1 2014-03-19 Richard Biener <rguenther@suse.de>
4 * tree-core.h (tree_type_common): Re-order pointer members
5 to reduce recursion depth during GC walks.
7 2014-03-19 Marek Polacek <polacek@redhat.com>
10 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
13 2014-03-19 Richard Biener <rguenther@suse.de>
16 * lto-streamer-in.c (input_function): In WPA stage do not drop
19 2014-03-19 Jakub Jelinek <jakub@redhat.com>
21 PR tree-optimization/60559
22 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
23 with build_zero_cst assignment.
25 2014-03-18 Kai Tietz <ktietz@redhat.com>
27 PR rtl-optimization/56356
28 * sdbout.c (sdbout_parms): Verify that parms'
29 incoming argument is valid.
30 (sdbout_reg_parms): Likewise.
32 2014-03-18 Richard Henderson <rth@redhat.com>
35 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
36 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
37 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
39 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
41 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
42 * doc/plugins.texi (Plugin callbacks): Mention
44 Italicize plugin event names in description. Explain that
45 PLUGIN_PRAGMAS has no sense for lto1. Explain
47 Remind that no GCC functions should be called after
49 Explain what pragmas with expansion are.
51 2014-03-18 Martin Liska <mliska@suse.cz>
53 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
54 gimple call statement is update.
55 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
56 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
58 2014-03-18 Jakub Jelinek <jakub@redhat.com>
61 * ubsan.c (ubsan_instrument_unreachable): Call
62 initialize_sanitizer_builtins.
63 (ubsan_pass): Likewise.
66 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
67 varpool_finalize_decl instead of rest_of_decl_compilation.
69 2014-03-18 Richard Biener <rguenther@suse.de>
71 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
72 by using bitmap_and_compl instead of bitmap_and_compl_into.
73 (df_rd_transfer_function): Likewise.
75 2014-03-18 Richard Biener <rguenther@suse.de>
77 * doc/lto.texi (fresolution): Fix typo.
79 2014-03-18 Richard Biener <rguenther@suse.de>
81 * doc/invoke.texi (flto): Update for changes in 4.9.
83 2014-03-18 Richard Biener <rguenther@suse.de>
85 * doc/loop.texi: Remove section on the removed lambda framework.
86 Update loop docs with recent changes in preserving loop structure.
88 2014-03-18 Richard Biener <rguenther@suse.de>
90 * doc/lto.texi (-fresolution): Document.
92 2014-03-18 Richard Biener <rguenther@suse.de>
94 * doc/contrib.texi: Adjust my name.
96 2014-03-18 Jakub Jelinek <jakub@redhat.com>
99 * internal-fn.c: Include diagnostic-core.h.
100 (expand_BUILTIN_EXPECT): New function.
101 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
102 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
103 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
104 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
106 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
107 Revert 3 argument __builtin_expect code.
108 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
109 * gimple-fold.c (gimple_fold_call): Likewise.
110 * tree.h (fold_builtin_expect): New prototype.
111 * builtins.c (build_builtin_expect_predicate): Add predictor
112 argument, if non-NULL, create 3 argument __builtin_expect.
113 (fold_builtin_expect): No longer static. Add ARG2 argument,
114 pass it through to build_builtin_expect_predicate.
115 (fold_builtin_2): Adjust caller.
116 (fold_builtin_3): Handle BUILT_IN_EXPECT.
117 * internal-fn.def (BUILTIN_EXPECT): New.
119 2014-03-18 Tobias Burnus <burnus@net-b.de>
122 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
123 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
124 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
126 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
129 * predict.c (combine_predictions_for_bb): Fix up formatting.
130 (expr_expected_value_1, expr_expected_value): Add predictor argument,
131 fill what it points to if non-NULL.
132 (tree_predict_by_opcode): Adjust caller, use the predictor.
133 * predict.def (PRED_COMPARE_AND_SWAP): Add.
135 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
137 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
138 proper constant for the store mode.
140 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
142 * symtab.c (change_decl_assembler_name): Fix transparent alias
145 2014-03-16 Renlin Li <Renlin.Li@arm.com>
147 * config/aarch64/aarch64.c: Correct the comments about the
148 aarch64 stack layout.
150 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
152 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
153 check for GF_OMP_FOR_KIND_FOR.
155 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
157 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
158 ymm and zmm register names.
160 2014-03-17 Jakub Jelinek <jakub@redhat.com>
163 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
164 note creation for the 2010-08-31 changes.
166 2014-03-17 Marek Polacek <polacek@redhat.com>
169 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
170 as -fno-tree-loop-vectorize.
171 (expand_omp_simd): Likewise.
173 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
175 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
176 (eligible_for_call_delay): New prototype.
177 * config/sparc/sparc.c (tls_call_delay): Rename into...
178 (eligible_for_call_delay): ...this. Return false if the instruction
179 cannot be put in the delay slot of a branch.
180 (eligible_for_restore_insn): Simplify.
181 (eligible_for_return_delay): Return false if the instruction cannot be
182 put in the delay slot of a branch and simplify.
183 (eligible_for_sibcall_delay): Return false if the instruction cannot be
184 put in the delay slot of a branch.
185 * config/sparc/sparc.md (fix_ut699): New attribute.
186 (tls_call_delay): Delete.
187 (in_call_delay): Reimplement.
188 (eligible_for_sibcall_delay): Rename into...
189 (in_sibcall_delay): ...this.
190 (eligible_for_return_delay): Rename into...
191 (in_return_delay): ...this.
192 (in_branch_delay): Reimplement.
193 (in_uncond_branch_delay): Delete.
194 (in_annul_branch_delay): Delete.
196 2014-03-14 Richard Henderson <rth@redhat.com>
199 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
200 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
201 (*floathi<X87MODEF>2_i387_with_temp): Remove.
202 (floathi splitters): Remove.
203 (float<SWI48x>xf2): New pattern.
204 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
205 code that tried to handle DImode for 32-bit, but which was excluded
206 by the pattern's condition. Drop allocation of stack temporary.
207 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
208 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
209 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
210 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
211 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
212 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
213 (*float<SWI48><MODEF>2_sse_interunit): Remove.
214 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
215 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
216 (*float<SWI48x><X87MODEF>2_i387): Remove.
217 (all float _with_temp splitters): Remove.
218 (*float<SWI48x><MODEF>2_i387): New pattern.
219 (*float<SWI48><MODEF>2_sse): New pattern.
220 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
221 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
223 2014-03-14 Jakub Jelinek <jakub@redhat.com>
224 Marek Polacek <polacek@redhat.com>
227 * common.opt (dump_base_name_prefixed): New Variable.
228 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
229 if x_dump_base_name_prefixed is already set, set it at the end.
231 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
233 PR rtl-optimization/60508
234 * lra-constraints.c (get_reload_reg): Add new parameter
236 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
237 Pass the new parameter values.
239 2014-03-14 Richard Biener <rguenther@suse.de>
241 * common.opt: Revert unintented changes from r205065.
244 2014-03-14 Richard Biener <rguenther@suse.de>
247 * cfghooks.c (split_block): Properly adjust all loops the
248 block was a latch of.
250 2014-03-14 Martin Jambor <mjambor@suse.cz>
253 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
256 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
259 * config/avr/avr.c (avr_set_current_function): Pass function name
260 through default_strip_name_encoding before sanity checking instead
261 of skipping the first char of the assembler name.
263 2014-03-13 Richard Henderson <rth@redhat.com>
266 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
267 (ix86_force_to_memory, ix86_free_from_memory): Remove.
268 * config/i386/i386-protos.h: Likewise.
269 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
270 in the expander instead of a splitter.
271 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
272 any possibility of requiring a memory.
273 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
274 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
275 (fp branch splitters): Update for ix86_split_fp_branch.
276 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
277 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
278 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
279 (*fop_<MODEF>_2_i387): Remove f/r alternative.
280 (*fop_<MODEF>_3_i387): Likewise.
281 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
282 (splitters for the fop_* register patterns): Remove.
283 (fscalexf4_i387): Rename from *fscalexf4_i387.
284 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
286 2014-03-13 Jakub Jelinek <jakub@redhat.com>
288 PR tree-optimization/59779
289 * tree-dfa.c (get_ref_base_and_extent): Use double_int
290 type for bitsize and maxsize instead of HOST_WIDE_INT.
292 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
294 PR rtl-optimization/57320
295 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
296 the CFG after thread_prologue_and_epilogue_insns.
298 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
300 PR rtl-optimization/57189
301 * lra-constraints.c (process_alt_operands): Disfavor spilling
304 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
306 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
309 2014-03-13 Jakub Jelinek <jakub@redhat.com>
311 PR tree-optimization/59025
313 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
314 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
316 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
319 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
320 calls of avr_out_plus_1.
322 2014-03-13 Bin Cheng <bin.cheng@arm.com>
324 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
325 BB's single pred and update the father loop's latch info later.
327 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
329 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
333 (VEC_base): Likewise.
334 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
335 registers, we need to swap double words in little endian mode.
337 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
338 to be a container mode for 128-bit integer operations added in ISA
339 2.07. Unlike TImode and PTImode, the preferred register set is
340 the Altivec/VMX registers for the 128-bit operations.
342 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
344 (rs6000_split_128bit_ok_p): Likewise.
346 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
347 macros for creating ISA 2.07 normal and overloaded builtin
348 functions with 3 arguments.
349 (BU_P8V_OVERLOAD_3): Likewise.
350 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
351 for use as overloaded functions.
352 (VPERM_1TI_UNS): Likewise.
353 (VSEL_1TI): Likewise.
354 (VSEL_1TI_UNS): Likewise.
355 (ST_INTERNAL_1ti): Likewise.
356 (LD_INTERNAL_1ti): Likewise.
357 (XXSEL_1TI): Likewise.
358 (XXSEL_1TI_UNS): Likewise.
359 (VPERM_1TI): Likewise.
360 (VPERM_1TI_UNS): Likewise.
361 (XXPERMDI_1TI): Likewise.
363 (LXVD2X_V1TI): Likewise.
364 (STXVD2X_V1TI): Likewise.
365 (VEC_INIT_V1TI): Likewise.
366 (VEC_SET_V1TI): Likewise.
367 (VEC_EXT_V1TI): Likewise.
368 (EQV_V1TI): Likewise.
369 (NAND_V1TI): Likewise.
370 (ORC_V1TI): Likewise.
371 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
372 added in ISA 2.07. Add both normal 'altivec' builtins, and the
376 (VADDEUQM): Likewise.
377 (VADDECUQ): Likewise.
378 (VSUBEUQM): Likewise.
379 (VSUBECUQ): Likewise.
381 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
382 __int128_t and __uint128_t types.
383 (__uint128_type): Likewise.
384 (altivec_categorize_keyword): Add support for vector __int128_t,
385 vector __uint128_t, vector __int128, and vector unsigned __int128
386 as a container type for TImode operations that need to be done in
387 VSX/Altivec registers.
388 (rs6000_macro_to_expand): Likewise.
389 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
390 to support 128-bit integer instructions vaddcuq, vadduqm,
391 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
392 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
394 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
395 for V1TImode, and set up preferences to use VSX/Altivec registers.
396 Setup VSX reload handlers.
397 (rs6000_debug_reg_global): Likewise.
398 (rs6000_init_hard_regno_mode_ok): Likewise.
399 (rs6000_preferred_simd_mode): Likewise.
400 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
401 (easy_altivec_constant): Likewise.
402 (output_vec_const_move): Likewise.
403 (rs6000_expand_vector_set): Convert V1TImode set and extract to
405 (rs6000_expand_vector_extract): Likewise.
406 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
408 (rs6000_const_vec): Add support for V1TImode.
409 (rs6000_emit_le_vsx_load): Swap double words when loading or
410 storing TImode/V1TImode.
411 (rs6000_emit_le_vsx_store): Likewise.
412 (rs6000_emit_le_vsx_move): Likewise.
413 (rs6000_emit_move): Add support for V1TImode.
414 (altivec_expand_ld_builtin): Likewise.
415 (altivec_expand_st_builtin): Likewise.
416 (altivec_expand_vec_init_builtin): Likewise.
417 (altivec_expand_builtin): Likewise.
418 (rs6000_init_builtins): Add support for V1TImode type. Add
419 support for ISA 2.07 128-bit integer builtins. Define type names
420 for the VSX/Altivec vector types.
421 (altivec_init_builtins): Add support for overloaded vector
422 functions with V1TImode type.
423 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
424 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
426 (rs6000_split_128bit_ok_p): Likewise.
427 (rs6000_handle_altivec_attribute): Create V1TImode from vector
428 __int128_t and vector __uint128_t.
430 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
438 (VS_scalar): Likewise.
439 (VS_double): Likewise.
440 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
442 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
443 we support the ISA 2.07 128-bit integer arithmetic instructions.
444 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
445 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
446 and TImode types for use with the builtin functions.
447 (V1TI_type_node): Likewise.
448 (unsigned_V1TI_type_node): Likewise.
449 (intTI_type_internal_node): Likewise.
450 (uintTI_type_internal_node): Likewise.
452 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
453 128-bit builtin functions.
454 (UNSPEC_VADDEUQM): Likewise.
455 (UNSPEC_VADDECUQ): Likewise.
456 (UNSPEC_VSUBCUQ): Likewise.
457 (UNSPEC_VSUBEUQM): Likewise.
458 (UNSPEC_VSUBECUQ): Likewise.
459 (VM): Add V1TImode to vector mode iterators.
462 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
463 (altivec_vaddcuq): Likewise.
464 (altivec_vsubuqm): Likewise.
465 (altivec_vsubcuq): Likewise.
466 (altivec_vaddeuqm): Likewise.
467 (altivec_vaddecuq): Likewise.
468 (altivec_vsubeuqm): Likewise.
469 (altivec_vsubecuq): Likewise.
471 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
473 (BOOL_128): Likewise.
474 (BOOL_REGS_OUTPUT): Likewise.
475 (BOOL_REGS_OP1): Likewise.
476 (BOOL_REGS_OP2): Likewise.
477 (BOOL_REGS_UNARY): Likewise.
478 (BOOL_REGS_AND_CR0): Likewise.
480 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
481 128-bit integer builtin support.
482 (vec_vadduqm): Likewise.
483 (vec_vaddecuq): Likewise.
484 (vec_vaddeuqm): Likewise.
485 (vec_vsubecuq): Likewise.
486 (vec_vsubeuqm): Likewise.
487 (vec_vsubcuq): Likewise.
488 (vec_vsubuqm): Likewise.
490 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
491 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
492 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
493 128-bit integer add/subtract to ISA 2.07.
495 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
497 * config/arc/arc.c (arc_predicate_delay_insns):
498 Fix third argument passed to conditionalize_nonjump.
500 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
502 * config/aarch64/aarch64-builtins.c
503 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
504 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
505 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
506 instead of __builtin_lfloor.
507 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
509 2014-03-12 Jakub Jelinek <jakub@redhat.com>
511 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
512 (tree_ssa_ifcombine_bb_1): New function.
513 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
514 is an empty forwarder block to then_bb or vice versa and then_bb
515 and else_bb are effectively swapped.
517 2014-03-12 Christian Bruel <christian.bruel@st.com>
520 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
521 REG_CFA_DEF_CFA note.
522 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
523 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
525 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
527 PR tree-optimization/60454
528 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
530 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
532 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
533 Do not define target_cpu_default2 to generic.
534 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
535 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
536 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
538 2014-03-12 Jakub Jelinek <jakub@redhat.com>
539 Marc Glisse <marc.glisse@inria.fr>
541 PR tree-optimization/60502
542 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
543 instead of build_low_bits_mask.
545 2014-03-12 Jakub Jelinek <jakub@redhat.com>
548 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
549 if there are multiple uses, but op doesn't live on E edge.
550 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
551 clobber stmts before __builtin_unreachable.
553 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
555 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
556 hard_frame_pointer_rtx.
557 * cse.c (cse_insn): Remove volatile check.
558 * cselib.c (cselib_process_insn): Likewise.
559 * dse.c (scan_insn): Likewise.
561 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
563 * config/arc/arc.c (conditionalize_nonjump): New function,
565 (arc_ifcvt): ... this.
566 (arc_predicate_delay_insns): Use it.
568 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
570 * config/arc/predicates.md (extend_operand): During/after reload,
571 allow const_int_operand.
572 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
573 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
574 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
576 (umulsi3_highpart_i): Likewise.
578 2014-03-11 Richard Biener <rguenther@suse.de>
580 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
581 Add asserts to guard possible wrong-code bugs.
583 2014-03-11 Richard Biener <rguenther@suse.de>
585 PR tree-optimization/60429
586 PR tree-optimization/60485
587 * tree-ssa-structalias.c (set_union_with_increment): Properly
588 take into account all fields that overlap the shifted vars.
589 (do_sd_constraint): Likewise.
590 (do_ds_constraint): Likewise.
591 (get_constraint_for_ptr_offset): Likewise.
593 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
595 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
596 (nios2_compute_frame_layout):
597 Add calculation of cfun->machine->fp_save_offset.
598 (nios2_expand_prologue): Correct setting of frame pointer register
600 (nios2_expand_epilogue): Update recovery of stack pointer from
601 frame pointer accordingly.
602 (nios2_initial_elimination_offset): Update calculation of offset
603 for eliminating to HARD_FRAME_POINTER_REGNUM.
605 2014-03-10 Jakub Jelinek <jakub@redhat.com>
608 * ipa.c (symtab_remove_unreachable_nodes): Don't call
609 cgraph_get_create_node on VAR_DECLs.
611 2014-03-10 Richard Biener <rguenther@suse.de>
614 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
616 2014-03-08 Douglas B Rupp <rupp@gnat.com>
618 * config/vms/vms.opt (vms_float_format): New variable.
620 2014-03-08 Tobias Burnus <burnus@net-b.de>
622 * doc/invoke.texi (-fcilkplus): Update implementation status.
624 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
625 Richard Biener <rguenther@suse.de>
627 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
628 consistently accross all TUs.
629 (run_gcc): Enable -fshort-double automatically at link at link-time
630 and disallow override.
632 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
635 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
636 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
637 if they can't be used.
639 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
641 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
642 for Solaris 11/x86 ld.
643 * configure: Regenerate.
645 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
647 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
648 (LIB_TLS_SPEC): Save as ld_tls_libs.
649 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
650 (HAVE_AS_IX86_TLSLDM): New test.
651 * configure, config.in: Regenerate.
652 * config/i386/i386.c (legitimize_tls_address): Fall back to
653 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
654 cannot support TLS_MODEL_LOCAL_DYNAMIC.
655 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
656 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
658 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
660 * common.opt (fira-loop-pressure): Mark as optimization.
662 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
664 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
665 an OpenMP mappable type.
667 2014-03-06 Matthias Klose <doko@ubuntu.com>
669 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
670 MULTILIB_OSDIRNAMES is not defined.
672 2014-03-06 Jakub Jelinek <jakub@redhat.com>
673 Meador Inge <meadori@codesourcery.com>
676 * config/arm/arm.c (arm_tls_symbol_p): Remove.
677 (arm_legitimize_address): Call legitimize_tls_address for any
678 arm_tls_referenced_p expression, handle constant addend. Call it
679 before testing for !TARGET_ARM.
680 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
682 2014-03-06 Richard Biener <rguenther@suse.de>
688 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
690 * tree-streamer.c (record_common_node): Assert we don't record
691 nodes with type double.
692 (preload_common_node): Skip type double, complex double and double
693 pointer since it is now frontend dependent due to fshort-double option.
695 2014-03-06 Richard Biener <rguenther@suse.de>
697 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
698 or -fno-lto is specified and the linker has full plugin support.
699 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
700 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
701 * lto-wrapper.c (merge_and_complain): Merge compile-time
703 (run_gcc): And pass it through to the link options.
705 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
709 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
711 * cselib.c (remove_useless_values): Skip to avoid quadratic
712 behavior if the condition moved from...
713 (cselib_process_insn): ... here holds.
715 2014-03-05 Jakub Jelinek <jakub@redhat.com>
718 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
719 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
722 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
723 (TM_H): Add x86-tune.def.
725 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
727 * config/aarch64/aarch64.c (generic_tunings):
728 Use cortexa57_extra_costs.
730 2014-03-05 Jakub Jelinek <jakub@redhat.com>
733 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
734 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
735 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
738 2014-03-04 Heiher <r@hev.cc>
740 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
741 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
743 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
745 * config/i386/predicates.md (const2356_operand): Change to ...
746 (const2367_operand): ... this.
747 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
749 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
750 (*avx512pf_scatterpf<mode>sf): Ditto.
751 (avx512pf_scatterpf<mode>df): Ditto.
752 (*avx512pf_scatterpf<mode>df_mask): Ditto.
753 (*avx512pf_scatterpf<mode>df): Ditto.
754 * config/i386/i386.c (ix86_expand_builtin): Update
755 incorrect hint operand error message.
757 2014-03-04 Richard Biener <rguenther@suse.de>
759 * lto-section-in.c (lto_get_section_data): Fix const cast.
761 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
763 * tree-streamer.c (record_common_node): Assert we don't record
764 nodes with type double.
765 (preload_common_node): Skip type double, complex double and double
766 pointer since it is now frontend dependent due to fshort-double option.
768 2014-03-04 Richard Biener <rguenther@suse.de>
771 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
772 (lto_input_toplevel_asms): Likewise.
773 * lto-section-in.c (lto_get_section_data): Instead do it here
776 2014-03-04 Richard Biener <rguenther@suse.de>
778 PR tree-optimization/60382
779 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
780 dead PHIs a reduction.
782 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
784 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
786 (_mm_prefetch): Move out of GCC target("sse") pragma.
787 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
788 GCC target("prfchw") pragma.
789 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
791 * config/i386/i386.c (ix86_option_override_internal): Enable
792 -mprfchw with -mprefetchwt1.
794 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
796 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
799 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
801 * opts.h (CL_PCH_IGNORE): Define.
802 * targhooks.c (option_affects_pch_p):
803 Return false for options that have CL_PCH_IGNORE set.
804 * opt-functions.awk: Process PchIgnore.
805 * doc/options.texi: Document PchIgnore.
807 * config/arc/arc.opt (misize): Add PchIgnore property.
809 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
811 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
812 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
813 constraint on constants to permit them being loaded into
814 GENERAL_REGS or BASE_REGS.
816 2014-03-03 Nick Clifton <nickc@redhat.com>
818 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
819 anti-cacnonical alternatives.
820 (negandhi3_real): New pattern.
821 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
823 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
825 * config/avr/avr-mcus.def: Remove atxmega16x1.
826 * config/avr/avr-tables.opt: Regenerate.
827 * config/avr/t-multilib: Regenerate.
828 * doc/avr-mmcu.texi: Regenerate.
830 2014-03-03 Tobias Grosser <tobias@grosser.es>
831 Mircea Namolaru <mircea.namolaru@inria.fr>
833 PR tree-optimization/58028
834 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
837 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
839 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
840 not handled by recognizers.
842 2014-03-03 Jakub Jelinek <jakub@redhat.com>
845 * function.c (expand_function_end): Don't emit
846 clobber_return_register sequence if clobber_after is a BARRIER.
847 * cfgexpand.c (construct_exit_block): Append instructions before
848 return_label to prev_bb.
850 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
852 * config/rs6000/constraints.md: Document reserved use of "wc".
854 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
857 * ipa.c (function_and_variable_visibility): When dissolving comdat
858 group, also set all symbols to local.
860 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
865 2013-12-14 Jan Hubicka <jh@suse.cz>
867 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
869 2014-03-02 Jon Beniston <jon@beniston.com>
875 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
876 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
877 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
878 (simple_return, *simple_return): New patterns
879 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
880 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
882 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
884 * dwarf2out.c (gen_subprogram_die): Tidy.
886 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
889 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
890 (*mov_t_msb_neg_negc): ... this new insn.
892 2014-02-28 Jason Merrill <jason@redhat.com>
895 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
898 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
901 * dwarf2out.c (decltype_auto_die): New static.
902 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
903 (gen_type_die_with_usage): Handle 'decltype(auto)'.
904 (is_cxx_auto): Likewise.
906 2014-02-28 Ian Bolton <ian.bolton@arm.com>
908 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
909 we are not using general regs only.
911 2014-02-28 Richard Biener <rguenther@suse.de>
914 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
915 previous fix and only allow to remove trivial pre-headers
916 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
917 (remove_forwarder_block): Properly update the latch of a loop.
919 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
922 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
923 (cselib_preserved_hash_table): New.
924 (preserve_constants_and_equivs): Move preserved vals to it.
925 (cselib_find_slot): Look it up first.
926 (cselib_init): Initialize it.
927 (cselib_finish): Release it.
928 (dump_cselib_table): Dump it.
930 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
933 * cselib.c (remove_useless_values): Skip to avoid quadratic
934 behavior if the condition moved from...
935 (cselib_process_insn): ... here holds.
937 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
940 * var-tracking.c (vt_initialize): Apply the same condition to
941 preserve the CFA base value.
943 2014-02-28 Joey Ye <joey.ye@arm.com>
946 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
947 if reload in progress or completed.
949 2014-02-28 Tobias Burnus <burnus@net-b.de>
952 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
955 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
957 * doc/tm.texi.in (Condition Code Status): Update documention for
958 relative locations of cc0-setter and cc0-user.
960 2014-02-27 Jeff Law <law@redhat.com>
962 PR rtl-optimization/52714
963 * combine.c (try_combine): When splitting an unrecognized PARALLEL
964 into two independent simple sets, if I3 is a jump, ensure the
965 pattern we place into I3 is a (set (pc) ...).
967 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
968 Jeff Law <law@redhat.com>
970 PR rtl-optimization/49847
971 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
972 are in different blocks.
973 * doc/tm.texi (Condition Code Status): Update documention for
974 relative locations of cc0-setter and cc0-user.
976 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
979 * lra.c (lra_emit_add): Check SUBREG too.
981 2014-02-27 Andreas Schwab <schwab@suse.de>
983 * config/m68k/m68k.c (m68k_option_override): Disable
984 -flive-range-shrinkage for classic m68k.
985 (m68k_override_options_after_change): Likewise.
987 2014-02-27 Marek Polacek <polacek@redhat.com>
990 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
991 -Wmaybe-uninitialized.
993 2014-02-27 Alan Modra <amodra@gmail.com>
996 * reload1.c (emit_input_reload_insns): When reload_override_in,
997 set old to rl->in_reg when rl->in_reg is a subreg.
999 2014-02-26 Richard Biener <rguenther@suse.de>
1002 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1004 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1006 * common/config/i386/predicates.md (const1256_operand): Remove.
1007 (const2356_operand): New.
1008 (const_1_to_2_operand): Remove.
1009 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1010 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1011 (*avx512pf_gatherpf<mode>sf): Ditto.
1012 (avx512pf_gatherpf<mode>df): Ditto.
1013 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1014 (*avx512pf_gatherpf<mode>df): Ditto.
1015 (avx512pf_scatterpf<mode>sf): Ditto.
1016 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1017 (*avx512pf_scatterpf<mode>sf): Ditto.
1018 (avx512pf_scatterpf<mode>df): Ditto.
1019 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1020 (*avx512pf_scatterpf<mode>df): Ditto.
1021 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1023 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1025 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1026 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1027 (_mm512_mask_testn_epi64_mask): Move to ...
1028 * config/i386/avx512cdintrin.h: Here.
1029 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1030 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1031 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1032 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1033 TARGET_AVX512F from TARGET_AVX512CD.
1035 2014-02-26 Richard Biener <rguenther@suse.de>
1038 * ipa.c (walk_polymorphic_call_targets): Properly guard
1039 call to inline_update_overall_summary.
1041 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1044 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1045 and latches only if requested. Fix latch if it is removed.
1046 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1047 LOOPS_HAVE_PREHEADERS.
1049 2014-02-25 Andrew Pinski <apinski@cavium.com>
1051 * builtins.c (expand_builtin_thread_pointer): Create a new target
1052 when the target is NULL.
1054 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1056 PR rtl-optimization/60317
1057 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1058 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1059 * lra-assigns.c: Include params.h.
1060 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1061 other reload pseudos considerations.
1063 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1065 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1066 to use canonical form for nor<mode>3.
1068 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1071 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1074 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1076 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1077 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1078 (ix86_handle_option): Handle OPT_mprefetchwt1.
1079 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1080 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1082 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1083 OPTION_MASK_ISA_PREFETCHWT1.
1084 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1085 (PTA_PREFETCHWT1): New.
1086 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1087 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1088 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1089 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1090 (*prefetch_avx512pf_<mode>_: Change into ...
1091 (*prefetch_prefetchwt1_<mode>: This.
1092 * config/i386/i386.opt (mprefetchwt1): New.
1093 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1094 (_mm_prefetch): Handle intent to write.
1095 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1097 2014-02-25 Richard Biener <rguenther@suse.de>
1100 * emit-rtl.c (mem_attrs_htab): Remove.
1101 (mem_attrs_htab_hash): Likewise.
1102 (mem_attrs_htab_eq): Likewise.
1103 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1104 (init_emit_once): Do not allocate mem_attrs_htab.
1106 2014-02-25 Richard Biener <rguenther@suse.de>
1109 * lto-opts.c (lto_write_options): Output non-explicit conservative
1110 -fwrapv, -fno-trapv and -fno-strict-overflow.
1111 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1112 (run_gcc): And pass them through.
1114 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1116 * sel-sched.c (calculate_new_fences): New parameter ptime.
1117 Calculate it as a maximum over all fence cycles.
1118 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1119 Print the final schedule timing when sched_verbose.
1121 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1123 PR rtl-optimization/60292
1124 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1125 bit fot the fence instruction.
1127 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1129 * calls.h: Fix typo in comment.
1131 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1133 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1134 adjusting offsetable addresses.
1136 2014-02-24 Guozhi Wei <carrot@google.com>
1138 * sparseset.h (sparseset_pop): Fix the wrong index.
1140 2014-02-24 Walter Lee <walt@tilera.com>
1142 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1143 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1145 * common/config/tilegx/tilegx-common.c
1146 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1147 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1149 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1150 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1151 (tilegx_gimplify_va_arg_expr): Handle big endian.
1152 (tilegx_expand_unaligned_load): Ditto.
1153 (tilegx_expand_unaligned_store): Ditto.
1154 (TARGET_RETURN_IN_MSB): New.
1155 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1156 (TARGET_ENDIAN_DEFAULT): New.
1157 (TARGET_BIG_ENDIAN): Handle big endian.
1158 (BYTES_BIG_ENDIAN): Ditto.
1159 (WORDS_BIG_ENDIAN): Ditto.
1160 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1163 * config/tilegx/tilegx.md (extv): Handle big endian.
1165 (insn_st<n>): Ditto.
1166 (insn_st<n>_add<bitsuffix>): Ditto.
1167 (insn_stnt<n>): Ditto.
1168 (insn_stnt<n>_add<bitsuffix>):Ditto.
1169 (vec_interleave_highv8qi): Handle big endian.
1170 (vec_interleave_highv8qi_be): New.
1171 (vec_interleave_highv8qi_le): New.
1172 (insn_v1int_h): Handle big endian.
1173 (vec_interleave_lowv8qi): Handle big endian.
1174 (vec_interleave_lowv8qi_be): New.
1175 (vec_interleave_lowv8qi_le): New.
1176 (insn_v1int_l): Handle big endian.
1177 (vec_interleave_highv4hi): Handle big endian.
1178 (vec_interleave_highv4hi_be): New.
1179 (vec_interleave_highv4hi_le): New.
1180 (insn_v2int_h): Handle big endian.
1181 (vec_interleave_lowv4hi): Handle big endian.
1182 (vec_interleave_lowv4hi_be): New.
1183 (vec_interleave_lowv4hi_le): New.
1184 (insn_v2int_l): Handle big endian.
1185 (vec_interleave_highv2si): Handle big endian.
1186 (vec_interleave_highv2si_be): New.
1187 (vec_interleave_highv2si_le): New.
1188 (insn_v4int_h): Handle big endian.
1189 (vec_interleave_lowv2si): Handle big endian.
1190 (vec_interleave_lowv2si_be): New.
1191 (vec_interleave_lowv2si_le): New.
1192 (insn_v4int_l): Handle big endian.
1193 * config/tilegx/tilegx.opt (mbig-endian): New option.
1194 (mlittle-endian): New option.
1195 * doc/install.texi: Document tilegxbe-linux.
1196 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1198 2014-02-24 Martin Jambor <mjambor@suse.cz>
1201 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1202 there are no parameter descriptors.
1204 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1206 PR rtl-optimization/60268
1207 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1208 initialization to ...
1209 (sched_rgn_init): ... here.
1210 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1212 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1214 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1217 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1219 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1222 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1224 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1225 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1227 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1229 * config/microblaze/predicates.md: Add cmp_op predicate.
1230 * config/microblaze/microblaze.md: Add branch_compare instruction
1231 which uses cmp_op predicate and emits cmp insn before branch.
1232 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1233 to microblaze_expand_conditional_branch and consolidate logic.
1234 (microblaze_expand_conditional_branch): emit branch_compare
1235 insn instead of handling cmp op separate from branch insn.
1237 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1239 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1242 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1244 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1245 define_insn with define_expand and new define_insn
1246 *altivec_lve<VI_char>x_internal.
1247 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1248 and new define_insn *altivec_stve<VI_char>x_internal.
1249 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1251 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1253 (altivec_expand_stvex_be): New function.
1255 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1257 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1258 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1259 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1260 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1262 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1265 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1266 instead of emit_move_insn.
1268 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1270 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1272 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1273 gen_altivec_vsumsws.
1275 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1277 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1278 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1279 (altivec_lvxl_<mode>): New define_expand incorporating
1280 -maltivec=be semantics where needed.
1281 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1282 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1283 semantics where needed.
1284 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1285 (altivec_stvx_<mode>): New define_expand incorporating
1286 -maltivec=be semantics where needed.
1287 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1288 VM2 iterator instead of V4SI.
1289 (altivec_stvxl_<mode>): New define_expand incorporating
1290 -maltivec=be semantics where needed.
1291 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1292 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1293 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1294 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1295 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1297 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1298 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1299 ALTIVEC_BUILTIN_STVXL.
1300 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1301 (altivec_expand_stvx_be): Likewise.
1302 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1303 (altivec_expand_lvx_be): Likewise.
1304 (altivec_expand_stvx_be): Likewise.
1305 (altivec_expand_builtin): Add cases for
1306 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1307 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1308 (altivec_init_builtins): Add definitions for
1309 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1310 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1312 2014-02-21 Catherine Moore <clm@codesourcery.com>
1314 * doc/invoke.texi (mvirt, mno-virt): Document.
1315 * config/mips/mips.opt (mvirt): New option.
1316 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1318 2014-02-21 Richard Biener <rguenther@suse.de>
1320 PR tree-optimization/60276
1321 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1322 (STMT_VINFO_MIN_NEG_DIST): New macro.
1323 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1324 STMT_VINFO_MIN_NEG_DIST.
1325 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1326 made for negative dependence distances still hold.
1328 2014-02-21 Richard Biener <rguenther@suse.de>
1331 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1332 DECL_INITIAL for globals not in the current function context.
1334 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1336 PR tree-optimization/56490
1337 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1338 * tree-ssa-uninit.c: Include params.h.
1339 (compute_control_dep_chain): Add num_calls argument, return false
1340 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1341 num_calls to recursive call.
1342 (find_predicates): Change dep_chain into normal array,
1343 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1344 variable and adjust compute_control_dep_chain caller.
1345 (find_def_preds): Likewise.
1347 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1349 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1350 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1352 2014-02-21 Nick Clifton <nickc@redhat.com>
1354 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1355 (pushhi1): Likewise.
1356 (popqi1): Add mode to pre_dec.
1359 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1361 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1362 mode for mask of V8SFmode permutation.
1364 2014-02-20 Richard Henderson <rth@redhat.com>
1367 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1368 a new pseudo for OLDVAL.
1370 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1373 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1374 gen_reg_rtx if d->testing_p.
1375 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1376 if d->testing_p and we will certainly return true.
1377 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1380 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1382 * emit-rtl.c (gen_reg_rtx): Assert that
1383 crtl->emit.regno_pointer_align_length is non-zero.
1385 2014-02-20 Richard Henderson <rth@redhat.com>
1388 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1389 on failure the store back into EXPECT.
1391 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1392 Sandra Loosemore <sandra@codesourcery.com>
1394 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1395 * config/nios2/nios2.c (nios2_function_profiler): Add
1396 -fPIC (flag_pic == 2) support.
1397 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1398 (nios2_large_offset_p): New function.
1399 (nios2_unspec_reloc_p): Move up position, update to use
1400 nios2_large_offset_p.
1401 (nios2_unspec_address): Remove function.
1402 (nios2_unspec_offset): New function.
1403 (nios2_large_got_address): New function.
1404 (nios2_got_address): Add large offset support.
1405 (nios2_legitimize_tls_address): Update usage of removed and new
1407 (nios2_symbol_binds_local_p): New function.
1408 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1409 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1410 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1411 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1412 processing for (const (unspec ...)).
1413 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1415 2014-02-20 Richard Biener <rguenther@suse.de>
1417 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1418 doing the substitution.
1419 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1421 2014-02-20 Martin Jambor <mjambor@suse.cz>
1424 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1425 info when checking whether lattices are bottom.
1427 2014-02-20 Richard Biener <rguenther@suse.de>
1430 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1433 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1436 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1437 parameter specifying the scaling.
1438 (inline_call): Update.
1439 (want_inline_recursively): Guard division by zero.
1440 (recursive_inlining): Update.
1441 * ipa-inline.h (clone_inlined_nodes): Update.
1443 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1446 * config/i386/i386.c (classify_argument): Pass structures of size
1447 64 bytes or less in register.
1449 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1450 Kirill Yukhin <kirill.yukhin@intel.com>
1452 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1453 (_mm_rcp28_round_ss): Ditto.
1454 (_mm_rsqrt28_round_sd): Ditto.
1455 (_mm_rsqrt28_round_ss): Ditto.
1456 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1457 (_mm_rcp14_round_ss): Ditto.
1458 (_mm_rsqrt14_round_sd): Ditto.
1459 (_mm_rsqrt14_round_ss): Ditto.
1460 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1461 the first input operand, get rid of match_dup.
1462 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1464 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1466 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1467 operand as the first input operand, set type attribute.
1468 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1470 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1471 operand as the first input operand, set type attribute.
1473 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1475 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1478 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1481 * config/i386/i386.c (construct_container): Remove TFmode check
1482 for X86_64_INTEGER_CLASS.
1484 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1487 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1488 only when -Wpsabi is enabled.
1490 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1493 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1494 passing arrays in registers are the same as for structs, so remove the
1495 special case for them.
1497 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1499 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1500 destination type, extract only the valid bits if the source type is not
1501 integral and has a different mode.
1503 2014-02-19 Richard Biener <rguenther@suse.de>
1506 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1509 2014-02-19 Richard Biener <rguenther@suse.de>
1512 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1513 (ipa_modify_call_arguments): Emit an argument load explicitely and
1514 preserve virtual SSA form there and for the replacement call.
1515 Do not update SSA form nor free dominance info.
1517 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1519 * ipa.c (function_and_variable_visibility): Also clear WEAK
1520 flag when disolving COMDAT_GROUP.
1522 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1524 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1525 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1527 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1528 do more sanity checks.
1529 (detect_type_change): Return true when giving up early.
1530 (compute_complex_assign_jump_func): Fix type parameter of
1531 ipa_set_ancestor_jf.
1532 (compute_complex_ancestor_jump_func): Likewise.
1533 (update_jump_functions_after_inlining): Fix updating of
1535 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1537 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1539 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1540 inline clones when edge disappears.
1542 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1545 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1546 Split 64-bit moves into 2 patterns. Do not allow the use of
1547 direct move for TDmode in little endian, since the decimal value
1548 has little endian bytes within a word, but the 64-bit pieces are
1549 ordered in a big endian fashion, and normal subreg's of TDmode are
1551 (mov<mode>_64bit_dm): Likewise.
1552 (movtd_64bit_nodm): Likewise.
1554 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1556 PR tree-optimization/60174
1557 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1558 statement of an SSA_NAME that occurs in an abnormal PHI node.
1560 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1563 * final.c (SEEN_BB): Remove.
1564 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1565 (final_scan_insn): Don't force_source_line on second
1566 NOTE_INSN_BASIC_BLOCK.
1568 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1571 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1572 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1573 (type_natural_mode): Warn ABI change when %zmm register is not
1574 available for AVX512F vector value passing.
1576 2014-02-18 Kai Tietz <ktietz@redhat.com>
1579 * config/i386/i386.c (ix86_expand_prologue): Use value in
1580 rax register as displacement when restoring %r10 or %rax.
1581 Fix wrong offset when restoring both registers.
1583 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1585 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1586 assertion with conditional return.
1588 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1589 Uros Bizjak <ubizjak@gmail.com>
1592 * config/i386/driver-i386.c (host_detect_local_cpu): If
1593 YMM state is not saved by the OS, also clear has_f16c. Move
1594 CPUID 0x80000001 handling before YMM state saving checking.
1596 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1598 PR rtl-optimization/58960
1599 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1600 factored out from ...
1601 (sched_init): ... here.
1602 (free_global_sched_pressure_data): New, factored out from ...
1603 (sched_finish): ... here.
1604 * sched-int.h (free_global_sched_pressure_data): Declare.
1605 * sched-rgn.c (nr_regions_initial): New static global.
1606 (haifa_find_rgns): Initialize it.
1607 (schedule_region): Disable sched-pressure for the newly
1610 2014-02-17 Richard Biener <rguenther@suse.de>
1612 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1613 release SSA defs of pattern stmts.
1615 2014-02-17 Richard Biener <rguenther@suse.de>
1617 * tree-inline.c (expand_call_inline): Release the virtual
1618 operand defined by the call we are about to inline.
1620 2014-02-17 Richard Biener <rguenther@suse.de>
1622 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1624 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1625 Ilya Tocar <ilya.tocar@intel.com>
1627 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1628 arguments order in builtin.
1629 (_mm512_permutexvar_epi64): Ditto.
1630 (_mm512_mask_permutexvar_epi64): Ditto
1631 (_mm512_maskz_permutexvar_epi32): Ditto
1632 (_mm512_permutexvar_epi32): Ditto
1633 (_mm512_mask_permutexvar_epi32): Ditto
1635 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1637 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1638 (p8_vmrgow): Likewise.
1640 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1642 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1645 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1648 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1649 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1650 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1651 using direct move instructions on ISA 2.07. Also adjust
1652 instruction length for 64-bit.
1653 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1654 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1656 2014-02-15 Alan Modra <amodra@gmail.com>
1660 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1661 find_replacement on parts of insn rtl that might be reloaded.
1663 2014-02-15 Richard Biener <rguenther@suse.de>
1665 PR tree-optimization/60183
1666 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1667 (tree_ssa_phiprop): Calculate and free post-dominators.
1669 2014-02-14 Jeff Law <law@redhat.com>
1671 PR rtl-optimization/60131
1672 * ree.c (get_extended_src_reg): New function.
1673 (combine_reaching_defs): Use it rather than assuming location of REG.
1674 (find_and_remove_re): Verify first operand of extension is
1675 a REG before adding the insns to the copy list.
1677 2014-02-14 Roland McGrath <mcgrathr@google.com>
1679 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1680 * configure: Regenerated.
1681 * config.in: Regenerated.
1682 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1683 instead of ASM_SHORT.
1685 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1686 Richard Earnshaw <rearnsha@arm.com>
1688 PR rtl-optimization/59535
1689 * lra-constraints.c (process_alt_operands): Encourage alternative
1690 when unassigned pseudo class is superset of the alternative class.
1691 (inherit_reload_reg): Don't inherit when optimizing for code size.
1692 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1693 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1694 modes not less than 4 for Thumb1.
1696 2014-02-14 Kyle McMartin <kyle@redhat.com>
1699 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1701 2014-02-14 Richard Biener <rguenther@suse.de>
1703 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1704 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1705 Do not strip INDIRECT_REFs.
1707 2014-02-14 Richard Biener <rguenther@suse.de>
1710 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1711 DECL_FUNCTION_SPECIFIC_TARGET.
1712 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1713 * tree-streamer-out.c (pack_ts_target_option): Remove.
1714 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1715 (write_ts_function_decl_tree_pointers): Do not stream
1716 DECL_FUNCTION_SPECIFIC_TARGET.
1717 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1718 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1719 (lto_input_ts_function_decl_tree_pointers): Do not stream
1720 DECL_FUNCTION_SPECIFIC_TARGET.
1722 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1724 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1725 (get_initial_def_for_induction, vectorizable_induction): Ignore
1726 debug stmts when looking for exit_phi.
1727 (vectorizable_live_operation): Fix up condition.
1729 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1731 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1732 nreverse() because it changes the content of original tree list.
1734 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1736 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1737 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1739 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1741 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1742 GNU coding standards.
1744 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1747 * dwarf2out.c (gen_subprogram_die): Don't call
1748 add_calling_convention_attribute if subr_die is old_die.
1750 2014-02-13 Sharad Singhai <singhai@google.com>
1752 * doc/optinfo.texi: Fix order of nodes.
1754 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1756 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1757 operands[2], not operands[3].
1759 2014-02-13 Richard Biener <rguenther@suse.de>
1762 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1763 mention the possibility of an in-tree build.
1764 (CLooG): Update recommended version to 0.18.1, mention the
1765 possibility of an in-tree build and clarify that the ISL
1766 bundled with CLooG does not work.
1768 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1771 * expr.c (compress_float_constant): If x is a hard register,
1772 extend into a pseudo and then move to x.
1774 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1776 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1777 caused by bad second argument to warning_at() with -mhotpatch and
1778 nested functions (e.g. with gfortran).
1780 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1782 * opts.c (option_name): Remove "enabled by default" rider.
1784 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1786 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1788 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1789 Uros Bizjak <ubizjak@gmail.com>
1792 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1793 * configure: Regenerated.
1795 2014-02-12 Richard Biener <rguenther@suse.de>
1797 * vec.c (vec_prefix::calculate_allocation): Move as
1798 inline variant to vec.h.
1799 (vec_prefix::calculate_allocation_1): New out-of-line version.
1800 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1801 (vec_prefix::m_has_auto_buf): Rename to ...
1802 (vec_prefix::m_using_auto_storage): ... this.
1803 (vec_prefix::calculate_allocation): Inline the easy cases
1804 and dispatch to calculate_allocation_1 which doesn't need the
1806 (va_heap::reserve): Use gcc_checking_assert.
1807 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1808 m_using_auto_storage.
1809 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1811 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1812 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1813 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1815 2014-02-12 Richard Biener <rguenther@suse.de>
1817 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1818 when we found a dependence.
1820 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1822 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1824 (maybe_fold_stmt): ... into this new function.
1825 * omp-low.c (lower_omp): Update comment.
1827 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1830 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1833 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1835 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1836 identifiers in comments.
1837 (cortexa53_extra_costs): Likewise.
1838 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1839 (cortexa7_extra_costs): Likewise.
1840 (cortexa12_extra_costs): Likewise.
1841 (cortexa15_extra_costs): Likewise.
1842 (v7m_extra_costs): Likewise.
1844 2014-02-12 Richard Biener <rguenther@suse.de>
1847 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1848 of posix_memalign being successful.
1849 (lower_stmt): Restrict lowering of posix_memalign to when
1850 -ftree-bit-ccp is enabled.
1852 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1854 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1856 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1858 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1860 PR rtl-optimization/60116
1861 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1862 other_insn once the combination has been validated.
1864 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1867 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1869 * ipa-devirt.c: Include demangle.h
1870 (odr_violation_reported): New static variable.
1871 (add_type_duplicate): Update odr_violations.
1872 (maybe_record_node): Add completep parameter; update it.
1873 (record_target_from_binfo): Add COMPLETEP parameter;
1874 update it as needed.
1875 (possible_polymorphic_call_targets_1): Likewise.
1876 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1877 rename FINAL to COMPLETE.
1878 (record_targets_from_bases): Sanity check we found the binfo;
1879 fix COMPLETEP updating.
1880 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1881 parameter, fix computing of COMPLETEP.
1882 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1883 at LTO time do demangling.
1884 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1885 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1887 (gimple_get_virt_method_for_binfo): Likewise.
1888 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1889 gimple_get_virt_method_for_vtable): Update prototypes.
1891 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1894 * genautomata.c (add_presence_absence): Fix typo with
1895 {final_}presence_list.
1897 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1900 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1901 for VSX/Altivec vectors that land in GPR registers.
1903 2014-02-11 Richard Henderson <rth@redhat.com>
1904 Jakub Jelinek <jakub@redhat.com>
1907 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1908 around drhs if type conversion to lacc->type is not useless.
1910 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1912 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1914 (cortex-a57.cortex-a53): Likewise.
1915 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1917 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1919 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1922 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1924 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1925 add_options_for_arm_vfp3.
1927 2014-02-11 Jeff Law <law@redhat.com>
1930 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1931 object with an undesirable mode.
1933 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1936 * config/i386/sol2-9.h: New file.
1937 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1938 *-*-solaris2.9*): Use it.
1940 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1942 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1943 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1945 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1947 * config/microblaze/microblaze.c: Extend mcpu version format
1949 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1951 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1953 2014-02-10 Richard Henderson <rth@redhat.com>
1956 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1957 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1958 ms-abi vs -mno-accumulate-outgoing-args.
1959 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1960 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1963 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1966 * cfgexpand.c (expand_asm_operands): Attach source location to
1967 ASM_INPUT rtx objects.
1968 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1970 2014-02-10 Nick Clifton <nickc@redhat.com>
1972 * config/mn10300/mn10300.c (popcount): New function.
1973 (mn10300_expand_prologue): Include saved registers in stack usage
1976 2014-02-10 Jeff Law <law@redhat.com>
1979 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1980 when changing the SET_DEST of a prior insn to avoid an input reload.
1982 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1984 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1985 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1986 -mcall-openbsd, or -mcall-linux.
1987 (CC1_ENDIAN_BIG_SPEC): Remove.
1988 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1989 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1990 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1991 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1992 and %cc1_endian_default.
1993 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1995 2014-02-10 Richard Biener <rguenther@suse.de>
1997 PR tree-optimization/60115
1998 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1999 MEM_REF handling. Properly verify that the accesses are not
2000 out of the objects bound.
2002 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2004 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2007 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2009 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2010 proper constants and fix formatting.
2011 (possible_polymorphic_call_targets): Fix formatting.
2013 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2014 Ilya Tocar <ilya.tocar@intel.com>
2016 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2017 (_mm512_loadu_epi32): Renamed into...
2018 (_mm512_loadu_si512): This.
2019 (_mm512_storeu_epi32): Renamed into...
2020 (_mm512_storeu_si512): This.
2021 (_mm512_maskz_ceil_ps): Removed.
2022 (_mm512_maskz_ceil_pd): Ditto.
2023 (_mm512_maskz_floor_ps): Ditto.
2024 (_mm512_maskz_floor_pd): Ditto.
2025 (_mm512_floor_round_ps): Ditto.
2026 (_mm512_floor_round_pd): Ditto.
2027 (_mm512_ceil_round_ps): Ditto.
2028 (_mm512_ceil_round_pd): Ditto.
2029 (_mm512_mask_floor_round_ps): Ditto.
2030 (_mm512_mask_floor_round_pd): Ditto.
2031 (_mm512_mask_ceil_round_ps): Ditto.
2032 (_mm512_mask_ceil_round_pd): Ditto.
2033 (_mm512_maskz_floor_round_ps): Ditto.
2034 (_mm512_maskz_floor_round_pd): Ditto.
2035 (_mm512_maskz_ceil_round_ps): Ditto.
2036 (_mm512_maskz_ceil_round_pd): Ditto.
2037 (_mm512_expand_pd): Ditto.
2038 (_mm512_expand_ps): Ditto.
2039 * config/i386/i386.c (ix86_builtins): Remove
2040 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2041 (bdesc_args): Ditto.
2042 * config/i386/predicates.md (const1256_operand): New.
2043 (const_1_to_2_operand): Ditto.
2044 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2045 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2046 (*avx512pf_gatherpf<mode>sf): Ditto.
2047 (avx512pf_gatherpf<mode>df): Ditto.
2048 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2049 (*avx512pf_gatherpf<mode>df): Ditto.
2050 (avx512pf_scatterpf<mode>sf): Ditto.
2051 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2052 (*avx512pf_scatterpf<mode>sf): Ditto.
2053 (avx512pf_scatterpf<mode>df): Ditto.
2054 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2055 (*avx512pf_scatterpf<mode>df): Ditto.
2056 (avx512f_expand<mode>): Removed.
2057 (<shift_insn><mode>3<mask_name>): Change predicate type.
2059 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2061 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2062 not at the end of datarefs vector use ordered_remove to avoid
2063 reordering datarefs vector.
2066 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2067 mark local addressable non-static vars as GOVD_PRIVATE
2068 instead of GOVD_LOCAL.
2069 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2070 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2074 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2075 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2076 assume_aligned or alloc_align attributes.
2077 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2078 arguments. Handle also assume_aligned and alloc_align attributes.
2079 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2080 calls to functions with assume_aligned or alloc_align attributes.
2081 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2083 2014-02-08 Terry Guo <terry.guo@arm.com>
2085 * doc/invoke.texi: Document ARM -march=armv7e-m.
2087 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2089 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2090 flag on __cilkrts_rethrow builtin.
2093 * ipa-cp.c (determine_versionability): Fail at -O0
2094 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2095 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2098 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2101 * tree-inline.c (copy_forbidden): Fail for
2102 __attribute__((optimize (0))) functions.
2104 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2106 * varpool.c: Include pointer-set.h.
2107 (varpool_remove_unreferenced_decls): Variables in other partitions
2108 will not be output; be however careful to not lose information
2111 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2113 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2114 lookup in the vtable constructor.
2116 2014-02-07 Jeff Law <law@redhat.com>
2119 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2120 define_insn_and_split.
2122 * ipa-inline.c (inline_small_functions): Fix typos.
2124 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2126 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2127 (s390_can_use_return_insn): Declare.
2128 * config/s390/s390.h (EPILOGUE_USES): Define.
2129 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2131 (s390_chunkify_start): Handle return JUMP_LABELs.
2132 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2133 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2134 (s390_can_use_return_insn): New functions.
2135 (s390_fix_long_loop_prediction): Handle conditional returns.
2136 (TARGET_SET_UP_BY_PROLOGUE): Define.
2137 * config/s390/s390.md (ANY_RETURN): New code iterator.
2138 (*creturn, *csimple_return, return, simple_return): New patterns.
2140 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2142 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2143 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2144 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2145 REG_CFA_RESTORE list when deciding not to restore a register.
2147 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2149 * config/s390/s390.c: Include tree-pass.h and context.h.
2150 (s390_early_mach): New function, split out from...
2151 (s390_emit_prologue): ...here.
2152 (pass_data_s390_early_mach): New pass structure.
2153 (pass_s390_early_mach): New class.
2154 (s390_option_override): Create and register early_mach pass.
2155 Move to end of file.
2157 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2159 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2160 to match for the exit block.
2162 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2164 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2165 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2166 Reject misaligned operands.
2168 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2170 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2172 2014-02-07 Richard Biener <rguenther@suse.de>
2175 * gimple-low.c (lower_builtin_posix_memalign): New function.
2176 (lower_stmt): Call it to lower posix_memalign in a way
2177 to make alignment info accessible.
2179 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2182 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2183 __builtin_setjmp_receiver.
2185 2014-02-07 Richard Biener <rguenther@suse.de>
2188 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2189 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2190 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2191 Handle BUILT_IN_POSIX_MEMALIGN.
2192 (find_func_clobbers): Likewise.
2193 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2194 (call_may_clobber_ref_p_1): Likewise.
2196 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2199 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2202 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2205 * lto-cgraph.c (lto_output_node): Use
2206 symtab_get_symbol_partitioning_class.
2207 (lto_output_varpool_node): likewise.
2208 (symtab_get_symbol_partitioning_class): Move here from
2210 * cgraph.h (symbol_partitioning_class): Likewise.
2211 (symtab_get_symbol_partitioning_class): Declare.
2213 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2215 * ggc.h (ggc_internal_cleared_alloc): New macro.
2216 * vec.h (vec_safe_copy): Handle memory stats.
2217 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2218 * target-globals.c (save_target_globals): Likewise.
2220 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2223 * expr.c (emit_move_resolve_push): Export; be bit more selective
2224 on when to clear alias set.
2225 * expr.h (emit_move_resolve_push): Declare.
2226 * function.h (struct function): Add tail_call_marked.
2227 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2228 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2229 * config/i386/i386.md (TImode move expander): De not call
2231 (FP push expanders): Preserve memory attributes.
2232 * config/i386/sse.md (push<mode>1): Remove.
2233 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2234 (ix86_expand_push): Remove.
2235 * config/i386/mmx.md (push<mode>1): Remove.
2237 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2239 PR rtl-optimization/60030
2240 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2241 lopart with paradoxical subreg before shifting it up by hprec.
2243 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2245 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2246 Remove extra newline at end of file.
2247 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2248 (arm_issue_rate): Handle cortexa57.
2249 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2250 (cortex-a57.cortex-a53): Likewise.
2252 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2255 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2256 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2258 (arm_expand_prologue): Adjust all callers.
2259 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2260 info, registers also at the lowest numbered registers side. Use
2261 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2265 * var-tracking.c (adjust_mems): Before adding a SET to
2266 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2268 2014-02-06 Alan Modra <amodra@gmail.com>
2271 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2272 change SDmode to DDmode when lra_in_progress.
2274 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2277 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2278 free_data_ref on the dr first, and before goto again also set dr
2279 to the next dr. For simd_lane_access, free old datarefs[i] before
2280 overwriting it. For get_vectype_for_scalar_type failure, don't
2281 free_data_ref if simd_lane_access.
2283 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2286 * tree.h (opts_for_fn): New inline function.
2287 (opt_for_fn): Define.
2288 * config/i386/i386.c (ix86_function_regparm): Use
2289 opt_for_fn (decl, optimize) instead of optimize.
2291 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2293 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2294 for SYMBOL_REF in large memory model.
2296 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2298 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2300 (cortex-a57): Likewise.
2301 (cortex-a57.cortex-a53): Likewise.
2303 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2304 Kugan Vivekanandarajah <kuganv@linaro.org>
2306 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2308 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2310 2014-02-06 Richard Biener <rguenther@suse.de>
2312 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2313 set_loop_copy and initialize_original_copy_tables.
2315 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2317 * config/aarch64/aarch64-simd.md
2318 (aarch64_ashr_simddi): Change QI to SI.
2320 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2321 Jakub Jelinek <jakub@redhat.com>
2324 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2327 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2329 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2330 CODE_FOR_altivec_vpku[hw]um to
2331 CODE_FOR_altivec_vpku[hw]um_direct.
2332 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2333 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2334 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2335 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2337 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2339 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2340 generation for -maltivec=be.
2341 (altivec_vsumsws): Simplify redundant test.
2343 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2345 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2346 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2347 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2348 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2349 gen_altivec_vpkuwum.
2350 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2352 (altivec_vpks<VI_char>ss): Likewise.
2353 (altivec_vpks<VI_char>us): Likewise.
2354 (altivec_vpku<VI_char>us): Likewise.
2355 (altivec_vpku<VI_char>um): Likewise.
2356 (altivec_vpku<VI_char>um_direct): New (copy of
2357 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2359 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2360 target is little endian and -maltivec=be is not specified.
2361 (*altivec_vupkhs<VU_char>_direct): New (copy of
2362 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2363 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2364 target is little endian and -maltivec=be is not specified.
2365 (*altivec_vupkls<VU_char>_direct): New (copy of
2366 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2367 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2368 little endian and -maltivec=be is not specified.
2369 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2370 little endian and -maltivec=be is not specified.
2372 2014-02-05 Richard Henderson <rth@redhat.com>
2375 * combine-stack-adj.c: Revert r206943.
2376 * sched-int.h (struct deps_desc): Add last_args_size.
2377 * sched-deps.c (init_deps): Initialize it.
2378 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2379 contain REG_ARGS_SIZE notes.
2381 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2383 * lto-cgraph.c (asm_nodes_output): Make global.
2384 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2385 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2386 (driver_handle_option): Handle OPT_fwpa.
2388 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2391 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2392 a comment typo and formatting issue. If odr_hash hasn't been
2393 created, return vNULL and set *completep to false.
2396 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2397 bb with no successors.
2399 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2402 * doc/invoke.texi (-march): Clarify documentation for ARM.
2406 2014-02-05 Richard Biener <rguenther@suse.de>
2408 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2409 when not vectorizing because of too many alias checks.
2410 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2411 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2413 2014-02-05 Nick Clifton <nickc@redhat.com>
2415 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2416 accept extended registers in any mode when compiling for the MN10300.
2418 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2420 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2421 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2422 sanitization attributes.
2423 (can_inline_edge_p): Likewise.
2424 (sanitize_attrs_match_for_inline_p): New function.
2426 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2428 * ipa-prop.c (detect_type_change): Shor circuit testing of
2429 type changes on THIS pointer.
2431 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2434 * config/pa/pa.c (legitimize_tls_address): Return original address
2435 if not passed a SYMBOL_REF rtx.
2436 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2438 (pa_emit_move_sequence): Simplify TLS source operands.
2439 (pa_legitimate_constant_p): Reject all TLS constants.
2440 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2441 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2443 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2445 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2446 groups when we know they are controlled by LTO.
2447 * varasm.c (default_binds_local_p_1): If object is in other partition,
2448 it will be resolved locally.
2450 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2452 * config/host-linux.c (linux_gt_pch_use_address): Don't
2453 use SSIZE_MAX because it is not always defined.
2455 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2458 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2459 threshold for pseudo splitting.
2460 (update_ebb_live_info): Process call argument hard registers and
2461 hard registers from insn definition too.
2462 (max_small_class_regs_num): New constant.
2463 (inherit_in_ebb): Update live hard regs through EBBs. Update
2464 reloads_num only for small register classes. Don't split for
2467 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2470 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2473 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2475 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2478 2014-02-04 Marek Polacek <polacek@redhat.com>
2480 * gdbinit.in (pel): Define.
2482 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2484 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2487 2014-02-04 Richard Biener <rguenther@suse.de>
2490 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2491 in function context local.
2492 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2493 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2494 similar to LTO_imported_decl_ref.
2496 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2498 PR tree-optimization/60002
2499 * cgraphclones.c (build_function_decl_skip_args): Clear
2502 PR tree-optimization/60023
2503 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2504 false to gsi_replace.
2505 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2506 has been in some EH region and vec_stmt could throw, add
2507 vec_stmt into the same EH region.
2508 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2509 has no lhs, ignore it.
2510 * internal-fn.c (expand_MASK_LOAD): Likewise.
2513 * tree-inline.c (copy_forbidden): Fail for
2514 __attribute__((optimize (0))) functions.
2517 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2518 is set, copy one less argument.
2519 (expand_simd_clones): Don't subtract clone_info->inbranch
2520 from simd_clone_struct_alloc argument.
2522 PR rtl-optimization/57915
2523 * recog.c (simplify_while_replacing): If all unary/binary/relational
2524 operation arguments are constant, attempt to simplify those.
2527 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2528 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2530 2014-02-04 Richard Biener <rguenther@suse.de>
2532 PR tree-optimization/60012
2533 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2534 TBAA disambiguation to all DDRs.
2536 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2539 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2540 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2542 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2545 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2547 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2549 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2550 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2552 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2555 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2556 to figure out targets of polymorphic calls with known decl.
2557 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2558 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2559 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2560 (get_polymorphic_call_info): ... here.
2561 (get_polymorphic_call_info_from_invariant): New function.
2563 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2565 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2566 lookup via vtable pointer; check for type consistency
2567 and turn inconsitent facts into UNREACHABLE.
2568 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2569 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2570 type inconsistent querries; return UNREACHABLE instead.
2572 2014-02-03 Richard Henderson <rth@twiddle.net>
2575 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2576 already processed this node.
2577 (normalize_one_pred_1): Pass along mark_set.
2578 (normalize_one_pred): Create and destroy a pointer_set_t.
2579 (normalize_one_pred_chain): Likewise.
2581 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2583 PR gcov-profile/58602
2584 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2586 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2589 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2590 -fno-devirtualize; try to devirtualize by the knowledge of
2591 virtual table pointer given by aggregate propagation.
2592 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2593 (ipa_print_node_jump_functions): Dump also offset that
2594 is relevant for polymorphic calls.
2595 (determine_known_aggregate_parts): Add arg_type parameter; use it
2596 instead of determining the type from pointer type.
2597 (ipa_compute_jump_functions_for_edge): Update call of
2598 determine_known_aggregate_parts.
2599 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2600 (gimple_get_virt_method_for_binfo): ... here; simplify using
2601 vtable_pointer_value_to_vtable.
2602 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2603 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2604 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2605 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2607 (vtable_pointer_value_to_binfo): ... here.
2608 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2610 2014-02-03 Teresa Johnson <tejohnson@google.com>
2612 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2613 redef of outer loop index variable.
2615 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2619 * doc/extend.texi (Function Attributes): Typo.
2621 2014-02-03 Cong Hou <congh@google.com>
2623 PR tree-optimization/60000
2624 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2625 if the vectorized statement is a store. A store statement can only
2626 appear at the end of pattern statements.
2628 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2630 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2631 (ix86_option_override_internal): Default long double to 64-bit for
2632 32-bit Bionic and to 128-bit for 64-bit Bionic.
2634 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2635 TARGET_LONG_DOUBLE_128 is true.
2636 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2638 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2639 (mlong-double-64): Negate -mlong-double-128.
2640 (mlong-double-128): New option.
2642 * config/i386/i386-c.c (ix86_target_macros): Define
2643 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2645 * doc/invoke.texi: Document -mlong-double-128.
2647 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2649 PR rtl-optimization/60024
2650 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2652 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2654 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2656 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2658 PR rtl-optimization/57662
2659 * sel-sched.c (code_motion_path_driver): Do not mark already not
2660 existing blocks in the visiting bitmap.
2662 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2664 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2665 on the insn being emitted.
2667 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2668 Will Deacon <will.deacon@arm.com>
2670 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2672 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2674 * config/arm/arm-tables.opt: Regenerate.
2676 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2678 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2679 for vector types other than V16QImode.
2680 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2681 define_expand, and call altivec_expand_vec_perm_le when producing
2682 code with little endian element order.
2683 (*altivec_vperm_<mode>_internal): New insn having previous
2684 behavior of altivec_vperm_<mode>.
2685 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2686 altivec_expand_vec_perm_le when producing code with little endian
2688 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2689 behavior of altivec_vperm_<mode>_uns.
2691 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2693 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2694 (altivec_vsumsws): Add handling for -maltivec=be with a little
2696 (altivec_vsumsws_direct): New.
2697 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2698 gen_altivec_vsumsws.
2700 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2702 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2703 vtable_pointer_value_to_binfo): New functions.
2704 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2705 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2707 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2709 * config/nios2/nios2.md (load_got_register): Initialize GOT
2710 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2711 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2713 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2715 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2716 preserverd by passthrough, do not propagate the type.
2718 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2720 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2721 (mips_atomic_assign_expand_fenv): New function.
2722 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2724 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2726 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2727 (__builtin_mips_set_fcsr): Likewise.
2728 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2729 MIPS_USI_FTYPE_VOID.
2730 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2731 (mips16_expand_set_fcsr): Likewise.
2732 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2733 (mips16_set_fcsr_stub): Likewise.
2734 (mips16_get_fcsr_one_only_stub): New class.
2735 (mips16_set_fcsr_one_only_stub): Likewise.
2736 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2737 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2738 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2739 (hard_float): New availability predicate.
2740 (mips_builtins): Add get_fcsr and set_fcsr.
2741 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2742 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2743 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2744 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2745 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2748 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2750 * config/mips/mips.c (mips_one_only_stub): New class.
2751 (mips_need_mips16_rdhwr_p): Replace with...
2752 (mips16_rdhwr_stub): ...this new variable.
2753 (mips16_stub_call_address): New function.
2754 (mips16_rdhwr_one_only_stub): New class.
2755 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2756 (mips_output_mips16_rdhwr): Delete.
2757 (mips_finish_stub): New function.
2758 (mips_code_end): Use it to handle rdhwr stubs.
2760 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2763 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2764 when calculating size of integer atomic types.
2766 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2768 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2770 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2772 PR tree-optimization/60003
2773 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2774 * profile.c (branch_prob): Use gimple_call_builtin_p
2775 to check for BUILT_IN_SETJMP_RECEIVER.
2776 * tree-inline.c (copy_bb): Call notice_special_calls.
2778 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2781 * lra-constraints.c (process_alt_operands): Update reload_sum only
2784 2014-01-31 Richard Henderson <rth@redhat.com>
2787 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2788 until after else_eh is processed.
2790 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2792 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2793 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2794 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2795 in smmintrin.h, remove them.
2796 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2797 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2798 * config/i386/i386.md (ROUND_SAE): Fix value.
2799 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2800 (const48_operand): New.
2801 * config/i386/subst.md (round), (round_expand): Use
2802 const_4_or_8_to_11_operand.
2803 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2805 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2807 * config/i386/constraints.md (Yk): Swap meaning with k.
2808 * config/i386/i386.md (movhi_internal): Change Yk to k.
2809 (movqi_internal): Ditto.
2810 (*k<logic><mode>): Ditto.
2813 (kandn<mode>): Ditto.
2814 (*<code>hi_1): Ditto.
2815 (*<code>qi_1): Ditto.
2816 (kxnor<mode>): Ditto.
2817 (kortestzhi): Ditto.
2818 (kortestchi): Ditto.
2820 (*one_cmplhi2_1): Ditto.
2821 (*one_cmplqi2_1): Ditto.
2822 * config/i386/sse.md (): Change k to Yk.
2823 (avx512f_load<mode>_mask): Ditto.
2824 (avx512f_blendm<mode>): Ditto.
2825 (avx512f_store<mode>_mask): Ditto.
2826 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2827 (avx512f_storedqu<mode>_mask): Ditto.
2828 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2830 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2831 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2832 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2833 (avx512f_maskcmp<mode>3): Ditto.
2834 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2835 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2836 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2837 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2838 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2839 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2840 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2841 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2842 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2843 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2844 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2845 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2846 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2847 (vec_extract_lo_<mode>_maskm): Ditto.
2848 (vec_extract_hi_<mode>_maskm): Ditto.
2849 (avx512f_vternlog<mode>_mask): Ditto.
2850 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2851 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2852 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2853 (avx512f_<code>v8div16qi2_mask): Ditto.
2854 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2855 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2856 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2857 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2858 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2859 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2860 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2861 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2862 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2863 (avx512cd_maskb_vec_dupv8di): Ditto.
2864 (avx512cd_maskw_vec_dupv16si): Ditto.
2865 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2866 (avx512f_vpermi2var<mode>3_mask): Ditto.
2867 (avx512f_vpermi2var<mode>3_mask): Ditto.
2868 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2869 (*avx512f_gathersi<mode>): Ditto.
2870 (*avx512f_gathersi<mode>_2): Ditto.
2871 (*avx512f_gatherdi<mode>): Ditto.
2872 (*avx512f_gatherdi<mode>_2): Ditto.
2873 (*avx512f_scattersi<mode>): Ditto.
2874 (*avx512f_scatterdi<mode>): Ditto.
2875 (avx512f_compress<mode>_mask): Ditto.
2876 (avx512f_compressstore<mode>_mask): Ditto.
2877 (avx512f_expand<mode>_mask): Ditto.
2878 * config/i386/subst.md (mask): Change k to Yk.
2879 (mask_scalar_merge): Ditto.
2882 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2884 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2886 2014-01-31 Richard Biener <rguenther@suse.de>
2889 * builtins.c (fold_builtin_memory_op): Make sure to not
2890 use a floating-point mode or a boolean or enumeral type for
2893 2014-01-30 DJ Delorie <dj@redhat.com>
2895 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2896 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2897 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2898 whenever main() has an epilogue.
2900 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2902 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2903 unused variable "field".
2904 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2905 (vsx_mergeh_<mode>): Likewise.
2906 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2907 (altivec_vmrghh): Likewise.
2908 (altivec_vmrghw): Likewise.
2909 (altivec_vmrglb): Likewise.
2910 (altivec_vmrglh): Likewise.
2911 (altivec_vmrglw): Likewise.
2912 (altivec_vspltb): Add missing uses.
2913 (altivec_vsplth): Likewise.
2914 (altivec_vspltw): Likewise.
2915 (altivec_vspltsf): Likewise.
2917 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2920 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2921 frame related instructions.
2923 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2925 PR rtl-optimization/59959
2926 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2927 any reload of register whose subreg is invalid.
2929 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2931 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2932 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2933 Add missing return type - void.
2935 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2937 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2938 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2939 remove element index adjustment for endian (now handled in vsx.md
2941 (altivec_expand_vec_perm_const): Use
2942 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2943 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2944 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2945 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2946 define_expand and a new define_insn *altivec_vspltb_internal;
2947 adjust for -maltivec=be on a little endian target.
2948 (altivec_vspltb_direct): New.
2949 (altivec_vsplth): Divide into a define_expand and a new
2950 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2951 little endian target.
2952 (altivec_vsplth_direct): New.
2953 (altivec_vspltw): Divide into a define_expand and a new
2954 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2955 little endian target.
2956 (altivec_vspltw_direct): New.
2957 (altivec_vspltsf): Divide into a define_expand and a new
2958 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2959 a little endian target.
2961 2014-01-30 Richard Biener <rguenther@suse.de>
2963 PR tree-optimization/59993
2964 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2965 can propagate form the earlier stmt and avoid the transform
2966 when the intermediate result is needed.
2968 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2970 * README.Portability: Fix typo.
2972 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2974 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2975 comparison_operator with ordered_comparison_operator.
2977 2014-01-30 Nick Clifton <nickc@redhat.com>
2979 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2980 Rename to mn10300_store_multiple_regs.
2981 * config/mn10300/mn10300.c: Likewise.
2982 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2983 store_multiple_regs.
2984 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2985 Call mn10300_store_multiple_regs.
2987 2014-01-30 Nick Clifton <nickc@redhat.com>
2988 DJ Delorie <dj@redhat.com>
2990 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2991 %fp 2 to keep registers after it properly word-aligned.
2992 (rl78_alloc_physical_registers_umul): Handle the case where both
2993 input operands are the same.
2995 2014-01-30 Richard Biener <rguenther@suse.de>
2997 PR tree-optimization/59903
2998 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3001 2014-01-30 Jason Merrill <jason@redhat.com>
3004 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3007 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3009 2014-01-30 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/59951
3012 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3014 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3017 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3018 SFmode to DFmode case.
3020 2014-01-29 DJ Delorie <dj@redhat.com>
3022 * config/msp430/msp430.opt (-minrt): New.
3023 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3025 (ENDFILE_SPEC): Likewise.
3027 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3029 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3030 (estimate_function_body_sizes): Use it.
3032 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3035 * dwarf2out.c (is_cxx_auto): New.
3036 (is_base_type): Use it.
3037 (gen_type_die_with_usage): Likewise.
3039 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3041 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3042 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3043 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3044 -maltivec=be with LE targets.
3045 (vsx_mergeh_<mode>): Likewise.
3046 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3047 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3048 (altivec_vmrghb): Replace with define_expand and new
3049 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3050 (altivec_vmrghb_direct): New define_insn.
3051 (altivec_vmrghh): Replace with define_expand and new
3052 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3053 (altivec_vmrghh_direct): New define_insn.
3054 (altivec_vmrghw): Replace with define_expand and new
3055 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3056 (altivec_vmrghw_direct): New define_insn.
3057 (*altivec_vmrghsf): Adjust for endianness.
3058 (altivec_vmrglb): Replace with define_expand and new
3059 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3060 (altivec_vmrglb_direct): New define_insn.
3061 (altivec_vmrglh): Replace with define_expand and new
3062 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3063 (altivec_vmrglh_direct): New define_insn.
3064 (altivec_vmrglw): Replace with define_expand and new
3065 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3066 (altivec_vmrglw_direct): New define_insn.
3067 (*altivec_vmrglsf): Adjust for endianness.
3068 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3069 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3070 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3071 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3072 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3073 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3074 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3075 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3077 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3079 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3080 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3083 2014-01-29 Richard Biener <rguenther@suse.de>
3085 PR tree-optimization/58742
3086 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3087 associate_pointerplus_align.
3088 (associate_pointerplus_diff): New function.
3089 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3090 and associate_pointerplus_diff.
3092 2014-01-29 Richard Biener <rguenther@suse.de>
3094 * lto-streamer.h (LTO_major_version): Bump to 3.
3095 (LTO_minor_version): Reset to 0.
3097 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3099 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3100 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3101 (arm_file_start): Generate correct asm header for armv7ve.
3102 * config/arm/bpabi.h: Add multilib support for armv7ve.
3103 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3104 and cortex-a15 to armv7ve.
3105 * config/arm/t-aprofile: Add multilib support for armv7ve.
3106 * doc/invoke.texi: Document -march=armv7ve.
3108 2014-01-29 Richard Biener <rguenther@suse.de>
3110 PR tree-optimization/58742
3111 * tree-ssa-forwprop.c (associate_plusminus): Return true
3112 if we changed sth, defer EH cleanup to ...
3113 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3114 (simplify_mult): New function.
3116 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3119 PR tree-optimization/59920
3120 * tree.c (build_common_builtin_nodes): Remove
3121 __builtin_setjmp_dispatcher initialization.
3122 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3123 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3124 instead of gsi_after_labels + manually skipping debug stmts.
3125 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3126 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3127 * tree-inline.c (copy_edges_for_bb): Remove
3128 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3129 argument. Ignore computed_goto_p stmts. Don't call
3130 make_abnormal_goto_edges. If a call might need abnormal edges
3131 for non-local gotos, see if it already has an edge to
3132 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3133 with true argument, don't do anything then, otherwise add
3134 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3135 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3137 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3138 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3139 (lower_stmt): Don't set data->calls_builtin_setjmp.
3140 (lower_builtin_setjmp): Adjust comment.
3141 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3142 * tree-cfg.c (found_computed_goto): Remove.
3143 (factor_computed_gotos): Remove.
3144 (make_goto_expr_edges): Return bool, true for computed gotos.
3145 Don't call make_abnormal_goto_edges.
3146 (build_gimple_cfg): Don't set found_computed_goto, don't call
3147 factor_computed_gotos.
3148 (computed_goto_p): No longer static.
3149 (make_blocks): Don't set found_computed_goto.
3150 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3151 (make_edges): If make_goto_expr_edges returns true, push bb
3152 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3153 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3154 vector. Record mapping between bbs and OpenMP regions if there
3155 are any, adjust make_gimple_omp_edges caller. Call
3156 handle_abnormal_edges.
3157 (make_abnormal_goto_edges): Remove.
3158 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3159 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3160 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3161 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3162 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3163 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3164 filling *region also set *region_idx to (*region)->entry->index.
3167 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3168 For REGs set ORIGINAL_REGNO.
3170 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3172 * doc/md.texi: Mention that a target shouldn't implement
3173 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3176 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3178 PR tree-optimization/59594
3179 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3180 a copy of the datarefs vector rather than the vector itself.
3182 2014-01-28 Jason Merrill <jason@redhat.com>
3185 * dwarf2out.c (auto_die): New static.
3186 (gen_type_die_with_usage): Handle C++1y 'auto'.
3187 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3190 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3193 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3194 (SPEC_X32): Likewise.
3195 (SPEC_64): Likewise.
3196 * config/i386/i386.c (ix86_option_override_internal): Turn off
3197 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3199 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3200 * config/i386/i386.h (TARGET_16BIT): New macro.
3201 (TARGET_16BIT_P): Likewise.
3202 * config/i386/i386.opt: Add m16.
3203 * doc/invoke.texi: Document -m16.
3205 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3207 PR preprocessor/59935
3208 * input.c (location_get_source_line): Bail out on when line number
3209 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3211 2014-01-28 Richard Biener <rguenther@suse.de>
3213 PR tree-optimization/58742
3214 * tree-ssa-forwprop.c (associate_plusminus): Handle
3215 pointer subtraction of the form (T)(P + A) - (T)P.
3217 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3219 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3222 2014-01-28 Richard Biener <rguenther@suse.de>
3225 2014-01-28 Richard Biener <rguenther@suse.de>
3227 PR rtl-optimization/45364
3228 PR rtl-optimization/59890
3229 * var-tracking.c (local_get_addr_clear_given_value): Handle
3230 already cleared slot.
3231 (val_reset): Handle not allocated local_get_addr_cache.
3232 (vt_find_locations): Use post-order on the inverted CFG.
3234 2014-01-28 Richard Biener <rguenther@suse.de>
3236 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3238 2014-01-28 Richard Biener <rguenther@suse.de>
3240 PR rtl-optimization/45364
3241 PR rtl-optimization/59890
3242 * var-tracking.c (local_get_addr_clear_given_value): Handle
3243 already cleared slot.
3244 (val_reset): Handle not allocated local_get_addr_cache.
3245 (vt_find_locations): Use post-order on the inverted CFG.
3247 2014-01-28 Alan Modra <amodra@gmail.com>
3249 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3250 * configure.ac <recursive call for build != host>: Define
3251 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3252 and LD_FOR_BUILD too.
3253 * configure: Regenerate.
3255 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3257 * config/i386/i386.c (get_builtin_code_for_version): Separate
3258 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3259 Broadwell from Haswell.
3261 2014-01-27 Steve Ellcey <sellcey@mips.com>
3263 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3264 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3265 * config/mips/mips.c (mips_option_override): Change setting
3267 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3268 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3269 Change from Mask to Var.
3271 2014-01-27 Jeff Law <law@redhat.com>
3273 * ipa-inline.c (inline_small_functions): Fix typo.
3275 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3277 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3278 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3279 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3280 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3281 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3282 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3283 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3284 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3285 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3286 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3287 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3288 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3289 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3290 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3291 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3292 (_mm512_storeu_epi64): Ditto.
3293 (_mm512_cmpge_epi32_mask): Ditto.
3294 (_mm512_cmpge_epu32_mask): Ditto.
3295 (_mm512_cmpge_epi64_mask): Ditto.
3296 (_mm512_cmpge_epu64_mask): Ditto.
3297 (_mm512_cmple_epi32_mask): Ditto.
3298 (_mm512_cmple_epu32_mask): Ditto.
3299 (_mm512_cmple_epi64_mask): Ditto.
3300 (_mm512_cmple_epu64_mask): Ditto.
3301 (_mm512_cmplt_epi32_mask): Ditto.
3302 (_mm512_cmplt_epu32_mask): Ditto.
3303 (_mm512_cmplt_epi64_mask): Ditto.
3304 (_mm512_cmplt_epu64_mask): Ditto.
3305 (_mm512_cmpneq_epi32_mask): Ditto.
3306 (_mm512_cmpneq_epu32_mask): Ditto.
3307 (_mm512_cmpneq_epi64_mask): Ditto.
3308 (_mm512_cmpneq_epu64_mask): Ditto.
3309 (_mm512_expand_pd): Ditto.
3310 (_mm512_expand_ps): Ditto.
3311 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3312 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3313 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3314 * config/i386/i386.c (ix86_builtins): Add
3315 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3316 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3317 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3318 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3319 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3320 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3321 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3322 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3323 IX86_BUILTIN_PMOVUSQW512_MEM.
3324 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3325 __builtin_ia32_pmovsqd512mem_mask,
3326 __builtin_ia32_pmovqd512mem_mask,
3327 __builtin_ia32_pmovusqw512mem_mask,
3328 __builtin_ia32_pmovsqw512mem_mask,
3329 __builtin_ia32_pmovqw512mem_mask,
3330 __builtin_ia32_pmovusdw512mem_mask,
3331 __builtin_ia32_pmovsdw512mem_mask,
3332 __builtin_ia32_pmovdw512mem_mask,
3333 __builtin_ia32_pmovqb512mem_mask,
3334 __builtin_ia32_pmovusqb512mem_mask,
3335 __builtin_ia32_pmovsqb512mem_mask,
3336 __builtin_ia32_pmovusdb512mem_mask,
3337 __builtin_ia32_pmovsdb512mem_mask,
3338 __builtin_ia32_pmovdb512mem_mask.
3339 (bdesc_args): Add __builtin_ia32_expanddf512,
3340 __builtin_ia32_expandsf512.
3341 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3342 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3343 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3344 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3345 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3346 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3347 (avx512f_<code>v8div16qi2_mask_store): This.
3348 (avx512f_expand<mode>): New.
3350 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3352 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3354 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3355 (_mm512_prefetch_i32scatter_pd): Ditto.
3356 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3357 (_mm512_prefetch_i64scatter_pd): Ditto.
3358 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3359 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3360 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3361 (_mm512_prefetch_i32scatter_ps): Ditto.
3362 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3363 (_mm512_prefetch_i64scatter_ps): Ditto.
3364 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3365 * config/i386/i386-builtin-types.def: Define
3366 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3367 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3368 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3369 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3370 IX86_BUILTIN_SCATTERPFQPD.
3371 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3372 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3373 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3374 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3375 __builtin_ia32_scatterpfqps.
3376 (ix86_expand_builtin): Expand new built-ins.
3377 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3378 fix memory access data type.
3379 (*avx512pf_gatherpf<mode>_mask): Ditto.
3380 (*avx512pf_gatherpf<mode>): Ditto.
3381 (avx512pf_scatterpf<mode>): Ditto.
3382 (*avx512pf_scatterpf<mode>_mask): Ditto.
3383 (*avx512pf_scatterpf<mode>): Ditto.
3384 (GATHER_SCATTER_SF_MEM_MODE): New.
3385 (avx512pf_gatherpf<mode>df): Ditto.
3386 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3387 (*avx512pf_scatterpf<mode>df): Ditto.
3389 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3392 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3393 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3396 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3398 * common/config/arm/arm-common.c
3399 (arm_rewrite_mcpu): Handle multiple names.
3401 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3403 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3405 * gimple-builder.h (create_gimple_tmp): Delete.
3407 2014-01-27 Christian Bruel <christian.bruel@st.com>
3409 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3412 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3414 * config/pa/pa.md (call): Generate indirect long calls to non-local
3415 functions when outputing 32-bit code.
3416 (call_value): Likewise except for special call to buggy powf function.
3418 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3419 portable runtime and PIC indirect calls.
3420 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3421 and PIC call sequences. Use ldo instead of blr to set return register
3422 in PIC call sequence.
3424 2014-01-25 Walter Lee <walt@tilera.com>
3426 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3427 avoid clobbering a live register.
3429 2014-01-25 Walter Lee <walt@tilera.com>
3431 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3432 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3433 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3434 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3436 2014-01-25 Walter Lee <walt@tilera.com>
3438 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3439 arguments on even registers.
3440 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3442 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3443 (BIGGEST_ALIGNMENT): Ditto.
3444 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3446 2014-01-25 Walter Lee <walt@tilera.com>
3448 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3449 insns before bundling.
3450 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3452 2014-01-25 Walter Lee <walt@tilera.com>
3454 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3455 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3456 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3458 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3460 * config/mips/constraints.md (kl): Delete.
3461 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3462 define expands, using...
3463 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3464 instructions for MIPS16.
3465 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3466 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3468 2014-01-25 Walter Lee <walt@tilera.com>
3470 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3474 2014-01-25 Walter Lee <walt@tilera.com>
3476 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3477 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3479 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3481 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3484 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3486 * print-rtl.c (in_call_function_usage): New var.
3487 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3488 EXPR_LIST mode as mode and not as reg note name.
3491 * cfgloopmanip.c (copy_loop_info): If
3492 loop->warned_aggressive_loop_optimizations, make sure
3493 the flag is set in target loop too.
3495 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3497 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3499 * builtins.def: Likewise.
3500 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3501 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3502 * ira.c (ira_setup_eliminable_regset): Likewise.
3503 * omp-low.c (gate_expand_omp): Likewise.
3504 (execute_lower_omp): Likewise.
3505 (diagnose_sb_0): Likewise.
3506 (gate_diagnose_omp_blocks): Likewise.
3507 (simd_clone_clauses_extract): Likewise.
3510 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3512 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3513 correction for little endian...
3514 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3517 2014-01-24 Jeff Law <law@redhat.com>
3519 PR tree-optimization/59919
3520 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3521 for non-returning calls.
3523 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3525 * common/config/aarch64/aarch64-common.c
3526 (aarch64_rewrite_mcpu): Handle multiple names.
3527 * config/aarch64/aarch64.h
3528 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3530 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3532 * input.c (add_file_to_cache_tab): Handle the case where fopen
3535 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3538 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3539 from push operand if code of push isn't PRE_DEC.
3541 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3544 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3545 -mquad-memory-atomic. Update -mquad-memory documentation to say
3546 it is only used for non-atomic loads/stores.
3548 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3549 -mquad-memory or -mquad-memory-atomic switches.
3551 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3552 -mquad-memory-atomic to ISA 2.07 support.
3554 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3555 to separate support of normal quad word memory operations (ldq, stq)
3556 from the atomic quad word memory operations.
3558 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3559 support to separate non-atomic quad word operations from atomic
3560 quad word operations. Disable non-atomic quad word operations in
3561 little endian mode so that we don't have to swap words after the
3562 load and before the store.
3563 (quad_load_store_p): Add comment about atomic quad word support.
3564 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3565 options printed with -mdebug=reg.
3567 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3568 -mquad-memory-atomic as the test for whether we have quad word
3569 atomic instructions.
3570 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3571 or -mp8-vector are used, allow byte/half-word atomic operations.
3573 * config/rs6000/sync.md (load_lockedti): Insure that the address
3574 is a proper indexed or indirect address for the lqarx instruction.
3575 On little endian systems, swap the hi/lo registers after the lqarx
3577 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3578 insure the address is valid for the lqarx instruction.
3579 (store_conditionalti): Insure that the address is a proper indexed
3580 or indirect address for the stqcrx. instruction. On little endian
3581 systems, swap the hi/lo registers before doing the stqcrx.
3583 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3584 insure the address is valid for the stqcrx. instruction.
3586 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3587 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3588 type of quad memory support is available.
3590 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3593 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3594 there is a danger of looping.
3596 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3598 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3599 force flag_ira_loop_pressure if set via command line.
3601 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3603 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3604 (ashr_simd): New builtin handling DI mode.
3605 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3606 (aarch64_sshr_simddi): New match pattern.
3607 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3608 (vshrd_n_s64): Likewise.
3609 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3611 2014-01-23 Nick Clifton <nickc@redhat.com>
3613 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3614 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3615 favour of mcu specific scripts.
3616 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3619 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3620 Alex Velenko <Alex.Velenko@arm.com>
3622 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3623 (vaddv_s16): Likewise.
3624 (vaddv_s32): Likewise.
3625 (vaddv_u8): Likewise.
3626 (vaddv_u16): Likewise.
3627 (vaddv_u32): Likewise.
3628 (vaddvq_s8): Likewise.
3629 (vaddvq_s16): Likewise.
3630 (vaddvq_s32): Likewise.
3631 (vaddvq_s64): Likewise.
3632 (vaddvq_u8): Likewise.
3633 (vaddvq_u16): Likewise.
3634 (vaddvq_u32): Likewise.
3635 (vaddvq_u64): Likewise.
3636 (vaddv_f32): Likewise.
3637 (vaddvq_f32): Likewise.
3638 (vaddvq_f64): Likewise.
3639 (vmaxv_f32): Likewise.
3640 (vmaxv_s8): Likewise.
3641 (vmaxv_s16): Likewise.
3642 (vmaxv_s32): Likewise.
3643 (vmaxv_u8): Likewise.
3644 (vmaxv_u16): Likewise.
3645 (vmaxv_u32): Likewise.
3646 (vmaxvq_f32): Likewise.
3647 (vmaxvq_f64): Likewise.
3648 (vmaxvq_s8): Likewise.
3649 (vmaxvq_s16): Likewise.
3650 (vmaxvq_s32): Likewise.
3651 (vmaxvq_u8): Likewise.
3652 (vmaxvq_u16): Likewise.
3653 (vmaxvq_u32): Likewise.
3654 (vmaxnmv_f32): Likewise.
3655 (vmaxnmvq_f32): Likewise.
3656 (vmaxnmvq_f64): Likewise.
3657 (vminv_f32): Likewise.
3658 (vminv_s8): Likewise.
3659 (vminv_s16): Likewise.
3660 (vminv_s32): Likewise.
3661 (vminv_u8): Likewise.
3662 (vminv_u16): Likewise.
3663 (vminv_u32): Likewise.
3664 (vminvq_f32): Likewise.
3665 (vminvq_f64): Likewise.
3666 (vminvq_s8): Likewise.
3667 (vminvq_s16): Likewise.
3668 (vminvq_s32): Likewise.
3669 (vminvq_u8): Likewise.
3670 (vminvq_u16): Likewise.
3671 (vminvq_u32): Likewise.
3672 (vminnmv_f32): Likewise.
3673 (vminnmvq_f32): Likewise.
3674 (vminnmvq_f64): Likewise.
3676 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3678 * config/aarch64/aarch64-simd.md
3679 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3680 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3681 (*aarch64_mul3_elt<mode>): Likewise.
3682 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3683 (*aarch64_mul3_elt_to_64v2df): Likewise.
3684 (*aarch64_mla_elt<mode>): Likewise.
3685 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3686 (*aarch64_mls_elt<mode>): Likewise.
3687 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3688 (*aarch64_fma4_elt<mode>): Likewise.
3689 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3690 (*aarch64_fma4_elt_to_64v2df): Likewise.
3691 (*aarch64_fnma4_elt<mode>): Likewise.
3692 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3693 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3694 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3695 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3696 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3697 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3698 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3699 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3700 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3702 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3704 * config/aarch64/aarch64-simd.md
3705 (aarch64_be_checked_get_lane<mode>): New define_expand.
3706 * config/aarch64/aarch64-simd-builtins.def
3707 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3708 New builtin definition.
3709 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3710 Use new safe be builtin.
3712 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3714 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3716 (aarch64_be_st1<mode>): Likewise.
3717 (aarch_ld1<VALL:mode>): Define_expand modified.
3718 (aarch_st1<VALL:mode>): Likewise.
3719 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3720 (UNSPEC_ST1): Likewise.
3722 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3724 * config/microblaze/microblaze.md: Add trap insn and attribute
3726 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3728 PR preprocessor/58580
3729 * input.h (location_get_source_line): Take an additional line_size
3731 (void diagnostics_file_cache_fini): Declare new function.
3732 * input.c (struct fcache): New type.
3733 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3734 New static constants.
3735 (diagnostic_file_cache_init, total_lines_num)
3736 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3737 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3738 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3739 (get_next_line, read_next_line, goto_next_line, read_line_num):
3740 New static function definitions.
3741 (diagnostic_file_cache_fini): New function.
3742 (location_get_source_line): Take an additional output line_len
3743 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3745 * diagnostic.c (diagnostic_finish): Call
3746 diagnostic_file_cache_fini.
3747 (adjust_line): Take an additional input parameter for the length
3748 of the line, rather than calculating it with strlen.
3749 (diagnostic_show_locus): Adjust the use of
3750 location_get_source_line and adjust_line with respect to their new
3751 signature. While displaying a line now, do not stop at the first
3752 null byte. Rather, display the zero byte as a space and keep
3753 going until we reach the size of the line.
3754 * Makefile.in: Add vec.o to OBJS-libcommon
3756 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3757 Ilya Tocar <ilya.tocar@intel.com>
3759 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3760 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3761 (__builtin_ia32_kmov16): Ditto.
3762 * config/i386/i386.md (UNSPEC_KMOV): New.
3765 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3767 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3768 (_mm512_storeu_si512): Ditto.
3770 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3773 * rtl.h (get_referenced_operands): Declare.
3774 * recog.c (get_referenced_operands): New function.
3775 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3776 operands have been referenced when recording LO_SUM references.
3778 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3780 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3782 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3784 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3785 Enable for generic and recent AMD targets.
3787 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3789 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3790 ARG_SIZE note when adjustment was eliminated.
3792 2014-01-22 Jeff Law <law@redhat.com>
3794 PR tree-optimization/59597
3795 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3796 in file. Accept new argument REGISTERING and use it to modify
3797 dump output appropriately.
3798 (register_jump_thread): Corresponding changes.
3799 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3800 thread paths involving joiner blocks. Add code to dump cancelled
3801 jump threading paths.
3803 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3805 PR rtl-optimization/59477
3806 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3807 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3809 2014-01-22 Tom Tromey <tromey@redhat.com>
3811 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3813 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3815 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3817 PR rtl-optimization/59896
3818 * lra-constraints.c (process_alt_operands): Check unused note for
3819 matched operands of insn with no output reloads.
3821 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3823 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3824 (mips_move_from_gpr_cost): Likewise.
3826 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3828 PR rtl-optimization/59858
3829 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3830 ira_class_hard_regs_num.
3831 (process_alt_operands): Increase reject for dying matched operand.
3833 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3836 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3837 smaller than size, perform several stores or loads and stores
3838 at dst + count - size to store or copy all of size bytes, rather
3839 than just last modesize bytes.
3841 2014-01-20 DJ Delorie <dj@redhat.com>
3843 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3844 that CLOBBERs are REGs before propogating their values.
3846 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3849 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3850 (cgraph_inline_failed_type): New function.
3851 * cgraph.h (DEFCIFCODE): Add type.
3852 (cgraph_inline_failed_type_t): New enum.
3853 (cgraph_inline_failed_type): New prototype.
3854 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3855 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3856 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3857 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3858 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3859 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3860 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3861 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3862 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3863 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3864 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3865 OPTIMIZATION_MISMATCH.
3866 * tree-inline.c (expand_call_inline): Emit errors during
3867 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3869 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3872 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3873 mode attribute in insn output.
3875 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3877 * output.h (output_constant): Delete.
3878 * varasm.c (output_constant): Make private.
3880 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3882 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3884 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3887 * tree.h (fold_builtin_strcat): New prototype.
3888 * builtins.c (fold_builtin_strcat): No longer static. Add len
3889 argument, if non-NULL, don't call c_strlen. Optimize
3890 directly into __builtin_memcpy instead of __builtin_strcpy.
3891 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3892 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3894 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3896 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3897 for SImode_address_operand operands, having only a REG argument.
3899 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3901 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3902 loader name using mbig-endian.
3903 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3905 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3907 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3911 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3913 * config/aarch64/aarch64-protos.h
3914 (aarch64_cannot_change_mode_class_ptr): Declare.
3915 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3916 aarch64_cannot_change_mode_class_ptr): New.
3917 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3918 backend hook aarch64_cannot_change_mode_class.
3920 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3922 * common/config/aarch64/aarch64-common.c
3923 (aarch64_handle_option): Don't handle any option order logic here.
3924 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3925 selected_cpu, warn on architecture version mismatch.
3926 (aarch64_override_options): Fix parsing order for option strings.
3928 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3929 Iain Sandoe <iain@codesourcery.com>
3932 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3933 warning. Amend comment to reflect current functionality.
3935 2014-01-20 Richard Biener <rguenther@suse.de>
3938 * builtins.c (fold_builtin_strcat): Remove case better handled
3939 by tree-ssa-strlen.c.
3941 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3943 * config/aarch64/aarch64.opt
3944 (mcpu, march, mtune): Make case-insensitive.
3946 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3949 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3950 if operands[1] is a REG or ZERO_EXTEND of a REG.
3952 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3954 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3956 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3958 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3959 long non-pic millicode calls.
3961 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3963 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3965 2014-01-19 Kito Cheng <kito@0xlab.org>
3967 * builtins.c (expand_movstr): Check movstr expand done or fail.
3969 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3970 H.J. Lu <hongjiu.lu@intel.com>
3973 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3974 to DImode for zero-extended addresses.
3976 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3978 PR rtl-optimization/57763
3979 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3980 on the new indirect jump_insn and increment LABEL_NUSES (label).
3982 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3986 * config.gcc (x86_archs): New variable.
3987 (x86_64_archs): Likewise.
3988 (x86_cpus): Likewise.
3989 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3990 --with-arch/--with-cpu= options.
3991 Support --with-arch=/--with-cpu={nehalem,westmere,
3992 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3994 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3996 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3997 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3999 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4001 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4003 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4006 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4007 clear cpp_get_options (parse_in)->warn_unused_macros for
4008 ix86_target_macros_internal with cpp_define.
4010 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4012 * jump.c (delete_related_insns): Keep (use (insn))s.
4013 * reorg.c (redundant_insn): Check for barriers too.
4015 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4017 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4019 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4021 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4022 call to $$dyncall when TARGET_LONG_CALLS is true.
4024 2014-01-17 Jeff Law <law@redhat.com>
4026 * ree.c (combine_set_extension): Temporarily disable test for
4027 changing number of hard registers.
4029 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4032 * ipa-inline-analysis.c (inline_free_summary):
4033 Do not free summary of aliases.
4035 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4038 * gimplify.c (gimplify_expr): Use create_tmp_var
4039 instead of create_tmp_var_raw. If cond doesn't have
4040 integral type, don't add the IFN_ANNOTATE builtin at all.
4042 2014-01-17 Martin Jambor <mjambor@suse.cz>
4045 * ipa-cp.c (prev_edge_clone): New variable.
4046 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4047 Also resize prev_edge_clone vector.
4048 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4049 (ipcp_edge_removal_hook): New function.
4050 (ipcp_driver): Register ipcp_edge_removal_hook.
4052 2014-01-17 Andrew Pinski <apinski@cavium.com>
4053 Steve Ellcey <sellcey@mips.com>
4056 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4059 2014-01-17 Jeff Law <law@redhat.com>
4062 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4063 so that pass_ccp runs first.
4065 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4067 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4068 (ix86_adjust_cost): Use !TARGET_XXX.
4069 (do_reorder_for_imul): Likewise.
4070 (swap_top_of_ready_list): Likewise.
4071 (ix86_sched_reorder): Likewise.
4073 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4075 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4076 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4077 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4078 (intel_memset): New. Duplicate slm_memset.
4079 (intel_cost): New. Duplicate slm_cost.
4080 (m_INTEL): New macro.
4081 (processor_target_table): Add "intel".
4082 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4083 with PROCESSOR_INTEL for "intel".
4084 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4085 PROCESSOR_SILVERMONT.
4086 (ix86_issue_rate): Likewise.
4087 (ix86_adjust_cost): Likewise.
4088 (ia32_multipass_dfa_lookahead): Likewise.
4089 (swap_top_of_ready_list): Likewise.
4090 (ix86_sched_reorder): Likewise.
4091 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4092 instead of TARGET_OPT_AGU.
4093 * config/i386/i386.h (TARGET_INTEL): New.
4094 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4095 (processor_type): Add PROCESSOR_INTEL.
4096 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4097 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4099 2014-01-17 Marek Polacek <polacek@redhat.com>
4102 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4105 2014-01-17 Richard Biener <rguenther@suse.de>
4107 PR tree-optimization/46590
4108 * opts.c (default_options_table): Add entries for
4109 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4110 all enabled at -O1 but not for -Og.
4111 * common.opt (fbranch-count-reg): Remove Init(1).
4112 (fmove-loop-invariants): Likewise.
4113 (ftree-pta): Likewise.
4115 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4117 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4118 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4119 decls to at least the GCC 4.8 used alignments.
4122 * tree-nested.c (convert_nonlocal_reference_stmt,
4123 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4124 of GIMPLE_BIND stmts, adjust associated decls.
4126 2014-01-17 Richard Biener <rguenther@suse.de>
4128 PR tree-optimization/46590
4129 * vec.h (vec<>::bseach): New member function implementing
4130 binary search according to C89 bsearch.
4131 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4132 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4133 bitmap pointer again. Make accesses_in_loop a flat array.
4134 (mem_ref_obstack): New global.
4135 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4136 (mark_ref_stored): Likewise.
4137 (ref_indep_loop_p_2): Likewise.
4138 (set_ref_stored_in_loop): New helper function.
4139 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4140 (memref_free): Adjust.
4141 (record_mem_ref_loc): Simplify.
4142 (gather_mem_refs_stmt): Adjust.
4143 (sort_locs_in_loop_postorder_cmp): New function.
4144 (analyze_memory_references): Sort accesses_in_loop after
4145 loop postorder number.
4146 (find_ref_loc_in_loop_cmp): New function.
4147 (for_all_locs_in_loop): Find relevant cluster of locs in
4148 accesses_in_loop and iterate without recursion.
4149 (execute_sm): Avoid uninit warning.
4150 (struct ref_always_accessed): Simplify.
4151 (ref_always_accessed::operator ()): Likewise.
4152 (ref_always_accessed_p): Likewise.
4153 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4154 loop postorder numbers here.
4155 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4158 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4161 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4162 on decls for which assemble_alias has been called.
4164 2014-01-17 Nick Clifton <nickc@redhat.com>
4166 * config/msp430/msp430.opt: (mcpu): New option.
4167 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4168 (msp430_option_override): Parse target_cpu. If the MCU name
4169 matches a generic string, clear target_mcu.
4170 (msp430_attr): Allow numeric interrupt values up to 63.
4171 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4172 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4174 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4176 * config/msp430/msp430.md (popm): Use %J rather than %I.
4177 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4178 (addhi_cy_i): Use immediate_operand for operand 2.
4179 * doc/invoke.texi: Document -mcpu option.
4181 2014-01-17 Richard Biener <rguenther@suse.de>
4183 PR rtl-optimization/38518
4184 * df.h (df_analyze_loop): Declare.
4185 * df-core.c: Include cfgloop.h.
4186 (df_analyze_1): Split out main part of df_analyze.
4187 (df_analyze): Adjust.
4188 (loop_inverted_post_order_compute): New function.
4189 (loop_post_order_compute): Likewise.
4190 (df_analyze_loop): New function avoiding whole-function
4192 * loop-invariant.c (find_defs): Use df_analyze_loop.
4193 (find_invariants): Adjust.
4194 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4196 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4198 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4199 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4201 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4203 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4204 traversal when removing references.
4206 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4209 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4211 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4214 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4215 pushing a reload for an autoinc when we had previously reloaded an
4216 inner part of the address.
4218 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4220 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4222 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4223 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4224 when not giving up or versioning for alias only because of
4226 (vect_analyze_data_ref_dependences): Set to true.
4227 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4229 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4230 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4234 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4237 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4238 operand 0 predicate for gathers, use a new pseudo as subtarget.
4240 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4243 * lra-constraints.c (process_alt_operands): Add printing debug info.
4244 Check absence of input/output reloads for matched operands too.
4246 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4248 PR rtl-optimization/59835
4249 * ira.c (ira_init_register_move_cost): Increase cost for
4252 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4254 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4256 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4259 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4260 non-register objects. Use gen_(high/low)part more consistently.
4263 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4266 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4267 endian support, remove tests for WORDS_BIG_ENDIAN.
4268 (p8_mfvsrd_3_<mode>): Likewise.
4269 (reload_gpr_from_vsx<mode>): Likewise.
4270 (reload_gpr_from_vsxsf): Likewise.
4271 (p8_mfvsrd_4_disf): Likewise.
4273 2014-01-16 Richard Biener <rguenther@suse.de>
4275 PR rtl-optimization/46590
4276 * lcm.c (compute_antinout_edge): Use postorder iteration.
4277 (compute_laterin): Use inverted postorder iteration.
4279 2014-01-16 Nick Clifton <nickc@redhat.com>
4282 * varasm.c (output_constant): Return the number of bytes actually
4284 (output_constructor_array_range): Update the field size with the
4285 number of bytes emitted by output_constant.
4286 (output_constructor_regular_field): Likewise. Also do not
4287 complain if the total number of bytes emitted is now greater
4288 than the expected fieldpos.
4289 * output.h (output_constant): Update prototype and descriptive comment.
4291 2014-01-16 Marek Polacek <polacek@redhat.com>
4294 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4295 it is error_mark_node.
4297 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4299 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4300 VALID_AVX256_REG_OR_OI_MODE.
4302 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4304 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4305 current procedure should be profiled.
4307 2014-01-15 Andrew Pinski <apinski@cavium.com>
4309 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4310 of moving from/to the STACK_REG register class.
4312 2014-01-15 Richard Henderson <rth@redhat.com>
4315 * reginfo.c (global_regs_decl): Globalize.
4316 * rtl.h (global_regs_decl): Declare.
4317 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4318 reserved via global_regs.
4320 2014-01-15 Teresa Johnson <tejohnson@google.com>
4322 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4324 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4326 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4327 and vmulosh rather than call gen_vec_widen_smult_*.
4328 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4329 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4330 (vec_widen_smult_even_v16qi): Likewise.
4331 (vec_widen_umult_even_v8hi): Likewise.
4332 (vec_widen_smult_even_v8hi): Likewise.
4333 (vec_widen_umult_odd_v16qi): Likewise.
4334 (vec_widen_smult_odd_v16qi): Likewise.
4335 (vec_widen_umult_odd_v8hi): Likewise.
4336 (vec_widen_smult_odd_v8hi): Likewise.
4337 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4338 vmuloub rather than call gen_vec_widen_umult_*.
4339 (vec_widen_umult_lo_v16qi): Likewise.
4340 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4341 vmulosb rather than call gen_vec_widen_smult_*.
4342 (vec_widen_smult_lo_v16qi): Likewise.
4343 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4344 rather than call gen_vec_widen_umult_*.
4345 (vec_widen_umult_lo_v8hi): Likewise.
4346 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4347 rather than call gen_vec_widen_smult_*.
4348 (vec_widen_smult_lo_v8hi): Likewise.
4350 2014-01-15 Jeff Law <law@redhat.com>
4352 PR tree-optimization/59747
4353 * ree.c (find_and_remove_re): Properly handle case where a second
4354 eliminated extension requires widening a copy created for elimination
4355 of a prior extension.
4356 (combine_set_extension): Ensure that the number of hard regs needed
4357 for a destination register does not change when we widen it.
4359 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4361 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4362 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4363 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4364 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4365 (avr-*-rtems*): Likewise.
4366 (bfin*-rtems*): Likewise.
4367 (moxie-*-rtems*): Likewise.
4368 (h8300-*-rtems*): Likewise.
4369 (i[34567]86-*-rtems*): Likewise.
4370 (lm32-*-rtems*): Likewise.
4371 (m32r-*-rtems*): Likewise.
4372 (m68k-*-rtems*): Likewise.
4373 (microblaze*-*-rtems*): Likewise.
4374 (mips*-*-rtems*): Likewise.
4375 (powerpc-*-rtems*): Likewise.
4376 (sh-*-rtems*): Likewise.
4377 (sparc-*-rtems*): Likewise.
4378 (sparc64-*-rtems*): Likewise.
4379 (v850-*-rtems*): Likewise.
4380 (m32c-*-rtems*): Likewise.
4382 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4384 PR rtl-optimization/59511
4385 * ira.c (ira_init_register_move_cost): Use memory costs for some
4386 cases of register move cost calculations.
4387 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4388 instead of BB frequency.
4389 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4390 * lra-assigns.c (find_hard_regno_for): Ditto.
4392 2014-01-15 Richard Biener <rguenther@suse.de>
4394 PR tree-optimization/59822
4395 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4396 (vectorizable_load): Use it to hoist defs of uses of invariant
4397 loads out of the loop.
4399 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4400 Kugan Vivekanandarajah <kuganv@linaro.org>
4403 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4406 2014-01-15 Richard Biener <rguenther@suse.de>
4408 PR rtl-optimization/59802
4409 * lcm.c (compute_available): Use inverted postorder to seed
4410 the initial worklist.
4412 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4415 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4416 ADDR_REGS for invalid symrefs in non-PIC code.
4418 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4421 * builtins.c (determine_block_size): Initialize *probable_max_size
4422 even if len_rtx is CONST_INT.
4424 2014-01-14 Andrew Pinski <apinski@cavium.com>
4426 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4427 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4428 (cortexa53_tunings): Likewise.
4429 (aarch64_sched_issue_rate): New function.
4430 (TARGET_SCHED_ISSUE_RATE): Define.
4432 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4434 * ira-costs.c (find_costs_and_classes): Add missed
4435 ira_init_register_move_cost_if_necessary.
4437 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4440 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4442 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4445 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4446 to indicate if type is used for function return value. Warn ABI
4447 change if the vector mode isn't available for function return value.
4448 (ix86_function_arg_advance): Pass false to type_natural_mode.
4449 (ix86_function_arg): Likewise.
4450 (ix86_gimplify_va_arg): Likewise.
4451 (function_arg_32): Don't warn ABI change.
4452 (ix86_function_value): Pass true to type_natural_mode.
4453 (ix86_return_in_memory): Likewise.
4454 (ix86_struct_value_rtx): Removed.
4455 (TARGET_STRUCT_VALUE_RTX): Likewise.
4457 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4459 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4460 converting a conditional jump into a conditional return.
4462 2014-01-14 Richard Biener <rguenther@suse.de>
4464 PR tree-optimization/58921
4465 PR tree-optimization/59006
4466 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4467 hoisting invariant stmts.
4468 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4469 invariant loads on the preheader edge if possible.
4471 2014-01-14 Joey Ye <joey.ye@arm.com>
4473 * doc/plugin.texi (Building GCC plugins): Update to C++.
4475 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4477 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4478 (_mm_rcp28_round_ss): Ditto.
4479 (_mm_rsqrt28_round_sd): Ditto.
4480 (_mm_rsqrt28_round_ss): Ditto.
4481 (_mm_rcp28_sd): Ditto.
4482 (_mm_rcp28_ss): Ditto.
4483 (_mm_rsqrt28_sd): Ditto.
4484 (_mm_rsqrt28_ss): Ditto.
4485 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4486 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4487 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4488 (IX86_BUILTIN_RCP28SD): Ditto.
4489 (IX86_BUILTIN_RCP28SS): Ditto.
4490 (IX86_BUILTIN_RSQRT28SD): Ditto.
4491 (IX86_BUILTIN_RSQRT28SS): Ditto.
4492 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4493 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4494 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4495 (ix86_expand_special_args_builtin): Expand new FTYPE.
4496 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4497 (srcp14<mode>): Make insn unary.
4498 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4499 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4500 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4501 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4502 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4503 Fix rounding: make it SAE only.
4504 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4506 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4508 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4509 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4510 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4511 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4512 (round_saeonly_mask_scalar_operand4): Ditto.
4513 (round_saeonly_mask_scalar_op3): Ditto.
4514 (round_saeonly_mask_scalar_op4): Ditto.
4516 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4518 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4519 Implement -maltivec=be for vec_insert and vec_extract.
4521 2014-01-10 DJ Delorie <dj@redhat.com>
4523 * config/msp430/msp430.md (call_internal): Don't allow memory
4524 references with SP as the base register.
4525 (call_value_internal): Likewise.
4526 * config/msp430/constraints.md (Yc): New. For memory references
4527 that don't use SP as a base register.
4529 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4530 "an integer without a # prefix"
4531 * config/msp430/msp430.md (epilogue_helper): Use it.
4533 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4536 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4537 AVX512F gather builtins.
4538 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4539 on gather decls with INTEGER_TYPE masktype.
4540 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4541 directly into the builtin rather than hoisting it before loop.
4543 PR tree-optimization/59387
4544 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4545 (scev_const_prop): If folded_casts and type has undefined overflow,
4546 use force_gimple_operand instead of force_gimple_operand_gsi and
4547 for each added stmt if it is assign with
4548 arith_code_with_undefined_signed_overflow, call
4549 rewrite_to_defined_overflow.
4550 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4551 gimple-fold.h instead.
4552 (arith_code_with_undefined_signed_overflow,
4553 rewrite_to_defined_overflow): Moved to ...
4554 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4555 rewrite_to_defined_overflow): ... here. No longer static.
4556 Include gimplify-me.h.
4557 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4558 rewrite_to_defined_overflow): New prototypes.
4560 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4562 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4564 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4566 * builtins.c (get_object_alignment_2): Minor tweak.
4567 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4569 2014-01-13 Christian Bruel <christian.bruel@st.com>
4571 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4572 optimized non constant lengths.
4574 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4577 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4578 load as __atomic_load_N if possible.
4580 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4582 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4584 (rs6000_expand_builtin): Adjust call.
4586 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4589 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4590 * config/rs6000/rs6000.c: Include target-globals.h.
4591 (rs6000_set_current_function): Instead of doing target_reinit
4592 unconditionally, use save_target_globals_default_opts and
4593 restore_target_globals.
4595 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4597 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4598 (rs6000_expand_builtin): Handle mffs and mtfsf.
4599 (rs6000_init_builtins): Define mffs and mtfsf.
4600 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4601 (rs6000_mffs): New pattern.
4602 (rs6000_mtfsf): New pattern.
4604 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4606 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4607 Start narrowing with START. Apply candidate-use pair
4608 and check overall cost in narrowing.
4609 (iv_ca_prune): Pass new argument.
4611 2014-01-10 Jeff Law <law@redhat.com>
4614 * ree.c (combine_reaching_defs): Ensure the defining statement
4615 occurs before the extension when optimizing extensions with
4616 different source and destination hard registers.
4618 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4621 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4622 vtables into the type inheritance graph.
4624 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4626 PR rtl-optimization/59754
4627 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4628 modes in the REGNO != REGNO case.
4630 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4632 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4634 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4636 PR tree-optimization/59745
4637 * tree-predcom.c (tree_predictive_commoning_loop): Call
4638 free_affine_expand_cache if giving up because components is NULL.
4640 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4641 GC in payload of target_globals struct instead of allocating them on
4642 the heap and the larger structs separately using GC.
4643 * target-globals.h (struct target_globals): Make regs, hard_regs,
4644 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4645 of GTY((skip)) and change type to void *.
4646 (reset_target_globals): Cast loads from those fields to corresponding
4649 2014-01-10 Steve Ellcey <sellcey@mips.com>
4652 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4653 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4654 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4656 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4659 * aarch64-modes.def (CC_Zmode): New flags mode.
4660 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4661 represents an equality.
4662 (aarch64_get_condition_code): Handle CC_Zmode.
4663 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4665 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4667 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4668 extraction in good case.
4670 2014-01-10 Richard Biener <rguenther@suse.de>
4672 PR tree-optimization/59374
4673 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4674 checking after SLP discovery. Mark stmts not participating
4675 in any SLP instance properly.
4677 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4679 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4680 when handling a SET rtx.
4682 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4684 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4685 (cortex-a57): Likewise.
4686 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4688 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4690 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4691 non-iwmmxt builtins.
4693 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4697 * ipa-devirt.c record_target_from_binfo): Take as argument
4698 stack of binfos and lookup matching one for virtual inheritance.
4699 (possible_polymorphic_call_targets_1): Update.
4701 2014-01-10 Huacai Chen <chenhc@lemote.com>
4703 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4704 kernel strings for Loongson-2E/2F/3A.
4706 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4709 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4710 is_gimple_call before calling gimple_call_internal_p.
4712 2014-01-09 Steve Ellcey <sellcey@mips.com>
4714 * Makefile.in (TREE_FLOW_H): Remove.
4715 (TREE_SSA_H): Add file names from tree-flow.h.
4716 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4717 * tree.h: Remove tree-flow.h reference.
4718 * hash-table.h: Remove tree-flow.h reference.
4719 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4720 reference with tree-ssa-loop.h.
4722 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4724 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4725 default element-order behavior for -maltivec.
4726 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4728 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4729 when targeting big endian, at least for now.
4730 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4732 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4735 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4736 var satisfies use_register_for_decl, just take into account type
4737 alignment, rather than decl alignment.
4739 PR tree-optimization/59622
4740 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4741 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4742 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4743 Don't devirtualize for inplace at all. For targets.length () == 1,
4744 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4746 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4748 * config/i386/i386.md (cpu): Remove the unused btver1.
4750 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4752 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4754 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4757 * tree-core.h (struct target_globals): New forward declaration.
4758 (struct tree_target_option): Add globals field.
4759 * tree.h (TREE_TARGET_GLOBALS): Define.
4760 (prepare_target_option_nodes_for_pch): New prototype.
4761 * target-globals.h (struct target_globals): Define even if
4763 * tree.c (prepare_target_option_node_for_pch,
4764 prepare_target_option_nodes_for_pch): New functions.
4765 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4766 * config/i386/i386.c: Include target-globals.h.
4767 (ix86_set_current_function): Instead of doing target_reinit
4768 unconditionally, use save_target_globals_default_opts and
4769 restore_target_globals.
4771 2014-01-09 Richard Biener <rguenther@suse.de>
4773 PR tree-optimization/59715
4774 * tree-cfg.h (split_critical_edges): Declare.
4775 * tree-cfg.c (split_critical_edges): Export.
4776 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4778 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4780 * cfgexpand.c (expand_stack_vars): Optionally disable
4781 asan stack protection.
4782 (expand_used_vars): Likewise.
4783 (partition_stack_vars): Likewise.
4784 * asan.c (asan_emit_stack_protection): Optionally disable
4785 after return stack usage.
4786 (instrument_derefs): Optionally disable memory access instrumentation.
4787 (instrument_builtin_call): Likewise.
4788 (instrument_strlen_call): Likewise.
4789 (asan_protect_global): Optionally disable global variables protection.
4790 * doc/invoke.texi: Added doc for new options.
4791 * params.def: Added new options.
4792 * params.h: Likewise.
4794 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4796 PR rtl-optimization/59724
4797 * ifcvt.c (cond_exec_process_if_block): Don't call
4798 flow_find_head_matching_sequence with 0 longest_match.
4799 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4800 non-active insns if !stop_after.
4801 (try_head_merge_bb): Revert 2014-01-07 changes.
4803 2014-01-08 Jeff Law <law@redhat.com>
4805 * ree.c (get_sub_rtx): New function, extracted from...
4806 (merge_def_and_ext): Here.
4807 (combine_reaching_defs): Use get_sub_rtx.
4809 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4811 * cgraph.h (varpool_variable_node): Do not choke on null node.
4813 2014-01-08 Catherine Moore <clm@codesourcery.com>
4815 * config/mips/mips.md (simple_return): Attempt to use JRC
4817 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4819 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4821 PR rtl-optimization/59137
4822 * reorg.c (steal_delay_list_from_target): Call update_block for
4824 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4826 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4828 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4829 two duplicate entries.
4831 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4834 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4836 * config/mips/mips.c (mips_truncated_op_cost): New function.
4837 (mips_rtx_costs): Adjust test for BADDU.
4838 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4840 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4842 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4843 (*baddu_si): ...this new pattern.
4845 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4848 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4850 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4853 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4855 (expand_expr, expand_normal): Adjust.
4856 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4857 inner_reference_p. Use inner_reference_p to expand inner references.
4858 (store_expr): Adjust.
4859 * cfgexpand.c (expand_call_stmt): Adjust.
4861 2014-01-08 Rong Xu <xur@google.com>
4863 * gcov-io.c (gcov_var): Move from gcov-io.h.
4864 (gcov_position): Ditto.
4865 (gcov_is_error): Ditto.
4866 (gcov_rewrite): Ditto.
4867 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4868 only part to libgcc/libgcov.h.
4870 2014-01-08 Marek Polacek <polacek@redhat.com>
4873 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4875 2014-01-08 Marek Polacek <polacek@redhat.com>
4878 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4880 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4882 PR rtl-optimization/59649
4883 * stor-layout.c (get_mode_bounds): For BImode return
4884 0 and STORE_FLAG_VALUE.
4886 2014-01-08 Richard Biener <rguenther@suse.de>
4889 * gimple.h (is_gimple_builtin_call): Remove.
4890 (gimple_builtin_call_types_compatible_p): New.
4891 (gimple_call_builtin_p): New overload.
4892 * gimple.c (is_gimple_builtin_call): Remove.
4893 (validate_call): Rename to ...
4894 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4896 (validate_type): New static function.
4897 (gimple_call_builtin_p): New overload and adjust.
4898 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4899 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4900 (gimple_fold_stmt_to_constant_1): Likewise.
4901 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4903 2014-01-08 Richard Biener <rguenther@suse.de>
4906 * gimplify.c (gimplify_expr): Gimplify register-register type
4907 VIEW_CONVERT_EXPRs to separate stmts.
4909 2014-01-07 Jeff Law <law@redhat.com>
4912 * ree.c (combine_set_extension): Handle case where source
4913 and destination registers in an extension insn are different.
4914 (combine_reaching_defs): Allow source and destination registers
4915 in extension to be different under limited circumstances.
4916 (add_removable_extension): Remove restriction that the
4917 source and destination registers in the extension are the same.
4918 (find_and_remove_re): Emit a copy from the extension's
4919 destination to its source after the defining insn if
4920 the source and destination registers are different.
4923 * ifcvt.c (merge_if_block): If we are merging a block with more than
4924 one successor with a block with no successors, remove any BARRIER
4925 after the second block.
4927 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4929 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4931 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4934 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4935 for 14-bit register offsets when INT14_OK_STRICT is false.
4937 2014-01-07 Roland Stigge <stigge@antcom.de>
4938 Michael Meissner <meissner@linux.vnet.ibm.com>
4941 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4942 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4944 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4946 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4949 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4951 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4952 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4953 rtx is const0_rtx or not.
4955 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4958 * target-globals.c (save_target_globals): Remove this_fn_optab
4960 * toplev.c: Include optabs.h.
4961 (target_reinit): Temporarily restore the global options if another
4962 set of options are in force.
4964 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4966 PR rtl-optimization/58668
4967 * cfgcleanup.c (flow_find_cross_jump): Don't count
4968 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4969 to determine what is counted.
4970 (flow_find_head_matching_sequence): Use active_insn_p to determine
4972 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4974 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4975 determine what is counted.
4977 PR tree-optimization/59643
4978 * tree-predcom.c (split_data_refs_to_components): If one dr is
4979 read and one write, determine_offset fails and the write isn't
4980 in the bad component, just put the read into the bad component.
4982 2014-01-07 Mike Stump <mikestump@comcast.net>
4983 Jakub Jelinek <jakub@redhat.com>
4986 * tree-core.h (struct tree_optimization_option): Change optabs
4987 type from unsigned char * to void *.
4988 * optabs.c (init_tree_optimization_optabs): Adjust
4989 TREE_OPTIMIZATION_OPTABS initialization.
4991 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4994 * config/i386/i386.h (struct machine_function): Add
4995 no_drap_save_restore field.
4996 * config/i386/i386.c (ix86_save_reg): Use
4997 !cfun->machine->no_drap_save_restore instead of
4998 crtl->stack_realign_needed.
4999 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5000 this function clears frame_pointer_needed. Set
5001 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5002 and DRAP reg is needed.
5004 2014-01-06 Marek Polacek <polacek@redhat.com>
5007 * doc/implement-c.texi: Mention that other integer types are
5008 permitted as bit-field types in strictly conforming mode.
5010 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5012 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5015 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5017 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5019 2014-01-06 Martin Jambor <mjambor@suse.cz>
5022 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5024 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5026 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5030 * var-tracking.c (add_stores): Preserve the value of the source even if
5031 we don't record the store.
5033 2014-01-06 Terry Guo <terry.guo@arm.com>
5035 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5037 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5040 * config/darwin.c (darwin_function_section): Adjust return values to
5041 correspond to optimisation changes made in r206070.
5043 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5045 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5046 from prefetch_block tune setting.
5047 (nocona_cost): Correct size of prefetch block to 64.
5049 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5051 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5052 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5053 used to save the static chain register in the computation of the offset
5054 from which the FP registers need to be restored.
5056 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5058 PR tree-optimization/59519
5059 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5060 ICE if get_current_def (current_new_name) is already non-NULL, as long
5061 as it is a phi result of some other phi in *new_exit_bb that has
5064 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5065 or vmovdqu* for misaligned_operand.
5066 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5067 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5068 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5069 aligned_mem for AVX512F masked aligned load and store builtins and for
5072 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5074 PR tree-optimization/59651
5075 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5076 Address range for negative step should be added by TYPE_SIZE_UNIT.
5078 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5080 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5081 overlapping registers also for registers other than the stack pointer.
5083 2014-01-03 Marek Polacek <polacek@redhat.com>
5086 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5089 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5092 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5095 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5096 (push splitter): Use <P:MODE_SIZE> instead of
5097 GET_MODE_SIZE (<P:MODE>mode).
5098 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5099 (mov -1, reg peephole2): Likewise.
5100 * config/i386/sse.md (*mov<mode>_internal,
5101 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5102 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5103 *<code><mode>3, *andnot<mode>3<mask_name>,
5104 <mask_codefor><code><mode>3<mask_name>): Likewise.
5105 * config/i386/subst.md (mask_mode512bit_condition,
5106 sd_mask_mode512bit_condition): Likewise.
5108 2014-01-02 Xinliang David Li <davidxl@google.com>
5110 PR tree-optimization/59303
5111 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5112 (dump_predicates): Better output format.
5113 (pred_equal_p): New function.
5114 (is_neq_relop_p): Ditto.
5115 (is_neq_zero_form_p): Ditto.
5116 (pred_expr_equal_p): Ditto.
5117 (pred_neg_p): Ditto.
5118 (simplify_pred): Ditto.
5119 (simplify_preds_2): Ditto.
5120 (simplify_preds_3): Ditto.
5121 (simplify_preds_4): Ditto.
5122 (simplify_preds): Ditto.
5124 (push_to_worklist): Ditto.
5125 (get_pred_info_from_cmp): Ditto.
5126 (is_degenerated_phi): Ditto.
5127 (normalize_one_pred_1): Ditto.
5128 (normalize_one_pred): Ditto.
5129 (normalize_one_pred_chain): Ditto.
5130 (normalize_preds): Ditto.
5131 (normalize_cond_1): Remove function.
5132 (normalize_cond): Ditto.
5133 (is_gcond_subset_of): Ditto.
5134 (is_subset_of_any): Ditto.
5135 (is_or_set_subset_of): Ditto.
5136 (is_and_set_subset_of): Ditto.
5137 (is_norm_cond_subset_of): Ditto.
5138 (pred_chain_length_cmp): Ditto.
5139 (convert_control_dep_chain_into_preds): Type change.
5140 (find_predicates): Ditto.
5141 (find_def_preds): Ditto.
5142 (destroy_predicates_vecs): Ditto.
5143 (find_matching_predicates_in_rest_chains): Ditto.
5144 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5145 (is_pred_expr_subset): Ditto.
5146 (is_pred_chain_subset_of): Ditto.
5147 (is_included_in): Ditto.
5148 (is_superset_of): Ditto.
5150 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5152 Update copyright years.
5154 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5156 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5157 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5158 config/arc/arc.md, config/arc/arc.opt,
5159 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5160 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5161 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5162 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5163 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5164 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5165 vtable-verify.c, vtable-verify.h: Use the standard form for the
5168 2014-01-02 Tobias Burnus <burnus@net-b.de>
5170 * gcc.c (process_command): Update copyright notice dates.
5171 * gcov-dump.c: Ditto.
5173 * doc/cpp.texi: Bump @copying's copyright year.
5174 * doc/cppinternals.texi: Ditto.
5175 * doc/gcc.texi: Ditto.
5176 * doc/gccint.texi: Ditto.
5177 * doc/gcov.texi: Ditto.
5178 * doc/install.texi: Ditto.
5179 * doc/invoke.texi: Ditto.
5181 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5183 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5185 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5187 * config/i386/sse.md (*mov<mode>_internal): Guard
5188 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5190 PR rtl-optimization/59647
5191 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5192 new_rtx into UNSIGNED_FLOAT rtxes.
5194 Copyright (C) 2014 Free Software Foundation, Inc.
5196 Copying and distribution of this file, with or without modification,
5197 are permitted in any medium without royalty provided the copyright
5198 notice and this notice are preserved.