather than %I.
4099 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4100 (addhi_cy_i): Use immediate_operand for operand 2.
4101 * doc/invoke.texi: Document -mcpu option.
4103 2014-01-17 Richard Biener <rguenther@suse.de>
4105 PR rtl-optimization/38518
4106 * df.h (df_analyze_loop): Declare.
4107 * df-core.c: Include cfgloop.h.
4108 (df_analyze_1): Split out main part of df_analyze.
4109 (df_analyze): Adjust.
4110 (loop_inverted_post_order_compute): New function.
4111 (loop_post_order_compute): Likewise.
4112 (df_analyze_loop): New function avoiding whole-function
4114 * loop-invariant.c (find_defs): Use df_analyze_loop.
4115 (find_invariants): Adjust.
4116 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4118 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4120 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4121 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4123 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4125 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4126 traversal when removing references.
4128 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4131 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4133 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4136 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4137 pushing a reload for an autoinc when we had previously reloaded an
4138 inner part of the address.
4140 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4142 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4144 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4145 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4146 when not giving up or versioning for alias only because of
4148 (vect_analyze_data_ref_dependences): Set to true.
4149 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4151 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4152 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4156 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4159 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4160 operand 0 predicate for gathers, use a new pseudo as subtarget.
4162 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4165 * lra-constraints.c (process_alt_operands): Add printing debug info.
4166 Check absence of input/output reloads for matched operands too.
4168 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4170 PR rtl-optimization/59835
4171 * ira.c (ira_init_register_move_cost): Increase cost for
4174 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4176 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4178 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4181 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4182 non-register objects. Use gen_(high/low)part more consistently.
4185 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4188 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4189 endian support, remove tests for WORDS_BIG_ENDIAN.
4190 (p8_mfvsrd_3_<mode>): Likewise.
4191 (reload_gpr_from_vsx<mode>): Likewise.
4192 (reload_gpr_from_vsxsf): Likewise.
4193 (p8_mfvsrd_4_disf): Likewise.
4195 2014-01-16 Richard Biener <rguenther@suse.de>
4197 PR rtl-optimization/46590
4198 * lcm.c (compute_antinout_edge): Use postorder iteration.
4199 (compute_laterin): Use inverted postorder iteration.
4201 2014-01-16 Nick Clifton <nickc@redhat.com>
4204 * varasm.c (output_constant): Return the number of bytes actually
4206 (output_constructor_array_range): Update the field size with the
4207 number of bytes emitted by output_constant.
4208 (output_constructor_regular_field): Likewise. Also do not
4209 complain if the total number of bytes emitted is now greater
4210 than the expected fieldpos.
4211 * output.h (output_constant): Update prototype and descriptive comment.
4213 2014-01-16 Marek Polacek <polacek@redhat.com>
4216 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4217 it is error_mark_node.
4219 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4221 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4222 VALID_AVX256_REG_OR_OI_MODE.
4224 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4226 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4227 current procedure should be profiled.
4229 2014-01-15 Andrew Pinski <apinski@cavium.com>
4231 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4232 of moving from/to the STACK_REG register class.
4234 2014-01-15 Richard Henderson <rth@redhat.com>
4237 * reginfo.c (global_regs_decl): Globalize.
4238 * rtl.h (global_regs_decl): Declare.
4239 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4240 reserved via global_regs.
4242 2014-01-15 Teresa Johnson <tejohnson@google.com>
4244 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4246 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4248 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4249 and vmulosh rather than call gen_vec_widen_smult_*.
4250 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4251 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4252 (vec_widen_smult_even_v16qi): Likewise.
4253 (vec_widen_umult_even_v8hi): Likewise.
4254 (vec_widen_smult_even_v8hi): Likewise.
4255 (vec_widen_umult_odd_v16qi): Likewise.
4256 (vec_widen_smult_odd_v16qi): Likewise.
4257 (vec_widen_umult_odd_v8hi): Likewise.
4258 (vec_widen_smult_odd_v8hi): Likewise.
4259 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4260 vmuloub rather than call gen_vec_widen_umult_*.
4261 (vec_widen_umult_lo_v16qi): Likewise.
4262 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4263 vmulosb rather than call gen_vec_widen_smult_*.
4264 (vec_widen_smult_lo_v16qi): Likewise.
4265 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4266 rather than call gen_vec_widen_umult_*.
4267 (vec_widen_umult_lo_v8hi): Likewise.
4268 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4269 rather than call gen_vec_widen_smult_*.
4270 (vec_widen_smult_lo_v8hi): Likewise.
4272 2014-01-15 Jeff Law <law@redhat.com>
4274 PR tree-optimization/59747
4275 * ree.c (find_and_remove_re): Properly handle case where a second
4276 eliminated extension requires widening a copy created for elimination
4277 of a prior extension.
4278 (combine_set_extension): Ensure that the number of hard regs needed
4279 for a destination register does not change when we widen it.
4281 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4283 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4284 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4285 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4286 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4287 (avr-*-rtems*): Likewise.
4288 (bfin*-rtems*): Likewise.
4289 (moxie-*-rtems*): Likewise.
4290 (h8300-*-rtems*): Likewise.
4291 (i[34567]86-*-rtems*): Likewise.
4292 (lm32-*-rtems*): Likewise.
4293 (m32r-*-rtems*): Likewise.
4294 (m68k-*-rtems*): Likewise.
4295 (microblaze*-*-rtems*): Likewise.
4296 (mips*-*-rtems*): Likewise.
4297 (powerpc-*-rtems*): Likewise.
4298 (sh-*-rtems*): Likewise.
4299 (sparc-*-rtems*): Likewise.
4300 (sparc64-*-rtems*): Likewise.
4301 (v850-*-rtems*): Likewise.
4302 (m32c-*-rtems*): Likewise.
4304 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4306 PR rtl-optimization/59511
4307 * ira.c (ira_init_register_move_cost): Use memory costs for some
4308 cases of register move cost calculations.
4309 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4310 instead of BB frequency.
4311 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4312 * lra-assigns.c (find_hard_regno_for): Ditto.
4314 2014-01-15 Richard Biener <rguenther@suse.de>
4316 PR tree-optimization/59822
4317 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4318 (vectorizable_load): Use it to hoist defs of uses of invariant
4319 loads out of the loop.
4321 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4322 Kugan Vivekanandarajah <kuganv@linaro.org>
4325 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4328 2014-01-15 Richard Biener <rguenther@suse.de>
4330 PR rtl-optimization/59802
4331 * lcm.c (compute_available): Use inverted postorder to seed
4332 the initial worklist.
4334 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4337 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4338 ADDR_REGS for invalid symrefs in non-PIC code.
4340 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4343 * builtins.c (determine_block_size): Initialize *probable_max_size
4344 even if len_rtx is CONST_INT.
4346 2014-01-14 Andrew Pinski <apinski@cavium.com>
4348 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4349 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4350 (cortexa53_tunings): Likewise.
4351 (aarch64_sched_issue_rate): New function.
4352 (TARGET_SCHED_ISSUE_RATE): Define.
4354 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4356 * ira-costs.c (find_costs_and_classes): Add missed
4357 ira_init_register_move_cost_if_necessary.
4359 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4362 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4364 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4367 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4368 to indicate if type is used for function return value. Warn ABI
4369 change if the vector mode isn't available for function return value.
4370 (ix86_function_arg_advance): Pass false to type_natural_mode.
4371 (ix86_function_arg): Likewise.
4372 (ix86_gimplify_va_arg): Likewise.
4373 (function_arg_32): Don't warn ABI change.
4374 (ix86_function_value): Pass true to type_natural_mode.
4375 (ix86_return_in_memory): Likewise.
4376 (ix86_struct_value_rtx): Removed.
4377 (TARGET_STRUCT_VALUE_RTX): Likewise.
4379 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4381 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4382 converting a conditional jump into a conditional return.
4384 2014-01-14 Richard Biener <rguenther@suse.de>
4386 PR tree-optimization/58921
4387 PR tree-optimization/59006
4388 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4389 hoisting invariant stmts.
4390 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4391 invariant loads on the preheader edge if possible.
4393 2014-01-14 Joey Ye <joey.ye@arm.com>
4395 * doc/plugin.texi (Building GCC plugins): Update to C++.
4397 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4399 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4400 (_mm_rcp28_round_ss): Ditto.
4401 (_mm_rsqrt28_round_sd): Ditto.
4402 (_mm_rsqrt28_round_ss): Ditto.
4403 (_mm_rcp28_sd): Ditto.
4404 (_mm_rcp28_ss): Ditto.
4405 (_mm_rsqrt28_sd): Ditto.
4406 (_mm_rsqrt28_ss): Ditto.
4407 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4408 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4409 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4410 (IX86_BUILTIN_RCP28SD): Ditto.
4411 (IX86_BUILTIN_RCP28SS): Ditto.
4412 (IX86_BUILTIN_RSQRT28SD): Ditto.
4413 (IX86_BUILTIN_RSQRT28SS): Ditto.
4414 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4415 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4416 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4417 (ix86_expand_special_args_builtin): Expand new FTYPE.
4418 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4419 (srcp14<mode>): Make insn unary.
4420 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4421 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4422 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4423 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4424 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4425 Fix rounding: make it SAE only.
4426 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4428 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4430 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4431 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4432 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4433 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4434 (round_saeonly_mask_scalar_operand4): Ditto.
4435 (round_saeonly_mask_scalar_op3): Ditto.
4436 (round_saeonly_mask_scalar_op4): Ditto.
4438 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4440 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4441 Implement -maltivec=be for vec_insert and vec_extract.
4443 2014-01-10 DJ Delorie <dj@redhat.com>
4445 * config/msp430/msp430.md (call_internal): Don't allow memory
4446 references with SP as the base register.
4447 (call_value_internal): Likewise.
4448 * config/msp430/constraints.md (Yc): New. For memory references
4449 that don't use SP as a base register.
4451 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4452 "an integer without a # prefix"
4453 * config/msp430/msp430.md (epilogue_helper): Use it.
4455 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4458 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4459 AVX512F gather builtins.
4460 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4461 on gather decls with INTEGER_TYPE masktype.
4462 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4463 directly into the builtin rather than hoisting it before loop.
4465 PR tree-optimization/59387
4466 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4467 (scev_const_prop): If folded_casts and type has undefined overflow,
4468 use force_gimple_operand instead of force_gimple_operand_gsi and
4469 for each added stmt if it is assign with
4470 arith_code_with_undefined_signed_overflow, call
4471 rewrite_to_defined_overflow.
4472 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4473 gimple-fold.h instead.
4474 (arith_code_with_undefined_signed_overflow,
4475 rewrite_to_defined_overflow): Moved to ...
4476 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4477 rewrite_to_defined_overflow): ... here. No longer static.
4478 Include gimplify-me.h.
4479 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4480 rewrite_to_defined_overflow): New prototypes.
4482 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4484 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4486 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4488 * builtins.c (get_object_alignment_2): Minor tweak.
4489 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4491 2014-01-13 Christian Bruel <christian.bruel@st.com>
4493 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4494 optimized non constant lengths.
4496 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4499 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4500 load as __atomic_load_N if possible.
4502 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4504 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4506 (rs6000_expand_builtin): Adjust call.
4508 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4511 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4512 * config/rs6000/rs6000.c: Include target-globals.h.
4513 (rs6000_set_current_function): Instead of doing target_reinit
4514 unconditionally, use save_target_globals_default_opts and
4515 restore_target_globals.
4517 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4519 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4520 (rs6000_expand_builtin): Handle mffs and mtfsf.
4521 (rs6000_init_builtins): Define mffs and mtfsf.
4522 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4523 (rs6000_mffs): New pattern.
4524 (rs6000_mtfsf): New pattern.
4526 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4528 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4529 Start narrowing with START. Apply candidate-use pair
4530 and check overall cost in narrowing.
4531 (iv_ca_prune): Pass new argument.
4533 2014-01-10 Jeff Law <law@redhat.com>
4536 * ree.c (combine_reaching_defs): Ensure the defining statement
4537 occurs before the extension when optimizing extensions with
4538 different source and destination hard registers.
4540 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4543 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4544 vtables into the type inheritance graph.
4546 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4548 PR rtl-optimization/59754
4549 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4550 modes in the REGNO != REGNO case.
4552 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4554 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4556 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4558 PR tree-optimization/59745
4559 * tree-predcom.c (tree_predictive_commoning_loop): Call
4560 free_affine_expand_cache if giving up because components is NULL.
4562 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4563 GC in payload of target_globals struct instead of allocating them on
4564 the heap and the larger structs separately using GC.
4565 * target-globals.h (struct target_globals): Make regs, hard_regs,
4566 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4567 of GTY((skip)) and change type to void *.
4568 (reset_target_globals): Cast loads from those fields to corresponding
4571 2014-01-10 Steve Ellcey <sellcey@mips.com>
4574 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4575 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4576 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4578 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4581 * aarch64-modes.def (CC_Zmode): New flags mode.
4582 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4583 represents an equality.
4584 (aarch64_get_condition_code): Handle CC_Zmode.
4585 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4587 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4589 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4590 extraction in good case.
4592 2014-01-10 Richard Biener <rguenther@suse.de>
4594 PR tree-optimization/59374
4595 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4596 checking after SLP discovery. Mark stmts not participating
4597 in any SLP instance properly.
4599 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4601 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4602 when handling a SET rtx.
4604 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4606 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4607 (cortex-a57): Likewise.
4608 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4610 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4612 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4613 non-iwmmxt builtins.
4615 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4619 * ipa-devirt.c record_target_from_binfo): Take as argument
4620 stack of binfos and lookup matching one for virtual inheritance.
4621 (possible_polymorphic_call_targets_1): Update.
4623 2014-01-10 Huacai Chen <chenhc@lemote.com>
4625 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4626 kernel strings for Loongson-2E/2F/3A.
4628 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4631 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4632 is_gimple_call before calling gimple_call_internal_p.
4634 2014-01-09 Steve Ellcey <sellcey@mips.com>
4636 * Makefile.in (TREE_FLOW_H): Remove.
4637 (TREE_SSA_H): Add file names from tree-flow.h.
4638 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4639 * tree.h: Remove tree-flow.h reference.
4640 * hash-table.h: Remove tree-flow.h reference.
4641 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4642 reference with tree-ssa-loop.h.
4644 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4646 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4647 default element-order behavior for -maltivec.
4648 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4649 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4650 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4651 when targeting big endian, at least for now.
4652 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4654 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4657 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4658 var satisfies use_register_for_decl, just take into account type
4659 alignment, rather than decl alignment.
4661 PR tree-optimization/59622
4662 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4663 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4664 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4665 Don't devirtualize for inplace at all. For targets.length () == 1,
4666 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4668 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4670 * config/i386/i386.md (cpu): Remove the unused btver1.
4672 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4674 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4676 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4679 * tree-core.h (struct target_globals): New forward declaration.
4680 (struct tree_target_option): Add globals field.
4681 * tree.h (TREE_TARGET_GLOBALS): Define.
4682 (prepare_target_option_nodes_for_pch): New prototype.
4683 * target-globals.h (struct target_globals): Define even if
4685 * tree.c (prepare_target_option_node_for_pch,
4686 prepare_target_option_nodes_for_pch): New functions.
4687 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4688 * config/i386/i386.c: Include target-globals.h.
4689 (ix86_set_current_function): Instead of doing target_reinit
4690 unconditionally, use save_target_globals_default_opts and
4691 restore_target_globals.
4693 2014-01-09 Richard Biener <rguenther@suse.de>
4695 PR tree-optimization/59715
4696 * tree-cfg.h (split_critical_edges): Declare.
4697 * tree-cfg.c (split_critical_edges): Export.
4698 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4700 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4702 * cfgexpand.c (expand_stack_vars): Optionally disable
4703 asan stack protection.
4704 (expand_used_vars): Likewise.
4705 (partition_stack_vars): Likewise.
4706 * asan.c (asan_emit_stack_protection): Optionally disable
4707 after return stack usage.
4708 (instrument_derefs): Optionally disable memory access instrumentation.
4709 (instrument_builtin_call): Likewise.
4710 (instrument_strlen_call): Likewise.
4711 (asan_protect_global): Optionally disable global variables protection.
4712 * doc/invoke.texi: Added doc for new options.
4713 * params.def: Added new options.
4714 * params.h: Likewise.
4716 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4718 PR rtl-optimization/59724
4719 * ifcvt.c (cond_exec_process_if_block): Don't call
4720 flow_find_head_matching_sequence with 0 longest_match.
4721 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4722 non-active insns if !stop_after.
4723 (try_head_merge_bb): Revert 2014-01-07 changes.
4725 2014-01-08 Jeff Law <law@redhat.com>
4727 * ree.c (get_sub_rtx): New function, extracted from...
4728 (merge_def_and_ext): Here.
4729 (combine_reaching_defs): Use get_sub_rtx.
4731 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4733 * cgraph.h (varpool_variable_node): Do not choke on null node.
4735 2014-01-08 Catherine Moore <clm@codesourcery.com>
4737 * config/mips/mips.md (simple_return): Attempt to use JRC
4739 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4741 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4743 PR rtl-optimization/59137
4744 * reorg.c (steal_delay_list_from_target): Call update_block for
4746 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4748 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4750 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4751 two duplicate entries.
4753 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4756 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4758 * config/mips/mips.c (mips_truncated_op_cost): New function.
4759 (mips_rtx_costs): Adjust test for BADDU.
4760 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4762 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4764 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4765 (*baddu_si): ...this new pattern.
4767 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4770 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4772 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4775 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4777 (expand_expr, expand_normal): Adjust.
4778 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4779 inner_reference_p. Use inner_reference_p to expand inner references.
4780 (store_expr): Adjust.
4781 * cfgexpand.c (expand_call_stmt): Adjust.
4783 2014-01-08 Rong Xu <xur@google.com>
4785 * gcov-io.c (gcov_var): Move from gcov-io.h.
4786 (gcov_position): Ditto.
4787 (gcov_is_error): Ditto.
4788 (gcov_rewrite): Ditto.
4789 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4790 only part to libgcc/libgcov.h.
4792 2014-01-08 Marek Polacek <polacek@redhat.com>
4795 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4797 2014-01-08 Marek Polacek <polacek@redhat.com>
4800 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4802 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4804 PR rtl-optimization/59649
4805 * stor-layout.c (get_mode_bounds): For BImode return
4806 0 and STORE_FLAG_VALUE.
4808 2014-01-08 Richard Biener <rguenther@suse.de>
4811 * gimple.h (is_gimple_builtin_call): Remove.
4812 (gimple_builtin_call_types_compatible_p): New.
4813 (gimple_call_builtin_p): New overload.
4814 * gimple.c (is_gimple_builtin_call): Remove.
4815 (validate_call): Rename to ...
4816 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4818 (validate_type): New static function.
4819 (gimple_call_builtin_p): New overload and adjust.
4820 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4821 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4822 (gimple_fold_stmt_to_constant_1): Likewise.
4823 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4825 2014-01-08 Richard Biener <rguenther@suse.de>
4828 * gimplify.c (gimplify_expr): Gimplify register-register type
4829 VIEW_CONVERT_EXPRs to separate stmts.
4831 2014-01-07 Jeff Law <law@redhat.com>
4834 * ree.c (combine_set_extension): Handle case where source
4835 and destination registers in an extension insn are different.
4836 (combine_reaching_defs): Allow source and destination registers
4837 in extension to be different under limited circumstances.
4838 (add_removable_extension): Remove restriction that the
4839 source and destination registers in the extension are the same.
4840 (find_and_remove_re): Emit a copy from the extension's
4841 destination to its source after the defining insn if
4842 the source and destination registers are different.
4845 * ifcvt.c (merge_if_block): If we are merging a block with more than
4846 one successor with a block with no successors, remove any BARRIER
4847 after the second block.
4849 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4851 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4853 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4856 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4857 for 14-bit register offsets when INT14_OK_STRICT is false.
4859 2014-01-07 Roland Stigge <stigge@antcom.de>
4860 Michael Meissner <meissner@linux.vnet.ibm.com>
4863 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4864 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4866 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4868 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4871 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4873 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4874 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4875 rtx is const0_rtx or not.
4877 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4880 * target-globals.c (save_target_globals): Remove this_fn_optab
4882 * toplev.c: Include optabs.h.
4883 (target_reinit): Temporarily restore the global options if another
4884 set of options are in force.
4886 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4888 PR rtl-optimization/58668
4889 * cfgcleanup.c (flow_find_cross_jump): Don't count
4890 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4891 to determine what is counted.
4892 (flow_find_head_matching_sequence): Use active_insn_p to determine
4894 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4896 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4897 determine what is counted.
4899 PR tree-optimization/59643
4900 * tree-predcom.c (split_data_refs_to_components): If one dr is
4901 read and one write, determine_offset fails and the write isn't
4902 in the bad component, just put the read into the bad component.
4904 2014-01-07 Mike Stump <mikestump@comcast.net>
4905 Jakub Jelinek <jakub@redhat.com>
4908 * tree-core.h (struct tree_optimization_option): Change optabs
4909 type from unsigned char * to void *.
4910 * optabs.c (init_tree_optimization_optabs): Adjust
4911 TREE_OPTIMIZATION_OPTABS initialization.
4913 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4916 * config/i386/i386.h (struct machine_function): Add
4917 no_drap_save_restore field.
4918 * config/i386/i386.c (ix86_save_reg): Use
4919 !cfun->machine->no_drap_save_restore instead of
4920 crtl->stack_realign_needed.
4921 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4922 this function clears frame_pointer_needed. Set
4923 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4924 and DRAP reg is needed.
4926 2014-01-06 Marek Polacek <polacek@redhat.com>
4929 * doc/implement-c.texi: Mention that other integer types are
4930 permitted as bit-field types in strictly conforming mode.
4932 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4934 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4937 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4939 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4941 2014-01-06 Martin Jambor <mjambor@suse.cz>
4944 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4946 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4948 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4952 * var-tracking.c (add_stores): Preserve the value of the source even if
4953 we don't record the store.
4955 2014-01-06 Terry Guo <terry.guo@arm.com>
4957 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4959 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4962 * config/darwin.c (darwin_function_section): Adjust return values to
4963 correspond to optimisation changes made in r206070.
4965 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4967 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4968 from prefetch_block tune setting.
4969 (nocona_cost): Correct size of prefetch block to 64.
4971 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4973 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4974 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4975 used to save the static chain register in the computation of the offset
4976 from which the FP registers need to be restored.
4978 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4980 PR tree-optimization/59519
4981 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4982 ICE if get_current_def (current_new_name) is already non-NULL, as long
4983 as it is a phi result of some other phi in *new_exit_bb that has
4986 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4987 or vmovdqu* for misaligned_operand.
4988 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4989 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4990 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4991 aligned_mem for AVX512F masked aligned load and store builtins and for
4994 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4996 PR tree-optimization/59651
4997 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4998 Address range for negative step should be added by TYPE_SIZE_UNIT.
5000 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5002 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5003 overlapping registers also for registers other than the stack pointer.
5005 2014-01-03 Marek Polacek <polacek@redhat.com>
5008 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5011 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5014 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5017 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5018 (push splitter): Use <P:MODE_SIZE> instead of
5019 GET_MODE_SIZE (<P:MODE>mode).
5020 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5021 (mov -1, reg peephole2): Likewise.
5022 * config/i386/sse.md (*mov<mode>_internal,
5023 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5024 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5025 *<code><mode>3, *andnot<mode>3<mask_name>,
5026 <mask_codefor><code><mode>3<mask_name>): Likewise.
5027 * config/i386/subst.md (mask_mode512bit_condition,
5028 sd_mask_mode512bit_condition): Likewise.
5030 2014-01-02 Xinliang David Li <davidxl@google.com>
5032 PR tree-optimization/59303
5033 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5034 (dump_predicates): Better output format.
5035 (pred_equal_p): New function.
5036 (is_neq_relop_p): Ditto.
5037 (is_neq_zero_form_p): Ditto.
5038 (pred_expr_equal_p): Ditto.
5039 (pred_neg_p): Ditto.
5040 (simplify_pred): Ditto.
5041 (simplify_preds_2): Ditto.
5042 (simplify_preds_3): Ditto.
5043 (simplify_preds_4): Ditto.
5044 (simplify_preds): Ditto.
5046 (push_to_worklist): Ditto.
5047 (get_pred_info_from_cmp): Ditto.
5048 (is_degenerated_phi): Ditto.
5049 (normalize_one_pred_1): Ditto.
5050 (normalize_one_pred): Ditto.
5051 (normalize_one_pred_chain): Ditto.
5052 (normalize_preds): Ditto.
5053 (normalize_cond_1): Remove function.
5054 (normalize_cond): Ditto.
5055 (is_gcond_subset_of): Ditto.
5056 (is_subset_of_any): Ditto.
5057 (is_or_set_subset_of): Ditto.
5058 (is_and_set_subset_of): Ditto.
5059 (is_norm_cond_subset_of): Ditto.
5060 (pred_chain_length_cmp): Ditto.
5061 (convert_control_dep_chain_into_preds): Type change.
5062 (find_predicates): Ditto.
5063 (find_def_preds): Ditto.
5064 (destroy_predicates_vecs): Ditto.
5065 (find_matching_predicates_in_rest_chains): Ditto.
5066 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5067 (is_pred_expr_subset): Ditto.
5068 (is_pred_chain_subset_of): Ditto.
5069 (is_included_in): Ditto.
5070 (is_superset_of): Ditto.
5072 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5074 Update copyright years.
5076 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5078 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5079 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5080 config/arc/arc.md, config/arc/arc.opt,
5081 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5082 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5083 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5084 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5085 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5086 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5087 vtable-verify.c, vtable-verify.h: Use the standard form for the
5090 2014-01-02 Tobias Burnus <burnus@net-b.de>
5092 * gcc.c (process_command): Update copyright notice dates.
5093 * gcov-dump.c: Ditto.
5095 * doc/cpp.texi: Bump @copying's copyright year.
5096 * doc/cppinternals.texi: Ditto.
5097 * doc/gcc.texi: Ditto.
5098 * doc/gccint.texi: Ditto.
5099 * doc/gcov.texi: Ditto.
5100 * doc/install.texi: Ditto.
5101 * doc/invoke.texi: Ditto.
5103 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5105 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5107 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5109 * config/i386/sse.md (*mov<mode>_internal): Guard
5110 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5112 PR rtl-optimization/59647
5113 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5114 new_rtx into UNSIGNED_FLOAT rtxes.
5116 Copyright (C) 2014 Free Software Foundation, Inc.
5118 Copying and distribution of this file, with or without modification,
5119 are permitted in any medium without royalty provided the copyright
5120 notice and this notice are preserved.