1 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
3 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
6 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
9 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
11 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
13 * config/arm/aarch-common-protos.h
14 (alu_cost_table): Fix spelling of "extend".
15 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
17 2014-03-19 Richard Biener <rguenther@suse.de>
20 * tree-core.h (tree_type_common): Re-order pointer members
21 to reduce recursion depth during GC walks.
23 2014-03-19 Marek Polacek <polacek@redhat.com>
26 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
29 2014-03-19 Richard Biener <rguenther@suse.de>
32 * lto-streamer-in.c (input_function): In WPA stage do not drop
35 2014-03-19 Jakub Jelinek <jakub@redhat.com>
37 PR tree-optimization/60559
38 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
39 with build_zero_cst assignment.
41 2014-03-18 Kai Tietz <ktietz@redhat.com>
43 PR rtl-optimization/56356
44 * sdbout.c (sdbout_parms): Verify that parms'
45 incoming argument is valid.
46 (sdbout_reg_parms): Likewise.
48 2014-03-18 Richard Henderson <rth@redhat.com>
51 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
52 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
53 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
55 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
57 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
58 * doc/plugins.texi (Plugin callbacks): Mention
60 Italicize plugin event names in description. Explain that
61 PLUGIN_PRAGMAS has no sense for lto1. Explain
63 Remind that no GCC functions should be called after
65 Explain what pragmas with expansion are.
67 2014-03-18 Martin Liska <mliska@suse.cz>
69 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
70 gimple call statement is update.
71 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
72 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
74 2014-03-18 Jakub Jelinek <jakub@redhat.com>
77 * ubsan.c (ubsan_instrument_unreachable): Call
78 initialize_sanitizer_builtins.
79 (ubsan_pass): Likewise.
82 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
83 varpool_finalize_decl instead of rest_of_decl_compilation.
85 2014-03-18 Richard Biener <rguenther@suse.de>
87 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
88 by using bitmap_and_compl instead of bitmap_and_compl_into.
89 (df_rd_transfer_function): Likewise.
91 2014-03-18 Richard Biener <rguenther@suse.de>
93 * doc/lto.texi (fresolution): Fix typo.
95 2014-03-18 Richard Biener <rguenther@suse.de>
97 * doc/invoke.texi (flto): Update for changes in 4.9.
99 2014-03-18 Richard Biener <rguenther@suse.de>
101 * doc/loop.texi: Remove section on the removed lambda framework.
102 Update loop docs with recent changes in preserving loop structure.
104 2014-03-18 Richard Biener <rguenther@suse.de>
106 * doc/lto.texi (-fresolution): Document.
108 2014-03-18 Richard Biener <rguenther@suse.de>
110 * doc/contrib.texi: Adjust my name.
112 2014-03-18 Jakub Jelinek <jakub@redhat.com>
115 * internal-fn.c: Include diagnostic-core.h.
116 (expand_BUILTIN_EXPECT): New function.
117 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
118 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
119 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
120 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
122 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
123 Revert 3 argument __builtin_expect code.
124 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
125 * gimple-fold.c (gimple_fold_call): Likewise.
126 * tree.h (fold_builtin_expect): New prototype.
127 * builtins.c (build_builtin_expect_predicate): Add predictor
128 argument, if non-NULL, create 3 argument __builtin_expect.
129 (fold_builtin_expect): No longer static. Add ARG2 argument,
130 pass it through to build_builtin_expect_predicate.
131 (fold_builtin_2): Adjust caller.
132 (fold_builtin_3): Handle BUILT_IN_EXPECT.
133 * internal-fn.def (BUILTIN_EXPECT): New.
135 2014-03-18 Tobias Burnus <burnus@net-b.de>
138 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
139 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
140 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
142 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
145 * predict.c (combine_predictions_for_bb): Fix up formatting.
146 (expr_expected_value_1, expr_expected_value): Add predictor argument,
147 fill what it points to if non-NULL.
148 (tree_predict_by_opcode): Adjust caller, use the predictor.
149 * predict.def (PRED_COMPARE_AND_SWAP): Add.
151 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
153 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
154 proper constant for the store mode.
156 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
158 * symtab.c (change_decl_assembler_name): Fix transparent alias
161 2014-03-16 Renlin Li <Renlin.Li@arm.com>
163 * config/aarch64/aarch64.c: Correct the comments about the
164 aarch64 stack layout.
166 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
168 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
169 check for GF_OMP_FOR_KIND_FOR.
171 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
173 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
174 ymm and zmm register names.
176 2014-03-17 Jakub Jelinek <jakub@redhat.com>
179 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
180 note creation for the 2010-08-31 changes.
182 2014-03-17 Marek Polacek <polacek@redhat.com>
185 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
186 as -fno-tree-loop-vectorize.
187 (expand_omp_simd): Likewise.
189 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
191 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
192 (eligible_for_call_delay): New prototype.
193 * config/sparc/sparc.c (tls_call_delay): Rename into...
194 (eligible_for_call_delay): ...this. Return false if the instruction
195 cannot be put in the delay slot of a branch.
196 (eligible_for_restore_insn): Simplify.
197 (eligible_for_return_delay): Return false if the instruction cannot be
198 put in the delay slot of a branch and simplify.
199 (eligible_for_sibcall_delay): Return false if the instruction cannot be
200 put in the delay slot of a branch.
201 * config/sparc/sparc.md (fix_ut699): New attribute.
202 (tls_call_delay): Delete.
203 (in_call_delay): Reimplement.
204 (eligible_for_sibcall_delay): Rename into...
205 (in_sibcall_delay): ...this.
206 (eligible_for_return_delay): Rename into...
207 (in_return_delay): ...this.
208 (in_branch_delay): Reimplement.
209 (in_uncond_branch_delay): Delete.
210 (in_annul_branch_delay): Delete.
212 2014-03-14 Richard Henderson <rth@redhat.com>
215 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
216 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
217 (*floathi<X87MODEF>2_i387_with_temp): Remove.
218 (floathi splitters): Remove.
219 (float<SWI48x>xf2): New pattern.
220 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
221 code that tried to handle DImode for 32-bit, but which was excluded
222 by the pattern's condition. Drop allocation of stack temporary.
223 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
224 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
225 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
226 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
227 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
228 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
229 (*float<SWI48><MODEF>2_sse_interunit): Remove.
230 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
231 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
232 (*float<SWI48x><X87MODEF>2_i387): Remove.
233 (all float _with_temp splitters): Remove.
234 (*float<SWI48x><MODEF>2_i387): New pattern.
235 (*float<SWI48><MODEF>2_sse): New pattern.
236 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
237 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
239 2014-03-14 Jakub Jelinek <jakub@redhat.com>
240 Marek Polacek <polacek@redhat.com>
243 * common.opt (dump_base_name_prefixed): New Variable.
244 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
245 if x_dump_base_name_prefixed is already set, set it at the end.
247 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
249 PR rtl-optimization/60508
250 * lra-constraints.c (get_reload_reg): Add new parameter
252 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
253 Pass the new parameter values.
255 2014-03-14 Richard Biener <rguenther@suse.de>
257 * common.opt: Revert unintented changes from r205065.
260 2014-03-14 Richard Biener <rguenther@suse.de>
263 * cfghooks.c (split_block): Properly adjust all loops the
264 block was a latch of.
266 2014-03-14 Martin Jambor <mjambor@suse.cz>
269 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
272 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
275 * config/avr/avr.c (avr_set_current_function): Pass function name
276 through default_strip_name_encoding before sanity checking instead
277 of skipping the first char of the assembler name.
279 2014-03-13 Richard Henderson <rth@redhat.com>
282 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
283 (ix86_force_to_memory, ix86_free_from_memory): Remove.
284 * config/i386/i386-protos.h: Likewise.
285 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
286 in the expander instead of a splitter.
287 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
288 any possibility of requiring a memory.
289 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
290 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
291 (fp branch splitters): Update for ix86_split_fp_branch.
292 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
293 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
294 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
295 (*fop_<MODEF>_2_i387): Remove f/r alternative.
296 (*fop_<MODEF>_3_i387): Likewise.
297 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
298 (splitters for the fop_* register patterns): Remove.
299 (fscalexf4_i387): Rename from *fscalexf4_i387.
300 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
302 2014-03-13 Jakub Jelinek <jakub@redhat.com>
304 PR tree-optimization/59779
305 * tree-dfa.c (get_ref_base_and_extent): Use double_int
306 type for bitsize and maxsize instead of HOST_WIDE_INT.
308 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
310 PR rtl-optimization/57320
311 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
312 the CFG after thread_prologue_and_epilogue_insns.
314 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
316 PR rtl-optimization/57189
317 * lra-constraints.c (process_alt_operands): Disfavor spilling
320 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
322 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
325 2014-03-13 Jakub Jelinek <jakub@redhat.com>
327 PR tree-optimization/59025
329 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
330 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
332 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
335 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
336 calls of avr_out_plus_1.
338 2014-03-13 Bin Cheng <bin.cheng@arm.com>
340 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
341 BB's single pred and update the father loop's latch info later.
343 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
345 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
349 (VEC_base): Likewise.
350 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
351 registers, we need to swap double words in little endian mode.
353 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
354 to be a container mode for 128-bit integer operations added in ISA
355 2.07. Unlike TImode and PTImode, the preferred register set is
356 the Altivec/VMX registers for the 128-bit operations.
358 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
360 (rs6000_split_128bit_ok_p): Likewise.
362 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
363 macros for creating ISA 2.07 normal and overloaded builtin
364 functions with 3 arguments.
365 (BU_P8V_OVERLOAD_3): Likewise.
366 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
367 for use as overloaded functions.
368 (VPERM_1TI_UNS): Likewise.
369 (VSEL_1TI): Likewise.
370 (VSEL_1TI_UNS): Likewise.
371 (ST_INTERNAL_1ti): Likewise.
372 (LD_INTERNAL_1ti): Likewise.
373 (XXSEL_1TI): Likewise.
374 (XXSEL_1TI_UNS): Likewise.
375 (VPERM_1TI): Likewise.
376 (VPERM_1TI_UNS): Likewise.
377 (XXPERMDI_1TI): Likewise.
379 (LXVD2X_V1TI): Likewise.
380 (STXVD2X_V1TI): Likewise.
381 (VEC_INIT_V1TI): Likewise.
382 (VEC_SET_V1TI): Likewise.
383 (VEC_EXT_V1TI): Likewise.
384 (EQV_V1TI): Likewise.
385 (NAND_V1TI): Likewise.
386 (ORC_V1TI): Likewise.
387 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
388 added in ISA 2.07. Add both normal 'altivec' builtins, and the
392 (VADDEUQM): Likewise.
393 (VADDECUQ): Likewise.
394 (VSUBEUQM): Likewise.
395 (VSUBECUQ): Likewise.
397 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
398 __int128_t and __uint128_t types.
399 (__uint128_type): Likewise.
400 (altivec_categorize_keyword): Add support for vector __int128_t,
401 vector __uint128_t, vector __int128, and vector unsigned __int128
402 as a container type for TImode operations that need to be done in
403 VSX/Altivec registers.
404 (rs6000_macro_to_expand): Likewise.
405 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
406 to support 128-bit integer instructions vaddcuq, vadduqm,
407 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
408 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
410 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
411 for V1TImode, and set up preferences to use VSX/Altivec registers.
412 Setup VSX reload handlers.
413 (rs6000_debug_reg_global): Likewise.
414 (rs6000_init_hard_regno_mode_ok): Likewise.
415 (rs6000_preferred_simd_mode): Likewise.
416 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
417 (easy_altivec_constant): Likewise.
418 (output_vec_const_move): Likewise.
419 (rs6000_expand_vector_set): Convert V1TImode set and extract to
421 (rs6000_expand_vector_extract): Likewise.
422 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
424 (rs6000_const_vec): Add support for V1TImode.
425 (rs6000_emit_le_vsx_load): Swap double words when loading or
426 storing TImode/V1TImode.
427 (rs6000_emit_le_vsx_store): Likewise.
428 (rs6000_emit_le_vsx_move): Likewise.
429 (rs6000_emit_move): Add support for V1TImode.
430 (altivec_expand_ld_builtin): Likewise.
431 (altivec_expand_st_builtin): Likewise.
432 (altivec_expand_vec_init_builtin): Likewise.
433 (altivec_expand_builtin): Likewise.
434 (rs6000_init_builtins): Add support for V1TImode type. Add
435 support for ISA 2.07 128-bit integer builtins. Define type names
436 for the VSX/Altivec vector types.
437 (altivec_init_builtins): Add support for overloaded vector
438 functions with V1TImode type.
439 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
440 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
442 (rs6000_split_128bit_ok_p): Likewise.
443 (rs6000_handle_altivec_attribute): Create V1TImode from vector
444 __int128_t and vector __uint128_t.
446 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
454 (VS_scalar): Likewise.
455 (VS_double): Likewise.
456 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
458 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
459 we support the ISA 2.07 128-bit integer arithmetic instructions.
460 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
461 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
462 and TImode types for use with the builtin functions.
463 (V1TI_type_node): Likewise.
464 (unsigned_V1TI_type_node): Likewise.
465 (intTI_type_internal_node): Likewise.
466 (uintTI_type_internal_node): Likewise.
468 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
469 128-bit builtin functions.
470 (UNSPEC_VADDEUQM): Likewise.
471 (UNSPEC_VADDECUQ): Likewise.
472 (UNSPEC_VSUBCUQ): Likewise.
473 (UNSPEC_VSUBEUQM): Likewise.
474 (UNSPEC_VSUBECUQ): Likewise.
475 (VM): Add V1TImode to vector mode iterators.
478 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
479 (altivec_vaddcuq): Likewise.
480 (altivec_vsubuqm): Likewise.
481 (altivec_vsubcuq): Likewise.
482 (altivec_vaddeuqm): Likewise.
483 (altivec_vaddecuq): Likewise.
484 (altivec_vsubeuqm): Likewise.
485 (altivec_vsubecuq): Likewise.
487 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
489 (BOOL_128): Likewise.
490 (BOOL_REGS_OUTPUT): Likewise.
491 (BOOL_REGS_OP1): Likewise.
492 (BOOL_REGS_OP2): Likewise.
493 (BOOL_REGS_UNARY): Likewise.
494 (BOOL_REGS_AND_CR0): Likewise.
496 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
497 128-bit integer builtin support.
498 (vec_vadduqm): Likewise.
499 (vec_vaddecuq): Likewise.
500 (vec_vaddeuqm): Likewise.
501 (vec_vsubecuq): Likewise.
502 (vec_vsubeuqm): Likewise.
503 (vec_vsubcuq): Likewise.
504 (vec_vsubuqm): Likewise.
506 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
507 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
508 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
509 128-bit integer add/subtract to ISA 2.07.
511 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
513 * config/arc/arc.c (arc_predicate_delay_insns):
514 Fix third argument passed to conditionalize_nonjump.
516 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
518 * config/aarch64/aarch64-builtins.c
519 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
520 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
521 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
522 instead of __builtin_lfloor.
523 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
525 2014-03-12 Jakub Jelinek <jakub@redhat.com>
527 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
528 (tree_ssa_ifcombine_bb_1): New function.
529 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
530 is an empty forwarder block to then_bb or vice versa and then_bb
531 and else_bb are effectively swapped.
533 2014-03-12 Christian Bruel <christian.bruel@st.com>
536 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
537 REG_CFA_DEF_CFA note.
538 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
539 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
541 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
543 PR tree-optimization/60454
544 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
546 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
548 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
549 Do not define target_cpu_default2 to generic.
550 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
551 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
552 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
554 2014-03-12 Jakub Jelinek <jakub@redhat.com>
555 Marc Glisse <marc.glisse@inria.fr>
557 PR tree-optimization/60502
558 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
559 instead of build_low_bits_mask.
561 2014-03-12 Jakub Jelinek <jakub@redhat.com>
564 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
565 if there are multiple uses, but op doesn't live on E edge.
566 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
567 clobber stmts before __builtin_unreachable.
569 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
571 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
572 hard_frame_pointer_rtx.
573 * cse.c (cse_insn): Remove volatile check.
574 * cselib.c (cselib_process_insn): Likewise.
575 * dse.c (scan_insn): Likewise.
577 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
579 * config/arc/arc.c (conditionalize_nonjump): New function,
581 (arc_ifcvt): ... this.
582 (arc_predicate_delay_insns): Use it.
584 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
586 * config/arc/predicates.md (extend_operand): During/after reload,
587 allow const_int_operand.
588 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
589 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
590 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
592 (umulsi3_highpart_i): Likewise.
594 2014-03-11 Richard Biener <rguenther@suse.de>
596 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
597 Add asserts to guard possible wrong-code bugs.
599 2014-03-11 Richard Biener <rguenther@suse.de>
601 PR tree-optimization/60429
602 PR tree-optimization/60485
603 * tree-ssa-structalias.c (set_union_with_increment): Properly
604 take into account all fields that overlap the shifted vars.
605 (do_sd_constraint): Likewise.
606 (do_ds_constraint): Likewise.
607 (get_constraint_for_ptr_offset): Likewise.
609 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
611 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
612 (nios2_compute_frame_layout):
613 Add calculation of cfun->machine->fp_save_offset.
614 (nios2_expand_prologue): Correct setting of frame pointer register
616 (nios2_expand_epilogue): Update recovery of stack pointer from
617 frame pointer accordingly.
618 (nios2_initial_elimination_offset): Update calculation of offset
619 for eliminating to HARD_FRAME_POINTER_REGNUM.
621 2014-03-10 Jakub Jelinek <jakub@redhat.com>
624 * ipa.c (symtab_remove_unreachable_nodes): Don't call
625 cgraph_get_create_node on VAR_DECLs.
627 2014-03-10 Richard Biener <rguenther@suse.de>
630 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
632 2014-03-08 Douglas B Rupp <rupp@gnat.com>
634 * config/vms/vms.opt (vms_float_format): New variable.
636 2014-03-08 Tobias Burnus <burnus@net-b.de>
638 * doc/invoke.texi (-fcilkplus): Update implementation status.
640 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
641 Richard Biener <rguenther@suse.de>
643 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
644 consistently accross all TUs.
645 (run_gcc): Enable -fshort-double automatically at link at link-time
646 and disallow override.
648 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
651 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
652 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
653 if they can't be used.
655 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
657 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
658 for Solaris 11/x86 ld.
659 * configure: Regenerate.
661 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
663 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
664 (LIB_TLS_SPEC): Save as ld_tls_libs.
665 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
666 (HAVE_AS_IX86_TLSLDM): New test.
667 * configure, config.in: Regenerate.
668 * config/i386/i386.c (legitimize_tls_address): Fall back to
669 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
670 cannot support TLS_MODEL_LOCAL_DYNAMIC.
671 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
672 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
674 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
676 * common.opt (fira-loop-pressure): Mark as optimization.
678 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
680 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
681 an OpenMP mappable type.
683 2014-03-06 Matthias Klose <doko@ubuntu.com>
685 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
686 MULTILIB_OSDIRNAMES is not defined.
688 2014-03-06 Jakub Jelinek <jakub@redhat.com>
689 Meador Inge <meadori@codesourcery.com>
692 * config/arm/arm.c (arm_tls_symbol_p): Remove.
693 (arm_legitimize_address): Call legitimize_tls_address for any
694 arm_tls_referenced_p expression, handle constant addend. Call it
695 before testing for !TARGET_ARM.
696 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
698 2014-03-06 Richard Biener <rguenther@suse.de>
704 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
706 * tree-streamer.c (record_common_node): Assert we don't record
707 nodes with type double.
708 (preload_common_node): Skip type double, complex double and double
709 pointer since it is now frontend dependent due to fshort-double option.
711 2014-03-06 Richard Biener <rguenther@suse.de>
713 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
714 or -fno-lto is specified and the linker has full plugin support.
715 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
716 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
717 * lto-wrapper.c (merge_and_complain): Merge compile-time
719 (run_gcc): And pass it through to the link options.
721 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
725 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
727 * cselib.c (remove_useless_values): Skip to avoid quadratic
728 behavior if the condition moved from...
729 (cselib_process_insn): ... here holds.
731 2014-03-05 Jakub Jelinek <jakub@redhat.com>
734 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
735 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
738 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
739 (TM_H): Add x86-tune.def.
741 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
743 * config/aarch64/aarch64.c (generic_tunings):
744 Use cortexa57_extra_costs.
746 2014-03-05 Jakub Jelinek <jakub@redhat.com>
749 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
750 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
751 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
754 2014-03-04 Heiher <r@hev.cc>
756 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
757 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
759 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
761 * config/i386/predicates.md (const2356_operand): Change to ...
762 (const2367_operand): ... this.
763 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
765 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
766 (*avx512pf_scatterpf<mode>sf): Ditto.
767 (avx512pf_scatterpf<mode>df): Ditto.
768 (*avx512pf_scatterpf<mode>df_mask): Ditto.
769 (*avx512pf_scatterpf<mode>df): Ditto.
770 * config/i386/i386.c (ix86_expand_builtin): Update
771 incorrect hint operand error message.
773 2014-03-04 Richard Biener <rguenther@suse.de>
775 * lto-section-in.c (lto_get_section_data): Fix const cast.
777 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
779 * tree-streamer.c (record_common_node): Assert we don't record
780 nodes with type double.
781 (preload_common_node): Skip type double, complex double and double
782 pointer since it is now frontend dependent due to fshort-double option.
784 2014-03-04 Richard Biener <rguenther@suse.de>
787 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
788 (lto_input_toplevel_asms): Likewise.
789 * lto-section-in.c (lto_get_section_data): Instead do it here
792 2014-03-04 Richard Biener <rguenther@suse.de>
794 PR tree-optimization/60382
795 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
796 dead PHIs a reduction.
798 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
800 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
802 (_mm_prefetch): Move out of GCC target("sse") pragma.
803 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
804 GCC target("prfchw") pragma.
805 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
807 * config/i386/i386.c (ix86_option_override_internal): Enable
808 -mprfchw with -mprefetchwt1.
810 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
812 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
815 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
817 * opts.h (CL_PCH_IGNORE): Define.
818 * targhooks.c (option_affects_pch_p):
819 Return false for options that have CL_PCH_IGNORE set.
820 * opt-functions.awk: Process PchIgnore.
821 * doc/options.texi: Document PchIgnore.
823 * config/arc/arc.opt (misize): Add PchIgnore property.
825 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
827 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
828 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
829 constraint on constants to permit them being loaded into
830 GENERAL_REGS or BASE_REGS.
832 2014-03-03 Nick Clifton <nickc@redhat.com>
834 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
835 anti-cacnonical alternatives.
836 (negandhi3_real): New pattern.
837 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
839 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
841 * config/avr/avr-mcus.def: Remove atxmega16x1.
842 * config/avr/avr-tables.opt: Regenerate.
843 * config/avr/t-multilib: Regenerate.
844 * doc/avr-mmcu.texi: Regenerate.
846 2014-03-03 Tobias Grosser <tobias@grosser.es>
847 Mircea Namolaru <mircea.namolaru@inria.fr>
849 PR tree-optimization/58028
850 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
853 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
855 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
856 not handled by recognizers.
858 2014-03-03 Jakub Jelinek <jakub@redhat.com>
861 * function.c (expand_function_end): Don't emit
862 clobber_return_register sequence if clobber_after is a BARRIER.
863 * cfgexpand.c (construct_exit_block): Append instructions before
864 return_label to prev_bb.
866 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
868 * config/rs6000/constraints.md: Document reserved use of "wc".
870 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
873 * ipa.c (function_and_variable_visibility): When dissolving comdat
874 group, also set all symbols to local.
876 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
881 2013-12-14 Jan Hubicka <jh@suse.cz>
883 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
885 2014-03-02 Jon Beniston <jon@beniston.com>
891 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
892 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
893 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
894 (simple_return, *simple_return): New patterns
895 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
896 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
898 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
900 * dwarf2out.c (gen_subprogram_die): Tidy.
902 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
905 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
906 (*mov_t_msb_neg_negc): ... this new insn.
908 2014-02-28 Jason Merrill <jason@redhat.com>
911 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
914 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
917 * dwarf2out.c (decltype_auto_die): New static.
918 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
919 (gen_type_die_with_usage): Handle 'decltype(auto)'.
920 (is_cxx_auto): Likewise.
922 2014-02-28 Ian Bolton <ian.bolton@arm.com>
924 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
925 we are not using general regs only.
927 2014-02-28 Richard Biener <rguenther@suse.de>
930 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
931 previous fix and only allow to remove trivial pre-headers
932 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
933 (remove_forwarder_block): Properly update the latch of a loop.
935 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
938 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
939 (cselib_preserved_hash_table): New.
940 (preserve_constants_and_equivs): Move preserved vals to it.
941 (cselib_find_slot): Look it up first.
942 (cselib_init): Initialize it.
943 (cselib_finish): Release it.
944 (dump_cselib_table): Dump it.
946 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
949 * cselib.c (remove_useless_values): Skip to avoid quadratic
950 behavior if the condition moved from...
951 (cselib_process_insn): ... here holds.
953 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
956 * var-tracking.c (vt_initialize): Apply the same condition to
957 preserve the CFA base value.
959 2014-02-28 Joey Ye <joey.ye@arm.com>
962 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
963 if reload in progress or completed.
965 2014-02-28 Tobias Burnus <burnus@net-b.de>
968 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
971 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
973 * doc/tm.texi.in (Condition Code Status): Update documention for
974 relative locations of cc0-setter and cc0-user.
976 2014-02-27 Jeff Law <law@redhat.com>
978 PR rtl-optimization/52714
979 * combine.c (try_combine): When splitting an unrecognized PARALLEL
980 into two independent simple sets, if I3 is a jump, ensure the
981 pattern we place into I3 is a (set (pc) ...).
983 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
984 Jeff Law <law@redhat.com>
986 PR rtl-optimization/49847
987 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
988 are in different blocks.
989 * doc/tm.texi (Condition Code Status): Update documention for
990 relative locations of cc0-setter and cc0-user.
992 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
995 * lra.c (lra_emit_add): Check SUBREG too.
997 2014-02-27 Andreas Schwab <schwab@suse.de>
999 * config/m68k/m68k.c (m68k_option_override): Disable
1000 -flive-range-shrinkage for classic m68k.
1001 (m68k_override_options_after_change): Likewise.
1003 2014-02-27 Marek Polacek <polacek@redhat.com>
1006 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1007 -Wmaybe-uninitialized.
1009 2014-02-27 Alan Modra <amodra@gmail.com>
1012 * reload1.c (emit_input_reload_insns): When reload_override_in,
1013 set old to rl->in_reg when rl->in_reg is a subreg.
1015 2014-02-26 Richard Biener <rguenther@suse.de>
1018 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1020 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1022 * common/config/i386/predicates.md (const1256_operand): Remove.
1023 (const2356_operand): New.
1024 (const_1_to_2_operand): Remove.
1025 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1026 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1027 (*avx512pf_gatherpf<mode>sf): Ditto.
1028 (avx512pf_gatherpf<mode>df): Ditto.
1029 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1030 (*avx512pf_gatherpf<mode>df): Ditto.
1031 (avx512pf_scatterpf<mode>sf): Ditto.
1032 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1033 (*avx512pf_scatterpf<mode>sf): Ditto.
1034 (avx512pf_scatterpf<mode>df): Ditto.
1035 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1036 (*avx512pf_scatterpf<mode>df): Ditto.
1037 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1039 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1041 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1042 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1043 (_mm512_mask_testn_epi64_mask): Move to ...
1044 * config/i386/avx512cdintrin.h: Here.
1045 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1046 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1047 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1048 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1049 TARGET_AVX512F from TARGET_AVX512CD.
1051 2014-02-26 Richard Biener <rguenther@suse.de>
1054 * ipa.c (walk_polymorphic_call_targets): Properly guard
1055 call to inline_update_overall_summary.
1057 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1060 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1061 and latches only if requested. Fix latch if it is removed.
1062 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1063 LOOPS_HAVE_PREHEADERS.
1065 2014-02-25 Andrew Pinski <apinski@cavium.com>
1067 * builtins.c (expand_builtin_thread_pointer): Create a new target
1068 when the target is NULL.
1070 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1072 PR rtl-optimization/60317
1073 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1074 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1075 * lra-assigns.c: Include params.h.
1076 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1077 other reload pseudos considerations.
1079 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1081 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1082 to use canonical form for nor<mode>3.
1084 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1087 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1090 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1092 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1093 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1094 (ix86_handle_option): Handle OPT_mprefetchwt1.
1095 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1096 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1098 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1099 OPTION_MASK_ISA_PREFETCHWT1.
1100 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1101 (PTA_PREFETCHWT1): New.
1102 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1103 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1104 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1105 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1106 (*prefetch_avx512pf_<mode>_: Change into ...
1107 (*prefetch_prefetchwt1_<mode>: This.
1108 * config/i386/i386.opt (mprefetchwt1): New.
1109 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1110 (_mm_prefetch): Handle intent to write.
1111 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1113 2014-02-25 Richard Biener <rguenther@suse.de>
1116 * emit-rtl.c (mem_attrs_htab): Remove.
1117 (mem_attrs_htab_hash): Likewise.
1118 (mem_attrs_htab_eq): Likewise.
1119 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1120 (init_emit_once): Do not allocate mem_attrs_htab.
1122 2014-02-25 Richard Biener <rguenther@suse.de>
1125 * lto-opts.c (lto_write_options): Output non-explicit conservative
1126 -fwrapv, -fno-trapv and -fno-strict-overflow.
1127 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1128 (run_gcc): And pass them through.
1130 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1132 * sel-sched.c (calculate_new_fences): New parameter ptime.
1133 Calculate it as a maximum over all fence cycles.
1134 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1135 Print the final schedule timing when sched_verbose.
1137 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1139 PR rtl-optimization/60292
1140 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1141 bit fot the fence instruction.
1143 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1145 * calls.h: Fix typo in comment.
1147 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1149 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1150 adjusting offsetable addresses.
1152 2014-02-24 Guozhi Wei <carrot@google.com>
1154 * sparseset.h (sparseset_pop): Fix the wrong index.
1156 2014-02-24 Walter Lee <walt@tilera.com>
1158 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1159 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1161 * common/config/tilegx/tilegx-common.c
1162 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1163 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1165 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1166 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1167 (tilegx_gimplify_va_arg_expr): Handle big endian.
1168 (tilegx_expand_unaligned_load): Ditto.
1169 (tilegx_expand_unaligned_store): Ditto.
1170 (TARGET_RETURN_IN_MSB): New.
1171 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1172 (TARGET_ENDIAN_DEFAULT): New.
1173 (TARGET_BIG_ENDIAN): Handle big endian.
1174 (BYTES_BIG_ENDIAN): Ditto.
1175 (WORDS_BIG_ENDIAN): Ditto.
1176 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1179 * config/tilegx/tilegx.md (extv): Handle big endian.
1181 (insn_st<n>): Ditto.
1182 (insn_st<n>_add<bitsuffix>): Ditto.
1183 (insn_stnt<n>): Ditto.
1184 (insn_stnt<n>_add<bitsuffix>):Ditto.
1185 (vec_interleave_highv8qi): Handle big endian.
1186 (vec_interleave_highv8qi_be): New.
1187 (vec_interleave_highv8qi_le): New.
1188 (insn_v1int_h): Handle big endian.
1189 (vec_interleave_lowv8qi): Handle big endian.
1190 (vec_interleave_lowv8qi_be): New.
1191 (vec_interleave_lowv8qi_le): New.
1192 (insn_v1int_l): Handle big endian.
1193 (vec_interleave_highv4hi): Handle big endian.
1194 (vec_interleave_highv4hi_be): New.
1195 (vec_interleave_highv4hi_le): New.
1196 (insn_v2int_h): Handle big endian.
1197 (vec_interleave_lowv4hi): Handle big endian.
1198 (vec_interleave_lowv4hi_be): New.
1199 (vec_interleave_lowv4hi_le): New.
1200 (insn_v2int_l): Handle big endian.
1201 (vec_interleave_highv2si): Handle big endian.
1202 (vec_interleave_highv2si_be): New.
1203 (vec_interleave_highv2si_le): New.
1204 (insn_v4int_h): Handle big endian.
1205 (vec_interleave_lowv2si): Handle big endian.
1206 (vec_interleave_lowv2si_be): New.
1207 (vec_interleave_lowv2si_le): New.
1208 (insn_v4int_l): Handle big endian.
1209 * config/tilegx/tilegx.opt (mbig-endian): New option.
1210 (mlittle-endian): New option.
1211 * doc/install.texi: Document tilegxbe-linux.
1212 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1214 2014-02-24 Martin Jambor <mjambor@suse.cz>
1217 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1218 there are no parameter descriptors.
1220 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1222 PR rtl-optimization/60268
1223 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1224 initialization to ...
1225 (sched_rgn_init): ... here.
1226 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1228 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1230 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1233 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1235 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1238 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1240 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1241 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1243 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1245 * config/microblaze/predicates.md: Add cmp_op predicate.
1246 * config/microblaze/microblaze.md: Add branch_compare instruction
1247 which uses cmp_op predicate and emits cmp insn before branch.
1248 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1249 to microblaze_expand_conditional_branch and consolidate logic.
1250 (microblaze_expand_conditional_branch): emit branch_compare
1251 insn instead of handling cmp op separate from branch insn.
1253 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1255 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1258 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1260 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1261 define_insn with define_expand and new define_insn
1262 *altivec_lve<VI_char>x_internal.
1263 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1264 and new define_insn *altivec_stve<VI_char>x_internal.
1265 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1267 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1269 (altivec_expand_stvex_be): New function.
1271 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1273 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1274 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1275 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1276 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1278 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1281 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1282 instead of emit_move_insn.
1284 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1286 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1288 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1289 gen_altivec_vsumsws.
1291 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1293 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1294 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1295 (altivec_lvxl_<mode>): New define_expand incorporating
1296 -maltivec=be semantics where needed.
1297 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1298 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1299 semantics where needed.
1300 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1301 (altivec_stvx_<mode>): New define_expand incorporating
1302 -maltivec=be semantics where needed.
1303 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1304 VM2 iterator instead of V4SI.
1305 (altivec_stvxl_<mode>): New define_expand incorporating
1306 -maltivec=be semantics where needed.
1307 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1308 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1309 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1310 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1311 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1313 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1314 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1315 ALTIVEC_BUILTIN_STVXL.
1316 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1317 (altivec_expand_stvx_be): Likewise.
1318 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1319 (altivec_expand_lvx_be): Likewise.
1320 (altivec_expand_stvx_be): Likewise.
1321 (altivec_expand_builtin): Add cases for
1322 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1323 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1324 (altivec_init_builtins): Add definitions for
1325 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1326 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1328 2014-02-21 Catherine Moore <clm@codesourcery.com>
1330 * doc/invoke.texi (mvirt, mno-virt): Document.
1331 * config/mips/mips.opt (mvirt): New option.
1332 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1334 2014-02-21 Richard Biener <rguenther@suse.de>
1336 PR tree-optimization/60276
1337 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1338 (STMT_VINFO_MIN_NEG_DIST): New macro.
1339 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1340 STMT_VINFO_MIN_NEG_DIST.
1341 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1342 made for negative dependence distances still hold.
1344 2014-02-21 Richard Biener <rguenther@suse.de>
1347 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1348 DECL_INITIAL for globals not in the current function context.
1350 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1352 PR tree-optimization/56490
1353 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1354 * tree-ssa-uninit.c: Include params.h.
1355 (compute_control_dep_chain): Add num_calls argument, return false
1356 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1357 num_calls to recursive call.
1358 (find_predicates): Change dep_chain into normal array,
1359 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1360 variable and adjust compute_control_dep_chain caller.
1361 (find_def_preds): Likewise.
1363 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1365 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1366 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1368 2014-02-21 Nick Clifton <nickc@redhat.com>
1370 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1371 (pushhi1): Likewise.
1372 (popqi1): Add mode to pre_dec.
1375 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1377 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1378 mode for mask of V8SFmode permutation.
1380 2014-02-20 Richard Henderson <rth@redhat.com>
1383 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1384 a new pseudo for OLDVAL.
1386 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1389 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1390 gen_reg_rtx if d->testing_p.
1391 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1392 if d->testing_p and we will certainly return true.
1393 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1396 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1398 * emit-rtl.c (gen_reg_rtx): Assert that
1399 crtl->emit.regno_pointer_align_length is non-zero.
1401 2014-02-20 Richard Henderson <rth@redhat.com>
1404 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1405 on failure the store back into EXPECT.
1407 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1408 Sandra Loosemore <sandra@codesourcery.com>
1410 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1411 * config/nios2/nios2.c (nios2_function_profiler): Add
1412 -fPIC (flag_pic == 2) support.
1413 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1414 (nios2_large_offset_p): New function.
1415 (nios2_unspec_reloc_p): Move up position, update to use
1416 nios2_large_offset_p.
1417 (nios2_unspec_address): Remove function.
1418 (nios2_unspec_offset): New function.
1419 (nios2_large_got_address): New function.
1420 (nios2_got_address): Add large offset support.
1421 (nios2_legitimize_tls_address): Update usage of removed and new
1423 (nios2_symbol_binds_local_p): New function.
1424 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1425 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1426 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1427 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1428 processing for (const (unspec ...)).
1429 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1431 2014-02-20 Richard Biener <rguenther@suse.de>
1433 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1434 doing the substitution.
1435 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1437 2014-02-20 Martin Jambor <mjambor@suse.cz>
1440 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1441 info when checking whether lattices are bottom.
1443 2014-02-20 Richard Biener <rguenther@suse.de>
1446 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1449 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1452 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1453 parameter specifying the scaling.
1454 (inline_call): Update.
1455 (want_inline_recursively): Guard division by zero.
1456 (recursive_inlining): Update.
1457 * ipa-inline.h (clone_inlined_nodes): Update.
1459 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1462 * config/i386/i386.c (classify_argument): Pass structures of size
1463 64 bytes or less in register.
1465 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1466 Kirill Yukhin <kirill.yukhin@intel.com>
1468 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1469 (_mm_rcp28_round_ss): Ditto.
1470 (_mm_rsqrt28_round_sd): Ditto.
1471 (_mm_rsqrt28_round_ss): Ditto.
1472 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1473 (_mm_rcp14_round_ss): Ditto.
1474 (_mm_rsqrt14_round_sd): Ditto.
1475 (_mm_rsqrt14_round_ss): Ditto.
1476 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1477 the first input operand, get rid of match_dup.
1478 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1480 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1482 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1483 operand as the first input operand, set type attribute.
1484 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1486 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1487 operand as the first input operand, set type attribute.
1489 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1491 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1494 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1497 * config/i386/i386.c (construct_container): Remove TFmode check
1498 for X86_64_INTEGER_CLASS.
1500 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1503 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1504 only when -Wpsabi is enabled.
1506 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1509 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1510 passing arrays in registers are the same as for structs, so remove the
1511 special case for them.
1513 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1515 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1516 destination type, extract only the valid bits if the source type is not
1517 integral and has a different mode.
1519 2014-02-19 Richard Biener <rguenther@suse.de>
1522 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1525 2014-02-19 Richard Biener <rguenther@suse.de>
1528 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1529 (ipa_modify_call_arguments): Emit an argument load explicitely and
1530 preserve virtual SSA form there and for the replacement call.
1531 Do not update SSA form nor free dominance info.
1533 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1535 * ipa.c (function_and_variable_visibility): Also clear WEAK
1536 flag when disolving COMDAT_GROUP.
1538 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1540 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1541 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1543 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1544 do more sanity checks.
1545 (detect_type_change): Return true when giving up early.
1546 (compute_complex_assign_jump_func): Fix type parameter of
1547 ipa_set_ancestor_jf.
1548 (compute_complex_ancestor_jump_func): Likewise.
1549 (update_jump_functions_after_inlining): Fix updating of
1551 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1553 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1555 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1556 inline clones when edge disappears.
1558 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1561 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1562 Split 64-bit moves into 2 patterns. Do not allow the use of
1563 direct move for TDmode in little endian, since the decimal value
1564 has little endian bytes within a word, but the 64-bit pieces are
1565 ordered in a big endian fashion, and normal subreg's of TDmode are
1567 (mov<mode>_64bit_dm): Likewise.
1568 (movtd_64bit_nodm): Likewise.
1570 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1572 PR tree-optimization/60174
1573 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1574 statement of an SSA_NAME that occurs in an abnormal PHI node.
1576 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1579 * final.c (SEEN_BB): Remove.
1580 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1581 (final_scan_insn): Don't force_source_line on second
1582 NOTE_INSN_BASIC_BLOCK.
1584 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1587 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1588 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1589 (type_natural_mode): Warn ABI change when %zmm register is not
1590 available for AVX512F vector value passing.
1592 2014-02-18 Kai Tietz <ktietz@redhat.com>
1595 * config/i386/i386.c (ix86_expand_prologue): Use value in
1596 rax register as displacement when restoring %r10 or %rax.
1597 Fix wrong offset when restoring both registers.
1599 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1601 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1602 assertion with conditional return.
1604 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1605 Uros Bizjak <ubizjak@gmail.com>
1608 * config/i386/driver-i386.c (host_detect_local_cpu): If
1609 YMM state is not saved by the OS, also clear has_f16c. Move
1610 CPUID 0x80000001 handling before YMM state saving checking.
1612 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1614 PR rtl-optimization/58960
1615 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1616 factored out from ...
1617 (sched_init): ... here.
1618 (free_global_sched_pressure_data): New, factored out from ...
1619 (sched_finish): ... here.
1620 * sched-int.h (free_global_sched_pressure_data): Declare.
1621 * sched-rgn.c (nr_regions_initial): New static global.
1622 (haifa_find_rgns): Initialize it.
1623 (schedule_region): Disable sched-pressure for the newly
1626 2014-02-17 Richard Biener <rguenther@suse.de>
1628 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1629 release SSA defs of pattern stmts.
1631 2014-02-17 Richard Biener <rguenther@suse.de>
1633 * tree-inline.c (expand_call_inline): Release the virtual
1634 operand defined by the call we are about to inline.
1636 2014-02-17 Richard Biener <rguenther@suse.de>
1638 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1640 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1641 Ilya Tocar <ilya.tocar@intel.com>
1643 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1644 arguments order in builtin.
1645 (_mm512_permutexvar_epi64): Ditto.
1646 (_mm512_mask_permutexvar_epi64): Ditto
1647 (_mm512_maskz_permutexvar_epi32): Ditto
1648 (_mm512_permutexvar_epi32): Ditto
1649 (_mm512_mask_permutexvar_epi32): Ditto
1651 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1653 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1654 (p8_vmrgow): Likewise.
1656 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1658 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1661 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1664 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1665 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1666 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1667 using direct move instructions on ISA 2.07. Also adjust
1668 instruction length for 64-bit.
1669 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1670 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1672 2014-02-15 Alan Modra <amodra@gmail.com>
1676 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1677 find_replacement on parts of insn rtl that might be reloaded.
1679 2014-02-15 Richard Biener <rguenther@suse.de>
1681 PR tree-optimization/60183
1682 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1683 (tree_ssa_phiprop): Calculate and free post-dominators.
1685 2014-02-14 Jeff Law <law@redhat.com>
1687 PR rtl-optimization/60131
1688 * ree.c (get_extended_src_reg): New function.
1689 (combine_reaching_defs): Use it rather than assuming location of REG.
1690 (find_and_remove_re): Verify first operand of extension is
1691 a REG before adding the insns to the copy list.
1693 2014-02-14 Roland McGrath <mcgrathr@google.com>
1695 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1696 * configure: Regenerated.
1697 * config.in: Regenerated.
1698 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1699 instead of ASM_SHORT.
1701 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1702 Richard Earnshaw <rearnsha@arm.com>
1704 PR rtl-optimization/59535
1705 * lra-constraints.c (process_alt_operands): Encourage alternative
1706 when unassigned pseudo class is superset of the alternative class.
1707 (inherit_reload_reg): Don't inherit when optimizing for code size.
1708 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1709 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1710 modes not less than 4 for Thumb1.
1712 2014-02-14 Kyle McMartin <kyle@redhat.com>
1715 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1717 2014-02-14 Richard Biener <rguenther@suse.de>
1719 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1720 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1721 Do not strip INDIRECT_REFs.
1723 2014-02-14 Richard Biener <rguenther@suse.de>
1726 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1727 DECL_FUNCTION_SPECIFIC_TARGET.
1728 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1729 * tree-streamer-out.c (pack_ts_target_option): Remove.
1730 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1731 (write_ts_function_decl_tree_pointers): Do not stream
1732 DECL_FUNCTION_SPECIFIC_TARGET.
1733 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1734 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1735 (lto_input_ts_function_decl_tree_pointers): Do not stream
1736 DECL_FUNCTION_SPECIFIC_TARGET.
1738 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1740 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1741 (get_initial_def_for_induction, vectorizable_induction): Ignore
1742 debug stmts when looking for exit_phi.
1743 (vectorizable_live_operation): Fix up condition.
1745 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1747 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1748 nreverse() because it changes the content of original tree list.
1750 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1752 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1753 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1755 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1757 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1758 GNU coding standards.
1760 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1763 * dwarf2out.c (gen_subprogram_die): Don't call
1764 add_calling_convention_attribute if subr_die is old_die.
1766 2014-02-13 Sharad Singhai <singhai@google.com>
1768 * doc/optinfo.texi: Fix order of nodes.
1770 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1772 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1773 operands[2], not operands[3].
1775 2014-02-13 Richard Biener <rguenther@suse.de>
1778 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1779 mention the possibility of an in-tree build.
1780 (CLooG): Update recommended version to 0.18.1, mention the
1781 possibility of an in-tree build and clarify that the ISL
1782 bundled with CLooG does not work.
1784 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1787 * expr.c (compress_float_constant): If x is a hard register,
1788 extend into a pseudo and then move to x.
1790 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1792 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1793 caused by bad second argument to warning_at() with -mhotpatch and
1794 nested functions (e.g. with gfortran).
1796 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1798 * opts.c (option_name): Remove "enabled by default" rider.
1800 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1802 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1804 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1805 Uros Bizjak <ubizjak@gmail.com>
1808 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1809 * configure: Regenerated.
1811 2014-02-12 Richard Biener <rguenther@suse.de>
1813 * vec.c (vec_prefix::calculate_allocation): Move as
1814 inline variant to vec.h.
1815 (vec_prefix::calculate_allocation_1): New out-of-line version.
1816 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1817 (vec_prefix::m_has_auto_buf): Rename to ...
1818 (vec_prefix::m_using_auto_storage): ... this.
1819 (vec_prefix::calculate_allocation): Inline the easy cases
1820 and dispatch to calculate_allocation_1 which doesn't need the
1822 (va_heap::reserve): Use gcc_checking_assert.
1823 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1824 m_using_auto_storage.
1825 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1827 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1828 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1829 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1831 2014-02-12 Richard Biener <rguenther@suse.de>
1833 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1834 when we found a dependence.
1836 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1838 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1840 (maybe_fold_stmt): ... into this new function.
1841 * omp-low.c (lower_omp): Update comment.
1843 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1846 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1849 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1851 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1852 identifiers in comments.
1853 (cortexa53_extra_costs): Likewise.
1854 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1855 (cortexa7_extra_costs): Likewise.
1856 (cortexa12_extra_costs): Likewise.
1857 (cortexa15_extra_costs): Likewise.
1858 (v7m_extra_costs): Likewise.
1860 2014-02-12 Richard Biener <rguenther@suse.de>
1863 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1864 of posix_memalign being successful.
1865 (lower_stmt): Restrict lowering of posix_memalign to when
1866 -ftree-bit-ccp is enabled.
1868 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1870 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1872 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1874 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1876 PR rtl-optimization/60116
1877 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1878 other_insn once the combination has been validated.
1880 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1883 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1885 * ipa-devirt.c: Include demangle.h
1886 (odr_violation_reported): New static variable.
1887 (add_type_duplicate): Update odr_violations.
1888 (maybe_record_node): Add completep parameter; update it.
1889 (record_target_from_binfo): Add COMPLETEP parameter;
1890 update it as needed.
1891 (possible_polymorphic_call_targets_1): Likewise.
1892 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1893 rename FINAL to COMPLETE.
1894 (record_targets_from_bases): Sanity check we found the binfo;
1895 fix COMPLETEP updating.
1896 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1897 parameter, fix computing of COMPLETEP.
1898 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1899 at LTO time do demangling.
1900 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1901 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1903 (gimple_get_virt_method_for_binfo): Likewise.
1904 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1905 gimple_get_virt_method_for_vtable): Update prototypes.
1907 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1910 * genautomata.c (add_presence_absence): Fix typo with
1911 {final_}presence_list.
1913 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1916 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1917 for VSX/Altivec vectors that land in GPR registers.
1919 2014-02-11 Richard Henderson <rth@redhat.com>
1920 Jakub Jelinek <jakub@redhat.com>
1923 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1924 around drhs if type conversion to lacc->type is not useless.
1926 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1928 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1930 (cortex-a57.cortex-a53): Likewise.
1931 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1933 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1935 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1938 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1940 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1941 add_options_for_arm_vfp3.
1943 2014-02-11 Jeff Law <law@redhat.com>
1946 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1947 object with an undesirable mode.
1949 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1952 * config/i386/sol2-9.h: New file.
1953 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1954 *-*-solaris2.9*): Use it.
1956 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1958 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1959 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1961 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1963 * config/microblaze/microblaze.c: Extend mcpu version format
1965 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1967 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1969 2014-02-10 Richard Henderson <rth@redhat.com>
1972 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1973 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1974 ms-abi vs -mno-accumulate-outgoing-args.
1975 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1976 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1979 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1982 * cfgexpand.c (expand_asm_operands): Attach source location to
1983 ASM_INPUT rtx objects.
1984 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1986 2014-02-10 Nick Clifton <nickc@redhat.com>
1988 * config/mn10300/mn10300.c (popcount): New function.
1989 (mn10300_expand_prologue): Include saved registers in stack usage
1992 2014-02-10 Jeff Law <law@redhat.com>
1995 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1996 when changing the SET_DEST of a prior insn to avoid an input reload.
1998 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2000 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2001 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2002 -mcall-openbsd, or -mcall-linux.
2003 (CC1_ENDIAN_BIG_SPEC): Remove.
2004 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2005 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2006 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2007 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2008 and %cc1_endian_default.
2009 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2011 2014-02-10 Richard Biener <rguenther@suse.de>
2013 PR tree-optimization/60115
2014 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2015 MEM_REF handling. Properly verify that the accesses are not
2016 out of the objects bound.
2018 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2020 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2023 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2025 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2026 proper constants and fix formatting.
2027 (possible_polymorphic_call_targets): Fix formatting.
2029 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2030 Ilya Tocar <ilya.tocar@intel.com>
2032 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2033 (_mm512_loadu_epi32): Renamed into...
2034 (_mm512_loadu_si512): This.
2035 (_mm512_storeu_epi32): Renamed into...
2036 (_mm512_storeu_si512): This.
2037 (_mm512_maskz_ceil_ps): Removed.
2038 (_mm512_maskz_ceil_pd): Ditto.
2039 (_mm512_maskz_floor_ps): Ditto.
2040 (_mm512_maskz_floor_pd): Ditto.
2041 (_mm512_floor_round_ps): Ditto.
2042 (_mm512_floor_round_pd): Ditto.
2043 (_mm512_ceil_round_ps): Ditto.
2044 (_mm512_ceil_round_pd): Ditto.
2045 (_mm512_mask_floor_round_ps): Ditto.
2046 (_mm512_mask_floor_round_pd): Ditto.
2047 (_mm512_mask_ceil_round_ps): Ditto.
2048 (_mm512_mask_ceil_round_pd): Ditto.
2049 (_mm512_maskz_floor_round_ps): Ditto.
2050 (_mm512_maskz_floor_round_pd): Ditto.
2051 (_mm512_maskz_ceil_round_ps): Ditto.
2052 (_mm512_maskz_ceil_round_pd): Ditto.
2053 (_mm512_expand_pd): Ditto.
2054 (_mm512_expand_ps): Ditto.
2055 * config/i386/i386.c (ix86_builtins): Remove
2056 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2057 (bdesc_args): Ditto.
2058 * config/i386/predicates.md (const1256_operand): New.
2059 (const_1_to_2_operand): Ditto.
2060 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2061 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2062 (*avx512pf_gatherpf<mode>sf): Ditto.
2063 (avx512pf_gatherpf<mode>df): Ditto.
2064 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2065 (*avx512pf_gatherpf<mode>df): Ditto.
2066 (avx512pf_scatterpf<mode>sf): Ditto.
2067 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2068 (*avx512pf_scatterpf<mode>sf): Ditto.
2069 (avx512pf_scatterpf<mode>df): Ditto.
2070 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2071 (*avx512pf_scatterpf<mode>df): Ditto.
2072 (avx512f_expand<mode>): Removed.
2073 (<shift_insn><mode>3<mask_name>): Change predicate type.
2075 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2077 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2078 not at the end of datarefs vector use ordered_remove to avoid
2079 reordering datarefs vector.
2082 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2083 mark local addressable non-static vars as GOVD_PRIVATE
2084 instead of GOVD_LOCAL.
2085 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2086 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2090 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2091 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2092 assume_aligned or alloc_align attributes.
2093 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2094 arguments. Handle also assume_aligned and alloc_align attributes.
2095 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2096 calls to functions with assume_aligned or alloc_align attributes.
2097 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2099 2014-02-08 Terry Guo <terry.guo@arm.com>
2101 * doc/invoke.texi: Document ARM -march=armv7e-m.
2103 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2105 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2106 flag on __cilkrts_rethrow builtin.
2109 * ipa-cp.c (determine_versionability): Fail at -O0
2110 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2111 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2114 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2117 * tree-inline.c (copy_forbidden): Fail for
2118 __attribute__((optimize (0))) functions.
2120 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2122 * varpool.c: Include pointer-set.h.
2123 (varpool_remove_unreferenced_decls): Variables in other partitions
2124 will not be output; be however careful to not lose information
2127 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2129 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2130 lookup in the vtable constructor.
2132 2014-02-07 Jeff Law <law@redhat.com>
2135 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2136 define_insn_and_split.
2138 * ipa-inline.c (inline_small_functions): Fix typos.
2140 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2142 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2143 (s390_can_use_return_insn): Declare.
2144 * config/s390/s390.h (EPILOGUE_USES): Define.
2145 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2147 (s390_chunkify_start): Handle return JUMP_LABELs.
2148 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2149 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2150 (s390_can_use_return_insn): New functions.
2151 (s390_fix_long_loop_prediction): Handle conditional returns.
2152 (TARGET_SET_UP_BY_PROLOGUE): Define.
2153 * config/s390/s390.md (ANY_RETURN): New code iterator.
2154 (*creturn, *csimple_return, return, simple_return): New patterns.
2156 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2158 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2159 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2160 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2161 REG_CFA_RESTORE list when deciding not to restore a register.
2163 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2165 * config/s390/s390.c: Include tree-pass.h and context.h.
2166 (s390_early_mach): New function, split out from...
2167 (s390_emit_prologue): ...here.
2168 (pass_data_s390_early_mach): New pass structure.
2169 (pass_s390_early_mach): New class.
2170 (s390_option_override): Create and register early_mach pass.
2171 Move to end of file.
2173 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2175 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2176 to match for the exit block.
2178 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2180 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2181 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2182 Reject misaligned operands.
2184 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2186 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2188 2014-02-07 Richard Biener <rguenther@suse.de>
2191 * gimple-low.c (lower_builtin_posix_memalign): New function.
2192 (lower_stmt): Call it to lower posix_memalign in a way
2193 to make alignment info accessible.
2195 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2198 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2199 __builtin_setjmp_receiver.
2201 2014-02-07 Richard Biener <rguenther@suse.de>
2204 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2205 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2206 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2207 Handle BUILT_IN_POSIX_MEMALIGN.
2208 (find_func_clobbers): Likewise.
2209 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2210 (call_may_clobber_ref_p_1): Likewise.
2212 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2215 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2218 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2221 * lto-cgraph.c (lto_output_node): Use
2222 symtab_get_symbol_partitioning_class.
2223 (lto_output_varpool_node): likewise.
2224 (symtab_get_symbol_partitioning_class): Move here from
2226 * cgraph.h (symbol_partitioning_class): Likewise.
2227 (symtab_get_symbol_partitioning_class): Declare.
2229 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2231 * ggc.h (ggc_internal_cleared_alloc): New macro.
2232 * vec.h (vec_safe_copy): Handle memory stats.
2233 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2234 * target-globals.c (save_target_globals): Likewise.
2236 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2239 * expr.c (emit_move_resolve_push): Export; be bit more selective
2240 on when to clear alias set.
2241 * expr.h (emit_move_resolve_push): Declare.
2242 * function.h (struct function): Add tail_call_marked.
2243 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2244 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2245 * config/i386/i386.md (TImode move expander): De not call
2247 (FP push expanders): Preserve memory attributes.
2248 * config/i386/sse.md (push<mode>1): Remove.
2249 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2250 (ix86_expand_push): Remove.
2251 * config/i386/mmx.md (push<mode>1): Remove.
2253 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2255 PR rtl-optimization/60030
2256 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2257 lopart with paradoxical subreg before shifting it up by hprec.
2259 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2261 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2262 Remove extra newline at end of file.
2263 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2264 (arm_issue_rate): Handle cortexa57.
2265 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2266 (cortex-a57.cortex-a53): Likewise.
2268 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2271 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2272 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2274 (arm_expand_prologue): Adjust all callers.
2275 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2276 info, registers also at the lowest numbered registers side. Use
2277 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2281 * var-tracking.c (adjust_mems): Before adding a SET to
2282 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2284 2014-02-06 Alan Modra <amodra@gmail.com>
2287 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2288 change SDmode to DDmode when lra_in_progress.
2290 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2293 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2294 free_data_ref on the dr first, and before goto again also set dr
2295 to the next dr. For simd_lane_access, free old datarefs[i] before
2296 overwriting it. For get_vectype_for_scalar_type failure, don't
2297 free_data_ref if simd_lane_access.
2299 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2302 * tree.h (opts_for_fn): New inline function.
2303 (opt_for_fn): Define.
2304 * config/i386/i386.c (ix86_function_regparm): Use
2305 opt_for_fn (decl, optimize) instead of optimize.
2307 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2309 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2310 for SYMBOL_REF in large memory model.
2312 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2314 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2316 (cortex-a57): Likewise.
2317 (cortex-a57.cortex-a53): Likewise.
2319 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2320 Kugan Vivekanandarajah <kuganv@linaro.org>
2322 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2324 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2326 2014-02-06 Richard Biener <rguenther@suse.de>
2328 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2329 set_loop_copy and initialize_original_copy_tables.
2331 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2333 * config/aarch64/aarch64-simd.md
2334 (aarch64_ashr_simddi): Change QI to SI.
2336 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2337 Jakub Jelinek <jakub@redhat.com>
2340 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2343 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2345 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2346 CODE_FOR_altivec_vpku[hw]um to
2347 CODE_FOR_altivec_vpku[hw]um_direct.
2348 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2349 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2350 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2351 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2353 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2355 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2356 generation for -maltivec=be.
2357 (altivec_vsumsws): Simplify redundant test.
2359 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2361 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2362 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2363 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2364 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2365 gen_altivec_vpkuwum.
2366 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2368 (altivec_vpks<VI_char>ss): Likewise.
2369 (altivec_vpks<VI_char>us): Likewise.
2370 (altivec_vpku<VI_char>us): Likewise.
2371 (altivec_vpku<VI_char>um): Likewise.
2372 (altivec_vpku<VI_char>um_direct): New (copy of
2373 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2375 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2376 target is little endian and -maltivec=be is not specified.
2377 (*altivec_vupkhs<VU_char>_direct): New (copy of
2378 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2379 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2380 target is little endian and -maltivec=be is not specified.
2381 (*altivec_vupkls<VU_char>_direct): New (copy of
2382 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2383 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2384 little endian and -maltivec=be is not specified.
2385 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2386 little endian and -maltivec=be is not specified.
2388 2014-02-05 Richard Henderson <rth@redhat.com>
2391 * combine-stack-adj.c: Revert r206943.
2392 * sched-int.h (struct deps_desc): Add last_args_size.
2393 * sched-deps.c (init_deps): Initialize it.
2394 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2395 contain REG_ARGS_SIZE notes.
2397 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2399 * lto-cgraph.c (asm_nodes_output): Make global.
2400 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2401 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2402 (driver_handle_option): Handle OPT_fwpa.
2404 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2407 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2408 a comment typo and formatting issue. If odr_hash hasn't been
2409 created, return vNULL and set *completep to false.
2412 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2413 bb with no successors.
2415 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2418 * doc/invoke.texi (-march): Clarify documentation for ARM.
2422 2014-02-05 Richard Biener <rguenther@suse.de>
2424 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2425 when not vectorizing because of too many alias checks.
2426 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2427 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2429 2014-02-05 Nick Clifton <nickc@redhat.com>
2431 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2432 accept extended registers in any mode when compiling for the MN10300.
2434 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2436 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2437 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2438 sanitization attributes.
2439 (can_inline_edge_p): Likewise.
2440 (sanitize_attrs_match_for_inline_p): New function.
2442 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2444 * ipa-prop.c (detect_type_change): Shor circuit testing of
2445 type changes on THIS pointer.
2447 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2450 * config/pa/pa.c (legitimize_tls_address): Return original address
2451 if not passed a SYMBOL_REF rtx.
2452 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2454 (pa_emit_move_sequence): Simplify TLS source operands.
2455 (pa_legitimate_constant_p): Reject all TLS constants.
2456 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2457 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2459 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2461 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2462 groups when we know they are controlled by LTO.
2463 * varasm.c (default_binds_local_p_1): If object is in other partition,
2464 it will be resolved locally.
2466 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2468 * config/host-linux.c (linux_gt_pch_use_address): Don't
2469 use SSIZE_MAX because it is not always defined.
2471 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2474 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2475 threshold for pseudo splitting.
2476 (update_ebb_live_info): Process call argument hard registers and
2477 hard registers from insn definition too.
2478 (max_small_class_regs_num): New constant.
2479 (inherit_in_ebb): Update live hard regs through EBBs. Update
2480 reloads_num only for small register classes. Don't split for
2483 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2486 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2489 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2491 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2494 2014-02-04 Marek Polacek <polacek@redhat.com>
2496 * gdbinit.in (pel): Define.
2498 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2500 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2503 2014-02-04 Richard Biener <rguenther@suse.de>
2506 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2507 in function context local.
2508 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2509 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2510 similar to LTO_imported_decl_ref.
2512 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2514 PR tree-optimization/60002
2515 * cgraphclones.c (build_function_decl_skip_args): Clear
2518 PR tree-optimization/60023
2519 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2520 false to gsi_replace.
2521 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2522 has been in some EH region and vec_stmt could throw, add
2523 vec_stmt into the same EH region.
2524 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2525 has no lhs, ignore it.
2526 * internal-fn.c (expand_MASK_LOAD): Likewise.
2529 * tree-inline.c (copy_forbidden): Fail for
2530 __attribute__((optimize (0))) functions.
2533 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2534 is set, copy one less argument.
2535 (expand_simd_clones): Don't subtract clone_info->inbranch
2536 from simd_clone_struct_alloc argument.
2538 PR rtl-optimization/57915
2539 * recog.c (simplify_while_replacing): If all unary/binary/relational
2540 operation arguments are constant, attempt to simplify those.
2543 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2544 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2546 2014-02-04 Richard Biener <rguenther@suse.de>
2548 PR tree-optimization/60012
2549 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2550 TBAA disambiguation to all DDRs.
2552 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2555 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2556 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2558 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2561 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2563 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2565 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2566 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2568 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2571 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2572 to figure out targets of polymorphic calls with known decl.
2573 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2574 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2575 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2576 (get_polymorphic_call_info): ... here.
2577 (get_polymorphic_call_info_from_invariant): New function.
2579 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2581 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2582 lookup via vtable pointer; check for type consistency
2583 and turn inconsitent facts into UNREACHABLE.
2584 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2585 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2586 type inconsistent querries; return UNREACHABLE instead.
2588 2014-02-03 Richard Henderson <rth@twiddle.net>
2591 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2592 already processed this node.
2593 (normalize_one_pred_1): Pass along mark_set.
2594 (normalize_one_pred): Create and destroy a pointer_set_t.
2595 (normalize_one_pred_chain): Likewise.
2597 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2599 PR gcov-profile/58602
2600 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2602 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2605 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2606 -fno-devirtualize; try to devirtualize by the knowledge of
2607 virtual table pointer given by aggregate propagation.
2608 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2609 (ipa_print_node_jump_functions): Dump also offset that
2610 is relevant for polymorphic calls.
2611 (determine_known_aggregate_parts): Add arg_type parameter; use it
2612 instead of determining the type from pointer type.
2613 (ipa_compute_jump_functions_for_edge): Update call of
2614 determine_known_aggregate_parts.
2615 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2616 (gimple_get_virt_method_for_binfo): ... here; simplify using
2617 vtable_pointer_value_to_vtable.
2618 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2619 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2620 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2621 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2623 (vtable_pointer_value_to_binfo): ... here.
2624 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2626 2014-02-03 Teresa Johnson <tejohnson@google.com>
2628 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2629 redef of outer loop index variable.
2631 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2635 * doc/extend.texi (Function Attributes): Typo.
2637 2014-02-03 Cong Hou <congh@google.com>
2639 PR tree-optimization/60000
2640 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2641 if the vectorized statement is a store. A store statement can only
2642 appear at the end of pattern statements.
2644 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2646 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2647 (ix86_option_override_internal): Default long double to 64-bit for
2648 32-bit Bionic and to 128-bit for 64-bit Bionic.
2650 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2651 TARGET_LONG_DOUBLE_128 is true.
2652 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2654 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2655 (mlong-double-64): Negate -mlong-double-128.
2656 (mlong-double-128): New option.
2658 * config/i386/i386-c.c (ix86_target_macros): Define
2659 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2661 * doc/invoke.texi: Document -mlong-double-128.
2663 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2665 PR rtl-optimization/60024
2666 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2668 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2670 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2672 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2674 PR rtl-optimization/57662
2675 * sel-sched.c (code_motion_path_driver): Do not mark already not
2676 existing blocks in the visiting bitmap.
2678 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2680 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2681 on the insn being emitted.
2683 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2684 Will Deacon <will.deacon@arm.com>
2686 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2688 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2690 * config/arm/arm-tables.opt: Regenerate.
2692 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2694 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2695 for vector types other than V16QImode.
2696 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2697 define_expand, and call altivec_expand_vec_perm_le when producing
2698 code with little endian element order.
2699 (*altivec_vperm_<mode>_internal): New insn having previous
2700 behavior of altivec_vperm_<mode>.
2701 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2702 altivec_expand_vec_perm_le when producing code with little endian
2704 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2705 behavior of altivec_vperm_<mode>_uns.
2707 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2709 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2710 (altivec_vsumsws): Add handling for -maltivec=be with a little
2712 (altivec_vsumsws_direct): New.
2713 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2714 gen_altivec_vsumsws.
2716 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2718 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2719 vtable_pointer_value_to_binfo): New functions.
2720 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2721 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2723 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2725 * config/nios2/nios2.md (load_got_register): Initialize GOT
2726 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2727 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2729 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2731 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2732 preserverd by passthrough, do not propagate the type.
2734 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2736 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2737 (mips_atomic_assign_expand_fenv): New function.
2738 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2740 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2742 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2743 (__builtin_mips_set_fcsr): Likewise.
2744 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2745 MIPS_USI_FTYPE_VOID.
2746 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2747 (mips16_expand_set_fcsr): Likewise.
2748 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2749 (mips16_set_fcsr_stub): Likewise.
2750 (mips16_get_fcsr_one_only_stub): New class.
2751 (mips16_set_fcsr_one_only_stub): Likewise.
2752 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2753 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2754 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2755 (hard_float): New availability predicate.
2756 (mips_builtins): Add get_fcsr and set_fcsr.
2757 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2758 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2759 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2760 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2761 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2764 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2766 * config/mips/mips.c (mips_one_only_stub): New class.
2767 (mips_need_mips16_rdhwr_p): Replace with...
2768 (mips16_rdhwr_stub): ...this new variable.
2769 (mips16_stub_call_address): New function.
2770 (mips16_rdhwr_one_only_stub): New class.
2771 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2772 (mips_output_mips16_rdhwr): Delete.
2773 (mips_finish_stub): New function.
2774 (mips_code_end): Use it to handle rdhwr stubs.
2776 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2779 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2780 when calculating size of integer atomic types.
2782 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2784 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2786 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2788 PR tree-optimization/60003
2789 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2790 * profile.c (branch_prob): Use gimple_call_builtin_p
2791 to check for BUILT_IN_SETJMP_RECEIVER.
2792 * tree-inline.c (copy_bb): Call notice_special_calls.
2794 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2797 * lra-constraints.c (process_alt_operands): Update reload_sum only
2800 2014-01-31 Richard Henderson <rth@redhat.com>
2803 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2804 until after else_eh is processed.
2806 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2808 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2809 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2810 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2811 in smmintrin.h, remove them.
2812 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2813 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2814 * config/i386/i386.md (ROUND_SAE): Fix value.
2815 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2816 (const48_operand): New.
2817 * config/i386/subst.md (round), (round_expand): Use
2818 const_4_or_8_to_11_operand.
2819 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2821 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2823 * config/i386/constraints.md (Yk): Swap meaning with k.
2824 * config/i386/i386.md (movhi_internal): Change Yk to k.
2825 (movqi_internal): Ditto.
2826 (*k<logic><mode>): Ditto.
2829 (kandn<mode>): Ditto.
2830 (*<code>hi_1): Ditto.
2831 (*<code>qi_1): Ditto.
2832 (kxnor<mode>): Ditto.
2833 (kortestzhi): Ditto.
2834 (kortestchi): Ditto.
2836 (*one_cmplhi2_1): Ditto.
2837 (*one_cmplqi2_1): Ditto.
2838 * config/i386/sse.md (): Change k to Yk.
2839 (avx512f_load<mode>_mask): Ditto.
2840 (avx512f_blendm<mode>): Ditto.
2841 (avx512f_store<mode>_mask): Ditto.
2842 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2843 (avx512f_storedqu<mode>_mask): Ditto.
2844 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2846 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2847 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2848 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2849 (avx512f_maskcmp<mode>3): Ditto.
2850 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2851 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2852 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2853 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2854 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2855 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2856 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2857 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2858 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2859 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2860 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2861 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2862 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2863 (vec_extract_lo_<mode>_maskm): Ditto.
2864 (vec_extract_hi_<mode>_maskm): Ditto.
2865 (avx512f_vternlog<mode>_mask): Ditto.
2866 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2867 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2868 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2869 (avx512f_<code>v8div16qi2_mask): Ditto.
2870 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2871 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2872 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2873 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2874 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2875 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2876 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2877 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2878 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2879 (avx512cd_maskb_vec_dupv8di): Ditto.
2880 (avx512cd_maskw_vec_dupv16si): Ditto.
2881 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2882 (avx512f_vpermi2var<mode>3_mask): Ditto.
2883 (avx512f_vpermi2var<mode>3_mask): Ditto.
2884 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2885 (*avx512f_gathersi<mode>): Ditto.
2886 (*avx512f_gathersi<mode>_2): Ditto.
2887 (*avx512f_gatherdi<mode>): Ditto.
2888 (*avx512f_gatherdi<mode>_2): Ditto.
2889 (*avx512f_scattersi<mode>): Ditto.
2890 (*avx512f_scatterdi<mode>): Ditto.
2891 (avx512f_compress<mode>_mask): Ditto.
2892 (avx512f_compressstore<mode>_mask): Ditto.
2893 (avx512f_expand<mode>_mask): Ditto.
2894 * config/i386/subst.md (mask): Change k to Yk.
2895 (mask_scalar_merge): Ditto.
2898 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2900 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2902 2014-01-31 Richard Biener <rguenther@suse.de>
2905 * builtins.c (fold_builtin_memory_op): Make sure to not
2906 use a floating-point mode or a boolean or enumeral type for
2909 2014-01-30 DJ Delorie <dj@redhat.com>
2911 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2912 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2913 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2914 whenever main() has an epilogue.
2916 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2918 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2919 unused variable "field".
2920 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2921 (vsx_mergeh_<mode>): Likewise.
2922 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2923 (altivec_vmrghh): Likewise.
2924 (altivec_vmrghw): Likewise.
2925 (altivec_vmrglb): Likewise.
2926 (altivec_vmrglh): Likewise.
2927 (altivec_vmrglw): Likewise.
2928 (altivec_vspltb): Add missing uses.
2929 (altivec_vsplth): Likewise.
2930 (altivec_vspltw): Likewise.
2931 (altivec_vspltsf): Likewise.
2933 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2936 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2937 frame related instructions.
2939 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2941 PR rtl-optimization/59959
2942 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2943 any reload of register whose subreg is invalid.
2945 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2947 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2948 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2949 Add missing return type - void.
2951 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2953 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2954 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2955 remove element index adjustment for endian (now handled in vsx.md
2957 (altivec_expand_vec_perm_const): Use
2958 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2959 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2960 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2961 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2962 define_expand and a new define_insn *altivec_vspltb_internal;
2963 adjust for -maltivec=be on a little endian target.
2964 (altivec_vspltb_direct): New.
2965 (altivec_vsplth): Divide into a define_expand and a new
2966 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2967 little endian target.
2968 (altivec_vsplth_direct): New.
2969 (altivec_vspltw): Divide into a define_expand and a new
2970 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2971 little endian target.
2972 (altivec_vspltw_direct): New.
2973 (altivec_vspltsf): Divide into a define_expand and a new
2974 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2975 a little endian target.
2977 2014-01-30 Richard Biener <rguenther@suse.de>
2979 PR tree-optimization/59993
2980 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2981 can propagate form the earlier stmt and avoid the transform
2982 when the intermediate result is needed.
2984 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2986 * README.Portability: Fix typo.
2988 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2990 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2991 comparison_operator with ordered_comparison_operator.
2993 2014-01-30 Nick Clifton <nickc@redhat.com>
2995 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2996 Rename to mn10300_store_multiple_regs.
2997 * config/mn10300/mn10300.c: Likewise.
2998 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2999 store_multiple_regs.
3000 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3001 Call mn10300_store_multiple_regs.
3003 2014-01-30 Nick Clifton <nickc@redhat.com>
3004 DJ Delorie <dj@redhat.com>
3006 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3007 %fp 2 to keep registers after it properly word-aligned.
3008 (rl78_alloc_physical_registers_umul): Handle the case where both
3009 input operands are the same.
3011 2014-01-30 Richard Biener <rguenther@suse.de>
3013 PR tree-optimization/59903
3014 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3017 2014-01-30 Jason Merrill <jason@redhat.com>
3020 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3023 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3025 2014-01-30 Richard Biener <rguenther@suse.de>
3027 PR tree-optimization/59951
3028 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3030 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3033 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3034 SFmode to DFmode case.
3036 2014-01-29 DJ Delorie <dj@redhat.com>
3038 * config/msp430/msp430.opt (-minrt): New.
3039 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3041 (ENDFILE_SPEC): Likewise.
3043 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3045 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3046 (estimate_function_body_sizes): Use it.
3048 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3051 * dwarf2out.c (is_cxx_auto): New.
3052 (is_base_type): Use it.
3053 (gen_type_die_with_usage): Likewise.
3055 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3057 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3058 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3059 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3060 -maltivec=be with LE targets.
3061 (vsx_mergeh_<mode>): Likewise.
3062 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3063 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3064 (altivec_vmrghb): Replace with define_expand and new
3065 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3066 (altivec_vmrghb_direct): New define_insn.
3067 (altivec_vmrghh): Replace with define_expand and new
3068 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3069 (altivec_vmrghh_direct): New define_insn.
3070 (altivec_vmrghw): Replace with define_expand and new
3071 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3072 (altivec_vmrghw_direct): New define_insn.
3073 (*altivec_vmrghsf): Adjust for endianness.
3074 (altivec_vmrglb): Replace with define_expand and new
3075 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3076 (altivec_vmrglb_direct): New define_insn.
3077 (altivec_vmrglh): Replace with define_expand and new
3078 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3079 (altivec_vmrglh_direct): New define_insn.
3080 (altivec_vmrglw): Replace with define_expand and new
3081 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3082 (altivec_vmrglw_direct): New define_insn.
3083 (*altivec_vmrglsf): Adjust for endianness.
3084 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3085 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3086 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3087 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3088 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3089 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3090 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3091 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3093 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3095 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3096 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3099 2014-01-29 Richard Biener <rguenther@suse.de>
3101 PR tree-optimization/58742
3102 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3103 associate_pointerplus_align.
3104 (associate_pointerplus_diff): New function.
3105 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3106 and associate_pointerplus_diff.
3108 2014-01-29 Richard Biener <rguenther@suse.de>
3110 * lto-streamer.h (LTO_major_version): Bump to 3.
3111 (LTO_minor_version): Reset to 0.
3113 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3115 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3116 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3117 (arm_file_start): Generate correct asm header for armv7ve.
3118 * config/arm/bpabi.h: Add multilib support for armv7ve.
3119 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3120 and cortex-a15 to armv7ve.
3121 * config/arm/t-aprofile: Add multilib support for armv7ve.
3122 * doc/invoke.texi: Document -march=armv7ve.
3124 2014-01-29 Richard Biener <rguenther@suse.de>
3126 PR tree-optimization/58742
3127 * tree-ssa-forwprop.c (associate_plusminus): Return true
3128 if we changed sth, defer EH cleanup to ...
3129 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3130 (simplify_mult): New function.
3132 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3135 PR tree-optimization/59920
3136 * tree.c (build_common_builtin_nodes): Remove
3137 __builtin_setjmp_dispatcher initialization.
3138 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3139 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3140 instead of gsi_after_labels + manually skipping debug stmts.
3141 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3142 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3143 * tree-inline.c (copy_edges_for_bb): Remove
3144 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3145 argument. Ignore computed_goto_p stmts. Don't call
3146 make_abnormal_goto_edges. If a call might need abnormal edges
3147 for non-local gotos, see if it already has an edge to
3148 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3149 with true argument, don't do anything then, otherwise add
3150 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3151 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3153 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3154 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3155 (lower_stmt): Don't set data->calls_builtin_setjmp.
3156 (lower_builtin_setjmp): Adjust comment.
3157 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3158 * tree-cfg.c (found_computed_goto): Remove.
3159 (factor_computed_gotos): Remove.
3160 (make_goto_expr_edges): Return bool, true for computed gotos.
3161 Don't call make_abnormal_goto_edges.
3162 (build_gimple_cfg): Don't set found_computed_goto, don't call
3163 factor_computed_gotos.
3164 (computed_goto_p): No longer static.
3165 (make_blocks): Don't set found_computed_goto.
3166 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3167 (make_edges): If make_goto_expr_edges returns true, push bb
3168 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3169 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3170 vector. Record mapping between bbs and OpenMP regions if there
3171 are any, adjust make_gimple_omp_edges caller. Call
3172 handle_abnormal_edges.
3173 (make_abnormal_goto_edges): Remove.
3174 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3175 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3176 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3177 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3178 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3179 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3180 filling *region also set *region_idx to (*region)->entry->index.
3183 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3184 For REGs set ORIGINAL_REGNO.
3186 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3188 * doc/md.texi: Mention that a target shouldn't implement
3189 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3192 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3194 PR tree-optimization/59594
3195 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3196 a copy of the datarefs vector rather than the vector itself.
3198 2014-01-28 Jason Merrill <jason@redhat.com>
3201 * dwarf2out.c (auto_die): New static.
3202 (gen_type_die_with_usage): Handle C++1y 'auto'.
3203 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3206 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3209 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3210 (SPEC_X32): Likewise.
3211 (SPEC_64): Likewise.
3212 * config/i386/i386.c (ix86_option_override_internal): Turn off
3213 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3215 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3216 * config/i386/i386.h (TARGET_16BIT): New macro.
3217 (TARGET_16BIT_P): Likewise.
3218 * config/i386/i386.opt: Add m16.
3219 * doc/invoke.texi: Document -m16.
3221 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3223 PR preprocessor/59935
3224 * input.c (location_get_source_line): Bail out on when line number
3225 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3227 2014-01-28 Richard Biener <rguenther@suse.de>
3229 PR tree-optimization/58742
3230 * tree-ssa-forwprop.c (associate_plusminus): Handle
3231 pointer subtraction of the form (T)(P + A) - (T)P.
3233 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3235 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3238 2014-01-28 Richard Biener <rguenther@suse.de>
3241 2014-01-28 Richard Biener <rguenther@suse.de>
3243 PR rtl-optimization/45364
3244 PR rtl-optimization/59890
3245 * var-tracking.c (local_get_addr_clear_given_value): Handle
3246 already cleared slot.
3247 (val_reset): Handle not allocated local_get_addr_cache.
3248 (vt_find_locations): Use post-order on the inverted CFG.
3250 2014-01-28 Richard Biener <rguenther@suse.de>
3252 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3254 2014-01-28 Richard Biener <rguenther@suse.de>
3256 PR rtl-optimization/45364
3257 PR rtl-optimization/59890
3258 * var-tracking.c (local_get_addr_clear_given_value): Handle
3259 already cleared slot.
3260 (val_reset): Handle not allocated local_get_addr_cache.
3261 (vt_find_locations): Use post-order on the inverted CFG.
3263 2014-01-28 Alan Modra <amodra@gmail.com>
3265 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3266 * configure.ac <recursive call for build != host>: Define
3267 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3268 and LD_FOR_BUILD too.
3269 * configure: Regenerate.
3271 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3273 * config/i386/i386.c (get_builtin_code_for_version): Separate
3274 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3275 Broadwell from Haswell.
3277 2014-01-27 Steve Ellcey <sellcey@mips.com>
3279 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3280 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3281 * config/mips/mips.c (mips_option_override): Change setting
3283 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3284 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3285 Change from Mask to Var.
3287 2014-01-27 Jeff Law <law@redhat.com>
3289 * ipa-inline.c (inline_small_functions): Fix typo.
3291 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3293 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3294 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3295 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3296 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3297 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3298 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3299 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3300 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3301 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3302 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3303 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3304 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3305 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3306 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3307 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3308 (_mm512_storeu_epi64): Ditto.
3309 (_mm512_cmpge_epi32_mask): Ditto.
3310 (_mm512_cmpge_epu32_mask): Ditto.
3311 (_mm512_cmpge_epi64_mask): Ditto.
3312 (_mm512_cmpge_epu64_mask): Ditto.
3313 (_mm512_cmple_epi32_mask): Ditto.
3314 (_mm512_cmple_epu32_mask): Ditto.
3315 (_mm512_cmple_epi64_mask): Ditto.
3316 (_mm512_cmple_epu64_mask): Ditto.
3317 (_mm512_cmplt_epi32_mask): Ditto.
3318 (_mm512_cmplt_epu32_mask): Ditto.
3319 (_mm512_cmplt_epi64_mask): Ditto.
3320 (_mm512_cmplt_epu64_mask): Ditto.
3321 (_mm512_cmpneq_epi32_mask): Ditto.
3322 (_mm512_cmpneq_epu32_mask): Ditto.
3323 (_mm512_cmpneq_epi64_mask): Ditto.
3324 (_mm512_cmpneq_epu64_mask): Ditto.
3325 (_mm512_expand_pd): Ditto.
3326 (_mm512_expand_ps): Ditto.
3327 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3328 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3329 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3330 * config/i386/i386.c (ix86_builtins): Add
3331 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3332 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3333 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3334 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3335 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3336 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3337 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3338 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3339 IX86_BUILTIN_PMOVUSQW512_MEM.
3340 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3341 __builtin_ia32_pmovsqd512mem_mask,
3342 __builtin_ia32_pmovqd512mem_mask,
3343 __builtin_ia32_pmovusqw512mem_mask,
3344 __builtin_ia32_pmovsqw512mem_mask,
3345 __builtin_ia32_pmovqw512mem_mask,
3346 __builtin_ia32_pmovusdw512mem_mask,
3347 __builtin_ia32_pmovsdw512mem_mask,
3348 __builtin_ia32_pmovdw512mem_mask,
3349 __builtin_ia32_pmovqb512mem_mask,
3350 __builtin_ia32_pmovusqb512mem_mask,
3351 __builtin_ia32_pmovsqb512mem_mask,
3352 __builtin_ia32_pmovusdb512mem_mask,
3353 __builtin_ia32_pmovsdb512mem_mask,
3354 __builtin_ia32_pmovdb512mem_mask.
3355 (bdesc_args): Add __builtin_ia32_expanddf512,
3356 __builtin_ia32_expandsf512.
3357 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3358 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3359 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3360 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3361 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3362 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3363 (avx512f_<code>v8div16qi2_mask_store): This.
3364 (avx512f_expand<mode>): New.
3366 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3368 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3370 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3371 (_mm512_prefetch_i32scatter_pd): Ditto.
3372 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3373 (_mm512_prefetch_i64scatter_pd): Ditto.
3374 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3375 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3376 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3377 (_mm512_prefetch_i32scatter_ps): Ditto.
3378 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3379 (_mm512_prefetch_i64scatter_ps): Ditto.
3380 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3381 * config/i386/i386-builtin-types.def: Define
3382 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3383 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3384 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3385 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3386 IX86_BUILTIN_SCATTERPFQPD.
3387 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3388 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3389 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3390 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3391 __builtin_ia32_scatterpfqps.
3392 (ix86_expand_builtin): Expand new built-ins.
3393 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3394 fix memory access data type.
3395 (*avx512pf_gatherpf<mode>_mask): Ditto.
3396 (*avx512pf_gatherpf<mode>): Ditto.
3397 (avx512pf_scatterpf<mode>): Ditto.
3398 (*avx512pf_scatterpf<mode>_mask): Ditto.
3399 (*avx512pf_scatterpf<mode>): Ditto.
3400 (GATHER_SCATTER_SF_MEM_MODE): New.
3401 (avx512pf_gatherpf<mode>df): Ditto.
3402 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3403 (*avx512pf_scatterpf<mode>df): Ditto.
3405 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3408 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3409 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3412 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3414 * common/config/arm/arm-common.c
3415 (arm_rewrite_mcpu): Handle multiple names.
3417 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3419 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3421 * gimple-builder.h (create_gimple_tmp): Delete.
3423 2014-01-27 Christian Bruel <christian.bruel@st.com>
3425 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3428 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3430 * config/pa/pa.md (call): Generate indirect long calls to non-local
3431 functions when outputing 32-bit code.
3432 (call_value): Likewise except for special call to buggy powf function.
3434 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3435 portable runtime and PIC indirect calls.
3436 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3437 and PIC call sequences. Use ldo instead of blr to set return register
3438 in PIC call sequence.
3440 2014-01-25 Walter Lee <walt@tilera.com>
3442 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3443 avoid clobbering a live register.
3445 2014-01-25 Walter Lee <walt@tilera.com>
3447 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3448 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3449 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3450 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3452 2014-01-25 Walter Lee <walt@tilera.com>
3454 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3455 arguments on even registers.
3456 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3458 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3459 (BIGGEST_ALIGNMENT): Ditto.
3460 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3462 2014-01-25 Walter Lee <walt@tilera.com>
3464 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3465 insns before bundling.
3466 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3468 2014-01-25 Walter Lee <walt@tilera.com>
3470 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3471 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3472 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3474 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3476 * config/mips/constraints.md (kl): Delete.
3477 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3478 define expands, using...
3479 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3480 instructions for MIPS16.
3481 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3482 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3484 2014-01-25 Walter Lee <walt@tilera.com>
3486 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3490 2014-01-25 Walter Lee <walt@tilera.com>
3492 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3493 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3495 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3497 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3500 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3502 * print-rtl.c (in_call_function_usage): New var.
3503 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3504 EXPR_LIST mode as mode and not as reg note name.
3507 * cfgloopmanip.c (copy_loop_info): If
3508 loop->warned_aggressive_loop_optimizations, make sure
3509 the flag is set in target loop too.
3511 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3513 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3515 * builtins.def: Likewise.
3516 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3517 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3518 * ira.c (ira_setup_eliminable_regset): Likewise.
3519 * omp-low.c (gate_expand_omp): Likewise.
3520 (execute_lower_omp): Likewise.
3521 (diagnose_sb_0): Likewise.
3522 (gate_diagnose_omp_blocks): Likewise.
3523 (simd_clone_clauses_extract): Likewise.
3526 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3528 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3529 correction for little endian...
3530 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3533 2014-01-24 Jeff Law <law@redhat.com>
3535 PR tree-optimization/59919
3536 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3537 for non-returning calls.
3539 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3541 * common/config/aarch64/aarch64-common.c
3542 (aarch64_rewrite_mcpu): Handle multiple names.
3543 * config/aarch64/aarch64.h
3544 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3546 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3548 * input.c (add_file_to_cache_tab): Handle the case where fopen
3551 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3554 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3555 from push operand if code of push isn't PRE_DEC.
3557 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3560 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3561 -mquad-memory-atomic. Update -mquad-memory documentation to say
3562 it is only used for non-atomic loads/stores.
3564 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3565 -mquad-memory or -mquad-memory-atomic switches.
3567 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3568 -mquad-memory-atomic to ISA 2.07 support.
3570 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3571 to separate support of normal quad word memory operations (ldq, stq)
3572 from the atomic quad word memory operations.
3574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3575 support to separate non-atomic quad word operations from atomic
3576 quad word operations. Disable non-atomic quad word operations in
3577 little endian mode so that we don't have to swap words after the
3578 load and before the store.
3579 (quad_load_store_p): Add comment about atomic quad word support.
3580 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3581 options printed with -mdebug=reg.
3583 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3584 -mquad-memory-atomic as the test for whether we have quad word
3585 atomic instructions.
3586 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3587 or -mp8-vector are used, allow byte/half-word atomic operations.
3589 * config/rs6000/sync.md (load_lockedti): Insure that the address
3590 is a proper indexed or indirect address for the lqarx instruction.
3591 On little endian systems, swap the hi/lo registers after the lqarx
3593 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3594 insure the address is valid for the lqarx instruction.
3595 (store_conditionalti): Insure that the address is a proper indexed
3596 or indirect address for the stqcrx. instruction. On little endian
3597 systems, swap the hi/lo registers before doing the stqcrx.
3599 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3600 insure the address is valid for the stqcrx. instruction.
3602 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3603 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3604 type of quad memory support is available.
3606 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3609 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3610 there is a danger of looping.
3612 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3615 force flag_ira_loop_pressure if set via command line.
3617 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3619 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3620 (ashr_simd): New builtin handling DI mode.
3621 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3622 (aarch64_sshr_simddi): New match pattern.
3623 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3624 (vshrd_n_s64): Likewise.
3625 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3627 2014-01-23 Nick Clifton <nickc@redhat.com>
3629 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3630 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3631 favour of mcu specific scripts.
3632 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3635 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3636 Alex Velenko <Alex.Velenko@arm.com>
3638 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3639 (vaddv_s16): Likewise.
3640 (vaddv_s32): Likewise.
3641 (vaddv_u8): Likewise.
3642 (vaddv_u16): Likewise.
3643 (vaddv_u32): Likewise.
3644 (vaddvq_s8): Likewise.
3645 (vaddvq_s16): Likewise.
3646 (vaddvq_s32): Likewise.
3647 (vaddvq_s64): Likewise.
3648 (vaddvq_u8): Likewise.
3649 (vaddvq_u16): Likewise.
3650 (vaddvq_u32): Likewise.
3651 (vaddvq_u64): Likewise.
3652 (vaddv_f32): Likewise.
3653 (vaddvq_f32): Likewise.
3654 (vaddvq_f64): Likewise.
3655 (vmaxv_f32): Likewise.
3656 (vmaxv_s8): Likewise.
3657 (vmaxv_s16): Likewise.
3658 (vmaxv_s32): Likewise.
3659 (vmaxv_u8): Likewise.
3660 (vmaxv_u16): Likewise.
3661 (vmaxv_u32): Likewise.
3662 (vmaxvq_f32): Likewise.
3663 (vmaxvq_f64): Likewise.
3664 (vmaxvq_s8): Likewise.
3665 (vmaxvq_s16): Likewise.
3666 (vmaxvq_s32): Likewise.
3667 (vmaxvq_u8): Likewise.
3668 (vmaxvq_u16): Likewise.
3669 (vmaxvq_u32): Likewise.
3670 (vmaxnmv_f32): Likewise.
3671 (vmaxnmvq_f32): Likewise.
3672 (vmaxnmvq_f64): Likewise.
3673 (vminv_f32): Likewise.
3674 (vminv_s8): Likewise.
3675 (vminv_s16): Likewise.
3676 (vminv_s32): Likewise.
3677 (vminv_u8): Likewise.
3678 (vminv_u16): Likewise.
3679 (vminv_u32): Likewise.
3680 (vminvq_f32): Likewise.
3681 (vminvq_f64): Likewise.
3682 (vminvq_s8): Likewise.
3683 (vminvq_s16): Likewise.
3684 (vminvq_s32): Likewise.
3685 (vminvq_u8): Likewise.
3686 (vminvq_u16): Likewise.
3687 (vminvq_u32): Likewise.
3688 (vminnmv_f32): Likewise.
3689 (vminnmvq_f32): Likewise.
3690 (vminnmvq_f64): Likewise.
3692 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3694 * config/aarch64/aarch64-simd.md
3695 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3696 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3697 (*aarch64_mul3_elt<mode>): Likewise.
3698 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3699 (*aarch64_mul3_elt_to_64v2df): Likewise.
3700 (*aarch64_mla_elt<mode>): Likewise.
3701 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3702 (*aarch64_mls_elt<mode>): Likewise.
3703 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3704 (*aarch64_fma4_elt<mode>): Likewise.
3705 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3706 (*aarch64_fma4_elt_to_64v2df): Likewise.
3707 (*aarch64_fnma4_elt<mode>): Likewise.
3708 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3709 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3710 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3711 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3712 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3713 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3714 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3715 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3716 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3718 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3720 * config/aarch64/aarch64-simd.md
3721 (aarch64_be_checked_get_lane<mode>): New define_expand.
3722 * config/aarch64/aarch64-simd-builtins.def
3723 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3724 New builtin definition.
3725 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3726 Use new safe be builtin.
3728 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3730 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3732 (aarch64_be_st1<mode>): Likewise.
3733 (aarch_ld1<VALL:mode>): Define_expand modified.
3734 (aarch_st1<VALL:mode>): Likewise.
3735 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3736 (UNSPEC_ST1): Likewise.
3738 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3740 * config/microblaze/microblaze.md: Add trap insn and attribute
3742 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3744 PR preprocessor/58580
3745 * input.h (location_get_source_line): Take an additional line_size
3747 (void diagnostics_file_cache_fini): Declare new function.
3748 * input.c (struct fcache): New type.
3749 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3750 New static constants.
3751 (diagnostic_file_cache_init, total_lines_num)
3752 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3753 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3754 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3755 (get_next_line, read_next_line, goto_next_line, read_line_num):
3756 New static function definitions.
3757 (diagnostic_file_cache_fini): New function.
3758 (location_get_source_line): Take an additional output line_len
3759 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3761 * diagnostic.c (diagnostic_finish): Call
3762 diagnostic_file_cache_fini.
3763 (adjust_line): Take an additional input parameter for the length
3764 of the line, rather than calculating it with strlen.
3765 (diagnostic_show_locus): Adjust the use of
3766 location_get_source_line and adjust_line with respect to their new
3767 signature. While displaying a line now, do not stop at the first
3768 null byte. Rather, display the zero byte as a space and keep
3769 going until we reach the size of the line.
3770 * Makefile.in: Add vec.o to OBJS-libcommon
3772 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3773 Ilya Tocar <ilya.tocar@intel.com>
3775 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3776 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3777 (__builtin_ia32_kmov16): Ditto.
3778 * config/i386/i386.md (UNSPEC_KMOV): New.
3781 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3783 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3784 (_mm512_storeu_si512): Ditto.
3786 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3789 * rtl.h (get_referenced_operands): Declare.
3790 * recog.c (get_referenced_operands): New function.
3791 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3792 operands have been referenced when recording LO_SUM references.
3794 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3796 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3798 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3800 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3801 Enable for generic and recent AMD targets.
3803 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3805 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3806 ARG_SIZE note when adjustment was eliminated.
3808 2014-01-22 Jeff Law <law@redhat.com>
3810 PR tree-optimization/59597
3811 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3812 in file. Accept new argument REGISTERING and use it to modify
3813 dump output appropriately.
3814 (register_jump_thread): Corresponding changes.
3815 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3816 thread paths involving joiner blocks. Add code to dump cancelled
3817 jump threading paths.
3819 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3821 PR rtl-optimization/59477
3822 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3823 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3825 2014-01-22 Tom Tromey <tromey@redhat.com>
3827 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3829 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3831 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3833 PR rtl-optimization/59896
3834 * lra-constraints.c (process_alt_operands): Check unused note for
3835 matched operands of insn with no output reloads.
3837 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3839 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3840 (mips_move_from_gpr_cost): Likewise.
3842 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3844 PR rtl-optimization/59858
3845 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3846 ira_class_hard_regs_num.
3847 (process_alt_operands): Increase reject for dying matched operand.
3849 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3852 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3853 smaller than size, perform several stores or loads and stores
3854 at dst + count - size to store or copy all of size bytes, rather
3855 than just last modesize bytes.
3857 2014-01-20 DJ Delorie <dj@redhat.com>
3859 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3860 that CLOBBERs are REGs before propogating their values.
3862 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3865 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3866 (cgraph_inline_failed_type): New function.
3867 * cgraph.h (DEFCIFCODE): Add type.
3868 (cgraph_inline_failed_type_t): New enum.
3869 (cgraph_inline_failed_type): New prototype.
3870 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3871 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3872 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3873 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3874 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3875 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3876 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3877 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3878 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3879 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3880 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3881 OPTIMIZATION_MISMATCH.
3882 * tree-inline.c (expand_call_inline): Emit errors during
3883 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3885 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3888 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3889 mode attribute in insn output.
3891 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3893 * output.h (output_constant): Delete.
3894 * varasm.c (output_constant): Make private.
3896 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3898 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3900 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3903 * tree.h (fold_builtin_strcat): New prototype.
3904 * builtins.c (fold_builtin_strcat): No longer static. Add len
3905 argument, if non-NULL, don't call c_strlen. Optimize
3906 directly into __builtin_memcpy instead of __builtin_strcpy.
3907 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3908 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3910 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3912 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3913 for SImode_address_operand operands, having only a REG argument.
3915 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3917 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3918 loader name using mbig-endian.
3919 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3921 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3923 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3927 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3929 * config/aarch64/aarch64-protos.h
3930 (aarch64_cannot_change_mode_class_ptr): Declare.
3931 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3932 aarch64_cannot_change_mode_class_ptr): New.
3933 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3934 backend hook aarch64_cannot_change_mode_class.
3936 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3938 * common/config/aarch64/aarch64-common.c
3939 (aarch64_handle_option): Don't handle any option order logic here.
3940 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3941 selected_cpu, warn on architecture version mismatch.
3942 (aarch64_override_options): Fix parsing order for option strings.
3944 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3945 Iain Sandoe <iain@codesourcery.com>
3948 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3949 warning. Amend comment to reflect current functionality.
3951 2014-01-20 Richard Biener <rguenther@suse.de>
3954 * builtins.c (fold_builtin_strcat): Remove case better handled
3955 by tree-ssa-strlen.c.
3957 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3959 * config/aarch64/aarch64.opt
3960 (mcpu, march, mtune): Make case-insensitive.
3962 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3965 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3966 if operands[1] is a REG or ZERO_EXTEND of a REG.
3968 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3970 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3972 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3974 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3975 long non-pic millicode calls.
3977 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3979 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3981 2014-01-19 Kito Cheng <kito@0xlab.org>
3983 * builtins.c (expand_movstr): Check movstr expand done or fail.
3985 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3986 H.J. Lu <hongjiu.lu@intel.com>
3989 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3990 to DImode for zero-extended addresses.
3992 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3994 PR rtl-optimization/57763
3995 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3996 on the new indirect jump_insn and increment LABEL_NUSES (label).
3998 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4002 * config.gcc (x86_archs): New variable.
4003 (x86_64_archs): Likewise.
4004 (x86_cpus): Likewise.
4005 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4006 --with-arch/--with-cpu= options.
4007 Support --with-arch=/--with-cpu={nehalem,westmere,
4008 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4010 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4012 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4013 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4015 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4017 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4019 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4022 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4023 clear cpp_get_options (parse_in)->warn_unused_macros for
4024 ix86_target_macros_internal with cpp_define.
4026 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4028 * jump.c (delete_related_insns): Keep (use (insn))s.
4029 * reorg.c (redundant_insn): Check for barriers too.
4031 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4033 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4035 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4037 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4038 call to $$dyncall when TARGET_LONG_CALLS is true.
4040 2014-01-17 Jeff Law <law@redhat.com>
4042 * ree.c (combine_set_extension): Temporarily disable test for
4043 changing number of hard registers.
4045 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4048 * ipa-inline-analysis.c (inline_free_summary):
4049 Do not free summary of aliases.
4051 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4054 * gimplify.c (gimplify_expr): Use create_tmp_var
4055 instead of create_tmp_var_raw. If cond doesn't have
4056 integral type, don't add the IFN_ANNOTATE builtin at all.
4058 2014-01-17 Martin Jambor <mjambor@suse.cz>
4061 * ipa-cp.c (prev_edge_clone): New variable.
4062 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4063 Also resize prev_edge_clone vector.
4064 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4065 (ipcp_edge_removal_hook): New function.
4066 (ipcp_driver): Register ipcp_edge_removal_hook.
4068 2014-01-17 Andrew Pinski <apinski@cavium.com>
4069 Steve Ellcey <sellcey@mips.com>
4072 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4075 2014-01-17 Jeff Law <law@redhat.com>
4078 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4079 so that pass_ccp runs first.
4081 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4083 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4084 (ix86_adjust_cost): Use !TARGET_XXX.
4085 (do_reorder_for_imul): Likewise.
4086 (swap_top_of_ready_list): Likewise.
4087 (ix86_sched_reorder): Likewise.
4089 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4091 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4092 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4093 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4094 (intel_memset): New. Duplicate slm_memset.
4095 (intel_cost): New. Duplicate slm_cost.
4096 (m_INTEL): New macro.
4097 (processor_target_table): Add "intel".
4098 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4099 with PROCESSOR_INTEL for "intel".
4100 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4101 PROCESSOR_SILVERMONT.
4102 (ix86_issue_rate): Likewise.
4103 (ix86_adjust_cost): Likewise.
4104 (ia32_multipass_dfa_lookahead): Likewise.
4105 (swap_top_of_ready_list): Likewise.
4106 (ix86_sched_reorder): Likewise.
4107 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4108 instead of TARGET_OPT_AGU.
4109 * config/i386/i386.h (TARGET_INTEL): New.
4110 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4111 (processor_type): Add PROCESSOR_INTEL.
4112 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4113 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4115 2014-01-17 Marek Polacek <polacek@redhat.com>
4118 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4121 2014-01-17 Richard Biener <rguenther@suse.de>
4123 PR tree-optimization/46590
4124 * opts.c (default_options_table): Add entries for
4125 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4126 all enabled at -O1 but not for -Og.
4127 * common.opt (fbranch-count-reg): Remove Init(1).
4128 (fmove-loop-invariants): Likewise.
4129 (ftree-pta): Likewise.
4131 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4133 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4134 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4135 decls to at least the GCC 4.8 used alignments.
4138 * tree-nested.c (convert_nonlocal_reference_stmt,
4139 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4140 of GIMPLE_BIND stmts, adjust associated decls.
4142 2014-01-17 Richard Biener <rguenther@suse.de>
4144 PR tree-optimization/46590
4145 * vec.h (vec<>::bseach): New member function implementing
4146 binary search according to C89 bsearch.
4147 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4148 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4149 bitmap pointer again. Make accesses_in_loop a flat array.
4150 (mem_ref_obstack): New global.
4151 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4152 (mark_ref_stored): Likewise.
4153 (ref_indep_loop_p_2): Likewise.
4154 (set_ref_stored_in_loop): New helper function.
4155 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4156 (memref_free): Adjust.
4157 (record_mem_ref_loc): Simplify.
4158 (gather_mem_refs_stmt): Adjust.
4159 (sort_locs_in_loop_postorder_cmp): New function.
4160 (analyze_memory_references): Sort accesses_in_loop after
4161 loop postorder number.
4162 (find_ref_loc_in_loop_cmp): New function.
4163 (for_all_locs_in_loop): Find relevant cluster of locs in
4164 accesses_in_loop and iterate without recursion.
4165 (execute_sm): Avoid uninit warning.
4166 (struct ref_always_accessed): Simplify.
4167 (ref_always_accessed::operator ()): Likewise.
4168 (ref_always_accessed_p): Likewise.
4169 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4170 loop postorder numbers here.
4171 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4174 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4177 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4178 on decls for which assemble_alias has been called.
4180 2014-01-17 Nick Clifton <nickc@redhat.com>
4182 * config/msp430/msp430.opt: (mcpu): New option.
4183 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4184 (msp430_option_override): Parse target_cpu. If the MCU name
4185 matches a generic string, clear target_mcu.
4186 (msp430_attr): Allow numeric interrupt values up to 63.
4187 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4188 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4190 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4192 * config/msp430/msp430.md (popm): Use %J rather than %I.
4193 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4194 (addhi_cy_i): Use immediate_operand for operand 2.
4195 * doc/invoke.texi: Document -mcpu option.
4197 2014-01-17 Richard Biener <rguenther@suse.de>
4199 PR rtl-optimization/38518
4200 * df.h (df_analyze_loop): Declare.
4201 * df-core.c: Include cfgloop.h.
4202 (df_analyze_1): Split out main part of df_analyze.
4203 (df_analyze): Adjust.
4204 (loop_inverted_post_order_compute): New function.
4205 (loop_post_order_compute): Likewise.
4206 (df_analyze_loop): New function avoiding whole-function
4208 * loop-invariant.c (find_defs): Use df_analyze_loop.
4209 (find_invariants): Adjust.
4210 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4212 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4214 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4215 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4217 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4219 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4220 traversal when removing references.
4222 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4225 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4227 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4230 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4231 pushing a reload for an autoinc when we had previously reloaded an
4232 inner part of the address.
4234 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4236 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4238 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4239 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4240 when not giving up or versioning for alias only because of
4242 (vect_analyze_data_ref_dependences): Set to true.
4243 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4245 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4246 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4250 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4253 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4254 operand 0 predicate for gathers, use a new pseudo as subtarget.
4256 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4259 * lra-constraints.c (process_alt_operands): Add printing debug info.
4260 Check absence of input/output reloads for matched operands too.
4262 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4264 PR rtl-optimization/59835
4265 * ira.c (ira_init_register_move_cost): Increase cost for
4268 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4270 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4272 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4275 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4276 non-register objects. Use gen_(high/low)part more consistently.
4279 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4282 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4283 endian support, remove tests for WORDS_BIG_ENDIAN.
4284 (p8_mfvsrd_3_<mode>): Likewise.
4285 (reload_gpr_from_vsx<mode>): Likewise.
4286 (reload_gpr_from_vsxsf): Likewise.
4287 (p8_mfvsrd_4_disf): Likewise.
4289 2014-01-16 Richard Biener <rguenther@suse.de>
4291 PR rtl-optimization/46590
4292 * lcm.c (compute_antinout_edge): Use postorder iteration.
4293 (compute_laterin): Use inverted postorder iteration.
4295 2014-01-16 Nick Clifton <nickc@redhat.com>
4298 * varasm.c (output_constant): Return the number of bytes actually
4300 (output_constructor_array_range): Update the field size with the
4301 number of bytes emitted by output_constant.
4302 (output_constructor_regular_field): Likewise. Also do not
4303 complain if the total number of bytes emitted is now greater
4304 than the expected fieldpos.
4305 * output.h (output_constant): Update prototype and descriptive comment.
4307 2014-01-16 Marek Polacek <polacek@redhat.com>
4310 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4311 it is error_mark_node.
4313 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4315 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4316 VALID_AVX256_REG_OR_OI_MODE.
4318 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4320 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4321 current procedure should be profiled.
4323 2014-01-15 Andrew Pinski <apinski@cavium.com>
4325 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4326 of moving from/to the STACK_REG register class.
4328 2014-01-15 Richard Henderson <rth@redhat.com>
4331 * reginfo.c (global_regs_decl): Globalize.
4332 * rtl.h (global_regs_decl): Declare.
4333 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4334 reserved via global_regs.
4336 2014-01-15 Teresa Johnson <tejohnson@google.com>
4338 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4340 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4342 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4343 and vmulosh rather than call gen_vec_widen_smult_*.
4344 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4345 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4346 (vec_widen_smult_even_v16qi): Likewise.
4347 (vec_widen_umult_even_v8hi): Likewise.
4348 (vec_widen_smult_even_v8hi): Likewise.
4349 (vec_widen_umult_odd_v16qi): Likewise.
4350 (vec_widen_smult_odd_v16qi): Likewise.
4351 (vec_widen_umult_odd_v8hi): Likewise.
4352 (vec_widen_smult_odd_v8hi): Likewise.
4353 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4354 vmuloub rather than call gen_vec_widen_umult_*.
4355 (vec_widen_umult_lo_v16qi): Likewise.
4356 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4357 vmulosb rather than call gen_vec_widen_smult_*.
4358 (vec_widen_smult_lo_v16qi): Likewise.
4359 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4360 rather than call gen_vec_widen_umult_*.
4361 (vec_widen_umult_lo_v8hi): Likewise.
4362 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4363 rather than call gen_vec_widen_smult_*.
4364 (vec_widen_smult_lo_v8hi): Likewise.
4366 2014-01-15 Jeff Law <law@redhat.com>
4368 PR tree-optimization/59747
4369 * ree.c (find_and_remove_re): Properly handle case where a second
4370 eliminated extension requires widening a copy created for elimination
4371 of a prior extension.
4372 (combine_set_extension): Ensure that the number of hard regs needed
4373 for a destination register does not change when we widen it.
4375 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4377 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4378 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4379 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4380 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4381 (avr-*-rtems*): Likewise.
4382 (bfin*-rtems*): Likewise.
4383 (moxie-*-rtems*): Likewise.
4384 (h8300-*-rtems*): Likewise.
4385 (i[34567]86-*-rtems*): Likewise.
4386 (lm32-*-rtems*): Likewise.
4387 (m32r-*-rtems*): Likewise.
4388 (m68k-*-rtems*): Likewise.
4389 (microblaze*-*-rtems*): Likewise.
4390 (mips*-*-rtems*): Likewise.
4391 (powerpc-*-rtems*): Likewise.
4392 (sh-*-rtems*): Likewise.
4393 (sparc-*-rtems*): Likewise.
4394 (sparc64-*-rtems*): Likewise.
4395 (v850-*-rtems*): Likewise.
4396 (m32c-*-rtems*): Likewise.
4398 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4400 PR rtl-optimization/59511
4401 * ira.c (ira_init_register_move_cost): Use memory costs for some
4402 cases of register move cost calculations.
4403 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4404 instead of BB frequency.
4405 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4406 * lra-assigns.c (find_hard_regno_for): Ditto.
4408 2014-01-15 Richard Biener <rguenther@suse.de>
4410 PR tree-optimization/59822
4411 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4412 (vectorizable_load): Use it to hoist defs of uses of invariant
4413 loads out of the loop.
4415 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4416 Kugan Vivekanandarajah <kuganv@linaro.org>
4419 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4422 2014-01-15 Richard Biener <rguenther@suse.de>
4424 PR rtl-optimization/59802
4425 * lcm.c (compute_available): Use inverted postorder to seed
4426 the initial worklist.
4428 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4431 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4432 ADDR_REGS for invalid symrefs in non-PIC code.
4434 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4437 * builtins.c (determine_block_size): Initialize *probable_max_size
4438 even if len_rtx is CONST_INT.
4440 2014-01-14 Andrew Pinski <apinski@cavium.com>
4442 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4443 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4444 (cortexa53_tunings): Likewise.
4445 (aarch64_sched_issue_rate): New function.
4446 (TARGET_SCHED_ISSUE_RATE): Define.
4448 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4450 * ira-costs.c (find_costs_and_classes): Add missed
4451 ira_init_register_move_cost_if_necessary.
4453 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4456 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4458 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4461 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4462 to indicate if type is used for function return value. Warn ABI
4463 change if the vector mode isn't available for function return value.
4464 (ix86_function_arg_advance): Pass false to type_natural_mode.
4465 (ix86_function_arg): Likewise.
4466 (ix86_gimplify_va_arg): Likewise.
4467 (function_arg_32): Don't warn ABI change.
4468 (ix86_function_value): Pass true to type_natural_mode.
4469 (ix86_return_in_memory): Likewise.
4470 (ix86_struct_value_rtx): Removed.
4471 (TARGET_STRUCT_VALUE_RTX): Likewise.
4473 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4475 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4476 converting a conditional jump into a conditional return.
4478 2014-01-14 Richard Biener <rguenther@suse.de>
4480 PR tree-optimization/58921
4481 PR tree-optimization/59006
4482 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4483 hoisting invariant stmts.
4484 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4485 invariant loads on the preheader edge if possible.
4487 2014-01-14 Joey Ye <joey.ye@arm.com>
4489 * doc/plugin.texi (Building GCC plugins): Update to C++.
4491 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4493 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4494 (_mm_rcp28_round_ss): Ditto.
4495 (_mm_rsqrt28_round_sd): Ditto.
4496 (_mm_rsqrt28_round_ss): Ditto.
4497 (_mm_rcp28_sd): Ditto.
4498 (_mm_rcp28_ss): Ditto.
4499 (_mm_rsqrt28_sd): Ditto.
4500 (_mm_rsqrt28_ss): Ditto.
4501 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4502 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4503 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4504 (IX86_BUILTIN_RCP28SD): Ditto.
4505 (IX86_BUILTIN_RCP28SS): Ditto.
4506 (IX86_BUILTIN_RSQRT28SD): Ditto.
4507 (IX86_BUILTIN_RSQRT28SS): Ditto.
4508 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4509 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4510 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4511 (ix86_expand_special_args_builtin): Expand new FTYPE.
4512 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4513 (srcp14<mode>): Make insn unary.
4514 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4515 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4516 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4517 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4518 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4519 Fix rounding: make it SAE only.
4520 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4522 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4524 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4525 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4526 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4527 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4528 (round_saeonly_mask_scalar_operand4): Ditto.
4529 (round_saeonly_mask_scalar_op3): Ditto.
4530 (round_saeonly_mask_scalar_op4): Ditto.
4532 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4534 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4535 Implement -maltivec=be for vec_insert and vec_extract.
4537 2014-01-10 DJ Delorie <dj@redhat.com>
4539 * config/msp430/msp430.md (call_internal): Don't allow memory
4540 references with SP as the base register.
4541 (call_value_internal): Likewise.
4542 * config/msp430/constraints.md (Yc): New. For memory references
4543 that don't use SP as a base register.
4545 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4546 "an integer without a # prefix"
4547 * config/msp430/msp430.md (epilogue_helper): Use it.
4549 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4552 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4553 AVX512F gather builtins.
4554 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4555 on gather decls with INTEGER_TYPE masktype.
4556 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4557 directly into the builtin rather than hoisting it before loop.
4559 PR tree-optimization/59387
4560 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4561 (scev_const_prop): If folded_casts and type has undefined overflow,
4562 use force_gimple_operand instead of force_gimple_operand_gsi and
4563 for each added stmt if it is assign with
4564 arith_code_with_undefined_signed_overflow, call
4565 rewrite_to_defined_overflow.
4566 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4567 gimple-fold.h instead.
4568 (arith_code_with_undefined_signed_overflow,
4569 rewrite_to_defined_overflow): Moved to ...
4570 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4571 rewrite_to_defined_overflow): ... here. No longer static.
4572 Include gimplify-me.h.
4573 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4574 rewrite_to_defined_overflow): New prototypes.
4576 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4578 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4580 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4582 * builtins.c (get_object_alignment_2): Minor tweak.
4583 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4585 2014-01-13 Christian Bruel <christian.bruel@st.com>
4587 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4588 optimized non constant lengths.
4590 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4593 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4594 load as __atomic_load_N if possible.
4596 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4598 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4600 (rs6000_expand_builtin): Adjust call.
4602 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4605 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4606 * config/rs6000/rs6000.c: Include target-globals.h.
4607 (rs6000_set_current_function): Instead of doing target_reinit
4608 unconditionally, use save_target_globals_default_opts and
4609 restore_target_globals.
4611 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4613 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4614 (rs6000_expand_builtin): Handle mffs and mtfsf.
4615 (rs6000_init_builtins): Define mffs and mtfsf.
4616 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4617 (rs6000_mffs): New pattern.
4618 (rs6000_mtfsf): New pattern.
4620 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4622 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4623 Start narrowing with START. Apply candidate-use pair
4624 and check overall cost in narrowing.
4625 (iv_ca_prune): Pass new argument.
4627 2014-01-10 Jeff Law <law@redhat.com>
4630 * ree.c (combine_reaching_defs): Ensure the defining statement
4631 occurs before the extension when optimizing extensions with
4632 different source and destination hard registers.
4634 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4637 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4638 vtables into the type inheritance graph.
4640 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4642 PR rtl-optimization/59754
4643 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4644 modes in the REGNO != REGNO case.
4646 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4648 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4650 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4652 PR tree-optimization/59745
4653 * tree-predcom.c (tree_predictive_commoning_loop): Call
4654 free_affine_expand_cache if giving up because components is NULL.
4656 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4657 GC in payload of target_globals struct instead of allocating them on
4658 the heap and the larger structs separately using GC.
4659 * target-globals.h (struct target_globals): Make regs, hard_regs,
4660 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4661 of GTY((skip)) and change type to void *.
4662 (reset_target_globals): Cast loads from those fields to corresponding
4665 2014-01-10 Steve Ellcey <sellcey@mips.com>
4668 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4669 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4670 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4672 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4675 * aarch64-modes.def (CC_Zmode): New flags mode.
4676 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4677 represents an equality.
4678 (aarch64_get_condition_code): Handle CC_Zmode.
4679 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4681 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4683 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4684 extraction in good case.
4686 2014-01-10 Richard Biener <rguenther@suse.de>
4688 PR tree-optimization/59374
4689 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4690 checking after SLP discovery. Mark stmts not participating
4691 in any SLP instance properly.
4693 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4695 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4696 when handling a SET rtx.
4698 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4700 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4701 (cortex-a57): Likewise.
4702 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4704 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4706 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4707 non-iwmmxt builtins.
4709 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4713 * ipa-devirt.c record_target_from_binfo): Take as argument
4714 stack of binfos and lookup matching one for virtual inheritance.
4715 (possible_polymorphic_call_targets_1): Update.
4717 2014-01-10 Huacai Chen <chenhc@lemote.com>
4719 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4720 kernel strings for Loongson-2E/2F/3A.
4722 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4725 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4726 is_gimple_call before calling gimple_call_internal_p.
4728 2014-01-09 Steve Ellcey <sellcey@mips.com>
4730 * Makefile.in (TREE_FLOW_H): Remove.
4731 (TREE_SSA_H): Add file names from tree-flow.h.
4732 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4733 * tree.h: Remove tree-flow.h reference.
4734 * hash-table.h: Remove tree-flow.h reference.
4735 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4736 reference with tree-ssa-loop.h.
4738 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4740 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4741 default element-order behavior for -maltivec.
4742 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4744 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4745 when targeting big endian, at least for now.
4746 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4748 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4751 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4752 var satisfies use_register_for_decl, just take into account type
4753 alignment, rather than decl alignment.
4755 PR tree-optimization/59622
4756 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4757 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4758 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4759 Don't devirtualize for inplace at all. For targets.length () == 1,
4760 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4762 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4764 * config/i386/i386.md (cpu): Remove the unused btver1.
4766 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4768 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4770 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4773 * tree-core.h (struct target_globals): New forward declaration.
4774 (struct tree_target_option): Add globals field.
4775 * tree.h (TREE_TARGET_GLOBALS): Define.
4776 (prepare_target_option_nodes_for_pch): New prototype.
4777 * target-globals.h (struct target_globals): Define even if
4779 * tree.c (prepare_target_option_node_for_pch,
4780 prepare_target_option_nodes_for_pch): New functions.
4781 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4782 * config/i386/i386.c: Include target-globals.h.
4783 (ix86_set_current_function): Instead of doing target_reinit
4784 unconditionally, use save_target_globals_default_opts and
4785 restore_target_globals.
4787 2014-01-09 Richard Biener <rguenther@suse.de>
4789 PR tree-optimization/59715
4790 * tree-cfg.h (split_critical_edges): Declare.
4791 * tree-cfg.c (split_critical_edges): Export.
4792 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4794 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4796 * cfgexpand.c (expand_stack_vars): Optionally disable
4797 asan stack protection.
4798 (expand_used_vars): Likewise.
4799 (partition_stack_vars): Likewise.
4800 * asan.c (asan_emit_stack_protection): Optionally disable
4801 after return stack usage.
4802 (instrument_derefs): Optionally disable memory access instrumentation.
4803 (instrument_builtin_call): Likewise.
4804 (instrument_strlen_call): Likewise.
4805 (asan_protect_global): Optionally disable global variables protection.
4806 * doc/invoke.texi: Added doc for new options.
4807 * params.def: Added new options.
4808 * params.h: Likewise.
4810 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4812 PR rtl-optimization/59724
4813 * ifcvt.c (cond_exec_process_if_block): Don't call
4814 flow_find_head_matching_sequence with 0 longest_match.
4815 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4816 non-active insns if !stop_after.
4817 (try_head_merge_bb): Revert 2014-01-07 changes.
4819 2014-01-08 Jeff Law <law@redhat.com>
4821 * ree.c (get_sub_rtx): New function, extracted from...
4822 (merge_def_and_ext): Here.
4823 (combine_reaching_defs): Use get_sub_rtx.
4825 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4827 * cgraph.h (varpool_variable_node): Do not choke on null node.
4829 2014-01-08 Catherine Moore <clm@codesourcery.com>
4831 * config/mips/mips.md (simple_return): Attempt to use JRC
4833 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4835 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4837 PR rtl-optimization/59137
4838 * reorg.c (steal_delay_list_from_target): Call update_block for
4840 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4842 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4844 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4845 two duplicate entries.
4847 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4850 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4852 * config/mips/mips.c (mips_truncated_op_cost): New function.
4853 (mips_rtx_costs): Adjust test for BADDU.
4854 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4856 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4858 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4859 (*baddu_si): ...this new pattern.
4861 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4864 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4866 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4869 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4871 (expand_expr, expand_normal): Adjust.
4872 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4873 inner_reference_p. Use inner_reference_p to expand inner references.
4874 (store_expr): Adjust.
4875 * cfgexpand.c (expand_call_stmt): Adjust.
4877 2014-01-08 Rong Xu <xur@google.com>
4879 * gcov-io.c (gcov_var): Move from gcov-io.h.
4880 (gcov_position): Ditto.
4881 (gcov_is_error): Ditto.
4882 (gcov_rewrite): Ditto.
4883 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4884 only part to libgcc/libgcov.h.
4886 2014-01-08 Marek Polacek <polacek@redhat.com>
4889 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4891 2014-01-08 Marek Polacek <polacek@redhat.com>
4894 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4896 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4898 PR rtl-optimization/59649
4899 * stor-layout.c (get_mode_bounds): For BImode return
4900 0 and STORE_FLAG_VALUE.
4902 2014-01-08 Richard Biener <rguenther@suse.de>
4905 * gimple.h (is_gimple_builtin_call): Remove.
4906 (gimple_builtin_call_types_compatible_p): New.
4907 (gimple_call_builtin_p): New overload.
4908 * gimple.c (is_gimple_builtin_call): Remove.
4909 (validate_call): Rename to ...
4910 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4912 (validate_type): New static function.
4913 (gimple_call_builtin_p): New overload and adjust.
4914 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4915 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4916 (gimple_fold_stmt_to_constant_1): Likewise.
4917 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4919 2014-01-08 Richard Biener <rguenther@suse.de>
4922 * gimplify.c (gimplify_expr): Gimplify register-register type
4923 VIEW_CONVERT_EXPRs to separate stmts.
4925 2014-01-07 Jeff Law <law@redhat.com>
4928 * ree.c (combine_set_extension): Handle case where source
4929 and destination registers in an extension insn are different.
4930 (combine_reaching_defs): Allow source and destination registers
4931 in extension to be different under limited circumstances.
4932 (add_removable_extension): Remove restriction that the
4933 source and destination registers in the extension are the same.
4934 (find_and_remove_re): Emit a copy from the extension's
4935 destination to its source after the defining insn if
4936 the source and destination registers are different.
4939 * ifcvt.c (merge_if_block): If we are merging a block with more than
4940 one successor with a block with no successors, remove any BARRIER
4941 after the second block.
4943 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4945 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4947 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4950 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4951 for 14-bit register offsets when INT14_OK_STRICT is false.
4953 2014-01-07 Roland Stigge <stigge@antcom.de>
4954 Michael Meissner <meissner@linux.vnet.ibm.com>
4957 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4958 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4960 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4962 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4965 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4967 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4968 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4969 rtx is const0_rtx or not.
4971 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4974 * target-globals.c (save_target_globals): Remove this_fn_optab
4976 * toplev.c: Include optabs.h.
4977 (target_reinit): Temporarily restore the global options if another
4978 set of options are in force.
4980 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4982 PR rtl-optimization/58668
4983 * cfgcleanup.c (flow_find_cross_jump): Don't count
4984 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4985 to determine what is counted.
4986 (flow_find_head_matching_sequence): Use active_insn_p to determine
4988 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4990 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4991 determine what is counted.
4993 PR tree-optimization/59643
4994 * tree-predcom.c (split_data_refs_to_components): If one dr is
4995 read and one write, determine_offset fails and the write isn't
4996 in the bad component, just put the read into the bad component.
4998 2014-01-07 Mike Stump <mikestump@comcast.net>
4999 Jakub Jelinek <jakub@redhat.com>
5002 * tree-core.h (struct tree_optimization_option): Change optabs
5003 type from unsigned char * to void *.
5004 * optabs.c (init_tree_optimization_optabs): Adjust
5005 TREE_OPTIMIZATION_OPTABS initialization.
5007 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5010 * config/i386/i386.h (struct machine_function): Add
5011 no_drap_save_restore field.
5012 * config/i386/i386.c (ix86_save_reg): Use
5013 !cfun->machine->no_drap_save_restore instead of
5014 crtl->stack_realign_needed.
5015 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5016 this function clears frame_pointer_needed. Set
5017 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5018 and DRAP reg is needed.
5020 2014-01-06 Marek Polacek <polacek@redhat.com>
5023 * doc/implement-c.texi: Mention that other integer types are
5024 permitted as bit-field types in strictly conforming mode.
5026 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5028 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5031 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5033 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5035 2014-01-06 Martin Jambor <mjambor@suse.cz>
5038 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5040 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5042 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5046 * var-tracking.c (add_stores): Preserve the value of the source even if
5047 we don't record the store.
5049 2014-01-06 Terry Guo <terry.guo@arm.com>
5051 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5053 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5056 * config/darwin.c (darwin_function_section): Adjust return values to
5057 correspond to optimisation changes made in r206070.
5059 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5061 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5062 from prefetch_block tune setting.
5063 (nocona_cost): Correct size of prefetch block to 64.
5065 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5067 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5068 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5069 used to save the static chain register in the computation of the offset
5070 from which the FP registers need to be restored.
5072 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5074 PR tree-optimization/59519
5075 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5076 ICE if get_current_def (current_new_name) is already non-NULL, as long
5077 as it is a phi result of some other phi in *new_exit_bb that has
5080 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5081 or vmovdqu* for misaligned_operand.
5082 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5083 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5084 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5085 aligned_mem for AVX512F masked aligned load and store builtins and for
5088 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5090 PR tree-optimization/59651
5091 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5092 Address range for negative step should be added by TYPE_SIZE_UNIT.
5094 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5096 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5097 overlapping registers also for registers other than the stack pointer.
5099 2014-01-03 Marek Polacek <polacek@redhat.com>
5102 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5105 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5108 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5111 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5112 (push splitter): Use <P:MODE_SIZE> instead of
5113 GET_MODE_SIZE (<P:MODE>mode).
5114 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5115 (mov -1, reg peephole2): Likewise.
5116 * config/i386/sse.md (*mov<mode>_internal,
5117 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5118 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5119 *<code><mode>3, *andnot<mode>3<mask_name>,
5120 <mask_codefor><code><mode>3<mask_name>): Likewise.
5121 * config/i386/subst.md (mask_mode512bit_condition,
5122 sd_mask_mode512bit_condition): Likewise.
5124 2014-01-02 Xinliang David Li <davidxl@google.com>
5126 PR tree-optimization/59303
5127 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5128 (dump_predicates): Better output format.
5129 (pred_equal_p): New function.
5130 (is_neq_relop_p): Ditto.
5131 (is_neq_zero_form_p): Ditto.
5132 (pred_expr_equal_p): Ditto.
5133 (pred_neg_p): Ditto.
5134 (simplify_pred): Ditto.
5135 (simplify_preds_2): Ditto.
5136 (simplify_preds_3): Ditto.
5137 (simplify_preds_4): Ditto.
5138 (simplify_preds): Ditto.
5140 (push_to_worklist): Ditto.
5141 (get_pred_info_from_cmp): Ditto.
5142 (is_degenerated_phi): Ditto.
5143 (normalize_one_pred_1): Ditto.
5144 (normalize_one_pred): Ditto.
5145 (normalize_one_pred_chain): Ditto.
5146 (normalize_preds): Ditto.
5147 (normalize_cond_1): Remove function.
5148 (normalize_cond): Ditto.
5149 (is_gcond_subset_of): Ditto.
5150 (is_subset_of_any): Ditto.
5151 (is_or_set_subset_of): Ditto.
5152 (is_and_set_subset_of): Ditto.
5153 (is_norm_cond_subset_of): Ditto.
5154 (pred_chain_length_cmp): Ditto.
5155 (convert_control_dep_chain_into_preds): Type change.
5156 (find_predicates): Ditto.
5157 (find_def_preds): Ditto.
5158 (destroy_predicates_vecs): Ditto.
5159 (find_matching_predicates_in_rest_chains): Ditto.
5160 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5161 (is_pred_expr_subset): Ditto.
5162 (is_pred_chain_subset_of): Ditto.
5163 (is_included_in): Ditto.
5164 (is_superset_of): Ditto.
5166 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5168 Update copyright years.
5170 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5172 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5173 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5174 config/arc/arc.md, config/arc/arc.opt,
5175 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5176 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5177 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5178 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5179 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5180 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5181 vtable-verify.c, vtable-verify.h: Use the standard form for the
5184 2014-01-02 Tobias Burnus <burnus@net-b.de>
5186 * gcc.c (process_command): Update copyright notice dates.
5187 * gcov-dump.c: Ditto.
5189 * doc/cpp.texi: Bump @copying's copyright year.
5190 * doc/cppinternals.texi: Ditto.
5191 * doc/gcc.texi: Ditto.
5192 * doc/gccint.texi: Ditto.
5193 * doc/gcov.texi: Ditto.
5194 * doc/install.texi: Ditto.
5195 * doc/invoke.texi: Ditto.
5197 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5199 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5201 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5203 * config/i386/sse.md (*mov<mode>_internal): Guard
5204 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5206 PR rtl-optimization/59647
5207 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5208 new_rtx into UNSIGNED_FLOAT rtxes.
5210 Copyright (C) 2014 Free Software Foundation, Inc.
5212 Copying and distribution of this file, with or without modification,
5213 are permitted in any medium without royalty provided the copyright
5214 notice and this notice are preserved.