1 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
4 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
6 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
8 * config/arm/aarch-common-protos.h
9 (alu_cost_table): Fix spelling of "extend".
10 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
12 2014-03-19 Richard Biener <rguenther@suse.de>
15 * tree-core.h (tree_type_common): Re-order pointer members
16 to reduce recursion depth during GC walks.
18 2014-03-19 Marek Polacek <polacek@redhat.com>
21 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
24 2014-03-19 Richard Biener <rguenther@suse.de>
27 * lto-streamer-in.c (input_function): In WPA stage do not drop
30 2014-03-19 Jakub Jelinek <jakub@redhat.com>
32 PR tree-optimization/60559
33 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
34 with build_zero_cst assignment.
36 2014-03-18 Kai Tietz <ktietz@redhat.com>
38 PR rtl-optimization/56356
39 * sdbout.c (sdbout_parms): Verify that parms'
40 incoming argument is valid.
41 (sdbout_reg_parms): Likewise.
43 2014-03-18 Richard Henderson <rth@redhat.com>
46 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
47 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
48 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
50 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
52 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
53 * doc/plugins.texi (Plugin callbacks): Mention
55 Italicize plugin event names in description. Explain that
56 PLUGIN_PRAGMAS has no sense for lto1. Explain
58 Remind that no GCC functions should be called after
60 Explain what pragmas with expansion are.
62 2014-03-18 Martin Liska <mliska@suse.cz>
64 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
65 gimple call statement is update.
66 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
67 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
69 2014-03-18 Jakub Jelinek <jakub@redhat.com>
72 * ubsan.c (ubsan_instrument_unreachable): Call
73 initialize_sanitizer_builtins.
74 (ubsan_pass): Likewise.
77 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
78 varpool_finalize_decl instead of rest_of_decl_compilation.
80 2014-03-18 Richard Biener <rguenther@suse.de>
82 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
83 by using bitmap_and_compl instead of bitmap_and_compl_into.
84 (df_rd_transfer_function): Likewise.
86 2014-03-18 Richard Biener <rguenther@suse.de>
88 * doc/lto.texi (fresolution): Fix typo.
90 2014-03-18 Richard Biener <rguenther@suse.de>
92 * doc/invoke.texi (flto): Update for changes in 4.9.
94 2014-03-18 Richard Biener <rguenther@suse.de>
96 * doc/loop.texi: Remove section on the removed lambda framework.
97 Update loop docs with recent changes in preserving loop structure.
99 2014-03-18 Richard Biener <rguenther@suse.de>
101 * doc/lto.texi (-fresolution): Document.
103 2014-03-18 Richard Biener <rguenther@suse.de>
105 * doc/contrib.texi: Adjust my name.
107 2014-03-18 Jakub Jelinek <jakub@redhat.com>
110 * internal-fn.c: Include diagnostic-core.h.
111 (expand_BUILTIN_EXPECT): New function.
112 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
113 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
114 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
115 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
117 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
118 Revert 3 argument __builtin_expect code.
119 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
120 * gimple-fold.c (gimple_fold_call): Likewise.
121 * tree.h (fold_builtin_expect): New prototype.
122 * builtins.c (build_builtin_expect_predicate): Add predictor
123 argument, if non-NULL, create 3 argument __builtin_expect.
124 (fold_builtin_expect): No longer static. Add ARG2 argument,
125 pass it through to build_builtin_expect_predicate.
126 (fold_builtin_2): Adjust caller.
127 (fold_builtin_3): Handle BUILT_IN_EXPECT.
128 * internal-fn.def (BUILTIN_EXPECT): New.
130 2014-03-18 Tobias Burnus <burnus@net-b.de>
133 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
134 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
135 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
137 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
140 * predict.c (combine_predictions_for_bb): Fix up formatting.
141 (expr_expected_value_1, expr_expected_value): Add predictor argument,
142 fill what it points to if non-NULL.
143 (tree_predict_by_opcode): Adjust caller, use the predictor.
144 * predict.def (PRED_COMPARE_AND_SWAP): Add.
146 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
148 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
149 proper constant for the store mode.
151 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
153 * symtab.c (change_decl_assembler_name): Fix transparent alias
156 2014-03-16 Renlin Li <Renlin.Li@arm.com>
158 * config/aarch64/aarch64.c: Correct the comments about the
159 aarch64 stack layout.
161 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
163 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
164 check for GF_OMP_FOR_KIND_FOR.
166 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
168 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
169 ymm and zmm register names.
171 2014-03-17 Jakub Jelinek <jakub@redhat.com>
174 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
175 note creation for the 2010-08-31 changes.
177 2014-03-17 Marek Polacek <polacek@redhat.com>
180 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
181 as -fno-tree-loop-vectorize.
182 (expand_omp_simd): Likewise.
184 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
186 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
187 (eligible_for_call_delay): New prototype.
188 * config/sparc/sparc.c (tls_call_delay): Rename into...
189 (eligible_for_call_delay): ...this. Return false if the instruction
190 cannot be put in the delay slot of a branch.
191 (eligible_for_restore_insn): Simplify.
192 (eligible_for_return_delay): Return false if the instruction cannot be
193 put in the delay slot of a branch and simplify.
194 (eligible_for_sibcall_delay): Return false if the instruction cannot be
195 put in the delay slot of a branch.
196 * config/sparc/sparc.md (fix_ut699): New attribute.
197 (tls_call_delay): Delete.
198 (in_call_delay): Reimplement.
199 (eligible_for_sibcall_delay): Rename into...
200 (in_sibcall_delay): ...this.
201 (eligible_for_return_delay): Rename into...
202 (in_return_delay): ...this.
203 (in_branch_delay): Reimplement.
204 (in_uncond_branch_delay): Delete.
205 (in_annul_branch_delay): Delete.
207 2014-03-14 Richard Henderson <rth@redhat.com>
210 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
211 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
212 (*floathi<X87MODEF>2_i387_with_temp): Remove.
213 (floathi splitters): Remove.
214 (float<SWI48x>xf2): New pattern.
215 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
216 code that tried to handle DImode for 32-bit, but which was excluded
217 by the pattern's condition. Drop allocation of stack temporary.
218 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
219 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
220 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
221 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
222 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
223 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
224 (*float<SWI48><MODEF>2_sse_interunit): Remove.
225 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
226 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
227 (*float<SWI48x><X87MODEF>2_i387): Remove.
228 (all float _with_temp splitters): Remove.
229 (*float<SWI48x><MODEF>2_i387): New pattern.
230 (*float<SWI48><MODEF>2_sse): New pattern.
231 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
232 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
234 2014-03-14 Jakub Jelinek <jakub@redhat.com>
235 Marek Polacek <polacek@redhat.com>
238 * common.opt (dump_base_name_prefixed): New Variable.
239 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
240 if x_dump_base_name_prefixed is already set, set it at the end.
242 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
244 PR rtl-optimization/60508
245 * lra-constraints.c (get_reload_reg): Add new parameter
247 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
248 Pass the new parameter values.
250 2014-03-14 Richard Biener <rguenther@suse.de>
252 * common.opt: Revert unintented changes from r205065.
255 2014-03-14 Richard Biener <rguenther@suse.de>
258 * cfghooks.c (split_block): Properly adjust all loops the
259 block was a latch of.
261 2014-03-14 Martin Jambor <mjambor@suse.cz>
264 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
267 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
270 * config/avr/avr.c (avr_set_current_function): Pass function name
271 through default_strip_name_encoding before sanity checking instead
272 of skipping the first char of the assembler name.
274 2014-03-13 Richard Henderson <rth@redhat.com>
277 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
278 (ix86_force_to_memory, ix86_free_from_memory): Remove.
279 * config/i386/i386-protos.h: Likewise.
280 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
281 in the expander instead of a splitter.
282 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
283 any possibility of requiring a memory.
284 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
285 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
286 (fp branch splitters): Update for ix86_split_fp_branch.
287 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
288 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
289 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
290 (*fop_<MODEF>_2_i387): Remove f/r alternative.
291 (*fop_<MODEF>_3_i387): Likewise.
292 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
293 (splitters for the fop_* register patterns): Remove.
294 (fscalexf4_i387): Rename from *fscalexf4_i387.
295 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
297 2014-03-13 Jakub Jelinek <jakub@redhat.com>
299 PR tree-optimization/59779
300 * tree-dfa.c (get_ref_base_and_extent): Use double_int
301 type for bitsize and maxsize instead of HOST_WIDE_INT.
303 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
305 PR rtl-optimization/57320
306 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
307 the CFG after thread_prologue_and_epilogue_insns.
309 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
311 PR rtl-optimization/57189
312 * lra-constraints.c (process_alt_operands): Disfavor spilling
315 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
317 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
320 2014-03-13 Jakub Jelinek <jakub@redhat.com>
322 PR tree-optimization/59025
324 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
325 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
327 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
330 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
331 calls of avr_out_plus_1.
333 2014-03-13 Bin Cheng <bin.cheng@arm.com>
335 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
336 BB's single pred and update the father loop's latch info later.
338 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
340 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
344 (VEC_base): Likewise.
345 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
346 registers, we need to swap double words in little endian mode.
348 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
349 to be a container mode for 128-bit integer operations added in ISA
350 2.07. Unlike TImode and PTImode, the preferred register set is
351 the Altivec/VMX registers for the 128-bit operations.
353 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
355 (rs6000_split_128bit_ok_p): Likewise.
357 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
358 macros for creating ISA 2.07 normal and overloaded builtin
359 functions with 3 arguments.
360 (BU_P8V_OVERLOAD_3): Likewise.
361 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
362 for use as overloaded functions.
363 (VPERM_1TI_UNS): Likewise.
364 (VSEL_1TI): Likewise.
365 (VSEL_1TI_UNS): Likewise.
366 (ST_INTERNAL_1ti): Likewise.
367 (LD_INTERNAL_1ti): Likewise.
368 (XXSEL_1TI): Likewise.
369 (XXSEL_1TI_UNS): Likewise.
370 (VPERM_1TI): Likewise.
371 (VPERM_1TI_UNS): Likewise.
372 (XXPERMDI_1TI): Likewise.
374 (LXVD2X_V1TI): Likewise.
375 (STXVD2X_V1TI): Likewise.
376 (VEC_INIT_V1TI): Likewise.
377 (VEC_SET_V1TI): Likewise.
378 (VEC_EXT_V1TI): Likewise.
379 (EQV_V1TI): Likewise.
380 (NAND_V1TI): Likewise.
381 (ORC_V1TI): Likewise.
382 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
383 added in ISA 2.07. Add both normal 'altivec' builtins, and the
387 (VADDEUQM): Likewise.
388 (VADDECUQ): Likewise.
389 (VSUBEUQM): Likewise.
390 (VSUBECUQ): Likewise.
392 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
393 __int128_t and __uint128_t types.
394 (__uint128_type): Likewise.
395 (altivec_categorize_keyword): Add support for vector __int128_t,
396 vector __uint128_t, vector __int128, and vector unsigned __int128
397 as a container type for TImode operations that need to be done in
398 VSX/Altivec registers.
399 (rs6000_macro_to_expand): Likewise.
400 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
401 to support 128-bit integer instructions vaddcuq, vadduqm,
402 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
403 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
405 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
406 for V1TImode, and set up preferences to use VSX/Altivec registers.
407 Setup VSX reload handlers.
408 (rs6000_debug_reg_global): Likewise.
409 (rs6000_init_hard_regno_mode_ok): Likewise.
410 (rs6000_preferred_simd_mode): Likewise.
411 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
412 (easy_altivec_constant): Likewise.
413 (output_vec_const_move): Likewise.
414 (rs6000_expand_vector_set): Convert V1TImode set and extract to
416 (rs6000_expand_vector_extract): Likewise.
417 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
419 (rs6000_const_vec): Add support for V1TImode.
420 (rs6000_emit_le_vsx_load): Swap double words when loading or
421 storing TImode/V1TImode.
422 (rs6000_emit_le_vsx_store): Likewise.
423 (rs6000_emit_le_vsx_move): Likewise.
424 (rs6000_emit_move): Add support for V1TImode.
425 (altivec_expand_ld_builtin): Likewise.
426 (altivec_expand_st_builtin): Likewise.
427 (altivec_expand_vec_init_builtin): Likewise.
428 (altivec_expand_builtin): Likewise.
429 (rs6000_init_builtins): Add support for V1TImode type. Add
430 support for ISA 2.07 128-bit integer builtins. Define type names
431 for the VSX/Altivec vector types.
432 (altivec_init_builtins): Add support for overloaded vector
433 functions with V1TImode type.
434 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
435 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
437 (rs6000_split_128bit_ok_p): Likewise.
438 (rs6000_handle_altivec_attribute): Create V1TImode from vector
439 __int128_t and vector __uint128_t.
441 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
449 (VS_scalar): Likewise.
450 (VS_double): Likewise.
451 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
453 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
454 we support the ISA 2.07 128-bit integer arithmetic instructions.
455 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
456 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
457 and TImode types for use with the builtin functions.
458 (V1TI_type_node): Likewise.
459 (unsigned_V1TI_type_node): Likewise.
460 (intTI_type_internal_node): Likewise.
461 (uintTI_type_internal_node): Likewise.
463 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
464 128-bit builtin functions.
465 (UNSPEC_VADDEUQM): Likewise.
466 (UNSPEC_VADDECUQ): Likewise.
467 (UNSPEC_VSUBCUQ): Likewise.
468 (UNSPEC_VSUBEUQM): Likewise.
469 (UNSPEC_VSUBECUQ): Likewise.
470 (VM): Add V1TImode to vector mode iterators.
473 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
474 (altivec_vaddcuq): Likewise.
475 (altivec_vsubuqm): Likewise.
476 (altivec_vsubcuq): Likewise.
477 (altivec_vaddeuqm): Likewise.
478 (altivec_vaddecuq): Likewise.
479 (altivec_vsubeuqm): Likewise.
480 (altivec_vsubecuq): Likewise.
482 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
484 (BOOL_128): Likewise.
485 (BOOL_REGS_OUTPUT): Likewise.
486 (BOOL_REGS_OP1): Likewise.
487 (BOOL_REGS_OP2): Likewise.
488 (BOOL_REGS_UNARY): Likewise.
489 (BOOL_REGS_AND_CR0): Likewise.
491 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
492 128-bit integer builtin support.
493 (vec_vadduqm): Likewise.
494 (vec_vaddecuq): Likewise.
495 (vec_vaddeuqm): Likewise.
496 (vec_vsubecuq): Likewise.
497 (vec_vsubeuqm): Likewise.
498 (vec_vsubcuq): Likewise.
499 (vec_vsubuqm): Likewise.
501 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
502 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
503 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
504 128-bit integer add/subtract to ISA 2.07.
506 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
508 * config/arc/arc.c (arc_predicate_delay_insns):
509 Fix third argument passed to conditionalize_nonjump.
511 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
513 * config/aarch64/aarch64-builtins.c
514 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
515 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
516 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
517 instead of __builtin_lfloor.
518 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
520 2014-03-12 Jakub Jelinek <jakub@redhat.com>
522 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
523 (tree_ssa_ifcombine_bb_1): New function.
524 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
525 is an empty forwarder block to then_bb or vice versa and then_bb
526 and else_bb are effectively swapped.
528 2014-03-12 Christian Bruel <christian.bruel@st.com>
531 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
532 REG_CFA_DEF_CFA note.
533 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
534 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
536 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
538 PR tree-optimization/60454
539 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
541 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
543 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
544 Do not define target_cpu_default2 to generic.
545 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
546 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
547 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
549 2014-03-12 Jakub Jelinek <jakub@redhat.com>
550 Marc Glisse <marc.glisse@inria.fr>
552 PR tree-optimization/60502
553 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
554 instead of build_low_bits_mask.
556 2014-03-12 Jakub Jelinek <jakub@redhat.com>
559 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
560 if there are multiple uses, but op doesn't live on E edge.
561 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
562 clobber stmts before __builtin_unreachable.
564 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
566 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
567 hard_frame_pointer_rtx.
568 * cse.c (cse_insn): Remove volatile check.
569 * cselib.c (cselib_process_insn): Likewise.
570 * dse.c (scan_insn): Likewise.
572 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
574 * config/arc/arc.c (conditionalize_nonjump): New function,
576 (arc_ifcvt): ... this.
577 (arc_predicate_delay_insns): Use it.
579 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
581 * config/arc/predicates.md (extend_operand): During/after reload,
582 allow const_int_operand.
583 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
584 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
585 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
587 (umulsi3_highpart_i): Likewise.
589 2014-03-11 Richard Biener <rguenther@suse.de>
591 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
592 Add asserts to guard possible wrong-code bugs.
594 2014-03-11 Richard Biener <rguenther@suse.de>
596 PR tree-optimization/60429
597 PR tree-optimization/60485
598 * tree-ssa-structalias.c (set_union_with_increment): Properly
599 take into account all fields that overlap the shifted vars.
600 (do_sd_constraint): Likewise.
601 (do_ds_constraint): Likewise.
602 (get_constraint_for_ptr_offset): Likewise.
604 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
606 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
607 (nios2_compute_frame_layout):
608 Add calculation of cfun->machine->fp_save_offset.
609 (nios2_expand_prologue): Correct setting of frame pointer register
611 (nios2_expand_epilogue): Update recovery of stack pointer from
612 frame pointer accordingly.
613 (nios2_initial_elimination_offset): Update calculation of offset
614 for eliminating to HARD_FRAME_POINTER_REGNUM.
616 2014-03-10 Jakub Jelinek <jakub@redhat.com>
619 * ipa.c (symtab_remove_unreachable_nodes): Don't call
620 cgraph_get_create_node on VAR_DECLs.
622 2014-03-10 Richard Biener <rguenther@suse.de>
625 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
627 2014-03-08 Douglas B Rupp <rupp@gnat.com>
629 * config/vms/vms.opt (vms_float_format): New variable.
631 2014-03-08 Tobias Burnus <burnus@net-b.de>
633 * doc/invoke.texi (-fcilkplus): Update implementation status.
635 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
636 Richard Biener <rguenther@suse.de>
638 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
639 consistently accross all TUs.
640 (run_gcc): Enable -fshort-double automatically at link at link-time
641 and disallow override.
643 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
646 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
647 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
648 if they can't be used.
650 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
652 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
653 for Solaris 11/x86 ld.
654 * configure: Regenerate.
656 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
658 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
659 (LIB_TLS_SPEC): Save as ld_tls_libs.
660 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
661 (HAVE_AS_IX86_TLSLDM): New test.
662 * configure, config.in: Regenerate.
663 * config/i386/i386.c (legitimize_tls_address): Fall back to
664 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
665 cannot support TLS_MODEL_LOCAL_DYNAMIC.
666 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
667 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
669 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
671 * common.opt (fira-loop-pressure): Mark as optimization.
673 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
675 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
676 an OpenMP mappable type.
678 2014-03-06 Matthias Klose <doko@ubuntu.com>
680 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
681 MULTILIB_OSDIRNAMES is not defined.
683 2014-03-06 Jakub Jelinek <jakub@redhat.com>
684 Meador Inge <meadori@codesourcery.com>
687 * config/arm/arm.c (arm_tls_symbol_p): Remove.
688 (arm_legitimize_address): Call legitimize_tls_address for any
689 arm_tls_referenced_p expression, handle constant addend. Call it
690 before testing for !TARGET_ARM.
691 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
693 2014-03-06 Richard Biener <rguenther@suse.de>
699 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
701 * tree-streamer.c (record_common_node): Assert we don't record
702 nodes with type double.
703 (preload_common_node): Skip type double, complex double and double
704 pointer since it is now frontend dependent due to fshort-double option.
706 2014-03-06 Richard Biener <rguenther@suse.de>
708 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
709 or -fno-lto is specified and the linker has full plugin support.
710 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
711 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
712 * lto-wrapper.c (merge_and_complain): Merge compile-time
714 (run_gcc): And pass it through to the link options.
716 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
720 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
722 * cselib.c (remove_useless_values): Skip to avoid quadratic
723 behavior if the condition moved from...
724 (cselib_process_insn): ... here holds.
726 2014-03-05 Jakub Jelinek <jakub@redhat.com>
729 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
730 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
733 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
734 (TM_H): Add x86-tune.def.
736 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
738 * config/aarch64/aarch64.c (generic_tunings):
739 Use cortexa57_extra_costs.
741 2014-03-05 Jakub Jelinek <jakub@redhat.com>
744 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
745 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
746 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
749 2014-03-04 Heiher <r@hev.cc>
751 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
752 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
754 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
756 * config/i386/predicates.md (const2356_operand): Change to ...
757 (const2367_operand): ... this.
758 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
760 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
761 (*avx512pf_scatterpf<mode>sf): Ditto.
762 (avx512pf_scatterpf<mode>df): Ditto.
763 (*avx512pf_scatterpf<mode>df_mask): Ditto.
764 (*avx512pf_scatterpf<mode>df): Ditto.
765 * config/i386/i386.c (ix86_expand_builtin): Update
766 incorrect hint operand error message.
768 2014-03-04 Richard Biener <rguenther@suse.de>
770 * lto-section-in.c (lto_get_section_data): Fix const cast.
772 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
774 * tree-streamer.c (record_common_node): Assert we don't record
775 nodes with type double.
776 (preload_common_node): Skip type double, complex double and double
777 pointer since it is now frontend dependent due to fshort-double option.
779 2014-03-04 Richard Biener <rguenther@suse.de>
782 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
783 (lto_input_toplevel_asms): Likewise.
784 * lto-section-in.c (lto_get_section_data): Instead do it here
787 2014-03-04 Richard Biener <rguenther@suse.de>
789 PR tree-optimization/60382
790 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
791 dead PHIs a reduction.
793 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
795 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
797 (_mm_prefetch): Move out of GCC target("sse") pragma.
798 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
799 GCC target("prfchw") pragma.
800 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
802 * config/i386/i386.c (ix86_option_override_internal): Enable
803 -mprfchw with -mprefetchwt1.
805 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
807 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
810 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
812 * opts.h (CL_PCH_IGNORE): Define.
813 * targhooks.c (option_affects_pch_p):
814 Return false for options that have CL_PCH_IGNORE set.
815 * opt-functions.awk: Process PchIgnore.
816 * doc/options.texi: Document PchIgnore.
818 * config/arc/arc.opt (misize): Add PchIgnore property.
820 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
822 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
823 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
824 constraint on constants to permit them being loaded into
825 GENERAL_REGS or BASE_REGS.
827 2014-03-03 Nick Clifton <nickc@redhat.com>
829 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
830 anti-cacnonical alternatives.
831 (negandhi3_real): New pattern.
832 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
834 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
836 * config/avr/avr-mcus.def: Remove atxmega16x1.
837 * config/avr/avr-tables.opt: Regenerate.
838 * config/avr/t-multilib: Regenerate.
839 * doc/avr-mmcu.texi: Regenerate.
841 2014-03-03 Tobias Grosser <tobias@grosser.es>
842 Mircea Namolaru <mircea.namolaru@inria.fr>
844 PR tree-optimization/58028
845 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
848 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
850 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
851 not handled by recognizers.
853 2014-03-03 Jakub Jelinek <jakub@redhat.com>
856 * function.c (expand_function_end): Don't emit
857 clobber_return_register sequence if clobber_after is a BARRIER.
858 * cfgexpand.c (construct_exit_block): Append instructions before
859 return_label to prev_bb.
861 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
863 * config/rs6000/constraints.md: Document reserved use of "wc".
865 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
868 * ipa.c (function_and_variable_visibility): When dissolving comdat
869 group, also set all symbols to local.
871 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
876 2013-12-14 Jan Hubicka <jh@suse.cz>
878 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
880 2014-03-02 Jon Beniston <jon@beniston.com>
886 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
887 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
888 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
889 (simple_return, *simple_return): New patterns
890 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
891 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
893 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
895 * dwarf2out.c (gen_subprogram_die): Tidy.
897 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
900 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
901 (*mov_t_msb_neg_negc): ... this new insn.
903 2014-02-28 Jason Merrill <jason@redhat.com>
906 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
909 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
912 * dwarf2out.c (decltype_auto_die): New static.
913 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
914 (gen_type_die_with_usage): Handle 'decltype(auto)'.
915 (is_cxx_auto): Likewise.
917 2014-02-28 Ian Bolton <ian.bolton@arm.com>
919 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
920 we are not using general regs only.
922 2014-02-28 Richard Biener <rguenther@suse.de>
925 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
926 previous fix and only allow to remove trivial pre-headers
927 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
928 (remove_forwarder_block): Properly update the latch of a loop.
930 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
933 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
934 (cselib_preserved_hash_table): New.
935 (preserve_constants_and_equivs): Move preserved vals to it.
936 (cselib_find_slot): Look it up first.
937 (cselib_init): Initialize it.
938 (cselib_finish): Release it.
939 (dump_cselib_table): Dump it.
941 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
944 * cselib.c (remove_useless_values): Skip to avoid quadratic
945 behavior if the condition moved from...
946 (cselib_process_insn): ... here holds.
948 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
951 * var-tracking.c (vt_initialize): Apply the same condition to
952 preserve the CFA base value.
954 2014-02-28 Joey Ye <joey.ye@arm.com>
957 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
958 if reload in progress or completed.
960 2014-02-28 Tobias Burnus <burnus@net-b.de>
963 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
966 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
968 * doc/tm.texi.in (Condition Code Status): Update documention for
969 relative locations of cc0-setter and cc0-user.
971 2014-02-27 Jeff Law <law@redhat.com>
973 PR rtl-optimization/52714
974 * combine.c (try_combine): When splitting an unrecognized PARALLEL
975 into two independent simple sets, if I3 is a jump, ensure the
976 pattern we place into I3 is a (set (pc) ...).
978 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
979 Jeff Law <law@redhat.com>
981 PR rtl-optimization/49847
982 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
983 are in different blocks.
984 * doc/tm.texi (Condition Code Status): Update documention for
985 relative locations of cc0-setter and cc0-user.
987 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
990 * lra.c (lra_emit_add): Check SUBREG too.
992 2014-02-27 Andreas Schwab <schwab@suse.de>
994 * config/m68k/m68k.c (m68k_option_override): Disable
995 -flive-range-shrinkage for classic m68k.
996 (m68k_override_options_after_change): Likewise.
998 2014-02-27 Marek Polacek <polacek@redhat.com>
1001 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1002 -Wmaybe-uninitialized.
1004 2014-02-27 Alan Modra <amodra@gmail.com>
1007 * reload1.c (emit_input_reload_insns): When reload_override_in,
1008 set old to rl->in_reg when rl->in_reg is a subreg.
1010 2014-02-26 Richard Biener <rguenther@suse.de>
1013 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1015 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1017 * common/config/i386/predicates.md (const1256_operand): Remove.
1018 (const2356_operand): New.
1019 (const_1_to_2_operand): Remove.
1020 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1021 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1022 (*avx512pf_gatherpf<mode>sf): Ditto.
1023 (avx512pf_gatherpf<mode>df): Ditto.
1024 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1025 (*avx512pf_gatherpf<mode>df): Ditto.
1026 (avx512pf_scatterpf<mode>sf): Ditto.
1027 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1028 (*avx512pf_scatterpf<mode>sf): Ditto.
1029 (avx512pf_scatterpf<mode>df): Ditto.
1030 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1031 (*avx512pf_scatterpf<mode>df): Ditto.
1032 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1034 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1036 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1037 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1038 (_mm512_mask_testn_epi64_mask): Move to ...
1039 * config/i386/avx512cdintrin.h: Here.
1040 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1041 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1042 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1043 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1044 TARGET_AVX512F from TARGET_AVX512CD.
1046 2014-02-26 Richard Biener <rguenther@suse.de>
1049 * ipa.c (walk_polymorphic_call_targets): Properly guard
1050 call to inline_update_overall_summary.
1052 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1055 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1056 and latches only if requested. Fix latch if it is removed.
1057 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1058 LOOPS_HAVE_PREHEADERS.
1060 2014-02-25 Andrew Pinski <apinski@cavium.com>
1062 * builtins.c (expand_builtin_thread_pointer): Create a new target
1063 when the target is NULL.
1065 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1067 PR rtl-optimization/60317
1068 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1069 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1070 * lra-assigns.c: Include params.h.
1071 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1072 other reload pseudos considerations.
1074 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1076 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1077 to use canonical form for nor<mode>3.
1079 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1082 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1085 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1087 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1088 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1089 (ix86_handle_option): Handle OPT_mprefetchwt1.
1090 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1091 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1093 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1094 OPTION_MASK_ISA_PREFETCHWT1.
1095 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1096 (PTA_PREFETCHWT1): New.
1097 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1098 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1099 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1100 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1101 (*prefetch_avx512pf_<mode>_: Change into ...
1102 (*prefetch_prefetchwt1_<mode>: This.
1103 * config/i386/i386.opt (mprefetchwt1): New.
1104 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1105 (_mm_prefetch): Handle intent to write.
1106 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1108 2014-02-25 Richard Biener <rguenther@suse.de>
1111 * emit-rtl.c (mem_attrs_htab): Remove.
1112 (mem_attrs_htab_hash): Likewise.
1113 (mem_attrs_htab_eq): Likewise.
1114 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1115 (init_emit_once): Do not allocate mem_attrs_htab.
1117 2014-02-25 Richard Biener <rguenther@suse.de>
1120 * lto-opts.c (lto_write_options): Output non-explicit conservative
1121 -fwrapv, -fno-trapv and -fno-strict-overflow.
1122 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1123 (run_gcc): And pass them through.
1125 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1127 * sel-sched.c (calculate_new_fences): New parameter ptime.
1128 Calculate it as a maximum over all fence cycles.
1129 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1130 Print the final schedule timing when sched_verbose.
1132 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1134 PR rtl-optimization/60292
1135 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1136 bit fot the fence instruction.
1138 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1140 * calls.h: Fix typo in comment.
1142 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1144 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1145 adjusting offsetable addresses.
1147 2014-02-24 Guozhi Wei <carrot@google.com>
1149 * sparseset.h (sparseset_pop): Fix the wrong index.
1151 2014-02-24 Walter Lee <walt@tilera.com>
1153 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1154 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1156 * common/config/tilegx/tilegx-common.c
1157 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1158 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1160 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1161 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1162 (tilegx_gimplify_va_arg_expr): Handle big endian.
1163 (tilegx_expand_unaligned_load): Ditto.
1164 (tilegx_expand_unaligned_store): Ditto.
1165 (TARGET_RETURN_IN_MSB): New.
1166 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1167 (TARGET_ENDIAN_DEFAULT): New.
1168 (TARGET_BIG_ENDIAN): Handle big endian.
1169 (BYTES_BIG_ENDIAN): Ditto.
1170 (WORDS_BIG_ENDIAN): Ditto.
1171 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1174 * config/tilegx/tilegx.md (extv): Handle big endian.
1176 (insn_st<n>): Ditto.
1177 (insn_st<n>_add<bitsuffix>): Ditto.
1178 (insn_stnt<n>): Ditto.
1179 (insn_stnt<n>_add<bitsuffix>):Ditto.
1180 (vec_interleave_highv8qi): Handle big endian.
1181 (vec_interleave_highv8qi_be): New.
1182 (vec_interleave_highv8qi_le): New.
1183 (insn_v1int_h): Handle big endian.
1184 (vec_interleave_lowv8qi): Handle big endian.
1185 (vec_interleave_lowv8qi_be): New.
1186 (vec_interleave_lowv8qi_le): New.
1187 (insn_v1int_l): Handle big endian.
1188 (vec_interleave_highv4hi): Handle big endian.
1189 (vec_interleave_highv4hi_be): New.
1190 (vec_interleave_highv4hi_le): New.
1191 (insn_v2int_h): Handle big endian.
1192 (vec_interleave_lowv4hi): Handle big endian.
1193 (vec_interleave_lowv4hi_be): New.
1194 (vec_interleave_lowv4hi_le): New.
1195 (insn_v2int_l): Handle big endian.
1196 (vec_interleave_highv2si): Handle big endian.
1197 (vec_interleave_highv2si_be): New.
1198 (vec_interleave_highv2si_le): New.
1199 (insn_v4int_h): Handle big endian.
1200 (vec_interleave_lowv2si): Handle big endian.
1201 (vec_interleave_lowv2si_be): New.
1202 (vec_interleave_lowv2si_le): New.
1203 (insn_v4int_l): Handle big endian.
1204 * config/tilegx/tilegx.opt (mbig-endian): New option.
1205 (mlittle-endian): New option.
1206 * doc/install.texi: Document tilegxbe-linux.
1207 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1209 2014-02-24 Martin Jambor <mjambor@suse.cz>
1212 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1213 there are no parameter descriptors.
1215 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1217 PR rtl-optimization/60268
1218 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1219 initialization to ...
1220 (sched_rgn_init): ... here.
1221 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1223 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1225 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1228 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1230 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1233 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1235 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1236 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1238 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1240 * config/microblaze/predicates.md: Add cmp_op predicate.
1241 * config/microblaze/microblaze.md: Add branch_compare instruction
1242 which uses cmp_op predicate and emits cmp insn before branch.
1243 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1244 to microblaze_expand_conditional_branch and consolidate logic.
1245 (microblaze_expand_conditional_branch): emit branch_compare
1246 insn instead of handling cmp op separate from branch insn.
1248 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1250 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1253 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1255 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1256 define_insn with define_expand and new define_insn
1257 *altivec_lve<VI_char>x_internal.
1258 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1259 and new define_insn *altivec_stve<VI_char>x_internal.
1260 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1262 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1264 (altivec_expand_stvex_be): New function.
1266 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1268 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1269 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1270 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1271 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1273 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1276 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1277 instead of emit_move_insn.
1279 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1281 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1283 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1284 gen_altivec_vsumsws.
1286 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1288 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1289 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1290 (altivec_lvxl_<mode>): New define_expand incorporating
1291 -maltivec=be semantics where needed.
1292 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1293 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1294 semantics where needed.
1295 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1296 (altivec_stvx_<mode>): New define_expand incorporating
1297 -maltivec=be semantics where needed.
1298 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1299 VM2 iterator instead of V4SI.
1300 (altivec_stvxl_<mode>): New define_expand incorporating
1301 -maltivec=be semantics where needed.
1302 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1303 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1304 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1305 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1306 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1308 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1309 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1310 ALTIVEC_BUILTIN_STVXL.
1311 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1312 (altivec_expand_stvx_be): Likewise.
1313 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1314 (altivec_expand_lvx_be): Likewise.
1315 (altivec_expand_stvx_be): Likewise.
1316 (altivec_expand_builtin): Add cases for
1317 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1318 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1319 (altivec_init_builtins): Add definitions for
1320 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1321 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1323 2014-02-21 Catherine Moore <clm@codesourcery.com>
1325 * doc/invoke.texi (mvirt, mno-virt): Document.
1326 * config/mips/mips.opt (mvirt): New option.
1327 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1329 2014-02-21 Richard Biener <rguenther@suse.de>
1331 PR tree-optimization/60276
1332 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1333 (STMT_VINFO_MIN_NEG_DIST): New macro.
1334 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1335 STMT_VINFO_MIN_NEG_DIST.
1336 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1337 made for negative dependence distances still hold.
1339 2014-02-21 Richard Biener <rguenther@suse.de>
1342 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1343 DECL_INITIAL for globals not in the current function context.
1345 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1347 PR tree-optimization/56490
1348 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1349 * tree-ssa-uninit.c: Include params.h.
1350 (compute_control_dep_chain): Add num_calls argument, return false
1351 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1352 num_calls to recursive call.
1353 (find_predicates): Change dep_chain into normal array,
1354 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1355 variable and adjust compute_control_dep_chain caller.
1356 (find_def_preds): Likewise.
1358 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1360 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1361 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1363 2014-02-21 Nick Clifton <nickc@redhat.com>
1365 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1366 (pushhi1): Likewise.
1367 (popqi1): Add mode to pre_dec.
1370 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1372 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1373 mode for mask of V8SFmode permutation.
1375 2014-02-20 Richard Henderson <rth@redhat.com>
1378 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1379 a new pseudo for OLDVAL.
1381 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1384 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1385 gen_reg_rtx if d->testing_p.
1386 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1387 if d->testing_p and we will certainly return true.
1388 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1391 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1393 * emit-rtl.c (gen_reg_rtx): Assert that
1394 crtl->emit.regno_pointer_align_length is non-zero.
1396 2014-02-20 Richard Henderson <rth@redhat.com>
1399 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1400 on failure the store back into EXPECT.
1402 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1403 Sandra Loosemore <sandra@codesourcery.com>
1405 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1406 * config/nios2/nios2.c (nios2_function_profiler): Add
1407 -fPIC (flag_pic == 2) support.
1408 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1409 (nios2_large_offset_p): New function.
1410 (nios2_unspec_reloc_p): Move up position, update to use
1411 nios2_large_offset_p.
1412 (nios2_unspec_address): Remove function.
1413 (nios2_unspec_offset): New function.
1414 (nios2_large_got_address): New function.
1415 (nios2_got_address): Add large offset support.
1416 (nios2_legitimize_tls_address): Update usage of removed and new
1418 (nios2_symbol_binds_local_p): New function.
1419 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1420 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1421 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1422 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1423 processing for (const (unspec ...)).
1424 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1426 2014-02-20 Richard Biener <rguenther@suse.de>
1428 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1429 doing the substitution.
1430 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1432 2014-02-20 Martin Jambor <mjambor@suse.cz>
1435 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1436 info when checking whether lattices are bottom.
1438 2014-02-20 Richard Biener <rguenther@suse.de>
1441 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1444 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1447 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1448 parameter specifying the scaling.
1449 (inline_call): Update.
1450 (want_inline_recursively): Guard division by zero.
1451 (recursive_inlining): Update.
1452 * ipa-inline.h (clone_inlined_nodes): Update.
1454 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1457 * config/i386/i386.c (classify_argument): Pass structures of size
1458 64 bytes or less in register.
1460 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1461 Kirill Yukhin <kirill.yukhin@intel.com>
1463 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1464 (_mm_rcp28_round_ss): Ditto.
1465 (_mm_rsqrt28_round_sd): Ditto.
1466 (_mm_rsqrt28_round_ss): Ditto.
1467 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1468 (_mm_rcp14_round_ss): Ditto.
1469 (_mm_rsqrt14_round_sd): Ditto.
1470 (_mm_rsqrt14_round_ss): Ditto.
1471 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1472 the first input operand, get rid of match_dup.
1473 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1475 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1477 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1478 operand as the first input operand, set type attribute.
1479 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1481 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1482 operand as the first input operand, set type attribute.
1484 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1486 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1489 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1492 * config/i386/i386.c (construct_container): Remove TFmode check
1493 for X86_64_INTEGER_CLASS.
1495 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1498 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1499 only when -Wpsabi is enabled.
1501 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1504 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1505 passing arrays in registers are the same as for structs, so remove the
1506 special case for them.
1508 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1510 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1511 destination type, extract only the valid bits if the source type is not
1512 integral and has a different mode.
1514 2014-02-19 Richard Biener <rguenther@suse.de>
1517 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1520 2014-02-19 Richard Biener <rguenther@suse.de>
1523 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1524 (ipa_modify_call_arguments): Emit an argument load explicitely and
1525 preserve virtual SSA form there and for the replacement call.
1526 Do not update SSA form nor free dominance info.
1528 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1530 * ipa.c (function_and_variable_visibility): Also clear WEAK
1531 flag when disolving COMDAT_GROUP.
1533 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1535 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1536 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1538 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1539 do more sanity checks.
1540 (detect_type_change): Return true when giving up early.
1541 (compute_complex_assign_jump_func): Fix type parameter of
1542 ipa_set_ancestor_jf.
1543 (compute_complex_ancestor_jump_func): Likewise.
1544 (update_jump_functions_after_inlining): Fix updating of
1546 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1548 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1550 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1551 inline clones when edge disappears.
1553 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1556 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1557 Split 64-bit moves into 2 patterns. Do not allow the use of
1558 direct move for TDmode in little endian, since the decimal value
1559 has little endian bytes within a word, but the 64-bit pieces are
1560 ordered in a big endian fashion, and normal subreg's of TDmode are
1562 (mov<mode>_64bit_dm): Likewise.
1563 (movtd_64bit_nodm): Likewise.
1565 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1567 PR tree-optimization/60174
1568 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1569 statement of an SSA_NAME that occurs in an abnormal PHI node.
1571 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1574 * final.c (SEEN_BB): Remove.
1575 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1576 (final_scan_insn): Don't force_source_line on second
1577 NOTE_INSN_BASIC_BLOCK.
1579 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1582 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1583 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1584 (type_natural_mode): Warn ABI change when %zmm register is not
1585 available for AVX512F vector value passing.
1587 2014-02-18 Kai Tietz <ktietz@redhat.com>
1590 * config/i386/i386.c (ix86_expand_prologue): Use value in
1591 rax register as displacement when restoring %r10 or %rax.
1592 Fix wrong offset when restoring both registers.
1594 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1596 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1597 assertion with conditional return.
1599 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1600 Uros Bizjak <ubizjak@gmail.com>
1603 * config/i386/driver-i386.c (host_detect_local_cpu): If
1604 YMM state is not saved by the OS, also clear has_f16c. Move
1605 CPUID 0x80000001 handling before YMM state saving checking.
1607 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1609 PR rtl-optimization/58960
1610 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1611 factored out from ...
1612 (sched_init): ... here.
1613 (free_global_sched_pressure_data): New, factored out from ...
1614 (sched_finish): ... here.
1615 * sched-int.h (free_global_sched_pressure_data): Declare.
1616 * sched-rgn.c (nr_regions_initial): New static global.
1617 (haifa_find_rgns): Initialize it.
1618 (schedule_region): Disable sched-pressure for the newly
1621 2014-02-17 Richard Biener <rguenther@suse.de>
1623 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1624 release SSA defs of pattern stmts.
1626 2014-02-17 Richard Biener <rguenther@suse.de>
1628 * tree-inline.c (expand_call_inline): Release the virtual
1629 operand defined by the call we are about to inline.
1631 2014-02-17 Richard Biener <rguenther@suse.de>
1633 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1635 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1636 Ilya Tocar <ilya.tocar@intel.com>
1638 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1639 arguments order in builtin.
1640 (_mm512_permutexvar_epi64): Ditto.
1641 (_mm512_mask_permutexvar_epi64): Ditto
1642 (_mm512_maskz_permutexvar_epi32): Ditto
1643 (_mm512_permutexvar_epi32): Ditto
1644 (_mm512_mask_permutexvar_epi32): Ditto
1646 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1648 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1649 (p8_vmrgow): Likewise.
1651 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1653 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1656 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1659 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1660 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1661 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1662 using direct move instructions on ISA 2.07. Also adjust
1663 instruction length for 64-bit.
1664 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1665 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1667 2014-02-15 Alan Modra <amodra@gmail.com>
1671 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1672 find_replacement on parts of insn rtl that might be reloaded.
1674 2014-02-15 Richard Biener <rguenther@suse.de>
1676 PR tree-optimization/60183
1677 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1678 (tree_ssa_phiprop): Calculate and free post-dominators.
1680 2014-02-14 Jeff Law <law@redhat.com>
1682 PR rtl-optimization/60131
1683 * ree.c (get_extended_src_reg): New function.
1684 (combine_reaching_defs): Use it rather than assuming location of REG.
1685 (find_and_remove_re): Verify first operand of extension is
1686 a REG before adding the insns to the copy list.
1688 2014-02-14 Roland McGrath <mcgrathr@google.com>
1690 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1691 * configure: Regenerated.
1692 * config.in: Regenerated.
1693 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1694 instead of ASM_SHORT.
1696 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1697 Richard Earnshaw <rearnsha@arm.com>
1699 PR rtl-optimization/59535
1700 * lra-constraints.c (process_alt_operands): Encourage alternative
1701 when unassigned pseudo class is superset of the alternative class.
1702 (inherit_reload_reg): Don't inherit when optimizing for code size.
1703 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1704 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1705 modes not less than 4 for Thumb1.
1707 2014-02-14 Kyle McMartin <kyle@redhat.com>
1710 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1712 2014-02-14 Richard Biener <rguenther@suse.de>
1714 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1715 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1716 Do not strip INDIRECT_REFs.
1718 2014-02-14 Richard Biener <rguenther@suse.de>
1721 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1722 DECL_FUNCTION_SPECIFIC_TARGET.
1723 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1724 * tree-streamer-out.c (pack_ts_target_option): Remove.
1725 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1726 (write_ts_function_decl_tree_pointers): Do not stream
1727 DECL_FUNCTION_SPECIFIC_TARGET.
1728 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1729 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1730 (lto_input_ts_function_decl_tree_pointers): Do not stream
1731 DECL_FUNCTION_SPECIFIC_TARGET.
1733 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1735 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1736 (get_initial_def_for_induction, vectorizable_induction): Ignore
1737 debug stmts when looking for exit_phi.
1738 (vectorizable_live_operation): Fix up condition.
1740 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1742 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1743 nreverse() because it changes the content of original tree list.
1745 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1747 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1748 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1750 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1752 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1753 GNU coding standards.
1755 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1758 * dwarf2out.c (gen_subprogram_die): Don't call
1759 add_calling_convention_attribute if subr_die is old_die.
1761 2014-02-13 Sharad Singhai <singhai@google.com>
1763 * doc/optinfo.texi: Fix order of nodes.
1765 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1767 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1768 operands[2], not operands[3].
1770 2014-02-13 Richard Biener <rguenther@suse.de>
1773 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1774 mention the possibility of an in-tree build.
1775 (CLooG): Update recommended version to 0.18.1, mention the
1776 possibility of an in-tree build and clarify that the ISL
1777 bundled with CLooG does not work.
1779 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1782 * expr.c (compress_float_constant): If x is a hard register,
1783 extend into a pseudo and then move to x.
1785 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1787 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1788 caused by bad second argument to warning_at() with -mhotpatch and
1789 nested functions (e.g. with gfortran).
1791 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1793 * opts.c (option_name): Remove "enabled by default" rider.
1795 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1797 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1799 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1800 Uros Bizjak <ubizjak@gmail.com>
1803 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1804 * configure: Regenerated.
1806 2014-02-12 Richard Biener <rguenther@suse.de>
1808 * vec.c (vec_prefix::calculate_allocation): Move as
1809 inline variant to vec.h.
1810 (vec_prefix::calculate_allocation_1): New out-of-line version.
1811 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1812 (vec_prefix::m_has_auto_buf): Rename to ...
1813 (vec_prefix::m_using_auto_storage): ... this.
1814 (vec_prefix::calculate_allocation): Inline the easy cases
1815 and dispatch to calculate_allocation_1 which doesn't need the
1817 (va_heap::reserve): Use gcc_checking_assert.
1818 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1819 m_using_auto_storage.
1820 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1822 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1823 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1824 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1826 2014-02-12 Richard Biener <rguenther@suse.de>
1828 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1829 when we found a dependence.
1831 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1833 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1835 (maybe_fold_stmt): ... into this new function.
1836 * omp-low.c (lower_omp): Update comment.
1838 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1841 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1844 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1846 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1847 identifiers in comments.
1848 (cortexa53_extra_costs): Likewise.
1849 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1850 (cortexa7_extra_costs): Likewise.
1851 (cortexa12_extra_costs): Likewise.
1852 (cortexa15_extra_costs): Likewise.
1853 (v7m_extra_costs): Likewise.
1855 2014-02-12 Richard Biener <rguenther@suse.de>
1858 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1859 of posix_memalign being successful.
1860 (lower_stmt): Restrict lowering of posix_memalign to when
1861 -ftree-bit-ccp is enabled.
1863 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1865 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1867 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1869 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1871 PR rtl-optimization/60116
1872 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1873 other_insn once the combination has been validated.
1875 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1878 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1880 * ipa-devirt.c: Include demangle.h
1881 (odr_violation_reported): New static variable.
1882 (add_type_duplicate): Update odr_violations.
1883 (maybe_record_node): Add completep parameter; update it.
1884 (record_target_from_binfo): Add COMPLETEP parameter;
1885 update it as needed.
1886 (possible_polymorphic_call_targets_1): Likewise.
1887 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1888 rename FINAL to COMPLETE.
1889 (record_targets_from_bases): Sanity check we found the binfo;
1890 fix COMPLETEP updating.
1891 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1892 parameter, fix computing of COMPLETEP.
1893 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1894 at LTO time do demangling.
1895 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1896 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1898 (gimple_get_virt_method_for_binfo): Likewise.
1899 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1900 gimple_get_virt_method_for_vtable): Update prototypes.
1902 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1905 * genautomata.c (add_presence_absence): Fix typo with
1906 {final_}presence_list.
1908 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1911 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1912 for VSX/Altivec vectors that land in GPR registers.
1914 2014-02-11 Richard Henderson <rth@redhat.com>
1915 Jakub Jelinek <jakub@redhat.com>
1918 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1919 around drhs if type conversion to lacc->type is not useless.
1921 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1923 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1925 (cortex-a57.cortex-a53): Likewise.
1926 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1928 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1930 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1933 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1935 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1936 add_options_for_arm_vfp3.
1938 2014-02-11 Jeff Law <law@redhat.com>
1941 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1942 object with an undesirable mode.
1944 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1947 * config/i386/sol2-9.h: New file.
1948 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1949 *-*-solaris2.9*): Use it.
1951 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1953 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1954 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1956 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1958 * config/microblaze/microblaze.c: Extend mcpu version format
1960 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1962 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1964 2014-02-10 Richard Henderson <rth@redhat.com>
1967 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1968 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1969 ms-abi vs -mno-accumulate-outgoing-args.
1970 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1971 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1974 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1977 * cfgexpand.c (expand_asm_operands): Attach source location to
1978 ASM_INPUT rtx objects.
1979 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1981 2014-02-10 Nick Clifton <nickc@redhat.com>
1983 * config/mn10300/mn10300.c (popcount): New function.
1984 (mn10300_expand_prologue): Include saved registers in stack usage
1987 2014-02-10 Jeff Law <law@redhat.com>
1990 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1991 when changing the SET_DEST of a prior insn to avoid an input reload.
1993 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1995 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1996 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1997 -mcall-openbsd, or -mcall-linux.
1998 (CC1_ENDIAN_BIG_SPEC): Remove.
1999 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2000 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2001 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2002 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2003 and %cc1_endian_default.
2004 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2006 2014-02-10 Richard Biener <rguenther@suse.de>
2008 PR tree-optimization/60115
2009 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2010 MEM_REF handling. Properly verify that the accesses are not
2011 out of the objects bound.
2013 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2015 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2018 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2020 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2021 proper constants and fix formatting.
2022 (possible_polymorphic_call_targets): Fix formatting.
2024 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2025 Ilya Tocar <ilya.tocar@intel.com>
2027 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2028 (_mm512_loadu_epi32): Renamed into...
2029 (_mm512_loadu_si512): This.
2030 (_mm512_storeu_epi32): Renamed into...
2031 (_mm512_storeu_si512): This.
2032 (_mm512_maskz_ceil_ps): Removed.
2033 (_mm512_maskz_ceil_pd): Ditto.
2034 (_mm512_maskz_floor_ps): Ditto.
2035 (_mm512_maskz_floor_pd): Ditto.
2036 (_mm512_floor_round_ps): Ditto.
2037 (_mm512_floor_round_pd): Ditto.
2038 (_mm512_ceil_round_ps): Ditto.
2039 (_mm512_ceil_round_pd): Ditto.
2040 (_mm512_mask_floor_round_ps): Ditto.
2041 (_mm512_mask_floor_round_pd): Ditto.
2042 (_mm512_mask_ceil_round_ps): Ditto.
2043 (_mm512_mask_ceil_round_pd): Ditto.
2044 (_mm512_maskz_floor_round_ps): Ditto.
2045 (_mm512_maskz_floor_round_pd): Ditto.
2046 (_mm512_maskz_ceil_round_ps): Ditto.
2047 (_mm512_maskz_ceil_round_pd): Ditto.
2048 (_mm512_expand_pd): Ditto.
2049 (_mm512_expand_ps): Ditto.
2050 * config/i386/i386.c (ix86_builtins): Remove
2051 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2052 (bdesc_args): Ditto.
2053 * config/i386/predicates.md (const1256_operand): New.
2054 (const_1_to_2_operand): Ditto.
2055 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2056 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2057 (*avx512pf_gatherpf<mode>sf): Ditto.
2058 (avx512pf_gatherpf<mode>df): Ditto.
2059 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2060 (*avx512pf_gatherpf<mode>df): Ditto.
2061 (avx512pf_scatterpf<mode>sf): Ditto.
2062 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2063 (*avx512pf_scatterpf<mode>sf): Ditto.
2064 (avx512pf_scatterpf<mode>df): Ditto.
2065 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2066 (*avx512pf_scatterpf<mode>df): Ditto.
2067 (avx512f_expand<mode>): Removed.
2068 (<shift_insn><mode>3<mask_name>): Change predicate type.
2070 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2072 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2073 not at the end of datarefs vector use ordered_remove to avoid
2074 reordering datarefs vector.
2077 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2078 mark local addressable non-static vars as GOVD_PRIVATE
2079 instead of GOVD_LOCAL.
2080 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2081 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2085 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2086 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2087 assume_aligned or alloc_align attributes.
2088 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2089 arguments. Handle also assume_aligned and alloc_align attributes.
2090 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2091 calls to functions with assume_aligned or alloc_align attributes.
2092 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2094 2014-02-08 Terry Guo <terry.guo@arm.com>
2096 * doc/invoke.texi: Document ARM -march=armv7e-m.
2098 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2100 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2101 flag on __cilkrts_rethrow builtin.
2104 * ipa-cp.c (determine_versionability): Fail at -O0
2105 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2106 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2109 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2112 * tree-inline.c (copy_forbidden): Fail for
2113 __attribute__((optimize (0))) functions.
2115 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2117 * varpool.c: Include pointer-set.h.
2118 (varpool_remove_unreferenced_decls): Variables in other partitions
2119 will not be output; be however careful to not lose information
2122 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2124 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2125 lookup in the vtable constructor.
2127 2014-02-07 Jeff Law <law@redhat.com>
2130 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2131 define_insn_and_split.
2133 * ipa-inline.c (inline_small_functions): Fix typos.
2135 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2137 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2138 (s390_can_use_return_insn): Declare.
2139 * config/s390/s390.h (EPILOGUE_USES): Define.
2140 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2142 (s390_chunkify_start): Handle return JUMP_LABELs.
2143 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2144 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2145 (s390_can_use_return_insn): New functions.
2146 (s390_fix_long_loop_prediction): Handle conditional returns.
2147 (TARGET_SET_UP_BY_PROLOGUE): Define.
2148 * config/s390/s390.md (ANY_RETURN): New code iterator.
2149 (*creturn, *csimple_return, return, simple_return): New patterns.
2151 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2153 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2154 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2155 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2156 REG_CFA_RESTORE list when deciding not to restore a register.
2158 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2160 * config/s390/s390.c: Include tree-pass.h and context.h.
2161 (s390_early_mach): New function, split out from...
2162 (s390_emit_prologue): ...here.
2163 (pass_data_s390_early_mach): New pass structure.
2164 (pass_s390_early_mach): New class.
2165 (s390_option_override): Create and register early_mach pass.
2166 Move to end of file.
2168 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2170 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2171 to match for the exit block.
2173 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2175 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2176 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2177 Reject misaligned operands.
2179 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2181 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2183 2014-02-07 Richard Biener <rguenther@suse.de>
2186 * gimple-low.c (lower_builtin_posix_memalign): New function.
2187 (lower_stmt): Call it to lower posix_memalign in a way
2188 to make alignment info accessible.
2190 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2193 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2194 __builtin_setjmp_receiver.
2196 2014-02-07 Richard Biener <rguenther@suse.de>
2199 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2200 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2201 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2202 Handle BUILT_IN_POSIX_MEMALIGN.
2203 (find_func_clobbers): Likewise.
2204 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2205 (call_may_clobber_ref_p_1): Likewise.
2207 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2210 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2213 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2216 * lto-cgraph.c (lto_output_node): Use
2217 symtab_get_symbol_partitioning_class.
2218 (lto_output_varpool_node): likewise.
2219 (symtab_get_symbol_partitioning_class): Move here from
2221 * cgraph.h (symbol_partitioning_class): Likewise.
2222 (symtab_get_symbol_partitioning_class): Declare.
2224 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2226 * ggc.h (ggc_internal_cleared_alloc): New macro.
2227 * vec.h (vec_safe_copy): Handle memory stats.
2228 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2229 * target-globals.c (save_target_globals): Likewise.
2231 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2234 * expr.c (emit_move_resolve_push): Export; be bit more selective
2235 on when to clear alias set.
2236 * expr.h (emit_move_resolve_push): Declare.
2237 * function.h (struct function): Add tail_call_marked.
2238 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2239 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2240 * config/i386/i386.md (TImode move expander): De not call
2242 (FP push expanders): Preserve memory attributes.
2243 * config/i386/sse.md (push<mode>1): Remove.
2244 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2245 (ix86_expand_push): Remove.
2246 * config/i386/mmx.md (push<mode>1): Remove.
2248 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2250 PR rtl-optimization/60030
2251 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2252 lopart with paradoxical subreg before shifting it up by hprec.
2254 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2256 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2257 Remove extra newline at end of file.
2258 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2259 (arm_issue_rate): Handle cortexa57.
2260 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2261 (cortex-a57.cortex-a53): Likewise.
2263 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2266 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2267 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2269 (arm_expand_prologue): Adjust all callers.
2270 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2271 info, registers also at the lowest numbered registers side. Use
2272 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2276 * var-tracking.c (adjust_mems): Before adding a SET to
2277 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2279 2014-02-06 Alan Modra <amodra@gmail.com>
2282 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2283 change SDmode to DDmode when lra_in_progress.
2285 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2288 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2289 free_data_ref on the dr first, and before goto again also set dr
2290 to the next dr. For simd_lane_access, free old datarefs[i] before
2291 overwriting it. For get_vectype_for_scalar_type failure, don't
2292 free_data_ref if simd_lane_access.
2294 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2297 * tree.h (opts_for_fn): New inline function.
2298 (opt_for_fn): Define.
2299 * config/i386/i386.c (ix86_function_regparm): Use
2300 opt_for_fn (decl, optimize) instead of optimize.
2302 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2304 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2305 for SYMBOL_REF in large memory model.
2307 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2309 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2311 (cortex-a57): Likewise.
2312 (cortex-a57.cortex-a53): Likewise.
2314 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2315 Kugan Vivekanandarajah <kuganv@linaro.org>
2317 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2319 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2321 2014-02-06 Richard Biener <rguenther@suse.de>
2323 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2324 set_loop_copy and initialize_original_copy_tables.
2326 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2328 * config/aarch64/aarch64-simd.md
2329 (aarch64_ashr_simddi): Change QI to SI.
2331 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2332 Jakub Jelinek <jakub@redhat.com>
2335 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2338 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2340 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2341 CODE_FOR_altivec_vpku[hw]um to
2342 CODE_FOR_altivec_vpku[hw]um_direct.
2343 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2344 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2345 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2346 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2348 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2350 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2351 generation for -maltivec=be.
2352 (altivec_vsumsws): Simplify redundant test.
2354 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2356 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2357 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2358 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2359 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2360 gen_altivec_vpkuwum.
2361 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2363 (altivec_vpks<VI_char>ss): Likewise.
2364 (altivec_vpks<VI_char>us): Likewise.
2365 (altivec_vpku<VI_char>us): Likewise.
2366 (altivec_vpku<VI_char>um): Likewise.
2367 (altivec_vpku<VI_char>um_direct): New (copy of
2368 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2370 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2371 target is little endian and -maltivec=be is not specified.
2372 (*altivec_vupkhs<VU_char>_direct): New (copy of
2373 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2374 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2375 target is little endian and -maltivec=be is not specified.
2376 (*altivec_vupkls<VU_char>_direct): New (copy of
2377 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2378 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2379 little endian and -maltivec=be is not specified.
2380 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2381 little endian and -maltivec=be is not specified.
2383 2014-02-05 Richard Henderson <rth@redhat.com>
2386 * combine-stack-adj.c: Revert r206943.
2387 * sched-int.h (struct deps_desc): Add last_args_size.
2388 * sched-deps.c (init_deps): Initialize it.
2389 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2390 contain REG_ARGS_SIZE notes.
2392 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2394 * lto-cgraph.c (asm_nodes_output): Make global.
2395 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2396 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2397 (driver_handle_option): Handle OPT_fwpa.
2399 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2402 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2403 a comment typo and formatting issue. If odr_hash hasn't been
2404 created, return vNULL and set *completep to false.
2407 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2408 bb with no successors.
2410 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2413 * doc/invoke.texi (-march): Clarify documentation for ARM.
2417 2014-02-05 Richard Biener <rguenther@suse.de>
2419 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2420 when not vectorizing because of too many alias checks.
2421 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2422 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2424 2014-02-05 Nick Clifton <nickc@redhat.com>
2426 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2427 accept extended registers in any mode when compiling for the MN10300.
2429 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2431 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2432 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2433 sanitization attributes.
2434 (can_inline_edge_p): Likewise.
2435 (sanitize_attrs_match_for_inline_p): New function.
2437 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2439 * ipa-prop.c (detect_type_change): Shor circuit testing of
2440 type changes on THIS pointer.
2442 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2445 * config/pa/pa.c (legitimize_tls_address): Return original address
2446 if not passed a SYMBOL_REF rtx.
2447 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2449 (pa_emit_move_sequence): Simplify TLS source operands.
2450 (pa_legitimate_constant_p): Reject all TLS constants.
2451 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2452 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2454 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2456 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2457 groups when we know they are controlled by LTO.
2458 * varasm.c (default_binds_local_p_1): If object is in other partition,
2459 it will be resolved locally.
2461 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2463 * config/host-linux.c (linux_gt_pch_use_address): Don't
2464 use SSIZE_MAX because it is not always defined.
2466 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2469 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2470 threshold for pseudo splitting.
2471 (update_ebb_live_info): Process call argument hard registers and
2472 hard registers from insn definition too.
2473 (max_small_class_regs_num): New constant.
2474 (inherit_in_ebb): Update live hard regs through EBBs. Update
2475 reloads_num only for small register classes. Don't split for
2478 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2481 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2484 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2486 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2489 2014-02-04 Marek Polacek <polacek@redhat.com>
2491 * gdbinit.in (pel): Define.
2493 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2495 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2498 2014-02-04 Richard Biener <rguenther@suse.de>
2501 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2502 in function context local.
2503 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2504 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2505 similar to LTO_imported_decl_ref.
2507 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2509 PR tree-optimization/60002
2510 * cgraphclones.c (build_function_decl_skip_args): Clear
2513 PR tree-optimization/60023
2514 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2515 false to gsi_replace.
2516 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2517 has been in some EH region and vec_stmt could throw, add
2518 vec_stmt into the same EH region.
2519 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2520 has no lhs, ignore it.
2521 * internal-fn.c (expand_MASK_LOAD): Likewise.
2524 * tree-inline.c (copy_forbidden): Fail for
2525 __attribute__((optimize (0))) functions.
2528 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2529 is set, copy one less argument.
2530 (expand_simd_clones): Don't subtract clone_info->inbranch
2531 from simd_clone_struct_alloc argument.
2533 PR rtl-optimization/57915
2534 * recog.c (simplify_while_replacing): If all unary/binary/relational
2535 operation arguments are constant, attempt to simplify those.
2538 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2539 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2541 2014-02-04 Richard Biener <rguenther@suse.de>
2543 PR tree-optimization/60012
2544 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2545 TBAA disambiguation to all DDRs.
2547 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2550 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2551 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2553 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2556 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2558 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2560 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2561 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2563 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2566 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2567 to figure out targets of polymorphic calls with known decl.
2568 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2569 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2570 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2571 (get_polymorphic_call_info): ... here.
2572 (get_polymorphic_call_info_from_invariant): New function.
2574 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2576 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2577 lookup via vtable pointer; check for type consistency
2578 and turn inconsitent facts into UNREACHABLE.
2579 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2580 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2581 type inconsistent querries; return UNREACHABLE instead.
2583 2014-02-03 Richard Henderson <rth@twiddle.net>
2586 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2587 already processed this node.
2588 (normalize_one_pred_1): Pass along mark_set.
2589 (normalize_one_pred): Create and destroy a pointer_set_t.
2590 (normalize_one_pred_chain): Likewise.
2592 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2594 PR gcov-profile/58602
2595 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2597 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2600 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2601 -fno-devirtualize; try to devirtualize by the knowledge of
2602 virtual table pointer given by aggregate propagation.
2603 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2604 (ipa_print_node_jump_functions): Dump also offset that
2605 is relevant for polymorphic calls.
2606 (determine_known_aggregate_parts): Add arg_type parameter; use it
2607 instead of determining the type from pointer type.
2608 (ipa_compute_jump_functions_for_edge): Update call of
2609 determine_known_aggregate_parts.
2610 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2611 (gimple_get_virt_method_for_binfo): ... here; simplify using
2612 vtable_pointer_value_to_vtable.
2613 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2614 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2615 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2616 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2618 (vtable_pointer_value_to_binfo): ... here.
2619 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2621 2014-02-03 Teresa Johnson <tejohnson@google.com>
2623 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2624 redef of outer loop index variable.
2626 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2630 * doc/extend.texi (Function Attributes): Typo.
2632 2014-02-03 Cong Hou <congh@google.com>
2634 PR tree-optimization/60000
2635 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2636 if the vectorized statement is a store. A store statement can only
2637 appear at the end of pattern statements.
2639 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2641 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2642 (ix86_option_override_internal): Default long double to 64-bit for
2643 32-bit Bionic and to 128-bit for 64-bit Bionic.
2645 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2646 TARGET_LONG_DOUBLE_128 is true.
2647 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2649 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2650 (mlong-double-64): Negate -mlong-double-128.
2651 (mlong-double-128): New option.
2653 * config/i386/i386-c.c (ix86_target_macros): Define
2654 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2656 * doc/invoke.texi: Document -mlong-double-128.
2658 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2660 PR rtl-optimization/60024
2661 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2663 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2665 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2667 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2669 PR rtl-optimization/57662
2670 * sel-sched.c (code_motion_path_driver): Do not mark already not
2671 existing blocks in the visiting bitmap.
2673 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2675 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2676 on the insn being emitted.
2678 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2679 Will Deacon <will.deacon@arm.com>
2681 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2683 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2685 * config/arm/arm-tables.opt: Regenerate.
2687 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2689 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2690 for vector types other than V16QImode.
2691 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2692 define_expand, and call altivec_expand_vec_perm_le when producing
2693 code with little endian element order.
2694 (*altivec_vperm_<mode>_internal): New insn having previous
2695 behavior of altivec_vperm_<mode>.
2696 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2697 altivec_expand_vec_perm_le when producing code with little endian
2699 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2700 behavior of altivec_vperm_<mode>_uns.
2702 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2704 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2705 (altivec_vsumsws): Add handling for -maltivec=be with a little
2707 (altivec_vsumsws_direct): New.
2708 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2709 gen_altivec_vsumsws.
2711 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2713 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2714 vtable_pointer_value_to_binfo): New functions.
2715 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2716 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2718 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2720 * config/nios2/nios2.md (load_got_register): Initialize GOT
2721 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2722 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2724 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2726 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2727 preserverd by passthrough, do not propagate the type.
2729 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2731 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2732 (mips_atomic_assign_expand_fenv): New function.
2733 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2735 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2737 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2738 (__builtin_mips_set_fcsr): Likewise.
2739 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2740 MIPS_USI_FTYPE_VOID.
2741 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2742 (mips16_expand_set_fcsr): Likewise.
2743 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2744 (mips16_set_fcsr_stub): Likewise.
2745 (mips16_get_fcsr_one_only_stub): New class.
2746 (mips16_set_fcsr_one_only_stub): Likewise.
2747 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2748 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2749 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2750 (hard_float): New availability predicate.
2751 (mips_builtins): Add get_fcsr and set_fcsr.
2752 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2753 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2754 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2755 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2756 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2759 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2761 * config/mips/mips.c (mips_one_only_stub): New class.
2762 (mips_need_mips16_rdhwr_p): Replace with...
2763 (mips16_rdhwr_stub): ...this new variable.
2764 (mips16_stub_call_address): New function.
2765 (mips16_rdhwr_one_only_stub): New class.
2766 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2767 (mips_output_mips16_rdhwr): Delete.
2768 (mips_finish_stub): New function.
2769 (mips_code_end): Use it to handle rdhwr stubs.
2771 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2774 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2775 when calculating size of integer atomic types.
2777 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2779 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2781 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2783 PR tree-optimization/60003
2784 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2785 * profile.c (branch_prob): Use gimple_call_builtin_p
2786 to check for BUILT_IN_SETJMP_RECEIVER.
2787 * tree-inline.c (copy_bb): Call notice_special_calls.
2789 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2792 * lra-constraints.c (process_alt_operands): Update reload_sum only
2795 2014-01-31 Richard Henderson <rth@redhat.com>
2798 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2799 until after else_eh is processed.
2801 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2803 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2804 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2805 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2806 in smmintrin.h, remove them.
2807 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2808 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2809 * config/i386/i386.md (ROUND_SAE): Fix value.
2810 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2811 (const48_operand): New.
2812 * config/i386/subst.md (round), (round_expand): Use
2813 const_4_or_8_to_11_operand.
2814 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2816 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2818 * config/i386/constraints.md (Yk): Swap meaning with k.
2819 * config/i386/i386.md (movhi_internal): Change Yk to k.
2820 (movqi_internal): Ditto.
2821 (*k<logic><mode>): Ditto.
2824 (kandn<mode>): Ditto.
2825 (*<code>hi_1): Ditto.
2826 (*<code>qi_1): Ditto.
2827 (kxnor<mode>): Ditto.
2828 (kortestzhi): Ditto.
2829 (kortestchi): Ditto.
2831 (*one_cmplhi2_1): Ditto.
2832 (*one_cmplqi2_1): Ditto.
2833 * config/i386/sse.md (): Change k to Yk.
2834 (avx512f_load<mode>_mask): Ditto.
2835 (avx512f_blendm<mode>): Ditto.
2836 (avx512f_store<mode>_mask): Ditto.
2837 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2838 (avx512f_storedqu<mode>_mask): Ditto.
2839 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2841 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2842 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2843 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2844 (avx512f_maskcmp<mode>3): Ditto.
2845 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2846 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2847 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2848 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2849 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2850 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2851 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2852 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2853 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2854 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2855 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2856 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2857 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2858 (vec_extract_lo_<mode>_maskm): Ditto.
2859 (vec_extract_hi_<mode>_maskm): Ditto.
2860 (avx512f_vternlog<mode>_mask): Ditto.
2861 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2862 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2863 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2864 (avx512f_<code>v8div16qi2_mask): Ditto.
2865 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2866 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2867 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2868 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2869 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2870 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2871 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2872 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2873 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2874 (avx512cd_maskb_vec_dupv8di): Ditto.
2875 (avx512cd_maskw_vec_dupv16si): Ditto.
2876 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2877 (avx512f_vpermi2var<mode>3_mask): Ditto.
2878 (avx512f_vpermi2var<mode>3_mask): Ditto.
2879 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2880 (*avx512f_gathersi<mode>): Ditto.
2881 (*avx512f_gathersi<mode>_2): Ditto.
2882 (*avx512f_gatherdi<mode>): Ditto.
2883 (*avx512f_gatherdi<mode>_2): Ditto.
2884 (*avx512f_scattersi<mode>): Ditto.
2885 (*avx512f_scatterdi<mode>): Ditto.
2886 (avx512f_compress<mode>_mask): Ditto.
2887 (avx512f_compressstore<mode>_mask): Ditto.
2888 (avx512f_expand<mode>_mask): Ditto.
2889 * config/i386/subst.md (mask): Change k to Yk.
2890 (mask_scalar_merge): Ditto.
2893 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2895 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2897 2014-01-31 Richard Biener <rguenther@suse.de>
2900 * builtins.c (fold_builtin_memory_op): Make sure to not
2901 use a floating-point mode or a boolean or enumeral type for
2904 2014-01-30 DJ Delorie <dj@redhat.com>
2906 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2907 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2908 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2909 whenever main() has an epilogue.
2911 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2913 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2914 unused variable "field".
2915 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2916 (vsx_mergeh_<mode>): Likewise.
2917 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2918 (altivec_vmrghh): Likewise.
2919 (altivec_vmrghw): Likewise.
2920 (altivec_vmrglb): Likewise.
2921 (altivec_vmrglh): Likewise.
2922 (altivec_vmrglw): Likewise.
2923 (altivec_vspltb): Add missing uses.
2924 (altivec_vsplth): Likewise.
2925 (altivec_vspltw): Likewise.
2926 (altivec_vspltsf): Likewise.
2928 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2931 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2932 frame related instructions.
2934 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2936 PR rtl-optimization/59959
2937 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2938 any reload of register whose subreg is invalid.
2940 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2942 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2943 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2944 Add missing return type - void.
2946 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2948 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2949 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2950 remove element index adjustment for endian (now handled in vsx.md
2952 (altivec_expand_vec_perm_const): Use
2953 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2954 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2955 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2956 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2957 define_expand and a new define_insn *altivec_vspltb_internal;
2958 adjust for -maltivec=be on a little endian target.
2959 (altivec_vspltb_direct): New.
2960 (altivec_vsplth): Divide into a define_expand and a new
2961 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2962 little endian target.
2963 (altivec_vsplth_direct): New.
2964 (altivec_vspltw): Divide into a define_expand and a new
2965 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2966 little endian target.
2967 (altivec_vspltw_direct): New.
2968 (altivec_vspltsf): Divide into a define_expand and a new
2969 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2970 a little endian target.
2972 2014-01-30 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/59993
2975 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2976 can propagate form the earlier stmt and avoid the transform
2977 when the intermediate result is needed.
2979 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2981 * README.Portability: Fix typo.
2983 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2985 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2986 comparison_operator with ordered_comparison_operator.
2988 2014-01-30 Nick Clifton <nickc@redhat.com>
2990 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2991 Rename to mn10300_store_multiple_regs.
2992 * config/mn10300/mn10300.c: Likewise.
2993 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2994 store_multiple_regs.
2995 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2996 Call mn10300_store_multiple_regs.
2998 2014-01-30 Nick Clifton <nickc@redhat.com>
2999 DJ Delorie <dj@redhat.com>
3001 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3002 %fp 2 to keep registers after it properly word-aligned.
3003 (rl78_alloc_physical_registers_umul): Handle the case where both
3004 input operands are the same.
3006 2014-01-30 Richard Biener <rguenther@suse.de>
3008 PR tree-optimization/59903
3009 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3012 2014-01-30 Jason Merrill <jason@redhat.com>
3015 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3018 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3020 2014-01-30 Richard Biener <rguenther@suse.de>
3022 PR tree-optimization/59951
3023 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3025 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3028 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3029 SFmode to DFmode case.
3031 2014-01-29 DJ Delorie <dj@redhat.com>
3033 * config/msp430/msp430.opt (-minrt): New.
3034 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3036 (ENDFILE_SPEC): Likewise.
3038 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3040 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3041 (estimate_function_body_sizes): Use it.
3043 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3046 * dwarf2out.c (is_cxx_auto): New.
3047 (is_base_type): Use it.
3048 (gen_type_die_with_usage): Likewise.
3050 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3052 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3053 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3054 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3055 -maltivec=be with LE targets.
3056 (vsx_mergeh_<mode>): Likewise.
3057 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3058 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3059 (altivec_vmrghb): Replace with define_expand and new
3060 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3061 (altivec_vmrghb_direct): New define_insn.
3062 (altivec_vmrghh): Replace with define_expand and new
3063 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3064 (altivec_vmrghh_direct): New define_insn.
3065 (altivec_vmrghw): Replace with define_expand and new
3066 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3067 (altivec_vmrghw_direct): New define_insn.
3068 (*altivec_vmrghsf): Adjust for endianness.
3069 (altivec_vmrglb): Replace with define_expand and new
3070 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3071 (altivec_vmrglb_direct): New define_insn.
3072 (altivec_vmrglh): Replace with define_expand and new
3073 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3074 (altivec_vmrglh_direct): New define_insn.
3075 (altivec_vmrglw): Replace with define_expand and new
3076 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3077 (altivec_vmrglw_direct): New define_insn.
3078 (*altivec_vmrglsf): Adjust for endianness.
3079 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3080 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3081 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3082 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3083 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3084 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3085 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3086 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3088 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3090 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3091 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3094 2014-01-29 Richard Biener <rguenther@suse.de>
3096 PR tree-optimization/58742
3097 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3098 associate_pointerplus_align.
3099 (associate_pointerplus_diff): New function.
3100 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3101 and associate_pointerplus_diff.
3103 2014-01-29 Richard Biener <rguenther@suse.de>
3105 * lto-streamer.h (LTO_major_version): Bump to 3.
3106 (LTO_minor_version): Reset to 0.
3108 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3110 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3111 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3112 (arm_file_start): Generate correct asm header for armv7ve.
3113 * config/arm/bpabi.h: Add multilib support for armv7ve.
3114 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3115 and cortex-a15 to armv7ve.
3116 * config/arm/t-aprofile: Add multilib support for armv7ve.
3117 * doc/invoke.texi: Document -march=armv7ve.
3119 2014-01-29 Richard Biener <rguenther@suse.de>
3121 PR tree-optimization/58742
3122 * tree-ssa-forwprop.c (associate_plusminus): Return true
3123 if we changed sth, defer EH cleanup to ...
3124 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3125 (simplify_mult): New function.
3127 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3130 PR tree-optimization/59920
3131 * tree.c (build_common_builtin_nodes): Remove
3132 __builtin_setjmp_dispatcher initialization.
3133 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3134 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3135 instead of gsi_after_labels + manually skipping debug stmts.
3136 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3137 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3138 * tree-inline.c (copy_edges_for_bb): Remove
3139 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3140 argument. Ignore computed_goto_p stmts. Don't call
3141 make_abnormal_goto_edges. If a call might need abnormal edges
3142 for non-local gotos, see if it already has an edge to
3143 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3144 with true argument, don't do anything then, otherwise add
3145 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3146 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3148 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3149 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3150 (lower_stmt): Don't set data->calls_builtin_setjmp.
3151 (lower_builtin_setjmp): Adjust comment.
3152 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3153 * tree-cfg.c (found_computed_goto): Remove.
3154 (factor_computed_gotos): Remove.
3155 (make_goto_expr_edges): Return bool, true for computed gotos.
3156 Don't call make_abnormal_goto_edges.
3157 (build_gimple_cfg): Don't set found_computed_goto, don't call
3158 factor_computed_gotos.
3159 (computed_goto_p): No longer static.
3160 (make_blocks): Don't set found_computed_goto.
3161 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3162 (make_edges): If make_goto_expr_edges returns true, push bb
3163 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3164 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3165 vector. Record mapping between bbs and OpenMP regions if there
3166 are any, adjust make_gimple_omp_edges caller. Call
3167 handle_abnormal_edges.
3168 (make_abnormal_goto_edges): Remove.
3169 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3170 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3171 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3172 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3173 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3174 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3175 filling *region also set *region_idx to (*region)->entry->index.
3178 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3179 For REGs set ORIGINAL_REGNO.
3181 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3183 * doc/md.texi: Mention that a target shouldn't implement
3184 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3187 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3189 PR tree-optimization/59594
3190 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3191 a copy of the datarefs vector rather than the vector itself.
3193 2014-01-28 Jason Merrill <jason@redhat.com>
3196 * dwarf2out.c (auto_die): New static.
3197 (gen_type_die_with_usage): Handle C++1y 'auto'.
3198 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3201 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3204 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3205 (SPEC_X32): Likewise.
3206 (SPEC_64): Likewise.
3207 * config/i386/i386.c (ix86_option_override_internal): Turn off
3208 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3210 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3211 * config/i386/i386.h (TARGET_16BIT): New macro.
3212 (TARGET_16BIT_P): Likewise.
3213 * config/i386/i386.opt: Add m16.
3214 * doc/invoke.texi: Document -m16.
3216 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3218 PR preprocessor/59935
3219 * input.c (location_get_source_line): Bail out on when line number
3220 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3222 2014-01-28 Richard Biener <rguenther@suse.de>
3224 PR tree-optimization/58742
3225 * tree-ssa-forwprop.c (associate_plusminus): Handle
3226 pointer subtraction of the form (T)(P + A) - (T)P.
3228 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3230 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3233 2014-01-28 Richard Biener <rguenther@suse.de>
3236 2014-01-28 Richard Biener <rguenther@suse.de>
3238 PR rtl-optimization/45364
3239 PR rtl-optimization/59890
3240 * var-tracking.c (local_get_addr_clear_given_value): Handle
3241 already cleared slot.
3242 (val_reset): Handle not allocated local_get_addr_cache.
3243 (vt_find_locations): Use post-order on the inverted CFG.
3245 2014-01-28 Richard Biener <rguenther@suse.de>
3247 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3249 2014-01-28 Richard Biener <rguenther@suse.de>
3251 PR rtl-optimization/45364
3252 PR rtl-optimization/59890
3253 * var-tracking.c (local_get_addr_clear_given_value): Handle
3254 already cleared slot.
3255 (val_reset): Handle not allocated local_get_addr_cache.
3256 (vt_find_locations): Use post-order on the inverted CFG.
3258 2014-01-28 Alan Modra <amodra@gmail.com>
3260 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3261 * configure.ac <recursive call for build != host>: Define
3262 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3263 and LD_FOR_BUILD too.
3264 * configure: Regenerate.
3266 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3268 * config/i386/i386.c (get_builtin_code_for_version): Separate
3269 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3270 Broadwell from Haswell.
3272 2014-01-27 Steve Ellcey <sellcey@mips.com>
3274 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3275 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3276 * config/mips/mips.c (mips_option_override): Change setting
3278 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3279 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3280 Change from Mask to Var.
3282 2014-01-27 Jeff Law <law@redhat.com>
3284 * ipa-inline.c (inline_small_functions): Fix typo.
3286 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3288 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3289 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3290 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3291 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3292 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3293 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3294 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3295 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3296 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3297 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3298 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3299 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3300 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3301 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3302 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3303 (_mm512_storeu_epi64): Ditto.
3304 (_mm512_cmpge_epi32_mask): Ditto.
3305 (_mm512_cmpge_epu32_mask): Ditto.
3306 (_mm512_cmpge_epi64_mask): Ditto.
3307 (_mm512_cmpge_epu64_mask): Ditto.
3308 (_mm512_cmple_epi32_mask): Ditto.
3309 (_mm512_cmple_epu32_mask): Ditto.
3310 (_mm512_cmple_epi64_mask): Ditto.
3311 (_mm512_cmple_epu64_mask): Ditto.
3312 (_mm512_cmplt_epi32_mask): Ditto.
3313 (_mm512_cmplt_epu32_mask): Ditto.
3314 (_mm512_cmplt_epi64_mask): Ditto.
3315 (_mm512_cmplt_epu64_mask): Ditto.
3316 (_mm512_cmpneq_epi32_mask): Ditto.
3317 (_mm512_cmpneq_epu32_mask): Ditto.
3318 (_mm512_cmpneq_epi64_mask): Ditto.
3319 (_mm512_cmpneq_epu64_mask): Ditto.
3320 (_mm512_expand_pd): Ditto.
3321 (_mm512_expand_ps): Ditto.
3322 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3323 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3324 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3325 * config/i386/i386.c (ix86_builtins): Add
3326 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3327 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3328 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3329 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3330 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3331 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3332 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3333 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3334 IX86_BUILTIN_PMOVUSQW512_MEM.
3335 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3336 __builtin_ia32_pmovsqd512mem_mask,
3337 __builtin_ia32_pmovqd512mem_mask,
3338 __builtin_ia32_pmovusqw512mem_mask,
3339 __builtin_ia32_pmovsqw512mem_mask,
3340 __builtin_ia32_pmovqw512mem_mask,
3341 __builtin_ia32_pmovusdw512mem_mask,
3342 __builtin_ia32_pmovsdw512mem_mask,
3343 __builtin_ia32_pmovdw512mem_mask,
3344 __builtin_ia32_pmovqb512mem_mask,
3345 __builtin_ia32_pmovusqb512mem_mask,
3346 __builtin_ia32_pmovsqb512mem_mask,
3347 __builtin_ia32_pmovusdb512mem_mask,
3348 __builtin_ia32_pmovsdb512mem_mask,
3349 __builtin_ia32_pmovdb512mem_mask.
3350 (bdesc_args): Add __builtin_ia32_expanddf512,
3351 __builtin_ia32_expandsf512.
3352 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3353 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3354 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3355 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3356 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3357 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3358 (avx512f_<code>v8div16qi2_mask_store): This.
3359 (avx512f_expand<mode>): New.
3361 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3363 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3365 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3366 (_mm512_prefetch_i32scatter_pd): Ditto.
3367 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3368 (_mm512_prefetch_i64scatter_pd): Ditto.
3369 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3370 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3371 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3372 (_mm512_prefetch_i32scatter_ps): Ditto.
3373 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3374 (_mm512_prefetch_i64scatter_ps): Ditto.
3375 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3376 * config/i386/i386-builtin-types.def: Define
3377 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3378 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3379 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3380 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3381 IX86_BUILTIN_SCATTERPFQPD.
3382 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3383 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3384 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3385 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3386 __builtin_ia32_scatterpfqps.
3387 (ix86_expand_builtin): Expand new built-ins.
3388 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3389 fix memory access data type.
3390 (*avx512pf_gatherpf<mode>_mask): Ditto.
3391 (*avx512pf_gatherpf<mode>): Ditto.
3392 (avx512pf_scatterpf<mode>): Ditto.
3393 (*avx512pf_scatterpf<mode>_mask): Ditto.
3394 (*avx512pf_scatterpf<mode>): Ditto.
3395 (GATHER_SCATTER_SF_MEM_MODE): New.
3396 (avx512pf_gatherpf<mode>df): Ditto.
3397 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3398 (*avx512pf_scatterpf<mode>df): Ditto.
3400 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3403 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3404 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3407 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3409 * common/config/arm/arm-common.c
3410 (arm_rewrite_mcpu): Handle multiple names.
3412 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3414 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3416 * gimple-builder.h (create_gimple_tmp): Delete.
3418 2014-01-27 Christian Bruel <christian.bruel@st.com>
3420 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3423 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3425 * config/pa/pa.md (call): Generate indirect long calls to non-local
3426 functions when outputing 32-bit code.
3427 (call_value): Likewise except for special call to buggy powf function.
3429 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3430 portable runtime and PIC indirect calls.
3431 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3432 and PIC call sequences. Use ldo instead of blr to set return register
3433 in PIC call sequence.
3435 2014-01-25 Walter Lee <walt@tilera.com>
3437 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3438 avoid clobbering a live register.
3440 2014-01-25 Walter Lee <walt@tilera.com>
3442 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3443 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3444 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3445 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3447 2014-01-25 Walter Lee <walt@tilera.com>
3449 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3450 arguments on even registers.
3451 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3453 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3454 (BIGGEST_ALIGNMENT): Ditto.
3455 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3457 2014-01-25 Walter Lee <walt@tilera.com>
3459 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3460 insns before bundling.
3461 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3463 2014-01-25 Walter Lee <walt@tilera.com>
3465 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3466 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3467 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3469 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3471 * config/mips/constraints.md (kl): Delete.
3472 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3473 define expands, using...
3474 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3475 instructions for MIPS16.
3476 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3477 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3479 2014-01-25 Walter Lee <walt@tilera.com>
3481 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3485 2014-01-25 Walter Lee <walt@tilera.com>
3487 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3488 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3490 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3492 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3495 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3497 * print-rtl.c (in_call_function_usage): New var.
3498 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3499 EXPR_LIST mode as mode and not as reg note name.
3502 * cfgloopmanip.c (copy_loop_info): If
3503 loop->warned_aggressive_loop_optimizations, make sure
3504 the flag is set in target loop too.
3506 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3508 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3510 * builtins.def: Likewise.
3511 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3512 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3513 * ira.c (ira_setup_eliminable_regset): Likewise.
3514 * omp-low.c (gate_expand_omp): Likewise.
3515 (execute_lower_omp): Likewise.
3516 (diagnose_sb_0): Likewise.
3517 (gate_diagnose_omp_blocks): Likewise.
3518 (simd_clone_clauses_extract): Likewise.
3521 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3523 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3524 correction for little endian...
3525 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3528 2014-01-24 Jeff Law <law@redhat.com>
3530 PR tree-optimization/59919
3531 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3532 for non-returning calls.
3534 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3536 * common/config/aarch64/aarch64-common.c
3537 (aarch64_rewrite_mcpu): Handle multiple names.
3538 * config/aarch64/aarch64.h
3539 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3541 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3543 * input.c (add_file_to_cache_tab): Handle the case where fopen
3546 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3549 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3550 from push operand if code of push isn't PRE_DEC.
3552 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3555 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3556 -mquad-memory-atomic. Update -mquad-memory documentation to say
3557 it is only used for non-atomic loads/stores.
3559 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3560 -mquad-memory or -mquad-memory-atomic switches.
3562 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3563 -mquad-memory-atomic to ISA 2.07 support.
3565 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3566 to separate support of normal quad word memory operations (ldq, stq)
3567 from the atomic quad word memory operations.
3569 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3570 support to separate non-atomic quad word operations from atomic
3571 quad word operations. Disable non-atomic quad word operations in
3572 little endian mode so that we don't have to swap words after the
3573 load and before the store.
3574 (quad_load_store_p): Add comment about atomic quad word support.
3575 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3576 options printed with -mdebug=reg.
3578 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3579 -mquad-memory-atomic as the test for whether we have quad word
3580 atomic instructions.
3581 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3582 or -mp8-vector are used, allow byte/half-word atomic operations.
3584 * config/rs6000/sync.md (load_lockedti): Insure that the address
3585 is a proper indexed or indirect address for the lqarx instruction.
3586 On little endian systems, swap the hi/lo registers after the lqarx
3588 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3589 insure the address is valid for the lqarx instruction.
3590 (store_conditionalti): Insure that the address is a proper indexed
3591 or indirect address for the stqcrx. instruction. On little endian
3592 systems, swap the hi/lo registers before doing the stqcrx.
3594 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3595 insure the address is valid for the stqcrx. instruction.
3597 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3598 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3599 type of quad memory support is available.
3601 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3604 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3605 there is a danger of looping.
3607 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3610 force flag_ira_loop_pressure if set via command line.
3612 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3614 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3615 (ashr_simd): New builtin handling DI mode.
3616 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3617 (aarch64_sshr_simddi): New match pattern.
3618 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3619 (vshrd_n_s64): Likewise.
3620 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3622 2014-01-23 Nick Clifton <nickc@redhat.com>
3624 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3625 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3626 favour of mcu specific scripts.
3627 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3630 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3631 Alex Velenko <Alex.Velenko@arm.com>
3633 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3634 (vaddv_s16): Likewise.
3635 (vaddv_s32): Likewise.
3636 (vaddv_u8): Likewise.
3637 (vaddv_u16): Likewise.
3638 (vaddv_u32): Likewise.
3639 (vaddvq_s8): Likewise.
3640 (vaddvq_s16): Likewise.
3641 (vaddvq_s32): Likewise.
3642 (vaddvq_s64): Likewise.
3643 (vaddvq_u8): Likewise.
3644 (vaddvq_u16): Likewise.
3645 (vaddvq_u32): Likewise.
3646 (vaddvq_u64): Likewise.
3647 (vaddv_f32): Likewise.
3648 (vaddvq_f32): Likewise.
3649 (vaddvq_f64): Likewise.
3650 (vmaxv_f32): Likewise.
3651 (vmaxv_s8): Likewise.
3652 (vmaxv_s16): Likewise.
3653 (vmaxv_s32): Likewise.
3654 (vmaxv_u8): Likewise.
3655 (vmaxv_u16): Likewise.
3656 (vmaxv_u32): Likewise.
3657 (vmaxvq_f32): Likewise.
3658 (vmaxvq_f64): Likewise.
3659 (vmaxvq_s8): Likewise.
3660 (vmaxvq_s16): Likewise.
3661 (vmaxvq_s32): Likewise.
3662 (vmaxvq_u8): Likewise.
3663 (vmaxvq_u16): Likewise.
3664 (vmaxvq_u32): Likewise.
3665 (vmaxnmv_f32): Likewise.
3666 (vmaxnmvq_f32): Likewise.
3667 (vmaxnmvq_f64): Likewise.
3668 (vminv_f32): Likewise.
3669 (vminv_s8): Likewise.
3670 (vminv_s16): Likewise.
3671 (vminv_s32): Likewise.
3672 (vminv_u8): Likewise.
3673 (vminv_u16): Likewise.
3674 (vminv_u32): Likewise.
3675 (vminvq_f32): Likewise.
3676 (vminvq_f64): Likewise.
3677 (vminvq_s8): Likewise.
3678 (vminvq_s16): Likewise.
3679 (vminvq_s32): Likewise.
3680 (vminvq_u8): Likewise.
3681 (vminvq_u16): Likewise.
3682 (vminvq_u32): Likewise.
3683 (vminnmv_f32): Likewise.
3684 (vminnmvq_f32): Likewise.
3685 (vminnmvq_f64): Likewise.
3687 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3689 * config/aarch64/aarch64-simd.md
3690 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3691 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3692 (*aarch64_mul3_elt<mode>): Likewise.
3693 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3694 (*aarch64_mul3_elt_to_64v2df): Likewise.
3695 (*aarch64_mla_elt<mode>): Likewise.
3696 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3697 (*aarch64_mls_elt<mode>): Likewise.
3698 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3699 (*aarch64_fma4_elt<mode>): Likewise.
3700 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3701 (*aarch64_fma4_elt_to_64v2df): Likewise.
3702 (*aarch64_fnma4_elt<mode>): Likewise.
3703 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3704 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3705 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3706 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3707 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3708 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3709 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3710 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3711 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3713 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3715 * config/aarch64/aarch64-simd.md
3716 (aarch64_be_checked_get_lane<mode>): New define_expand.
3717 * config/aarch64/aarch64-simd-builtins.def
3718 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3719 New builtin definition.
3720 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3721 Use new safe be builtin.
3723 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3725 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3727 (aarch64_be_st1<mode>): Likewise.
3728 (aarch_ld1<VALL:mode>): Define_expand modified.
3729 (aarch_st1<VALL:mode>): Likewise.
3730 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3731 (UNSPEC_ST1): Likewise.
3733 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3735 * config/microblaze/microblaze.md: Add trap insn and attribute
3737 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3739 PR preprocessor/58580
3740 * input.h (location_get_source_line): Take an additional line_size
3742 (void diagnostics_file_cache_fini): Declare new function.
3743 * input.c (struct fcache): New type.
3744 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3745 New static constants.
3746 (diagnostic_file_cache_init, total_lines_num)
3747 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3748 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3749 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3750 (get_next_line, read_next_line, goto_next_line, read_line_num):
3751 New static function definitions.
3752 (diagnostic_file_cache_fini): New function.
3753 (location_get_source_line): Take an additional output line_len
3754 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3756 * diagnostic.c (diagnostic_finish): Call
3757 diagnostic_file_cache_fini.
3758 (adjust_line): Take an additional input parameter for the length
3759 of the line, rather than calculating it with strlen.
3760 (diagnostic_show_locus): Adjust the use of
3761 location_get_source_line and adjust_line with respect to their new
3762 signature. While displaying a line now, do not stop at the first
3763 null byte. Rather, display the zero byte as a space and keep
3764 going until we reach the size of the line.
3765 * Makefile.in: Add vec.o to OBJS-libcommon
3767 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3768 Ilya Tocar <ilya.tocar@intel.com>
3770 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3771 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3772 (__builtin_ia32_kmov16): Ditto.
3773 * config/i386/i386.md (UNSPEC_KMOV): New.
3776 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3778 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3779 (_mm512_storeu_si512): Ditto.
3781 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3784 * rtl.h (get_referenced_operands): Declare.
3785 * recog.c (get_referenced_operands): New function.
3786 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3787 operands have been referenced when recording LO_SUM references.
3789 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3791 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3793 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3795 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3796 Enable for generic and recent AMD targets.
3798 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3800 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3801 ARG_SIZE note when adjustment was eliminated.
3803 2014-01-22 Jeff Law <law@redhat.com>
3805 PR tree-optimization/59597
3806 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3807 in file. Accept new argument REGISTERING and use it to modify
3808 dump output appropriately.
3809 (register_jump_thread): Corresponding changes.
3810 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3811 thread paths involving joiner blocks. Add code to dump cancelled
3812 jump threading paths.
3814 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3816 PR rtl-optimization/59477
3817 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3818 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3820 2014-01-22 Tom Tromey <tromey@redhat.com>
3822 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3824 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3826 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3828 PR rtl-optimization/59896
3829 * lra-constraints.c (process_alt_operands): Check unused note for
3830 matched operands of insn with no output reloads.
3832 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3834 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3835 (mips_move_from_gpr_cost): Likewise.
3837 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3839 PR rtl-optimization/59858
3840 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3841 ira_class_hard_regs_num.
3842 (process_alt_operands): Increase reject for dying matched operand.
3844 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3847 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3848 smaller than size, perform several stores or loads and stores
3849 at dst + count - size to store or copy all of size bytes, rather
3850 than just last modesize bytes.
3852 2014-01-20 DJ Delorie <dj@redhat.com>
3854 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3855 that CLOBBERs are REGs before propogating their values.
3857 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3860 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3861 (cgraph_inline_failed_type): New function.
3862 * cgraph.h (DEFCIFCODE): Add type.
3863 (cgraph_inline_failed_type_t): New enum.
3864 (cgraph_inline_failed_type): New prototype.
3865 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3866 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3867 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3868 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3869 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3870 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3871 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3872 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3873 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3874 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3875 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3876 OPTIMIZATION_MISMATCH.
3877 * tree-inline.c (expand_call_inline): Emit errors during
3878 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3880 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3883 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3884 mode attribute in insn output.
3886 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3888 * output.h (output_constant): Delete.
3889 * varasm.c (output_constant): Make private.
3891 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3893 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3895 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3898 * tree.h (fold_builtin_strcat): New prototype.
3899 * builtins.c (fold_builtin_strcat): No longer static. Add len
3900 argument, if non-NULL, don't call c_strlen. Optimize
3901 directly into __builtin_memcpy instead of __builtin_strcpy.
3902 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3903 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3905 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3907 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3908 for SImode_address_operand operands, having only a REG argument.
3910 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3912 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3913 loader name using mbig-endian.
3914 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3916 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3918 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3922 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3924 * config/aarch64/aarch64-protos.h
3925 (aarch64_cannot_change_mode_class_ptr): Declare.
3926 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3927 aarch64_cannot_change_mode_class_ptr): New.
3928 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3929 backend hook aarch64_cannot_change_mode_class.
3931 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3933 * common/config/aarch64/aarch64-common.c
3934 (aarch64_handle_option): Don't handle any option order logic here.
3935 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3936 selected_cpu, warn on architecture version mismatch.
3937 (aarch64_override_options): Fix parsing order for option strings.
3939 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3940 Iain Sandoe <iain@codesourcery.com>
3943 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3944 warning. Amend comment to reflect current functionality.
3946 2014-01-20 Richard Biener <rguenther@suse.de>
3949 * builtins.c (fold_builtin_strcat): Remove case better handled
3950 by tree-ssa-strlen.c.
3952 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3954 * config/aarch64/aarch64.opt
3955 (mcpu, march, mtune): Make case-insensitive.
3957 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3960 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3961 if operands[1] is a REG or ZERO_EXTEND of a REG.
3963 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3965 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3967 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3969 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3970 long non-pic millicode calls.
3972 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3974 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3976 2014-01-19 Kito Cheng <kito@0xlab.org>
3978 * builtins.c (expand_movstr): Check movstr expand done or fail.
3980 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3981 H.J. Lu <hongjiu.lu@intel.com>
3984 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3985 to DImode for zero-extended addresses.
3987 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3989 PR rtl-optimization/57763
3990 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3991 on the new indirect jump_insn and increment LABEL_NUSES (label).
3993 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3997 * config.gcc (x86_archs): New variable.
3998 (x86_64_archs): Likewise.
3999 (x86_cpus): Likewise.
4000 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4001 --with-arch/--with-cpu= options.
4002 Support --with-arch=/--with-cpu={nehalem,westmere,
4003 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4005 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4007 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4008 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4010 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4012 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4014 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4017 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4018 clear cpp_get_options (parse_in)->warn_unused_macros for
4019 ix86_target_macros_internal with cpp_define.
4021 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4023 * jump.c (delete_related_insns): Keep (use (insn))s.
4024 * reorg.c (redundant_insn): Check for barriers too.
4026 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4028 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4030 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4032 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4033 call to $$dyncall when TARGET_LONG_CALLS is true.
4035 2014-01-17 Jeff Law <law@redhat.com>
4037 * ree.c (combine_set_extension): Temporarily disable test for
4038 changing number of hard registers.
4040 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4043 * ipa-inline-analysis.c (inline_free_summary):
4044 Do not free summary of aliases.
4046 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4049 * gimplify.c (gimplify_expr): Use create_tmp_var
4050 instead of create_tmp_var_raw. If cond doesn't have
4051 integral type, don't add the IFN_ANNOTATE builtin at all.
4053 2014-01-17 Martin Jambor <mjambor@suse.cz>
4056 * ipa-cp.c (prev_edge_clone): New variable.
4057 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4058 Also resize prev_edge_clone vector.
4059 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4060 (ipcp_edge_removal_hook): New function.
4061 (ipcp_driver): Register ipcp_edge_removal_hook.
4063 2014-01-17 Andrew Pinski <apinski@cavium.com>
4064 Steve Ellcey <sellcey@mips.com>
4067 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4070 2014-01-17 Jeff Law <law@redhat.com>
4073 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4074 so that pass_ccp runs first.
4076 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4078 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4079 (ix86_adjust_cost): Use !TARGET_XXX.
4080 (do_reorder_for_imul): Likewise.
4081 (swap_top_of_ready_list): Likewise.
4082 (ix86_sched_reorder): Likewise.
4084 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4086 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4087 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4088 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4089 (intel_memset): New. Duplicate slm_memset.
4090 (intel_cost): New. Duplicate slm_cost.
4091 (m_INTEL): New macro.
4092 (processor_target_table): Add "intel".
4093 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4094 with PROCESSOR_INTEL for "intel".
4095 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4096 PROCESSOR_SILVERMONT.
4097 (ix86_issue_rate): Likewise.
4098 (ix86_adjust_cost): Likewise.
4099 (ia32_multipass_dfa_lookahead): Likewise.
4100 (swap_top_of_ready_list): Likewise.
4101 (ix86_sched_reorder): Likewise.
4102 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4103 instead of TARGET_OPT_AGU.
4104 * config/i386/i386.h (TARGET_INTEL): New.
4105 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4106 (processor_type): Add PROCESSOR_INTEL.
4107 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4108 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4110 2014-01-17 Marek Polacek <polacek@redhat.com>
4113 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4116 2014-01-17 Richard Biener <rguenther@suse.de>
4118 PR tree-optimization/46590
4119 * opts.c (default_options_table): Add entries for
4120 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4121 all enabled at -O1 but not for -Og.
4122 * common.opt (fbranch-count-reg): Remove Init(1).
4123 (fmove-loop-invariants): Likewise.
4124 (ftree-pta): Likewise.
4126 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4128 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4129 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4130 decls to at least the GCC 4.8 used alignments.
4133 * tree-nested.c (convert_nonlocal_reference_stmt,
4134 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4135 of GIMPLE_BIND stmts, adjust associated decls.
4137 2014-01-17 Richard Biener <rguenther@suse.de>
4139 PR tree-optimization/46590
4140 * vec.h (vec<>::bseach): New member function implementing
4141 binary search according to C89 bsearch.
4142 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4143 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4144 bitmap pointer again. Make accesses_in_loop a flat array.
4145 (mem_ref_obstack): New global.
4146 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4147 (mark_ref_stored): Likewise.
4148 (ref_indep_loop_p_2): Likewise.
4149 (set_ref_stored_in_loop): New helper function.
4150 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4151 (memref_free): Adjust.
4152 (record_mem_ref_loc): Simplify.
4153 (gather_mem_refs_stmt): Adjust.
4154 (sort_locs_in_loop_postorder_cmp): New function.
4155 (analyze_memory_references): Sort accesses_in_loop after
4156 loop postorder number.
4157 (find_ref_loc_in_loop_cmp): New function.
4158 (for_all_locs_in_loop): Find relevant cluster of locs in
4159 accesses_in_loop and iterate without recursion.
4160 (execute_sm): Avoid uninit warning.
4161 (struct ref_always_accessed): Simplify.
4162 (ref_always_accessed::operator ()): Likewise.
4163 (ref_always_accessed_p): Likewise.
4164 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4165 loop postorder numbers here.
4166 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4169 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4172 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4173 on decls for which assemble_alias has been called.
4175 2014-01-17 Nick Clifton <nickc@redhat.com>
4177 * config/msp430/msp430.opt: (mcpu): New option.
4178 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4179 (msp430_option_override): Parse target_cpu. If the MCU name
4180 matches a generic string, clear target_mcu.
4181 (msp430_attr): Allow numeric interrupt values up to 63.
4182 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4183 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4185 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4187 * config/msp430/msp430.md (popm): Use %J rather than %I.
4188 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4189 (addhi_cy_i): Use immediate_operand for operand 2.
4190 * doc/invoke.texi: Document -mcpu option.
4192 2014-01-17 Richard Biener <rguenther@suse.de>
4194 PR rtl-optimization/38518
4195 * df.h (df_analyze_loop): Declare.
4196 * df-core.c: Include cfgloop.h.
4197 (df_analyze_1): Split out main part of df_analyze.
4198 (df_analyze): Adjust.
4199 (loop_inverted_post_order_compute): New function.
4200 (loop_post_order_compute): Likewise.
4201 (df_analyze_loop): New function avoiding whole-function
4203 * loop-invariant.c (find_defs): Use df_analyze_loop.
4204 (find_invariants): Adjust.
4205 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4207 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4209 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4210 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4212 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4214 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4215 traversal when removing references.
4217 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4220 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4222 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4225 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4226 pushing a reload for an autoinc when we had previously reloaded an
4227 inner part of the address.
4229 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4231 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4233 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4234 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4235 when not giving up or versioning for alias only because of
4237 (vect_analyze_data_ref_dependences): Set to true.
4238 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4240 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4241 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4245 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4248 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4249 operand 0 predicate for gathers, use a new pseudo as subtarget.
4251 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4254 * lra-constraints.c (process_alt_operands): Add printing debug info.
4255 Check absence of input/output reloads for matched operands too.
4257 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4259 PR rtl-optimization/59835
4260 * ira.c (ira_init_register_move_cost): Increase cost for
4263 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4265 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4267 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4270 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4271 non-register objects. Use gen_(high/low)part more consistently.
4274 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4277 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4278 endian support, remove tests for WORDS_BIG_ENDIAN.
4279 (p8_mfvsrd_3_<mode>): Likewise.
4280 (reload_gpr_from_vsx<mode>): Likewise.
4281 (reload_gpr_from_vsxsf): Likewise.
4282 (p8_mfvsrd_4_disf): Likewise.
4284 2014-01-16 Richard Biener <rguenther@suse.de>
4286 PR rtl-optimization/46590
4287 * lcm.c (compute_antinout_edge): Use postorder iteration.
4288 (compute_laterin): Use inverted postorder iteration.
4290 2014-01-16 Nick Clifton <nickc@redhat.com>
4293 * varasm.c (output_constant): Return the number of bytes actually
4295 (output_constructor_array_range): Update the field size with the
4296 number of bytes emitted by output_constant.
4297 (output_constructor_regular_field): Likewise. Also do not
4298 complain if the total number of bytes emitted is now greater
4299 than the expected fieldpos.
4300 * output.h (output_constant): Update prototype and descriptive comment.
4302 2014-01-16 Marek Polacek <polacek@redhat.com>
4305 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4306 it is error_mark_node.
4308 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4310 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4311 VALID_AVX256_REG_OR_OI_MODE.
4313 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4315 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4316 current procedure should be profiled.
4318 2014-01-15 Andrew Pinski <apinski@cavium.com>
4320 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4321 of moving from/to the STACK_REG register class.
4323 2014-01-15 Richard Henderson <rth@redhat.com>
4326 * reginfo.c (global_regs_decl): Globalize.
4327 * rtl.h (global_regs_decl): Declare.
4328 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4329 reserved via global_regs.
4331 2014-01-15 Teresa Johnson <tejohnson@google.com>
4333 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4335 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4337 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4338 and vmulosh rather than call gen_vec_widen_smult_*.
4339 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4340 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4341 (vec_widen_smult_even_v16qi): Likewise.
4342 (vec_widen_umult_even_v8hi): Likewise.
4343 (vec_widen_smult_even_v8hi): Likewise.
4344 (vec_widen_umult_odd_v16qi): Likewise.
4345 (vec_widen_smult_odd_v16qi): Likewise.
4346 (vec_widen_umult_odd_v8hi): Likewise.
4347 (vec_widen_smult_odd_v8hi): Likewise.
4348 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4349 vmuloub rather than call gen_vec_widen_umult_*.
4350 (vec_widen_umult_lo_v16qi): Likewise.
4351 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4352 vmulosb rather than call gen_vec_widen_smult_*.
4353 (vec_widen_smult_lo_v16qi): Likewise.
4354 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4355 rather than call gen_vec_widen_umult_*.
4356 (vec_widen_umult_lo_v8hi): Likewise.
4357 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4358 rather than call gen_vec_widen_smult_*.
4359 (vec_widen_smult_lo_v8hi): Likewise.
4361 2014-01-15 Jeff Law <law@redhat.com>
4363 PR tree-optimization/59747
4364 * ree.c (find_and_remove_re): Properly handle case where a second
4365 eliminated extension requires widening a copy created for elimination
4366 of a prior extension.
4367 (combine_set_extension): Ensure that the number of hard regs needed
4368 for a destination register does not change when we widen it.
4370 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4372 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4373 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4374 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4375 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4376 (avr-*-rtems*): Likewise.
4377 (bfin*-rtems*): Likewise.
4378 (moxie-*-rtems*): Likewise.
4379 (h8300-*-rtems*): Likewise.
4380 (i[34567]86-*-rtems*): Likewise.
4381 (lm32-*-rtems*): Likewise.
4382 (m32r-*-rtems*): Likewise.
4383 (m68k-*-rtems*): Likewise.
4384 (microblaze*-*-rtems*): Likewise.
4385 (mips*-*-rtems*): Likewise.
4386 (powerpc-*-rtems*): Likewise.
4387 (sh-*-rtems*): Likewise.
4388 (sparc-*-rtems*): Likewise.
4389 (sparc64-*-rtems*): Likewise.
4390 (v850-*-rtems*): Likewise.
4391 (m32c-*-rtems*): Likewise.
4393 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4395 PR rtl-optimization/59511
4396 * ira.c (ira_init_register_move_cost): Use memory costs for some
4397 cases of register move cost calculations.
4398 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4399 instead of BB frequency.
4400 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4401 * lra-assigns.c (find_hard_regno_for): Ditto.
4403 2014-01-15 Richard Biener <rguenther@suse.de>
4405 PR tree-optimization/59822
4406 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4407 (vectorizable_load): Use it to hoist defs of uses of invariant
4408 loads out of the loop.
4410 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4411 Kugan Vivekanandarajah <kuganv@linaro.org>
4414 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4417 2014-01-15 Richard Biener <rguenther@suse.de>
4419 PR rtl-optimization/59802
4420 * lcm.c (compute_available): Use inverted postorder to seed
4421 the initial worklist.
4423 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4426 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4427 ADDR_REGS for invalid symrefs in non-PIC code.
4429 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4432 * builtins.c (determine_block_size): Initialize *probable_max_size
4433 even if len_rtx is CONST_INT.
4435 2014-01-14 Andrew Pinski <apinski@cavium.com>
4437 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4438 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4439 (cortexa53_tunings): Likewise.
4440 (aarch64_sched_issue_rate): New function.
4441 (TARGET_SCHED_ISSUE_RATE): Define.
4443 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4445 * ira-costs.c (find_costs_and_classes): Add missed
4446 ira_init_register_move_cost_if_necessary.
4448 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4451 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4453 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4456 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4457 to indicate if type is used for function return value. Warn ABI
4458 change if the vector mode isn't available for function return value.
4459 (ix86_function_arg_advance): Pass false to type_natural_mode.
4460 (ix86_function_arg): Likewise.
4461 (ix86_gimplify_va_arg): Likewise.
4462 (function_arg_32): Don't warn ABI change.
4463 (ix86_function_value): Pass true to type_natural_mode.
4464 (ix86_return_in_memory): Likewise.
4465 (ix86_struct_value_rtx): Removed.
4466 (TARGET_STRUCT_VALUE_RTX): Likewise.
4468 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4470 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4471 converting a conditional jump into a conditional return.
4473 2014-01-14 Richard Biener <rguenther@suse.de>
4475 PR tree-optimization/58921
4476 PR tree-optimization/59006
4477 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4478 hoisting invariant stmts.
4479 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4480 invariant loads on the preheader edge if possible.
4482 2014-01-14 Joey Ye <joey.ye@arm.com>
4484 * doc/plugin.texi (Building GCC plugins): Update to C++.
4486 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4488 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4489 (_mm_rcp28_round_ss): Ditto.
4490 (_mm_rsqrt28_round_sd): Ditto.
4491 (_mm_rsqrt28_round_ss): Ditto.
4492 (_mm_rcp28_sd): Ditto.
4493 (_mm_rcp28_ss): Ditto.
4494 (_mm_rsqrt28_sd): Ditto.
4495 (_mm_rsqrt28_ss): Ditto.
4496 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4497 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4498 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4499 (IX86_BUILTIN_RCP28SD): Ditto.
4500 (IX86_BUILTIN_RCP28SS): Ditto.
4501 (IX86_BUILTIN_RSQRT28SD): Ditto.
4502 (IX86_BUILTIN_RSQRT28SS): Ditto.
4503 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4504 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4505 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4506 (ix86_expand_special_args_builtin): Expand new FTYPE.
4507 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4508 (srcp14<mode>): Make insn unary.
4509 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4510 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4511 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4512 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4513 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4514 Fix rounding: make it SAE only.
4515 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4517 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4519 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4520 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4521 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4522 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4523 (round_saeonly_mask_scalar_operand4): Ditto.
4524 (round_saeonly_mask_scalar_op3): Ditto.
4525 (round_saeonly_mask_scalar_op4): Ditto.
4527 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4530 Implement -maltivec=be for vec_insert and vec_extract.
4532 2014-01-10 DJ Delorie <dj@redhat.com>
4534 * config/msp430/msp430.md (call_internal): Don't allow memory
4535 references with SP as the base register.
4536 (call_value_internal): Likewise.
4537 * config/msp430/constraints.md (Yc): New. For memory references
4538 that don't use SP as a base register.
4540 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4541 "an integer without a # prefix"
4542 * config/msp430/msp430.md (epilogue_helper): Use it.
4544 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4547 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4548 AVX512F gather builtins.
4549 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4550 on gather decls with INTEGER_TYPE masktype.
4551 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4552 directly into the builtin rather than hoisting it before loop.
4554 PR tree-optimization/59387
4555 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4556 (scev_const_prop): If folded_casts and type has undefined overflow,
4557 use force_gimple_operand instead of force_gimple_operand_gsi and
4558 for each added stmt if it is assign with
4559 arith_code_with_undefined_signed_overflow, call
4560 rewrite_to_defined_overflow.
4561 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4562 gimple-fold.h instead.
4563 (arith_code_with_undefined_signed_overflow,
4564 rewrite_to_defined_overflow): Moved to ...
4565 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4566 rewrite_to_defined_overflow): ... here. No longer static.
4567 Include gimplify-me.h.
4568 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4569 rewrite_to_defined_overflow): New prototypes.
4571 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4573 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4575 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4577 * builtins.c (get_object_alignment_2): Minor tweak.
4578 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4580 2014-01-13 Christian Bruel <christian.bruel@st.com>
4582 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4583 optimized non constant lengths.
4585 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4588 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4589 load as __atomic_load_N if possible.
4591 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4593 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4595 (rs6000_expand_builtin): Adjust call.
4597 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4600 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4601 * config/rs6000/rs6000.c: Include target-globals.h.
4602 (rs6000_set_current_function): Instead of doing target_reinit
4603 unconditionally, use save_target_globals_default_opts and
4604 restore_target_globals.
4606 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4608 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4609 (rs6000_expand_builtin): Handle mffs and mtfsf.
4610 (rs6000_init_builtins): Define mffs and mtfsf.
4611 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4612 (rs6000_mffs): New pattern.
4613 (rs6000_mtfsf): New pattern.
4615 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4617 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4618 Start narrowing with START. Apply candidate-use pair
4619 and check overall cost in narrowing.
4620 (iv_ca_prune): Pass new argument.
4622 2014-01-10 Jeff Law <law@redhat.com>
4625 * ree.c (combine_reaching_defs): Ensure the defining statement
4626 occurs before the extension when optimizing extensions with
4627 different source and destination hard registers.
4629 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4632 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4633 vtables into the type inheritance graph.
4635 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4637 PR rtl-optimization/59754
4638 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4639 modes in the REGNO != REGNO case.
4641 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4643 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4645 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4647 PR tree-optimization/59745
4648 * tree-predcom.c (tree_predictive_commoning_loop): Call
4649 free_affine_expand_cache if giving up because components is NULL.
4651 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4652 GC in payload of target_globals struct instead of allocating them on
4653 the heap and the larger structs separately using GC.
4654 * target-globals.h (struct target_globals): Make regs, hard_regs,
4655 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4656 of GTY((skip)) and change type to void *.
4657 (reset_target_globals): Cast loads from those fields to corresponding
4660 2014-01-10 Steve Ellcey <sellcey@mips.com>
4663 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4664 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4665 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4667 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4670 * aarch64-modes.def (CC_Zmode): New flags mode.
4671 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4672 represents an equality.
4673 (aarch64_get_condition_code): Handle CC_Zmode.
4674 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4676 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4678 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4679 extraction in good case.
4681 2014-01-10 Richard Biener <rguenther@suse.de>
4683 PR tree-optimization/59374
4684 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4685 checking after SLP discovery. Mark stmts not participating
4686 in any SLP instance properly.
4688 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4690 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4691 when handling a SET rtx.
4693 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4695 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4696 (cortex-a57): Likewise.
4697 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4699 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4701 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4702 non-iwmmxt builtins.
4704 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4708 * ipa-devirt.c record_target_from_binfo): Take as argument
4709 stack of binfos and lookup matching one for virtual inheritance.
4710 (possible_polymorphic_call_targets_1): Update.
4712 2014-01-10 Huacai Chen <chenhc@lemote.com>
4714 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4715 kernel strings for Loongson-2E/2F/3A.
4717 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4720 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4721 is_gimple_call before calling gimple_call_internal_p.
4723 2014-01-09 Steve Ellcey <sellcey@mips.com>
4725 * Makefile.in (TREE_FLOW_H): Remove.
4726 (TREE_SSA_H): Add file names from tree-flow.h.
4727 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4728 * tree.h: Remove tree-flow.h reference.
4729 * hash-table.h: Remove tree-flow.h reference.
4730 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4731 reference with tree-ssa-loop.h.
4733 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4735 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4736 default element-order behavior for -maltivec.
4737 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4739 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4740 when targeting big endian, at least for now.
4741 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4743 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4746 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4747 var satisfies use_register_for_decl, just take into account type
4748 alignment, rather than decl alignment.
4750 PR tree-optimization/59622
4751 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4752 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4753 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4754 Don't devirtualize for inplace at all. For targets.length () == 1,
4755 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4757 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4759 * config/i386/i386.md (cpu): Remove the unused btver1.
4761 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4763 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4765 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4768 * tree-core.h (struct target_globals): New forward declaration.
4769 (struct tree_target_option): Add globals field.
4770 * tree.h (TREE_TARGET_GLOBALS): Define.
4771 (prepare_target_option_nodes_for_pch): New prototype.
4772 * target-globals.h (struct target_globals): Define even if
4774 * tree.c (prepare_target_option_node_for_pch,
4775 prepare_target_option_nodes_for_pch): New functions.
4776 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4777 * config/i386/i386.c: Include target-globals.h.
4778 (ix86_set_current_function): Instead of doing target_reinit
4779 unconditionally, use save_target_globals_default_opts and
4780 restore_target_globals.
4782 2014-01-09 Richard Biener <rguenther@suse.de>
4784 PR tree-optimization/59715
4785 * tree-cfg.h (split_critical_edges): Declare.
4786 * tree-cfg.c (split_critical_edges): Export.
4787 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4789 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4791 * cfgexpand.c (expand_stack_vars): Optionally disable
4792 asan stack protection.
4793 (expand_used_vars): Likewise.
4794 (partition_stack_vars): Likewise.
4795 * asan.c (asan_emit_stack_protection): Optionally disable
4796 after return stack usage.
4797 (instrument_derefs): Optionally disable memory access instrumentation.
4798 (instrument_builtin_call): Likewise.
4799 (instrument_strlen_call): Likewise.
4800 (asan_protect_global): Optionally disable global variables protection.
4801 * doc/invoke.texi: Added doc for new options.
4802 * params.def: Added new options.
4803 * params.h: Likewise.
4805 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4807 PR rtl-optimization/59724
4808 * ifcvt.c (cond_exec_process_if_block): Don't call
4809 flow_find_head_matching_sequence with 0 longest_match.
4810 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4811 non-active insns if !stop_after.
4812 (try_head_merge_bb): Revert 2014-01-07 changes.
4814 2014-01-08 Jeff Law <law@redhat.com>
4816 * ree.c (get_sub_rtx): New function, extracted from...
4817 (merge_def_and_ext): Here.
4818 (combine_reaching_defs): Use get_sub_rtx.
4820 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4822 * cgraph.h (varpool_variable_node): Do not choke on null node.
4824 2014-01-08 Catherine Moore <clm@codesourcery.com>
4826 * config/mips/mips.md (simple_return): Attempt to use JRC
4828 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4830 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4832 PR rtl-optimization/59137
4833 * reorg.c (steal_delay_list_from_target): Call update_block for
4835 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4837 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4839 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4840 two duplicate entries.
4842 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4845 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4847 * config/mips/mips.c (mips_truncated_op_cost): New function.
4848 (mips_rtx_costs): Adjust test for BADDU.
4849 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4851 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4853 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4854 (*baddu_si): ...this new pattern.
4856 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4859 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4861 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4864 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4866 (expand_expr, expand_normal): Adjust.
4867 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4868 inner_reference_p. Use inner_reference_p to expand inner references.
4869 (store_expr): Adjust.
4870 * cfgexpand.c (expand_call_stmt): Adjust.
4872 2014-01-08 Rong Xu <xur@google.com>
4874 * gcov-io.c (gcov_var): Move from gcov-io.h.
4875 (gcov_position): Ditto.
4876 (gcov_is_error): Ditto.
4877 (gcov_rewrite): Ditto.
4878 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4879 only part to libgcc/libgcov.h.
4881 2014-01-08 Marek Polacek <polacek@redhat.com>
4884 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4886 2014-01-08 Marek Polacek <polacek@redhat.com>
4889 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4891 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4893 PR rtl-optimization/59649
4894 * stor-layout.c (get_mode_bounds): For BImode return
4895 0 and STORE_FLAG_VALUE.
4897 2014-01-08 Richard Biener <rguenther@suse.de>
4900 * gimple.h (is_gimple_builtin_call): Remove.
4901 (gimple_builtin_call_types_compatible_p): New.
4902 (gimple_call_builtin_p): New overload.
4903 * gimple.c (is_gimple_builtin_call): Remove.
4904 (validate_call): Rename to ...
4905 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4907 (validate_type): New static function.
4908 (gimple_call_builtin_p): New overload and adjust.
4909 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4910 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4911 (gimple_fold_stmt_to_constant_1): Likewise.
4912 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4914 2014-01-08 Richard Biener <rguenther@suse.de>
4917 * gimplify.c (gimplify_expr): Gimplify register-register type
4918 VIEW_CONVERT_EXPRs to separate stmts.
4920 2014-01-07 Jeff Law <law@redhat.com>
4923 * ree.c (combine_set_extension): Handle case where source
4924 and destination registers in an extension insn are different.
4925 (combine_reaching_defs): Allow source and destination registers
4926 in extension to be different under limited circumstances.
4927 (add_removable_extension): Remove restriction that the
4928 source and destination registers in the extension are the same.
4929 (find_and_remove_re): Emit a copy from the extension's
4930 destination to its source after the defining insn if
4931 the source and destination registers are different.
4934 * ifcvt.c (merge_if_block): If we are merging a block with more than
4935 one successor with a block with no successors, remove any BARRIER
4936 after the second block.
4938 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4940 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4942 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4945 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4946 for 14-bit register offsets when INT14_OK_STRICT is false.
4948 2014-01-07 Roland Stigge <stigge@antcom.de>
4949 Michael Meissner <meissner@linux.vnet.ibm.com>
4952 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4953 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4955 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4957 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4960 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4962 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4963 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4964 rtx is const0_rtx or not.
4966 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4969 * target-globals.c (save_target_globals): Remove this_fn_optab
4971 * toplev.c: Include optabs.h.
4972 (target_reinit): Temporarily restore the global options if another
4973 set of options are in force.
4975 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4977 PR rtl-optimization/58668
4978 * cfgcleanup.c (flow_find_cross_jump): Don't count
4979 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4980 to determine what is counted.
4981 (flow_find_head_matching_sequence): Use active_insn_p to determine
4983 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4985 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4986 determine what is counted.
4988 PR tree-optimization/59643
4989 * tree-predcom.c (split_data_refs_to_components): If one dr is
4990 read and one write, determine_offset fails and the write isn't
4991 in the bad component, just put the read into the bad component.
4993 2014-01-07 Mike Stump <mikestump@comcast.net>
4994 Jakub Jelinek <jakub@redhat.com>
4997 * tree-core.h (struct tree_optimization_option): Change optabs
4998 type from unsigned char * to void *.
4999 * optabs.c (init_tree_optimization_optabs): Adjust
5000 TREE_OPTIMIZATION_OPTABS initialization.
5002 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5005 * config/i386/i386.h (struct machine_function): Add
5006 no_drap_save_restore field.
5007 * config/i386/i386.c (ix86_save_reg): Use
5008 !cfun->machine->no_drap_save_restore instead of
5009 crtl->stack_realign_needed.
5010 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5011 this function clears frame_pointer_needed. Set
5012 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5013 and DRAP reg is needed.
5015 2014-01-06 Marek Polacek <polacek@redhat.com>
5018 * doc/implement-c.texi: Mention that other integer types are
5019 permitted as bit-field types in strictly conforming mode.
5021 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5023 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5026 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5028 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5030 2014-01-06 Martin Jambor <mjambor@suse.cz>
5033 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5035 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5037 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5041 * var-tracking.c (add_stores): Preserve the value of the source even if
5042 we don't record the store.
5044 2014-01-06 Terry Guo <terry.guo@arm.com>
5046 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5048 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5051 * config/darwin.c (darwin_function_section): Adjust return values to
5052 correspond to optimisation changes made in r206070.
5054 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5056 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5057 from prefetch_block tune setting.
5058 (nocona_cost): Correct size of prefetch block to 64.
5060 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5062 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5063 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5064 used to save the static chain register in the computation of the offset
5065 from which the FP registers need to be restored.
5067 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5069 PR tree-optimization/59519
5070 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5071 ICE if get_current_def (current_new_name) is already non-NULL, as long
5072 as it is a phi result of some other phi in *new_exit_bb that has
5075 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5076 or vmovdqu* for misaligned_operand.
5077 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5078 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5079 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5080 aligned_mem for AVX512F masked aligned load and store builtins and for
5083 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5085 PR tree-optimization/59651
5086 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5087 Address range for negative step should be added by TYPE_SIZE_UNIT.
5089 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5091 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5092 overlapping registers also for registers other than the stack pointer.
5094 2014-01-03 Marek Polacek <polacek@redhat.com>
5097 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5100 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5103 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5106 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5107 (push splitter): Use <P:MODE_SIZE> instead of
5108 GET_MODE_SIZE (<P:MODE>mode).
5109 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5110 (mov -1, reg peephole2): Likewise.
5111 * config/i386/sse.md (*mov<mode>_internal,
5112 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5113 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5114 *<code><mode>3, *andnot<mode>3<mask_name>,
5115 <mask_codefor><code><mode>3<mask_name>): Likewise.
5116 * config/i386/subst.md (mask_mode512bit_condition,
5117 sd_mask_mode512bit_condition): Likewise.
5119 2014-01-02 Xinliang David Li <davidxl@google.com>
5121 PR tree-optimization/59303
5122 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5123 (dump_predicates): Better output format.
5124 (pred_equal_p): New function.
5125 (is_neq_relop_p): Ditto.
5126 (is_neq_zero_form_p): Ditto.
5127 (pred_expr_equal_p): Ditto.
5128 (pred_neg_p): Ditto.
5129 (simplify_pred): Ditto.
5130 (simplify_preds_2): Ditto.
5131 (simplify_preds_3): Ditto.
5132 (simplify_preds_4): Ditto.
5133 (simplify_preds): Ditto.
5135 (push_to_worklist): Ditto.
5136 (get_pred_info_from_cmp): Ditto.
5137 (is_degenerated_phi): Ditto.
5138 (normalize_one_pred_1): Ditto.
5139 (normalize_one_pred): Ditto.
5140 (normalize_one_pred_chain): Ditto.
5141 (normalize_preds): Ditto.
5142 (normalize_cond_1): Remove function.
5143 (normalize_cond): Ditto.
5144 (is_gcond_subset_of): Ditto.
5145 (is_subset_of_any): Ditto.
5146 (is_or_set_subset_of): Ditto.
5147 (is_and_set_subset_of): Ditto.
5148 (is_norm_cond_subset_of): Ditto.
5149 (pred_chain_length_cmp): Ditto.
5150 (convert_control_dep_chain_into_preds): Type change.
5151 (find_predicates): Ditto.
5152 (find_def_preds): Ditto.
5153 (destroy_predicates_vecs): Ditto.
5154 (find_matching_predicates_in_rest_chains): Ditto.
5155 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5156 (is_pred_expr_subset): Ditto.
5157 (is_pred_chain_subset_of): Ditto.
5158 (is_included_in): Ditto.
5159 (is_superset_of): Ditto.
5161 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5163 Update copyright years.
5165 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5167 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5168 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5169 config/arc/arc.md, config/arc/arc.opt,
5170 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5171 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5172 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5173 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5174 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5175 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5176 vtable-verify.c, vtable-verify.h: Use the standard form for the
5179 2014-01-02 Tobias Burnus <burnus@net-b.de>
5181 * gcc.c (process_command): Update copyright notice dates.
5182 * gcov-dump.c: Ditto.
5184 * doc/cpp.texi: Bump @copying's copyright year.
5185 * doc/cppinternals.texi: Ditto.
5186 * doc/gcc.texi: Ditto.
5187 * doc/gccint.texi: Ditto.
5188 * doc/gcov.texi: Ditto.
5189 * doc/install.texi: Ditto.
5190 * doc/invoke.texi: Ditto.
5192 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5194 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5196 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5198 * config/i386/sse.md (*mov<mode>_internal): Guard
5199 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5201 PR rtl-optimization/59647
5202 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5203 new_rtx into UNSIGNED_FLOAT rtxes.
5205 Copyright (C) 2014 Free Software Foundation, Inc.
5207 Copying and distribution of this file, with or without modification,
5208 are permitted in any medium without royalty provided the copyright
5209 notice and this notice are preserved.