1 2014-03-18 Kai Tietz <ktietz@redhat.com>
3 PR rtl-optimization/56356
4 * sdbout.c (sdbout_parms): Verify that parms'
5 incoming argument is valid.
6 (sdbout_reg_parms): Likewise.
8 2014-03-18 Richard Henderson <rth@redhat.com>
11 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
12 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
13 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
15 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
17 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
18 * doc/plugins.texi (Plugin callbacks): Mention
20 Italicize plugin event names in description. Explain that
21 PLUGIN_PRAGMAS has no sense for lto1. Explain
23 Remind that no GCC functions should be called after
25 Explain what pragmas with expansion are.
27 2014-03-18 Martin Liska <mliska@suse.cz>
29 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
30 gimple call statement is update.
31 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
32 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
34 2014-03-18 Jakub Jelinek <jakub@redhat.com>
37 * ubsan.c (ubsan_instrument_unreachable): Call
38 initialize_sanitizer_builtins.
39 (ubsan_pass): Likewise.
42 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
43 varpool_finalize_decl instead of rest_of_decl_compilation.
45 2014-03-18 Richard Biener <rguenther@suse.de>
47 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
48 by using bitmap_and_compl instead of bitmap_and_compl_into.
49 (df_rd_transfer_function): Likewise.
51 2014-03-18 Richard Biener <rguenther@suse.de>
53 * doc/lto.texi (fresolution): Fix typo.
55 2014-03-18 Richard Biener <rguenther@suse.de>
57 * doc/invoke.texi (flto): Update for changes in 4.9.
59 2014-03-18 Richard Biener <rguenther@suse.de>
61 * doc/loop.texi: Remove section on the removed lambda framework.
62 Update loop docs with recent changes in preserving loop structure.
64 2014-03-18 Richard Biener <rguenther@suse.de>
66 * doc/lto.texi (-fresolution): Document.
68 2014-03-18 Richard Biener <rguenther@suse.de>
70 * doc/contrib.texi: Adjust my name.
72 2014-03-18 Jakub Jelinek <jakub@redhat.com>
75 * internal-fn.c: Include diagnostic-core.h.
76 (expand_BUILTIN_EXPECT): New function.
77 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
78 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
79 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
80 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
82 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
83 Revert 3 argument __builtin_expect code.
84 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
85 * gimple-fold.c (gimple_fold_call): Likewise.
86 * tree.h (fold_builtin_expect): New prototype.
87 * builtins.c (build_builtin_expect_predicate): Add predictor
88 argument, if non-NULL, create 3 argument __builtin_expect.
89 (fold_builtin_expect): No longer static. Add ARG2 argument,
90 pass it through to build_builtin_expect_predicate.
91 (fold_builtin_2): Adjust caller.
92 (fold_builtin_3): Handle BUILT_IN_EXPECT.
93 * internal-fn.def (BUILTIN_EXPECT): New.
95 2014-03-18 Tobias Burnus <burnus@net-b.de>
98 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
99 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
100 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
102 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
105 * predict.c (combine_predictions_for_bb): Fix up formatting.
106 (expr_expected_value_1, expr_expected_value): Add predictor argument,
107 fill what it points to if non-NULL.
108 (tree_predict_by_opcode): Adjust caller, use the predictor.
109 * predict.def (PRED_COMPARE_AND_SWAP): Add.
111 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
113 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
114 proper constant for the store mode.
116 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
118 * symtab.c (change_decl_assembler_name): Fix transparent alias
121 2014-03-16 Renlin Li <Renlin.Li@arm.com>
123 * config/aarch64/aarch64.c: Correct the comments about the
124 aarch64 stack layout.
126 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
128 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
129 check for GF_OMP_FOR_KIND_FOR.
131 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
133 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
134 ymm and zmm register names.
136 2014-03-17 Jakub Jelinek <jakub@redhat.com>
139 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
140 note creation for the 2010-08-31 changes.
142 2014-03-17 Marek Polacek <polacek@redhat.com>
145 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
146 as -fno-tree-loop-vectorize.
147 (expand_omp_simd): Likewise.
149 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
151 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
152 (eligible_for_call_delay): New prototype.
153 * config/sparc/sparc.c (tls_call_delay): Rename into...
154 (eligible_for_call_delay): ...this. Return false if the instruction
155 cannot be put in the delay slot of a branch.
156 (eligible_for_restore_insn): Simplify.
157 (eligible_for_return_delay): Return false if the instruction cannot be
158 put in the delay slot of a branch and simplify.
159 (eligible_for_sibcall_delay): Return false if the instruction cannot be
160 put in the delay slot of a branch.
161 * config/sparc/sparc.md (fix_ut699): New attribute.
162 (tls_call_delay): Delete.
163 (in_call_delay): Reimplement.
164 (eligible_for_sibcall_delay): Rename into...
165 (in_sibcall_delay): ...this.
166 (eligible_for_return_delay): Rename into...
167 (in_return_delay): ...this.
168 (in_branch_delay): Reimplement.
169 (in_uncond_branch_delay): Delete.
170 (in_annul_branch_delay): Delete.
172 2014-03-14 Richard Henderson <rth@redhat.com>
175 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
176 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
177 (*floathi<X87MODEF>2_i387_with_temp): Remove.
178 (floathi splitters): Remove.
179 (float<SWI48x>xf2): New pattern.
180 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
181 code that tried to handle DImode for 32-bit, but which was excluded
182 by the pattern's condition. Drop allocation of stack temporary.
183 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
184 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
185 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
186 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
187 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
188 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
189 (*float<SWI48><MODEF>2_sse_interunit): Remove.
190 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
191 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
192 (*float<SWI48x><X87MODEF>2_i387): Remove.
193 (all float _with_temp splitters): Remove.
194 (*float<SWI48x><MODEF>2_i387): New pattern.
195 (*float<SWI48><MODEF>2_sse): New pattern.
196 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
197 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
199 2014-03-14 Jakub Jelinek <jakub@redhat.com>
200 Marek Polacek <polacek@redhat.com>
203 * common.opt (dump_base_name_prefixed): New Variable.
204 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
205 if x_dump_base_name_prefixed is already set, set it at the end.
207 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
209 PR rtl-optimization/60508
210 * lra-constraints.c (get_reload_reg): Add new parameter
212 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
213 Pass the new parameter values.
215 2014-03-14 Richard Biener <rguenther@suse.de>
217 * common.opt: Revert unintented changes from r205065.
220 2014-03-14 Richard Biener <rguenther@suse.de>
223 * cfghooks.c (split_block): Properly adjust all loops the
224 block was a latch of.
226 2014-03-14 Martin Jambor <mjambor@suse.cz>
229 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
232 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
235 * config/avr/avr.c (avr_set_current_function): Pass function name
236 through default_strip_name_encoding before sanity checking instead
237 of skipping the first char of the assembler name.
239 2014-03-13 Richard Henderson <rth@redhat.com>
242 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
243 (ix86_force_to_memory, ix86_free_from_memory): Remove.
244 * config/i386/i386-protos.h: Likewise.
245 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
246 in the expander instead of a splitter.
247 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
248 any possibility of requiring a memory.
249 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
250 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
251 (fp branch splitters): Update for ix86_split_fp_branch.
252 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
253 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
254 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
255 (*fop_<MODEF>_2_i387): Remove f/r alternative.
256 (*fop_<MODEF>_3_i387): Likewise.
257 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
258 (splitters for the fop_* register patterns): Remove.
259 (fscalexf4_i387): Rename from *fscalexf4_i387.
260 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
262 2014-03-13 Jakub Jelinek <jakub@redhat.com>
264 PR tree-optimization/59779
265 * tree-dfa.c (get_ref_base_and_extent): Use double_int
266 type for bitsize and maxsize instead of HOST_WIDE_INT.
268 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
270 PR rtl-optimization/57320
271 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
272 the CFG after thread_prologue_and_epilogue_insns.
274 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
276 PR rtl-optimization/57189
277 * lra-constraints.c (process_alt_operands): Disfavor spilling
280 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
282 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
285 2014-03-13 Jakub Jelinek <jakub@redhat.com>
287 PR tree-optimization/59025
289 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
290 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
292 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
295 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
296 calls of avr_out_plus_1.
298 2014-03-13 Bin Cheng <bin.cheng@arm.com>
300 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
301 BB's single pred and update the father loop's latch info later.
303 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
305 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
309 (VEC_base): Likewise.
310 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
311 registers, we need to swap double words in little endian mode.
313 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
314 to be a container mode for 128-bit integer operations added in ISA
315 2.07. Unlike TImode and PTImode, the preferred register set is
316 the Altivec/VMX registers for the 128-bit operations.
318 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
320 (rs6000_split_128bit_ok_p): Likewise.
322 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
323 macros for creating ISA 2.07 normal and overloaded builtin
324 functions with 3 arguments.
325 (BU_P8V_OVERLOAD_3): Likewise.
326 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
327 for use as overloaded functions.
328 (VPERM_1TI_UNS): Likewise.
329 (VSEL_1TI): Likewise.
330 (VSEL_1TI_UNS): Likewise.
331 (ST_INTERNAL_1ti): Likewise.
332 (LD_INTERNAL_1ti): Likewise.
333 (XXSEL_1TI): Likewise.
334 (XXSEL_1TI_UNS): Likewise.
335 (VPERM_1TI): Likewise.
336 (VPERM_1TI_UNS): Likewise.
337 (XXPERMDI_1TI): Likewise.
339 (LXVD2X_V1TI): Likewise.
340 (STXVD2X_V1TI): Likewise.
341 (VEC_INIT_V1TI): Likewise.
342 (VEC_SET_V1TI): Likewise.
343 (VEC_EXT_V1TI): Likewise.
344 (EQV_V1TI): Likewise.
345 (NAND_V1TI): Likewise.
346 (ORC_V1TI): Likewise.
347 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
348 added in ISA 2.07. Add both normal 'altivec' builtins, and the
352 (VADDEUQM): Likewise.
353 (VADDECUQ): Likewise.
354 (VSUBEUQM): Likewise.
355 (VSUBECUQ): Likewise.
357 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
358 __int128_t and __uint128_t types.
359 (__uint128_type): Likewise.
360 (altivec_categorize_keyword): Add support for vector __int128_t,
361 vector __uint128_t, vector __int128, and vector unsigned __int128
362 as a container type for TImode operations that need to be done in
363 VSX/Altivec registers.
364 (rs6000_macro_to_expand): Likewise.
365 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
366 to support 128-bit integer instructions vaddcuq, vadduqm,
367 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
368 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
370 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
371 for V1TImode, and set up preferences to use VSX/Altivec registers.
372 Setup VSX reload handlers.
373 (rs6000_debug_reg_global): Likewise.
374 (rs6000_init_hard_regno_mode_ok): Likewise.
375 (rs6000_preferred_simd_mode): Likewise.
376 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
377 (easy_altivec_constant): Likewise.
378 (output_vec_const_move): Likewise.
379 (rs6000_expand_vector_set): Convert V1TImode set and extract to
381 (rs6000_expand_vector_extract): Likewise.
382 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
384 (rs6000_const_vec): Add support for V1TImode.
385 (rs6000_emit_le_vsx_load): Swap double words when loading or
386 storing TImode/V1TImode.
387 (rs6000_emit_le_vsx_store): Likewise.
388 (rs6000_emit_le_vsx_move): Likewise.
389 (rs6000_emit_move): Add support for V1TImode.
390 (altivec_expand_ld_builtin): Likewise.
391 (altivec_expand_st_builtin): Likewise.
392 (altivec_expand_vec_init_builtin): Likewise.
393 (altivec_expand_builtin): Likewise.
394 (rs6000_init_builtins): Add support for V1TImode type. Add
395 support for ISA 2.07 128-bit integer builtins. Define type names
396 for the VSX/Altivec vector types.
397 (altivec_init_builtins): Add support for overloaded vector
398 functions with V1TImode type.
399 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
400 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
402 (rs6000_split_128bit_ok_p): Likewise.
403 (rs6000_handle_altivec_attribute): Create V1TImode from vector
404 __int128_t and vector __uint128_t.
406 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
414 (VS_scalar): Likewise.
415 (VS_double): Likewise.
416 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
418 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
419 we support the ISA 2.07 128-bit integer arithmetic instructions.
420 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
421 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
422 and TImode types for use with the builtin functions.
423 (V1TI_type_node): Likewise.
424 (unsigned_V1TI_type_node): Likewise.
425 (intTI_type_internal_node): Likewise.
426 (uintTI_type_internal_node): Likewise.
428 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
429 128-bit builtin functions.
430 (UNSPEC_VADDEUQM): Likewise.
431 (UNSPEC_VADDECUQ): Likewise.
432 (UNSPEC_VSUBCUQ): Likewise.
433 (UNSPEC_VSUBEUQM): Likewise.
434 (UNSPEC_VSUBECUQ): Likewise.
435 (VM): Add V1TImode to vector mode iterators.
438 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
439 (altivec_vaddcuq): Likewise.
440 (altivec_vsubuqm): Likewise.
441 (altivec_vsubcuq): Likewise.
442 (altivec_vaddeuqm): Likewise.
443 (altivec_vaddecuq): Likewise.
444 (altivec_vsubeuqm): Likewise.
445 (altivec_vsubecuq): Likewise.
447 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
449 (BOOL_128): Likewise.
450 (BOOL_REGS_OUTPUT): Likewise.
451 (BOOL_REGS_OP1): Likewise.
452 (BOOL_REGS_OP2): Likewise.
453 (BOOL_REGS_UNARY): Likewise.
454 (BOOL_REGS_AND_CR0): Likewise.
456 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
457 128-bit integer builtin support.
458 (vec_vadduqm): Likewise.
459 (vec_vaddecuq): Likewise.
460 (vec_vaddeuqm): Likewise.
461 (vec_vsubecuq): Likewise.
462 (vec_vsubeuqm): Likewise.
463 (vec_vsubcuq): Likewise.
464 (vec_vsubuqm): Likewise.
466 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
467 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
468 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
469 128-bit integer add/subtract to ISA 2.07.
471 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
473 * config/arc/arc.c (arc_predicate_delay_insns):
474 Fix third argument passed to conditionalize_nonjump.
476 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
478 * config/aarch64/aarch64-builtins.c
479 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
480 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
481 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
482 instead of __builtin_lfloor.
483 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
485 2014-03-12 Jakub Jelinek <jakub@redhat.com>
487 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
488 (tree_ssa_ifcombine_bb_1): New function.
489 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
490 is an empty forwarder block to then_bb or vice versa and then_bb
491 and else_bb are effectively swapped.
493 2014-03-12 Christian Bruel <christian.bruel@st.com>
496 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
497 REG_CFA_DEF_CFA note.
498 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
499 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
501 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
503 PR tree-optimization/60454
504 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
506 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
508 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
509 Do not define target_cpu_default2 to generic.
510 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
511 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
512 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
514 2014-03-12 Jakub Jelinek <jakub@redhat.com>
515 Marc Glisse <marc.glisse@inria.fr>
517 PR tree-optimization/60502
518 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
519 instead of build_low_bits_mask.
521 2014-03-12 Jakub Jelinek <jakub@redhat.com>
524 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
525 if there are multiple uses, but op doesn't live on E edge.
526 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
527 clobber stmts before __builtin_unreachable.
529 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
531 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
532 hard_frame_pointer_rtx.
533 * cse.c (cse_insn): Remove volatile check.
534 * cselib.c (cselib_process_insn): Likewise.
535 * dse.c (scan_insn): Likewise.
537 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
539 * config/arc/arc.c (conditionalize_nonjump): New function,
541 (arc_ifcvt): ... this.
542 (arc_predicate_delay_insns): Use it.
544 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
546 * config/arc/predicates.md (extend_operand): During/after reload,
547 allow const_int_operand.
548 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
549 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
550 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
552 (umulsi3_highpart_i): Likewise.
554 2014-03-11 Richard Biener <rguenther@suse.de>
556 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
557 Add asserts to guard possible wrong-code bugs.
559 2014-03-11 Richard Biener <rguenther@suse.de>
561 PR tree-optimization/60429
562 PR tree-optimization/60485
563 * tree-ssa-structalias.c (set_union_with_increment): Properly
564 take into account all fields that overlap the shifted vars.
565 (do_sd_constraint): Likewise.
566 (do_ds_constraint): Likewise.
567 (get_constraint_for_ptr_offset): Likewise.
569 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
571 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
572 (nios2_compute_frame_layout):
573 Add calculation of cfun->machine->fp_save_offset.
574 (nios2_expand_prologue): Correct setting of frame pointer register
576 (nios2_expand_epilogue): Update recovery of stack pointer from
577 frame pointer accordingly.
578 (nios2_initial_elimination_offset): Update calculation of offset
579 for eliminating to HARD_FRAME_POINTER_REGNUM.
581 2014-03-10 Jakub Jelinek <jakub@redhat.com>
584 * ipa.c (symtab_remove_unreachable_nodes): Don't call
585 cgraph_get_create_node on VAR_DECLs.
587 2014-03-10 Richard Biener <rguenther@suse.de>
590 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
592 2014-03-08 Douglas B Rupp <rupp@gnat.com>
594 * config/vms/vms.opt (vms_float_format): New variable.
596 2014-03-08 Tobias Burnus <burnus@net-b.de>
598 * doc/invoke.texi (-fcilkplus): Update implementation status.
600 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
601 Richard Biener <rguenther@suse.de>
603 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
604 consistently accross all TUs.
605 (run_gcc): Enable -fshort-double automatically at link at link-time
606 and disallow override.
608 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
611 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
612 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
613 if they can't be used.
615 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
617 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
618 for Solaris 11/x86 ld.
619 * configure: Regenerate.
621 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
623 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
624 (LIB_TLS_SPEC): Save as ld_tls_libs.
625 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
626 (HAVE_AS_IX86_TLSLDM): New test.
627 * configure, config.in: Regenerate.
628 * config/i386/i386.c (legitimize_tls_address): Fall back to
629 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
630 cannot support TLS_MODEL_LOCAL_DYNAMIC.
631 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
632 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
634 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
636 * common.opt (fira-loop-pressure): Mark as optimization.
638 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
640 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
641 an OpenMP mappable type.
643 2014-03-06 Matthias Klose <doko@ubuntu.com>
645 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
646 MULTILIB_OSDIRNAMES is not defined.
648 2014-03-06 Jakub Jelinek <jakub@redhat.com>
649 Meador Inge <meadori@codesourcery.com>
652 * config/arm/arm.c (arm_tls_symbol_p): Remove.
653 (arm_legitimize_address): Call legitimize_tls_address for any
654 arm_tls_referenced_p expression, handle constant addend. Call it
655 before testing for !TARGET_ARM.
656 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
658 2014-03-06 Richard Biener <rguenther@suse.de>
664 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
666 * tree-streamer.c (record_common_node): Assert we don't record
667 nodes with type double.
668 (preload_common_node): Skip type double, complex double and double
669 pointer since it is now frontend dependent due to fshort-double option.
671 2014-03-06 Richard Biener <rguenther@suse.de>
673 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
674 or -fno-lto is specified and the linker has full plugin support.
675 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
676 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
677 * lto-wrapper.c (merge_and_complain): Merge compile-time
679 (run_gcc): And pass it through to the link options.
681 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
685 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
687 * cselib.c (remove_useless_values): Skip to avoid quadratic
688 behavior if the condition moved from...
689 (cselib_process_insn): ... here holds.
691 2014-03-05 Jakub Jelinek <jakub@redhat.com>
694 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
695 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
698 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
699 (TM_H): Add x86-tune.def.
701 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
703 * config/aarch64/aarch64.c (generic_tunings):
704 Use cortexa57_extra_costs.
706 2014-03-05 Jakub Jelinek <jakub@redhat.com>
709 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
710 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
711 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
714 2014-03-04 Heiher <r@hev.cc>
716 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
717 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
719 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
721 * config/i386/predicates.md (const2356_operand): Change to ...
722 (const2367_operand): ... this.
723 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
725 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
726 (*avx512pf_scatterpf<mode>sf): Ditto.
727 (avx512pf_scatterpf<mode>df): Ditto.
728 (*avx512pf_scatterpf<mode>df_mask): Ditto.
729 (*avx512pf_scatterpf<mode>df): Ditto.
730 * config/i386/i386.c (ix86_expand_builtin): Update
731 incorrect hint operand error message.
733 2014-03-04 Richard Biener <rguenther@suse.de>
735 * lto-section-in.c (lto_get_section_data): Fix const cast.
737 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
739 * tree-streamer.c (record_common_node): Assert we don't record
740 nodes with type double.
741 (preload_common_node): Skip type double, complex double and double
742 pointer since it is now frontend dependent due to fshort-double option.
744 2014-03-04 Richard Biener <rguenther@suse.de>
747 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
748 (lto_input_toplevel_asms): Likewise.
749 * lto-section-in.c (lto_get_section_data): Instead do it here
752 2014-03-04 Richard Biener <rguenther@suse.de>
754 PR tree-optimization/60382
755 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
756 dead PHIs a reduction.
758 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
760 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
762 (_mm_prefetch): Move out of GCC target("sse") pragma.
763 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
764 GCC target("prfchw") pragma.
765 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
767 * config/i386/i386.c (ix86_option_override_internal): Enable
768 -mprfchw with -mprefetchwt1.
770 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
772 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
775 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
777 * opts.h (CL_PCH_IGNORE): Define.
778 * targhooks.c (option_affects_pch_p):
779 Return false for options that have CL_PCH_IGNORE set.
780 * opt-functions.awk: Process PchIgnore.
781 * doc/options.texi: Document PchIgnore.
783 * config/arc/arc.opt (misize): Add PchIgnore property.
785 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
787 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
788 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
789 constraint on constants to permit them being loaded into
790 GENERAL_REGS or BASE_REGS.
792 2014-03-03 Nick Clifton <nickc@redhat.com>
794 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
795 anti-cacnonical alternatives.
796 (negandhi3_real): New pattern.
797 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
799 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
801 * config/avr/avr-mcus.def: Remove atxmega16x1.
802 * config/avr/avr-tables.opt: Regenerate.
803 * config/avr/t-multilib: Regenerate.
804 * doc/avr-mmcu.texi: Regenerate.
806 2014-03-03 Tobias Grosser <tobias@grosser.es>
807 Mircea Namolaru <mircea.namolaru@inria.fr>
809 PR tree-optimization/58028
810 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
813 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
815 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
816 not handled by recognizers.
818 2014-03-03 Jakub Jelinek <jakub@redhat.com>
821 * function.c (expand_function_end): Don't emit
822 clobber_return_register sequence if clobber_after is a BARRIER.
823 * cfgexpand.c (construct_exit_block): Append instructions before
824 return_label to prev_bb.
826 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
828 * config/rs6000/constraints.md: Document reserved use of "wc".
830 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
833 * ipa.c (function_and_variable_visibility): When dissolving comdat
834 group, also set all symbols to local.
836 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
841 2013-12-14 Jan Hubicka <jh@suse.cz>
843 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
845 2014-03-02 Jon Beniston <jon@beniston.com>
851 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
852 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
853 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
854 (simple_return, *simple_return): New patterns
855 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
856 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
858 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
860 * dwarf2out.c (gen_subprogram_die): Tidy.
862 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
865 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
866 (*mov_t_msb_neg_negc): ... this new insn.
868 2014-02-28 Jason Merrill <jason@redhat.com>
871 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
874 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
877 * dwarf2out.c (decltype_auto_die): New static.
878 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
879 (gen_type_die_with_usage): Handle 'decltype(auto)'.
880 (is_cxx_auto): Likewise.
882 2014-02-28 Ian Bolton <ian.bolton@arm.com>
884 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
885 we are not using general regs only.
887 2014-02-28 Richard Biener <rguenther@suse.de>
890 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
891 previous fix and only allow to remove trivial pre-headers
892 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
893 (remove_forwarder_block): Properly update the latch of a loop.
895 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
898 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
899 (cselib_preserved_hash_table): New.
900 (preserve_constants_and_equivs): Move preserved vals to it.
901 (cselib_find_slot): Look it up first.
902 (cselib_init): Initialize it.
903 (cselib_finish): Release it.
904 (dump_cselib_table): Dump it.
906 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
909 * cselib.c (remove_useless_values): Skip to avoid quadratic
910 behavior if the condition moved from...
911 (cselib_process_insn): ... here holds.
913 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
916 * var-tracking.c (vt_initialize): Apply the same condition to
917 preserve the CFA base value.
919 2014-02-28 Joey Ye <joey.ye@arm.com>
922 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
923 if reload in progress or completed.
925 2014-02-28 Tobias Burnus <burnus@net-b.de>
928 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
931 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
933 * doc/tm.texi.in (Condition Code Status): Update documention for
934 relative locations of cc0-setter and cc0-user.
936 2014-02-27 Jeff Law <law@redhat.com>
938 PR rtl-optimization/52714
939 * combine.c (try_combine): When splitting an unrecognized PARALLEL
940 into two independent simple sets, if I3 is a jump, ensure the
941 pattern we place into I3 is a (set (pc) ...).
943 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
944 Jeff Law <law@redhat.com>
946 PR rtl-optimization/49847
947 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
948 are in different blocks.
949 * doc/tm.texi (Condition Code Status): Update documention for
950 relative locations of cc0-setter and cc0-user.
952 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
955 * lra.c (lra_emit_add): Check SUBREG too.
957 2014-02-27 Andreas Schwab <schwab@suse.de>
959 * config/m68k/m68k.c (m68k_option_override): Disable
960 -flive-range-shrinkage for classic m68k.
961 (m68k_override_options_after_change): Likewise.
963 2014-02-27 Marek Polacek <polacek@redhat.com>
966 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
967 -Wmaybe-uninitialized.
969 2014-02-27 Alan Modra <amodra@gmail.com>
972 * reload1.c (emit_input_reload_insns): When reload_override_in,
973 set old to rl->in_reg when rl->in_reg is a subreg.
975 2014-02-26 Richard Biener <rguenther@suse.de>
978 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
980 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
982 * common/config/i386/predicates.md (const1256_operand): Remove.
983 (const2356_operand): New.
984 (const_1_to_2_operand): Remove.
985 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
986 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
987 (*avx512pf_gatherpf<mode>sf): Ditto.
988 (avx512pf_gatherpf<mode>df): Ditto.
989 (*avx512pf_gatherpf<mode>df_mask): Ditto.
990 (*avx512pf_gatherpf<mode>df): Ditto.
991 (avx512pf_scatterpf<mode>sf): Ditto.
992 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
993 (*avx512pf_scatterpf<mode>sf): Ditto.
994 (avx512pf_scatterpf<mode>df): Ditto.
995 (*avx512pf_scatterpf<mode>df_mask): Ditto.
996 (*avx512pf_scatterpf<mode>df): Ditto.
997 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
999 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1001 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1002 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1003 (_mm512_mask_testn_epi64_mask): Move to ...
1004 * config/i386/avx512cdintrin.h: Here.
1005 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1006 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1007 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1008 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1009 TARGET_AVX512F from TARGET_AVX512CD.
1011 2014-02-26 Richard Biener <rguenther@suse.de>
1014 * ipa.c (walk_polymorphic_call_targets): Properly guard
1015 call to inline_update_overall_summary.
1017 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1020 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1021 and latches only if requested. Fix latch if it is removed.
1022 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1023 LOOPS_HAVE_PREHEADERS.
1025 2014-02-25 Andrew Pinski <apinski@cavium.com>
1027 * builtins.c (expand_builtin_thread_pointer): Create a new target
1028 when the target is NULL.
1030 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1032 PR rtl-optimization/60317
1033 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1034 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1035 * lra-assigns.c: Include params.h.
1036 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1037 other reload pseudos considerations.
1039 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1041 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1042 to use canonical form for nor<mode>3.
1044 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1047 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1050 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1052 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1053 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1054 (ix86_handle_option): Handle OPT_mprefetchwt1.
1055 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1056 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1058 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1059 OPTION_MASK_ISA_PREFETCHWT1.
1060 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1061 (PTA_PREFETCHWT1): New.
1062 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1063 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1064 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1065 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1066 (*prefetch_avx512pf_<mode>_: Change into ...
1067 (*prefetch_prefetchwt1_<mode>: This.
1068 * config/i386/i386.opt (mprefetchwt1): New.
1069 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1070 (_mm_prefetch): Handle intent to write.
1071 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1073 2014-02-25 Richard Biener <rguenther@suse.de>
1076 * emit-rtl.c (mem_attrs_htab): Remove.
1077 (mem_attrs_htab_hash): Likewise.
1078 (mem_attrs_htab_eq): Likewise.
1079 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1080 (init_emit_once): Do not allocate mem_attrs_htab.
1082 2014-02-25 Richard Biener <rguenther@suse.de>
1085 * lto-opts.c (lto_write_options): Output non-explicit conservative
1086 -fwrapv, -fno-trapv and -fno-strict-overflow.
1087 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1088 (run_gcc): And pass them through.
1090 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1092 * sel-sched.c (calculate_new_fences): New parameter ptime.
1093 Calculate it as a maximum over all fence cycles.
1094 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1095 Print the final schedule timing when sched_verbose.
1097 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1099 PR rtl-optimization/60292
1100 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1101 bit fot the fence instruction.
1103 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1105 * calls.h: Fix typo in comment.
1107 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1109 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1110 adjusting offsetable addresses.
1112 2014-02-24 Guozhi Wei <carrot@google.com>
1114 * sparseset.h (sparseset_pop): Fix the wrong index.
1116 2014-02-24 Walter Lee <walt@tilera.com>
1118 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1119 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1121 * common/config/tilegx/tilegx-common.c
1122 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1123 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1125 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1126 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1127 (tilegx_gimplify_va_arg_expr): Handle big endian.
1128 (tilegx_expand_unaligned_load): Ditto.
1129 (tilegx_expand_unaligned_store): Ditto.
1130 (TARGET_RETURN_IN_MSB): New.
1131 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1132 (TARGET_ENDIAN_DEFAULT): New.
1133 (TARGET_BIG_ENDIAN): Handle big endian.
1134 (BYTES_BIG_ENDIAN): Ditto.
1135 (WORDS_BIG_ENDIAN): Ditto.
1136 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1139 * config/tilegx/tilegx.md (extv): Handle big endian.
1141 (insn_st<n>): Ditto.
1142 (insn_st<n>_add<bitsuffix>): Ditto.
1143 (insn_stnt<n>): Ditto.
1144 (insn_stnt<n>_add<bitsuffix>):Ditto.
1145 (vec_interleave_highv8qi): Handle big endian.
1146 (vec_interleave_highv8qi_be): New.
1147 (vec_interleave_highv8qi_le): New.
1148 (insn_v1int_h): Handle big endian.
1149 (vec_interleave_lowv8qi): Handle big endian.
1150 (vec_interleave_lowv8qi_be): New.
1151 (vec_interleave_lowv8qi_le): New.
1152 (insn_v1int_l): Handle big endian.
1153 (vec_interleave_highv4hi): Handle big endian.
1154 (vec_interleave_highv4hi_be): New.
1155 (vec_interleave_highv4hi_le): New.
1156 (insn_v2int_h): Handle big endian.
1157 (vec_interleave_lowv4hi): Handle big endian.
1158 (vec_interleave_lowv4hi_be): New.
1159 (vec_interleave_lowv4hi_le): New.
1160 (insn_v2int_l): Handle big endian.
1161 (vec_interleave_highv2si): Handle big endian.
1162 (vec_interleave_highv2si_be): New.
1163 (vec_interleave_highv2si_le): New.
1164 (insn_v4int_h): Handle big endian.
1165 (vec_interleave_lowv2si): Handle big endian.
1166 (vec_interleave_lowv2si_be): New.
1167 (vec_interleave_lowv2si_le): New.
1168 (insn_v4int_l): Handle big endian.
1169 * config/tilegx/tilegx.opt (mbig-endian): New option.
1170 (mlittle-endian): New option.
1171 * doc/install.texi: Document tilegxbe-linux.
1172 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1174 2014-02-24 Martin Jambor <mjambor@suse.cz>
1177 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1178 there are no parameter descriptors.
1180 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1182 PR rtl-optimization/60268
1183 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1184 initialization to ...
1185 (sched_rgn_init): ... here.
1186 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1188 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1190 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1193 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1195 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1198 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1200 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1201 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1203 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1205 * config/microblaze/predicates.md: Add cmp_op predicate.
1206 * config/microblaze/microblaze.md: Add branch_compare instruction
1207 which uses cmp_op predicate and emits cmp insn before branch.
1208 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1209 to microblaze_expand_conditional_branch and consolidate logic.
1210 (microblaze_expand_conditional_branch): emit branch_compare
1211 insn instead of handling cmp op separate from branch insn.
1213 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1215 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1218 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1220 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1221 define_insn with define_expand and new define_insn
1222 *altivec_lve<VI_char>x_internal.
1223 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1224 and new define_insn *altivec_stve<VI_char>x_internal.
1225 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1227 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1229 (altivec_expand_stvex_be): New function.
1231 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1233 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1234 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1235 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1236 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1238 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1241 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1242 instead of emit_move_insn.
1244 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1246 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1248 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1249 gen_altivec_vsumsws.
1251 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1253 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1254 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1255 (altivec_lvxl_<mode>): New define_expand incorporating
1256 -maltivec=be semantics where needed.
1257 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1258 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1259 semantics where needed.
1260 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1261 (altivec_stvx_<mode>): New define_expand incorporating
1262 -maltivec=be semantics where needed.
1263 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1264 VM2 iterator instead of V4SI.
1265 (altivec_stvxl_<mode>): New define_expand incorporating
1266 -maltivec=be semantics where needed.
1267 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1268 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1269 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1270 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1271 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1272 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1273 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1274 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1275 ALTIVEC_BUILTIN_STVXL.
1276 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1277 (altivec_expand_stvx_be): Likewise.
1278 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1279 (altivec_expand_lvx_be): Likewise.
1280 (altivec_expand_stvx_be): Likewise.
1281 (altivec_expand_builtin): Add cases for
1282 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1283 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1284 (altivec_init_builtins): Add definitions for
1285 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1286 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1288 2014-02-21 Catherine Moore <clm@codesourcery.com>
1290 * doc/invoke.texi (mvirt, mno-virt): Document.
1291 * config/mips/mips.opt (mvirt): New option.
1292 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1294 2014-02-21 Richard Biener <rguenther@suse.de>
1296 PR tree-optimization/60276
1297 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1298 (STMT_VINFO_MIN_NEG_DIST): New macro.
1299 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1300 STMT_VINFO_MIN_NEG_DIST.
1301 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1302 made for negative dependence distances still hold.
1304 2014-02-21 Richard Biener <rguenther@suse.de>
1307 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1308 DECL_INITIAL for globals not in the current function context.
1310 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1312 PR tree-optimization/56490
1313 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1314 * tree-ssa-uninit.c: Include params.h.
1315 (compute_control_dep_chain): Add num_calls argument, return false
1316 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1317 num_calls to recursive call.
1318 (find_predicates): Change dep_chain into normal array,
1319 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1320 variable and adjust compute_control_dep_chain caller.
1321 (find_def_preds): Likewise.
1323 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1325 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1326 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1328 2014-02-21 Nick Clifton <nickc@redhat.com>
1330 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1331 (pushhi1): Likewise.
1332 (popqi1): Add mode to pre_dec.
1335 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1337 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1338 mode for mask of V8SFmode permutation.
1340 2014-02-20 Richard Henderson <rth@redhat.com>
1343 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1344 a new pseudo for OLDVAL.
1346 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1349 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1350 gen_reg_rtx if d->testing_p.
1351 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1352 if d->testing_p and we will certainly return true.
1353 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1356 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1358 * emit-rtl.c (gen_reg_rtx): Assert that
1359 crtl->emit.regno_pointer_align_length is non-zero.
1361 2014-02-20 Richard Henderson <rth@redhat.com>
1364 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1365 on failure the store back into EXPECT.
1367 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1368 Sandra Loosemore <sandra@codesourcery.com>
1370 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1371 * config/nios2/nios2.c (nios2_function_profiler): Add
1372 -fPIC (flag_pic == 2) support.
1373 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1374 (nios2_large_offset_p): New function.
1375 (nios2_unspec_reloc_p): Move up position, update to use
1376 nios2_large_offset_p.
1377 (nios2_unspec_address): Remove function.
1378 (nios2_unspec_offset): New function.
1379 (nios2_large_got_address): New function.
1380 (nios2_got_address): Add large offset support.
1381 (nios2_legitimize_tls_address): Update usage of removed and new
1383 (nios2_symbol_binds_local_p): New function.
1384 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1385 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1386 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1387 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1388 processing for (const (unspec ...)).
1389 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1391 2014-02-20 Richard Biener <rguenther@suse.de>
1393 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1394 doing the substitution.
1395 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1397 2014-02-20 Martin Jambor <mjambor@suse.cz>
1400 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1401 info when checking whether lattices are bottom.
1403 2014-02-20 Richard Biener <rguenther@suse.de>
1406 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1409 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1412 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1413 parameter specifying the scaling.
1414 (inline_call): Update.
1415 (want_inline_recursively): Guard division by zero.
1416 (recursive_inlining): Update.
1417 * ipa-inline.h (clone_inlined_nodes): Update.
1419 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1422 * config/i386/i386.c (classify_argument): Pass structures of size
1423 64 bytes or less in register.
1425 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1426 Kirill Yukhin <kirill.yukhin@intel.com>
1428 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1429 (_mm_rcp28_round_ss): Ditto.
1430 (_mm_rsqrt28_round_sd): Ditto.
1431 (_mm_rsqrt28_round_ss): Ditto.
1432 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1433 (_mm_rcp14_round_ss): Ditto.
1434 (_mm_rsqrt14_round_sd): Ditto.
1435 (_mm_rsqrt14_round_ss): Ditto.
1436 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1437 the first input operand, get rid of match_dup.
1438 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1440 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1442 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1443 operand as the first input operand, set type attribute.
1444 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1446 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1447 operand as the first input operand, set type attribute.
1449 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1451 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1454 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1457 * config/i386/i386.c (construct_container): Remove TFmode check
1458 for X86_64_INTEGER_CLASS.
1460 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1463 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1464 only when -Wpsabi is enabled.
1466 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1469 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1470 passing arrays in registers are the same as for structs, so remove the
1471 special case for them.
1473 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1475 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1476 destination type, extract only the valid bits if the source type is not
1477 integral and has a different mode.
1479 2014-02-19 Richard Biener <rguenther@suse.de>
1482 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1485 2014-02-19 Richard Biener <rguenther@suse.de>
1488 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1489 (ipa_modify_call_arguments): Emit an argument load explicitely and
1490 preserve virtual SSA form there and for the replacement call.
1491 Do not update SSA form nor free dominance info.
1493 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1495 * ipa.c (function_and_variable_visibility): Also clear WEAK
1496 flag when disolving COMDAT_GROUP.
1498 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1500 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1501 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1503 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1504 do more sanity checks.
1505 (detect_type_change): Return true when giving up early.
1506 (compute_complex_assign_jump_func): Fix type parameter of
1507 ipa_set_ancestor_jf.
1508 (compute_complex_ancestor_jump_func): Likewise.
1509 (update_jump_functions_after_inlining): Fix updating of
1511 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1513 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1515 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1516 inline clones when edge disappears.
1518 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1521 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1522 Split 64-bit moves into 2 patterns. Do not allow the use of
1523 direct move for TDmode in little endian, since the decimal value
1524 has little endian bytes within a word, but the 64-bit pieces are
1525 ordered in a big endian fashion, and normal subreg's of TDmode are
1527 (mov<mode>_64bit_dm): Likewise.
1528 (movtd_64bit_nodm): Likewise.
1530 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1532 PR tree-optimization/60174
1533 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1534 statement of an SSA_NAME that occurs in an abnormal PHI node.
1536 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1539 * final.c (SEEN_BB): Remove.
1540 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1541 (final_scan_insn): Don't force_source_line on second
1542 NOTE_INSN_BASIC_BLOCK.
1544 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1547 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1548 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1549 (type_natural_mode): Warn ABI change when %zmm register is not
1550 available for AVX512F vector value passing.
1552 2014-02-18 Kai Tietz <ktietz@redhat.com>
1555 * config/i386/i386.c (ix86_expand_prologue): Use value in
1556 rax register as displacement when restoring %r10 or %rax.
1557 Fix wrong offset when restoring both registers.
1559 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1561 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1562 assertion with conditional return.
1564 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1565 Uros Bizjak <ubizjak@gmail.com>
1568 * config/i386/driver-i386.c (host_detect_local_cpu): If
1569 YMM state is not saved by the OS, also clear has_f16c. Move
1570 CPUID 0x80000001 handling before YMM state saving checking.
1572 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1574 PR rtl-optimization/58960
1575 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1576 factored out from ...
1577 (sched_init): ... here.
1578 (free_global_sched_pressure_data): New, factored out from ...
1579 (sched_finish): ... here.
1580 * sched-int.h (free_global_sched_pressure_data): Declare.
1581 * sched-rgn.c (nr_regions_initial): New static global.
1582 (haifa_find_rgns): Initialize it.
1583 (schedule_region): Disable sched-pressure for the newly
1586 2014-02-17 Richard Biener <rguenther@suse.de>
1588 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1589 release SSA defs of pattern stmts.
1591 2014-02-17 Richard Biener <rguenther@suse.de>
1593 * tree-inline.c (expand_call_inline): Release the virtual
1594 operand defined by the call we are about to inline.
1596 2014-02-17 Richard Biener <rguenther@suse.de>
1598 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1600 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1601 Ilya Tocar <ilya.tocar@intel.com>
1603 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1604 arguments order in builtin.
1605 (_mm512_permutexvar_epi64): Ditto.
1606 (_mm512_mask_permutexvar_epi64): Ditto
1607 (_mm512_maskz_permutexvar_epi32): Ditto
1608 (_mm512_permutexvar_epi32): Ditto
1609 (_mm512_mask_permutexvar_epi32): Ditto
1611 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1613 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1614 (p8_vmrgow): Likewise.
1616 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1618 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1621 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1624 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1625 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1626 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1627 using direct move instructions on ISA 2.07. Also adjust
1628 instruction length for 64-bit.
1629 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1630 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1632 2014-02-15 Alan Modra <amodra@gmail.com>
1636 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1637 find_replacement on parts of insn rtl that might be reloaded.
1639 2014-02-15 Richard Biener <rguenther@suse.de>
1641 PR tree-optimization/60183
1642 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1643 (tree_ssa_phiprop): Calculate and free post-dominators.
1645 2014-02-14 Jeff Law <law@redhat.com>
1647 PR rtl-optimization/60131
1648 * ree.c (get_extended_src_reg): New function.
1649 (combine_reaching_defs): Use it rather than assuming location of REG.
1650 (find_and_remove_re): Verify first operand of extension is
1651 a REG before adding the insns to the copy list.
1653 2014-02-14 Roland McGrath <mcgrathr@google.com>
1655 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1656 * configure: Regenerated.
1657 * config.in: Regenerated.
1658 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1659 instead of ASM_SHORT.
1661 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1662 Richard Earnshaw <rearnsha@arm.com>
1664 PR rtl-optimization/59535
1665 * lra-constraints.c (process_alt_operands): Encourage alternative
1666 when unassigned pseudo class is superset of the alternative class.
1667 (inherit_reload_reg): Don't inherit when optimizing for code size.
1668 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1669 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1670 modes not less than 4 for Thumb1.
1672 2014-02-14 Kyle McMartin <kyle@redhat.com>
1675 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1677 2014-02-14 Richard Biener <rguenther@suse.de>
1679 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1680 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1681 Do not strip INDIRECT_REFs.
1683 2014-02-14 Richard Biener <rguenther@suse.de>
1686 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1687 DECL_FUNCTION_SPECIFIC_TARGET.
1688 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1689 * tree-streamer-out.c (pack_ts_target_option): Remove.
1690 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1691 (write_ts_function_decl_tree_pointers): Do not stream
1692 DECL_FUNCTION_SPECIFIC_TARGET.
1693 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1694 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1695 (lto_input_ts_function_decl_tree_pointers): Do not stream
1696 DECL_FUNCTION_SPECIFIC_TARGET.
1698 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1700 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1701 (get_initial_def_for_induction, vectorizable_induction): Ignore
1702 debug stmts when looking for exit_phi.
1703 (vectorizable_live_operation): Fix up condition.
1705 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1707 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1708 nreverse() because it changes the content of original tree list.
1710 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1712 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1713 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1715 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1717 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1718 GNU coding standards.
1720 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1723 * dwarf2out.c (gen_subprogram_die): Don't call
1724 add_calling_convention_attribute if subr_die is old_die.
1726 2014-02-13 Sharad Singhai <singhai@google.com>
1728 * doc/optinfo.texi: Fix order of nodes.
1730 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1732 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1733 operands[2], not operands[3].
1735 2014-02-13 Richard Biener <rguenther@suse.de>
1738 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1739 mention the possibility of an in-tree build.
1740 (CLooG): Update recommended version to 0.18.1, mention the
1741 possibility of an in-tree build and clarify that the ISL
1742 bundled with CLooG does not work.
1744 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1747 * expr.c (compress_float_constant): If x is a hard register,
1748 extend into a pseudo and then move to x.
1750 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1752 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1753 caused by bad second argument to warning_at() with -mhotpatch and
1754 nested functions (e.g. with gfortran).
1756 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1758 * opts.c (option_name): Remove "enabled by default" rider.
1760 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1762 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1764 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1765 Uros Bizjak <ubizjak@gmail.com>
1768 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1769 * configure: Regenerated.
1771 2014-02-12 Richard Biener <rguenther@suse.de>
1773 * vec.c (vec_prefix::calculate_allocation): Move as
1774 inline variant to vec.h.
1775 (vec_prefix::calculate_allocation_1): New out-of-line version.
1776 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1777 (vec_prefix::m_has_auto_buf): Rename to ...
1778 (vec_prefix::m_using_auto_storage): ... this.
1779 (vec_prefix::calculate_allocation): Inline the easy cases
1780 and dispatch to calculate_allocation_1 which doesn't need the
1782 (va_heap::reserve): Use gcc_checking_assert.
1783 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1784 m_using_auto_storage.
1785 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1787 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1788 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1789 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1791 2014-02-12 Richard Biener <rguenther@suse.de>
1793 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1794 when we found a dependence.
1796 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1798 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1800 (maybe_fold_stmt): ... into this new function.
1801 * omp-low.c (lower_omp): Update comment.
1803 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1806 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1809 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1811 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1812 identifiers in comments.
1813 (cortexa53_extra_costs): Likewise.
1814 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1815 (cortexa7_extra_costs): Likewise.
1816 (cortexa12_extra_costs): Likewise.
1817 (cortexa15_extra_costs): Likewise.
1818 (v7m_extra_costs): Likewise.
1820 2014-02-12 Richard Biener <rguenther@suse.de>
1823 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1824 of posix_memalign being successful.
1825 (lower_stmt): Restrict lowering of posix_memalign to when
1826 -ftree-bit-ccp is enabled.
1828 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1830 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1832 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1834 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1836 PR rtl-optimization/60116
1837 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1838 other_insn once the combination has been validated.
1840 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1843 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1845 * ipa-devirt.c: Include demangle.h
1846 (odr_violation_reported): New static variable.
1847 (add_type_duplicate): Update odr_violations.
1848 (maybe_record_node): Add completep parameter; update it.
1849 (record_target_from_binfo): Add COMPLETEP parameter;
1850 update it as needed.
1851 (possible_polymorphic_call_targets_1): Likewise.
1852 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1853 rename FINAL to COMPLETE.
1854 (record_targets_from_bases): Sanity check we found the binfo;
1855 fix COMPLETEP updating.
1856 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1857 parameter, fix computing of COMPLETEP.
1858 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1859 at LTO time do demangling.
1860 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1861 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1863 (gimple_get_virt_method_for_binfo): Likewise.
1864 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1865 gimple_get_virt_method_for_vtable): Update prototypes.
1867 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1870 * genautomata.c (add_presence_absence): Fix typo with
1871 {final_}presence_list.
1873 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1876 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1877 for VSX/Altivec vectors that land in GPR registers.
1879 2014-02-11 Richard Henderson <rth@redhat.com>
1880 Jakub Jelinek <jakub@redhat.com>
1883 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1884 around drhs if type conversion to lacc->type is not useless.
1886 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1888 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1890 (cortex-a57.cortex-a53): Likewise.
1891 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1893 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1895 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1898 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1900 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1901 add_options_for_arm_vfp3.
1903 2014-02-11 Jeff Law <law@redhat.com>
1906 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1907 object with an undesirable mode.
1909 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1912 * config/i386/sol2-9.h: New file.
1913 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1914 *-*-solaris2.9*): Use it.
1916 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1918 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1919 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1921 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1923 * config/microblaze/microblaze.c: Extend mcpu version format
1925 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1927 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1929 2014-02-10 Richard Henderson <rth@redhat.com>
1932 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1933 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1934 ms-abi vs -mno-accumulate-outgoing-args.
1935 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1936 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1939 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1942 * cfgexpand.c (expand_asm_operands): Attach source location to
1943 ASM_INPUT rtx objects.
1944 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1946 2014-02-10 Nick Clifton <nickc@redhat.com>
1948 * config/mn10300/mn10300.c (popcount): New function.
1949 (mn10300_expand_prologue): Include saved registers in stack usage
1952 2014-02-10 Jeff Law <law@redhat.com>
1955 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1956 when changing the SET_DEST of a prior insn to avoid an input reload.
1958 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1960 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1961 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1962 -mcall-openbsd, or -mcall-linux.
1963 (CC1_ENDIAN_BIG_SPEC): Remove.
1964 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1965 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1966 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1967 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1968 and %cc1_endian_default.
1969 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1971 2014-02-10 Richard Biener <rguenther@suse.de>
1973 PR tree-optimization/60115
1974 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1975 MEM_REF handling. Properly verify that the accesses are not
1976 out of the objects bound.
1978 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1980 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1983 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1985 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1986 proper constants and fix formatting.
1987 (possible_polymorphic_call_targets): Fix formatting.
1989 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1990 Ilya Tocar <ilya.tocar@intel.com>
1992 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1993 (_mm512_loadu_epi32): Renamed into...
1994 (_mm512_loadu_si512): This.
1995 (_mm512_storeu_epi32): Renamed into...
1996 (_mm512_storeu_si512): This.
1997 (_mm512_maskz_ceil_ps): Removed.
1998 (_mm512_maskz_ceil_pd): Ditto.
1999 (_mm512_maskz_floor_ps): Ditto.
2000 (_mm512_maskz_floor_pd): Ditto.
2001 (_mm512_floor_round_ps): Ditto.
2002 (_mm512_floor_round_pd): Ditto.
2003 (_mm512_ceil_round_ps): Ditto.
2004 (_mm512_ceil_round_pd): Ditto.
2005 (_mm512_mask_floor_round_ps): Ditto.
2006 (_mm512_mask_floor_round_pd): Ditto.
2007 (_mm512_mask_ceil_round_ps): Ditto.
2008 (_mm512_mask_ceil_round_pd): Ditto.
2009 (_mm512_maskz_floor_round_ps): Ditto.
2010 (_mm512_maskz_floor_round_pd): Ditto.
2011 (_mm512_maskz_ceil_round_ps): Ditto.
2012 (_mm512_maskz_ceil_round_pd): Ditto.
2013 (_mm512_expand_pd): Ditto.
2014 (_mm512_expand_ps): Ditto.
2015 * config/i386/i386.c (ix86_builtins): Remove
2016 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2017 (bdesc_args): Ditto.
2018 * config/i386/predicates.md (const1256_operand): New.
2019 (const_1_to_2_operand): Ditto.
2020 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2021 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2022 (*avx512pf_gatherpf<mode>sf): Ditto.
2023 (avx512pf_gatherpf<mode>df): Ditto.
2024 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2025 (*avx512pf_gatherpf<mode>df): Ditto.
2026 (avx512pf_scatterpf<mode>sf): Ditto.
2027 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2028 (*avx512pf_scatterpf<mode>sf): Ditto.
2029 (avx512pf_scatterpf<mode>df): Ditto.
2030 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2031 (*avx512pf_scatterpf<mode>df): Ditto.
2032 (avx512f_expand<mode>): Removed.
2033 (<shift_insn><mode>3<mask_name>): Change predicate type.
2035 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2037 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2038 not at the end of datarefs vector use ordered_remove to avoid
2039 reordering datarefs vector.
2042 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2043 mark local addressable non-static vars as GOVD_PRIVATE
2044 instead of GOVD_LOCAL.
2045 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2046 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2050 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2051 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2052 assume_aligned or alloc_align attributes.
2053 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2054 arguments. Handle also assume_aligned and alloc_align attributes.
2055 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2056 calls to functions with assume_aligned or alloc_align attributes.
2057 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2059 2014-02-08 Terry Guo <terry.guo@arm.com>
2061 * doc/invoke.texi: Document ARM -march=armv7e-m.
2063 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2065 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2066 flag on __cilkrts_rethrow builtin.
2069 * ipa-cp.c (determine_versionability): Fail at -O0
2070 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2071 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2074 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2077 * tree-inline.c (copy_forbidden): Fail for
2078 __attribute__((optimize (0))) functions.
2080 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2082 * varpool.c: Include pointer-set.h.
2083 (varpool_remove_unreferenced_decls): Variables in other partitions
2084 will not be output; be however careful to not lose information
2087 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2089 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2090 lookup in the vtable constructor.
2092 2014-02-07 Jeff Law <law@redhat.com>
2095 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2096 define_insn_and_split.
2098 * ipa-inline.c (inline_small_functions): Fix typos.
2100 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2102 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2103 (s390_can_use_return_insn): Declare.
2104 * config/s390/s390.h (EPILOGUE_USES): Define.
2105 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2107 (s390_chunkify_start): Handle return JUMP_LABELs.
2108 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2109 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2110 (s390_can_use_return_insn): New functions.
2111 (s390_fix_long_loop_prediction): Handle conditional returns.
2112 (TARGET_SET_UP_BY_PROLOGUE): Define.
2113 * config/s390/s390.md (ANY_RETURN): New code iterator.
2114 (*creturn, *csimple_return, return, simple_return): New patterns.
2116 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2118 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2119 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2120 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2121 REG_CFA_RESTORE list when deciding not to restore a register.
2123 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2125 * config/s390/s390.c: Include tree-pass.h and context.h.
2126 (s390_early_mach): New function, split out from...
2127 (s390_emit_prologue): ...here.
2128 (pass_data_s390_early_mach): New pass structure.
2129 (pass_s390_early_mach): New class.
2130 (s390_option_override): Create and register early_mach pass.
2131 Move to end of file.
2133 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2135 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2136 to match for the exit block.
2138 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2140 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2141 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2142 Reject misaligned operands.
2144 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2146 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2148 2014-02-07 Richard Biener <rguenther@suse.de>
2151 * gimple-low.c (lower_builtin_posix_memalign): New function.
2152 (lower_stmt): Call it to lower posix_memalign in a way
2153 to make alignment info accessible.
2155 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2158 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2159 __builtin_setjmp_receiver.
2161 2014-02-07 Richard Biener <rguenther@suse.de>
2164 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2165 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2166 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2167 Handle BUILT_IN_POSIX_MEMALIGN.
2168 (find_func_clobbers): Likewise.
2169 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2170 (call_may_clobber_ref_p_1): Likewise.
2172 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2175 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2178 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2181 * lto-cgraph.c (lto_output_node): Use
2182 symtab_get_symbol_partitioning_class.
2183 (lto_output_varpool_node): likewise.
2184 (symtab_get_symbol_partitioning_class): Move here from
2186 * cgraph.h (symbol_partitioning_class): Likewise.
2187 (symtab_get_symbol_partitioning_class): Declare.
2189 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2191 * ggc.h (ggc_internal_cleared_alloc): New macro.
2192 * vec.h (vec_safe_copy): Handle memory stats.
2193 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2194 * target-globals.c (save_target_globals): Likewise.
2196 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2199 * expr.c (emit_move_resolve_push): Export; be bit more selective
2200 on when to clear alias set.
2201 * expr.h (emit_move_resolve_push): Declare.
2202 * function.h (struct function): Add tail_call_marked.
2203 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2204 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2205 * config/i386/i386.md (TImode move expander): De not call
2207 (FP push expanders): Preserve memory attributes.
2208 * config/i386/sse.md (push<mode>1): Remove.
2209 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2210 (ix86_expand_push): Remove.
2211 * config/i386/mmx.md (push<mode>1): Remove.
2213 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2215 PR rtl-optimization/60030
2216 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2217 lopart with paradoxical subreg before shifting it up by hprec.
2219 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2221 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2222 Remove extra newline at end of file.
2223 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2224 (arm_issue_rate): Handle cortexa57.
2225 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2226 (cortex-a57.cortex-a53): Likewise.
2228 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2231 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2232 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2234 (arm_expand_prologue): Adjust all callers.
2235 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2236 info, registers also at the lowest numbered registers side. Use
2237 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2241 * var-tracking.c (adjust_mems): Before adding a SET to
2242 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2244 2014-02-06 Alan Modra <amodra@gmail.com>
2247 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2248 change SDmode to DDmode when lra_in_progress.
2250 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2253 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2254 free_data_ref on the dr first, and before goto again also set dr
2255 to the next dr. For simd_lane_access, free old datarefs[i] before
2256 overwriting it. For get_vectype_for_scalar_type failure, don't
2257 free_data_ref if simd_lane_access.
2259 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2262 * tree.h (opts_for_fn): New inline function.
2263 (opt_for_fn): Define.
2264 * config/i386/i386.c (ix86_function_regparm): Use
2265 opt_for_fn (decl, optimize) instead of optimize.
2267 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2269 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2270 for SYMBOL_REF in large memory model.
2272 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2274 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2276 (cortex-a57): Likewise.
2277 (cortex-a57.cortex-a53): Likewise.
2279 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2280 Kugan Vivekanandarajah <kuganv@linaro.org>
2282 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2284 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2286 2014-02-06 Richard Biener <rguenther@suse.de>
2288 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2289 set_loop_copy and initialize_original_copy_tables.
2291 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2293 * config/aarch64/aarch64-simd.md
2294 (aarch64_ashr_simddi): Change QI to SI.
2296 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2297 Jakub Jelinek <jakub@redhat.com>
2300 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2303 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2305 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2306 CODE_FOR_altivec_vpku[hw]um to
2307 CODE_FOR_altivec_vpku[hw]um_direct.
2308 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2309 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2310 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2311 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2313 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2315 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2316 generation for -maltivec=be.
2317 (altivec_vsumsws): Simplify redundant test.
2319 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2321 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2322 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2323 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2324 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2325 gen_altivec_vpkuwum.
2326 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2328 (altivec_vpks<VI_char>ss): Likewise.
2329 (altivec_vpks<VI_char>us): Likewise.
2330 (altivec_vpku<VI_char>us): Likewise.
2331 (altivec_vpku<VI_char>um): Likewise.
2332 (altivec_vpku<VI_char>um_direct): New (copy of
2333 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2335 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2336 target is little endian and -maltivec=be is not specified.
2337 (*altivec_vupkhs<VU_char>_direct): New (copy of
2338 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2339 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2340 target is little endian and -maltivec=be is not specified.
2341 (*altivec_vupkls<VU_char>_direct): New (copy of
2342 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2343 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2344 little endian and -maltivec=be is not specified.
2345 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2346 little endian and -maltivec=be is not specified.
2348 2014-02-05 Richard Henderson <rth@redhat.com>
2351 * combine-stack-adj.c: Revert r206943.
2352 * sched-int.h (struct deps_desc): Add last_args_size.
2353 * sched-deps.c (init_deps): Initialize it.
2354 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2355 contain REG_ARGS_SIZE notes.
2357 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2359 * lto-cgraph.c (asm_nodes_output): Make global.
2360 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2361 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2362 (driver_handle_option): Handle OPT_fwpa.
2364 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2367 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2368 a comment typo and formatting issue. If odr_hash hasn't been
2369 created, return vNULL and set *completep to false.
2372 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2373 bb with no successors.
2375 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2378 * doc/invoke.texi (-march): Clarify documentation for ARM.
2382 2014-02-05 Richard Biener <rguenther@suse.de>
2384 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2385 when not vectorizing because of too many alias checks.
2386 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2387 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2389 2014-02-05 Nick Clifton <nickc@redhat.com>
2391 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2392 accept extended registers in any mode when compiling for the MN10300.
2394 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2396 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2397 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2398 sanitization attributes.
2399 (can_inline_edge_p): Likewise.
2400 (sanitize_attrs_match_for_inline_p): New function.
2402 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2404 * ipa-prop.c (detect_type_change): Shor circuit testing of
2405 type changes on THIS pointer.
2407 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2410 * config/pa/pa.c (legitimize_tls_address): Return original address
2411 if not passed a SYMBOL_REF rtx.
2412 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2414 (pa_emit_move_sequence): Simplify TLS source operands.
2415 (pa_legitimate_constant_p): Reject all TLS constants.
2416 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2417 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2419 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2421 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2422 groups when we know they are controlled by LTO.
2423 * varasm.c (default_binds_local_p_1): If object is in other partition,
2424 it will be resolved locally.
2426 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2428 * config/host-linux.c (linux_gt_pch_use_address): Don't
2429 use SSIZE_MAX because it is not always defined.
2431 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2434 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2435 threshold for pseudo splitting.
2436 (update_ebb_live_info): Process call argument hard registers and
2437 hard registers from insn definition too.
2438 (max_small_class_regs_num): New constant.
2439 (inherit_in_ebb): Update live hard regs through EBBs. Update
2440 reloads_num only for small register classes. Don't split for
2443 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2446 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2449 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2451 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2454 2014-02-04 Marek Polacek <polacek@redhat.com>
2456 * gdbinit.in (pel): Define.
2458 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2460 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2463 2014-02-04 Richard Biener <rguenther@suse.de>
2466 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2467 in function context local.
2468 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2469 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2470 similar to LTO_imported_decl_ref.
2472 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2474 PR tree-optimization/60002
2475 * cgraphclones.c (build_function_decl_skip_args): Clear
2478 PR tree-optimization/60023
2479 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2480 false to gsi_replace.
2481 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2482 has been in some EH region and vec_stmt could throw, add
2483 vec_stmt into the same EH region.
2484 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2485 has no lhs, ignore it.
2486 * internal-fn.c (expand_MASK_LOAD): Likewise.
2489 * tree-inline.c (copy_forbidden): Fail for
2490 __attribute__((optimize (0))) functions.
2493 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2494 is set, copy one less argument.
2495 (expand_simd_clones): Don't subtract clone_info->inbranch
2496 from simd_clone_struct_alloc argument.
2498 PR rtl-optimization/57915
2499 * recog.c (simplify_while_replacing): If all unary/binary/relational
2500 operation arguments are constant, attempt to simplify those.
2503 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2504 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2506 2014-02-04 Richard Biener <rguenther@suse.de>
2508 PR tree-optimization/60012
2509 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2510 TBAA disambiguation to all DDRs.
2512 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2515 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2516 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2518 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2521 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2523 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2525 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2526 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2528 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2531 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2532 to figure out targets of polymorphic calls with known decl.
2533 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2534 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2535 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2536 (get_polymorphic_call_info): ... here.
2537 (get_polymorphic_call_info_from_invariant): New function.
2539 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2541 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2542 lookup via vtable pointer; check for type consistency
2543 and turn inconsitent facts into UNREACHABLE.
2544 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2545 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2546 type inconsistent querries; return UNREACHABLE instead.
2548 2014-02-03 Richard Henderson <rth@twiddle.net>
2551 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2552 already processed this node.
2553 (normalize_one_pred_1): Pass along mark_set.
2554 (normalize_one_pred): Create and destroy a pointer_set_t.
2555 (normalize_one_pred_chain): Likewise.
2557 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2559 PR gcov-profile/58602
2560 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2562 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2565 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2566 -fno-devirtualize; try to devirtualize by the knowledge of
2567 virtual table pointer given by aggregate propagation.
2568 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2569 (ipa_print_node_jump_functions): Dump also offset that
2570 is relevant for polymorphic calls.
2571 (determine_known_aggregate_parts): Add arg_type parameter; use it
2572 instead of determining the type from pointer type.
2573 (ipa_compute_jump_functions_for_edge): Update call of
2574 determine_known_aggregate_parts.
2575 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2576 (gimple_get_virt_method_for_binfo): ... here; simplify using
2577 vtable_pointer_value_to_vtable.
2578 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2579 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2580 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2581 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2583 (vtable_pointer_value_to_binfo): ... here.
2584 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2586 2014-02-03 Teresa Johnson <tejohnson@google.com>
2588 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2589 redef of outer loop index variable.
2591 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2595 * doc/extend.texi (Function Attributes): Typo.
2597 2014-02-03 Cong Hou <congh@google.com>
2599 PR tree-optimization/60000
2600 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2601 if the vectorized statement is a store. A store statement can only
2602 appear at the end of pattern statements.
2604 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2606 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2607 (ix86_option_override_internal): Default long double to 64-bit for
2608 32-bit Bionic and to 128-bit for 64-bit Bionic.
2610 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2611 TARGET_LONG_DOUBLE_128 is true.
2612 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2614 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2615 (mlong-double-64): Negate -mlong-double-128.
2616 (mlong-double-128): New option.
2618 * config/i386/i386-c.c (ix86_target_macros): Define
2619 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2621 * doc/invoke.texi: Document -mlong-double-128.
2623 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2625 PR rtl-optimization/60024
2626 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2628 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2630 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2632 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2634 PR rtl-optimization/57662
2635 * sel-sched.c (code_motion_path_driver): Do not mark already not
2636 existing blocks in the visiting bitmap.
2638 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2640 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2641 on the insn being emitted.
2643 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2644 Will Deacon <will.deacon@arm.com>
2646 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2648 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2650 * config/arm/arm-tables.opt: Regenerate.
2652 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2654 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2655 for vector types other than V16QImode.
2656 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2657 define_expand, and call altivec_expand_vec_perm_le when producing
2658 code with little endian element order.
2659 (*altivec_vperm_<mode>_internal): New insn having previous
2660 behavior of altivec_vperm_<mode>.
2661 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2662 altivec_expand_vec_perm_le when producing code with little endian
2664 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2665 behavior of altivec_vperm_<mode>_uns.
2667 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2669 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2670 (altivec_vsumsws): Add handling for -maltivec=be with a little
2672 (altivec_vsumsws_direct): New.
2673 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2674 gen_altivec_vsumsws.
2676 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2678 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2679 vtable_pointer_value_to_binfo): New functions.
2680 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2681 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2683 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2685 * config/nios2/nios2.md (load_got_register): Initialize GOT
2686 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2687 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2689 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2691 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2692 preserverd by passthrough, do not propagate the type.
2694 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2696 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2697 (mips_atomic_assign_expand_fenv): New function.
2698 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2700 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2702 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2703 (__builtin_mips_set_fcsr): Likewise.
2704 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2705 MIPS_USI_FTYPE_VOID.
2706 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2707 (mips16_expand_set_fcsr): Likewise.
2708 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2709 (mips16_set_fcsr_stub): Likewise.
2710 (mips16_get_fcsr_one_only_stub): New class.
2711 (mips16_set_fcsr_one_only_stub): Likewise.
2712 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2713 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2714 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2715 (hard_float): New availability predicate.
2716 (mips_builtins): Add get_fcsr and set_fcsr.
2717 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2718 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2719 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2720 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2721 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2724 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2726 * config/mips/mips.c (mips_one_only_stub): New class.
2727 (mips_need_mips16_rdhwr_p): Replace with...
2728 (mips16_rdhwr_stub): ...this new variable.
2729 (mips16_stub_call_address): New function.
2730 (mips16_rdhwr_one_only_stub): New class.
2731 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2732 (mips_output_mips16_rdhwr): Delete.
2733 (mips_finish_stub): New function.
2734 (mips_code_end): Use it to handle rdhwr stubs.
2736 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2739 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2740 when calculating size of integer atomic types.
2742 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2744 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2746 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2748 PR tree-optimization/60003
2749 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2750 * profile.c (branch_prob): Use gimple_call_builtin_p
2751 to check for BUILT_IN_SETJMP_RECEIVER.
2752 * tree-inline.c (copy_bb): Call notice_special_calls.
2754 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2757 * lra-constraints.c (process_alt_operands): Update reload_sum only
2760 2014-01-31 Richard Henderson <rth@redhat.com>
2763 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2764 until after else_eh is processed.
2766 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2768 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2769 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2770 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2771 in smmintrin.h, remove them.
2772 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2773 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2774 * config/i386/i386.md (ROUND_SAE): Fix value.
2775 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2776 (const48_operand): New.
2777 * config/i386/subst.md (round), (round_expand): Use
2778 const_4_or_8_to_11_operand.
2779 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2781 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2783 * config/i386/constraints.md (Yk): Swap meaning with k.
2784 * config/i386/i386.md (movhi_internal): Change Yk to k.
2785 (movqi_internal): Ditto.
2786 (*k<logic><mode>): Ditto.
2789 (kandn<mode>): Ditto.
2790 (*<code>hi_1): Ditto.
2791 (*<code>qi_1): Ditto.
2792 (kxnor<mode>): Ditto.
2793 (kortestzhi): Ditto.
2794 (kortestchi): Ditto.
2796 (*one_cmplhi2_1): Ditto.
2797 (*one_cmplqi2_1): Ditto.
2798 * config/i386/sse.md (): Change k to Yk.
2799 (avx512f_load<mode>_mask): Ditto.
2800 (avx512f_blendm<mode>): Ditto.
2801 (avx512f_store<mode>_mask): Ditto.
2802 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2803 (avx512f_storedqu<mode>_mask): Ditto.
2804 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2806 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2807 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2808 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2809 (avx512f_maskcmp<mode>3): Ditto.
2810 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2811 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2812 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2813 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2814 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2815 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2816 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2817 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2818 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2819 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2820 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2821 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2822 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2823 (vec_extract_lo_<mode>_maskm): Ditto.
2824 (vec_extract_hi_<mode>_maskm): Ditto.
2825 (avx512f_vternlog<mode>_mask): Ditto.
2826 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2827 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2828 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2829 (avx512f_<code>v8div16qi2_mask): Ditto.
2830 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2831 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2832 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2833 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2834 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2835 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2836 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2837 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2838 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2839 (avx512cd_maskb_vec_dupv8di): Ditto.
2840 (avx512cd_maskw_vec_dupv16si): Ditto.
2841 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2842 (avx512f_vpermi2var<mode>3_mask): Ditto.
2843 (avx512f_vpermi2var<mode>3_mask): Ditto.
2844 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2845 (*avx512f_gathersi<mode>): Ditto.
2846 (*avx512f_gathersi<mode>_2): Ditto.
2847 (*avx512f_gatherdi<mode>): Ditto.
2848 (*avx512f_gatherdi<mode>_2): Ditto.
2849 (*avx512f_scattersi<mode>): Ditto.
2850 (*avx512f_scatterdi<mode>): Ditto.
2851 (avx512f_compress<mode>_mask): Ditto.
2852 (avx512f_compressstore<mode>_mask): Ditto.
2853 (avx512f_expand<mode>_mask): Ditto.
2854 * config/i386/subst.md (mask): Change k to Yk.
2855 (mask_scalar_merge): Ditto.
2858 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2860 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2862 2014-01-31 Richard Biener <rguenther@suse.de>
2865 * builtins.c (fold_builtin_memory_op): Make sure to not
2866 use a floating-point mode or a boolean or enumeral type for
2869 2014-01-30 DJ Delorie <dj@redhat.com>
2871 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2872 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2873 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2874 whenever main() has an epilogue.
2876 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2878 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2879 unused variable "field".
2880 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2881 (vsx_mergeh_<mode>): Likewise.
2882 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2883 (altivec_vmrghh): Likewise.
2884 (altivec_vmrghw): Likewise.
2885 (altivec_vmrglb): Likewise.
2886 (altivec_vmrglh): Likewise.
2887 (altivec_vmrglw): Likewise.
2888 (altivec_vspltb): Add missing uses.
2889 (altivec_vsplth): Likewise.
2890 (altivec_vspltw): Likewise.
2891 (altivec_vspltsf): Likewise.
2893 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2896 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2897 frame related instructions.
2899 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2901 PR rtl-optimization/59959
2902 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2903 any reload of register whose subreg is invalid.
2905 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2907 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2908 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2909 Add missing return type - void.
2911 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2913 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2914 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2915 remove element index adjustment for endian (now handled in vsx.md
2917 (altivec_expand_vec_perm_const): Use
2918 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2919 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2920 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2921 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2922 define_expand and a new define_insn *altivec_vspltb_internal;
2923 adjust for -maltivec=be on a little endian target.
2924 (altivec_vspltb_direct): New.
2925 (altivec_vsplth): Divide into a define_expand and a new
2926 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2927 little endian target.
2928 (altivec_vsplth_direct): New.
2929 (altivec_vspltw): Divide into a define_expand and a new
2930 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2931 little endian target.
2932 (altivec_vspltw_direct): New.
2933 (altivec_vspltsf): Divide into a define_expand and a new
2934 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2935 a little endian target.
2937 2014-01-30 Richard Biener <rguenther@suse.de>
2939 PR tree-optimization/59993
2940 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2941 can propagate form the earlier stmt and avoid the transform
2942 when the intermediate result is needed.
2944 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2946 * README.Portability: Fix typo.
2948 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2950 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2951 comparison_operator with ordered_comparison_operator.
2953 2014-01-30 Nick Clifton <nickc@redhat.com>
2955 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2956 Rename to mn10300_store_multiple_regs.
2957 * config/mn10300/mn10300.c: Likewise.
2958 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2959 store_multiple_regs.
2960 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2961 Call mn10300_store_multiple_regs.
2963 2014-01-30 Nick Clifton <nickc@redhat.com>
2964 DJ Delorie <dj@redhat.com>
2966 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2967 %fp 2 to keep registers after it properly word-aligned.
2968 (rl78_alloc_physical_registers_umul): Handle the case where both
2969 input operands are the same.
2971 2014-01-30 Richard Biener <rguenther@suse.de>
2973 PR tree-optimization/59903
2974 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2977 2014-01-30 Jason Merrill <jason@redhat.com>
2980 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2983 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2985 2014-01-30 Richard Biener <rguenther@suse.de>
2987 PR tree-optimization/59951
2988 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2990 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2993 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2994 SFmode to DFmode case.
2996 2014-01-29 DJ Delorie <dj@redhat.com>
2998 * config/msp430/msp430.opt (-minrt): New.
2999 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3001 (ENDFILE_SPEC): Likewise.
3003 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3005 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3006 (estimate_function_body_sizes): Use it.
3008 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3011 * dwarf2out.c (is_cxx_auto): New.
3012 (is_base_type): Use it.
3013 (gen_type_die_with_usage): Likewise.
3015 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3017 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3018 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3019 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3020 -maltivec=be with LE targets.
3021 (vsx_mergeh_<mode>): Likewise.
3022 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3023 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3024 (altivec_vmrghb): Replace with define_expand and new
3025 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3026 (altivec_vmrghb_direct): New define_insn.
3027 (altivec_vmrghh): Replace with define_expand and new
3028 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3029 (altivec_vmrghh_direct): New define_insn.
3030 (altivec_vmrghw): Replace with define_expand and new
3031 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3032 (altivec_vmrghw_direct): New define_insn.
3033 (*altivec_vmrghsf): Adjust for endianness.
3034 (altivec_vmrglb): Replace with define_expand and new
3035 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3036 (altivec_vmrglb_direct): New define_insn.
3037 (altivec_vmrglh): Replace with define_expand and new
3038 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3039 (altivec_vmrglh_direct): New define_insn.
3040 (altivec_vmrglw): Replace with define_expand and new
3041 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3042 (altivec_vmrglw_direct): New define_insn.
3043 (*altivec_vmrglsf): Adjust for endianness.
3044 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3045 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3046 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3047 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3048 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3049 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3050 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3051 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3053 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3055 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3056 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3059 2014-01-29 Richard Biener <rguenther@suse.de>
3061 PR tree-optimization/58742
3062 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3063 associate_pointerplus_align.
3064 (associate_pointerplus_diff): New function.
3065 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3066 and associate_pointerplus_diff.
3068 2014-01-29 Richard Biener <rguenther@suse.de>
3070 * lto-streamer.h (LTO_major_version): Bump to 3.
3071 (LTO_minor_version): Reset to 0.
3073 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3075 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3076 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3077 (arm_file_start): Generate correct asm header for armv7ve.
3078 * config/arm/bpabi.h: Add multilib support for armv7ve.
3079 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3080 and cortex-a15 to armv7ve.
3081 * config/arm/t-aprofile: Add multilib support for armv7ve.
3082 * doc/invoke.texi: Document -march=armv7ve.
3084 2014-01-29 Richard Biener <rguenther@suse.de>
3086 PR tree-optimization/58742
3087 * tree-ssa-forwprop.c (associate_plusminus): Return true
3088 if we changed sth, defer EH cleanup to ...
3089 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3090 (simplify_mult): New function.
3092 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3095 PR tree-optimization/59920
3096 * tree.c (build_common_builtin_nodes): Remove
3097 __builtin_setjmp_dispatcher initialization.
3098 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3099 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3100 instead of gsi_after_labels + manually skipping debug stmts.
3101 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3102 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3103 * tree-inline.c (copy_edges_for_bb): Remove
3104 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3105 argument. Ignore computed_goto_p stmts. Don't call
3106 make_abnormal_goto_edges. If a call might need abnormal edges
3107 for non-local gotos, see if it already has an edge to
3108 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3109 with true argument, don't do anything then, otherwise add
3110 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3111 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3113 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3114 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3115 (lower_stmt): Don't set data->calls_builtin_setjmp.
3116 (lower_builtin_setjmp): Adjust comment.
3117 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3118 * tree-cfg.c (found_computed_goto): Remove.
3119 (factor_computed_gotos): Remove.
3120 (make_goto_expr_edges): Return bool, true for computed gotos.
3121 Don't call make_abnormal_goto_edges.
3122 (build_gimple_cfg): Don't set found_computed_goto, don't call
3123 factor_computed_gotos.
3124 (computed_goto_p): No longer static.
3125 (make_blocks): Don't set found_computed_goto.
3126 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3127 (make_edges): If make_goto_expr_edges returns true, push bb
3128 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3129 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3130 vector. Record mapping between bbs and OpenMP regions if there
3131 are any, adjust make_gimple_omp_edges caller. Call
3132 handle_abnormal_edges.
3133 (make_abnormal_goto_edges): Remove.
3134 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3135 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3136 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3137 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3138 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3139 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3140 filling *region also set *region_idx to (*region)->entry->index.
3143 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3144 For REGs set ORIGINAL_REGNO.
3146 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3148 * doc/md.texi: Mention that a target shouldn't implement
3149 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3152 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3154 PR tree-optimization/59594
3155 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3156 a copy of the datarefs vector rather than the vector itself.
3158 2014-01-28 Jason Merrill <jason@redhat.com>
3161 * dwarf2out.c (auto_die): New static.
3162 (gen_type_die_with_usage): Handle C++1y 'auto'.
3163 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3166 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3169 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3170 (SPEC_X32): Likewise.
3171 (SPEC_64): Likewise.
3172 * config/i386/i386.c (ix86_option_override_internal): Turn off
3173 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3175 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3176 * config/i386/i386.h (TARGET_16BIT): New macro.
3177 (TARGET_16BIT_P): Likewise.
3178 * config/i386/i386.opt: Add m16.
3179 * doc/invoke.texi: Document -m16.
3181 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3183 PR preprocessor/59935
3184 * input.c (location_get_source_line): Bail out on when line number
3185 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3187 2014-01-28 Richard Biener <rguenther@suse.de>
3189 PR tree-optimization/58742
3190 * tree-ssa-forwprop.c (associate_plusminus): Handle
3191 pointer subtraction of the form (T)(P + A) - (T)P.
3193 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3195 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3198 2014-01-28 Richard Biener <rguenther@suse.de>
3201 2014-01-28 Richard Biener <rguenther@suse.de>
3203 PR rtl-optimization/45364
3204 PR rtl-optimization/59890
3205 * var-tracking.c (local_get_addr_clear_given_value): Handle
3206 already cleared slot.
3207 (val_reset): Handle not allocated local_get_addr_cache.
3208 (vt_find_locations): Use post-order on the inverted CFG.
3210 2014-01-28 Richard Biener <rguenther@suse.de>
3212 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3214 2014-01-28 Richard Biener <rguenther@suse.de>
3216 PR rtl-optimization/45364
3217 PR rtl-optimization/59890
3218 * var-tracking.c (local_get_addr_clear_given_value): Handle
3219 already cleared slot.
3220 (val_reset): Handle not allocated local_get_addr_cache.
3221 (vt_find_locations): Use post-order on the inverted CFG.
3223 2014-01-28 Alan Modra <amodra@gmail.com>
3225 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3226 * configure.ac <recursive call for build != host>: Define
3227 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3228 and LD_FOR_BUILD too.
3229 * configure: Regenerate.
3231 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3233 * config/i386/i386.c (get_builtin_code_for_version): Separate
3234 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3235 Broadwell from Haswell.
3237 2014-01-27 Steve Ellcey <sellcey@mips.com>
3239 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3240 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3241 * config/mips/mips.c (mips_option_override): Change setting
3243 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3244 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3245 Change from Mask to Var.
3247 2014-01-27 Jeff Law <law@redhat.com>
3249 * ipa-inline.c (inline_small_functions): Fix typo.
3251 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3253 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3254 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3255 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3256 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3257 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3258 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3259 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3260 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3261 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3262 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3263 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3264 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3265 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3266 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3267 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3268 (_mm512_storeu_epi64): Ditto.
3269 (_mm512_cmpge_epi32_mask): Ditto.
3270 (_mm512_cmpge_epu32_mask): Ditto.
3271 (_mm512_cmpge_epi64_mask): Ditto.
3272 (_mm512_cmpge_epu64_mask): Ditto.
3273 (_mm512_cmple_epi32_mask): Ditto.
3274 (_mm512_cmple_epu32_mask): Ditto.
3275 (_mm512_cmple_epi64_mask): Ditto.
3276 (_mm512_cmple_epu64_mask): Ditto.
3277 (_mm512_cmplt_epi32_mask): Ditto.
3278 (_mm512_cmplt_epu32_mask): Ditto.
3279 (_mm512_cmplt_epi64_mask): Ditto.
3280 (_mm512_cmplt_epu64_mask): Ditto.
3281 (_mm512_cmpneq_epi32_mask): Ditto.
3282 (_mm512_cmpneq_epu32_mask): Ditto.
3283 (_mm512_cmpneq_epi64_mask): Ditto.
3284 (_mm512_cmpneq_epu64_mask): Ditto.
3285 (_mm512_expand_pd): Ditto.
3286 (_mm512_expand_ps): Ditto.
3287 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3288 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3289 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3290 * config/i386/i386.c (ix86_builtins): Add
3291 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3292 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3293 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3294 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3295 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3296 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3297 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3298 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3299 IX86_BUILTIN_PMOVUSQW512_MEM.
3300 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3301 __builtin_ia32_pmovsqd512mem_mask,
3302 __builtin_ia32_pmovqd512mem_mask,
3303 __builtin_ia32_pmovusqw512mem_mask,
3304 __builtin_ia32_pmovsqw512mem_mask,
3305 __builtin_ia32_pmovqw512mem_mask,
3306 __builtin_ia32_pmovusdw512mem_mask,
3307 __builtin_ia32_pmovsdw512mem_mask,
3308 __builtin_ia32_pmovdw512mem_mask,
3309 __builtin_ia32_pmovqb512mem_mask,
3310 __builtin_ia32_pmovusqb512mem_mask,
3311 __builtin_ia32_pmovsqb512mem_mask,
3312 __builtin_ia32_pmovusdb512mem_mask,
3313 __builtin_ia32_pmovsdb512mem_mask,
3314 __builtin_ia32_pmovdb512mem_mask.
3315 (bdesc_args): Add __builtin_ia32_expanddf512,
3316 __builtin_ia32_expandsf512.
3317 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3318 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3319 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3320 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3321 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3322 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3323 (avx512f_<code>v8div16qi2_mask_store): This.
3324 (avx512f_expand<mode>): New.
3326 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3328 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3330 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3331 (_mm512_prefetch_i32scatter_pd): Ditto.
3332 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3333 (_mm512_prefetch_i64scatter_pd): Ditto.
3334 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3335 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3336 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3337 (_mm512_prefetch_i32scatter_ps): Ditto.
3338 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3339 (_mm512_prefetch_i64scatter_ps): Ditto.
3340 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3341 * config/i386/i386-builtin-types.def: Define
3342 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3343 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3344 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3345 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3346 IX86_BUILTIN_SCATTERPFQPD.
3347 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3348 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3349 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3350 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3351 __builtin_ia32_scatterpfqps.
3352 (ix86_expand_builtin): Expand new built-ins.
3353 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3354 fix memory access data type.
3355 (*avx512pf_gatherpf<mode>_mask): Ditto.
3356 (*avx512pf_gatherpf<mode>): Ditto.
3357 (avx512pf_scatterpf<mode>): Ditto.
3358 (*avx512pf_scatterpf<mode>_mask): Ditto.
3359 (*avx512pf_scatterpf<mode>): Ditto.
3360 (GATHER_SCATTER_SF_MEM_MODE): New.
3361 (avx512pf_gatherpf<mode>df): Ditto.
3362 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3363 (*avx512pf_scatterpf<mode>df): Ditto.
3365 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3368 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3369 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3372 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3374 * common/config/arm/arm-common.c
3375 (arm_rewrite_mcpu): Handle multiple names.
3377 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3379 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3381 * gimple-builder.h (create_gimple_tmp): Delete.
3383 2014-01-27 Christian Bruel <christian.bruel@st.com>
3385 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3388 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3390 * config/pa/pa.md (call): Generate indirect long calls to non-local
3391 functions when outputing 32-bit code.
3392 (call_value): Likewise except for special call to buggy powf function.
3394 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3395 portable runtime and PIC indirect calls.
3396 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3397 and PIC call sequences. Use ldo instead of blr to set return register
3398 in PIC call sequence.
3400 2014-01-25 Walter Lee <walt@tilera.com>
3402 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3403 avoid clobbering a live register.
3405 2014-01-25 Walter Lee <walt@tilera.com>
3407 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3408 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3409 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3410 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3412 2014-01-25 Walter Lee <walt@tilera.com>
3414 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3415 arguments on even registers.
3416 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3418 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3419 (BIGGEST_ALIGNMENT): Ditto.
3420 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3422 2014-01-25 Walter Lee <walt@tilera.com>
3424 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3425 insns before bundling.
3426 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3428 2014-01-25 Walter Lee <walt@tilera.com>
3430 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3431 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3432 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3434 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3436 * config/mips/constraints.md (kl): Delete.
3437 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3438 define expands, using...
3439 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3440 instructions for MIPS16.
3441 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3442 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3444 2014-01-25 Walter Lee <walt@tilera.com>
3446 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3450 2014-01-25 Walter Lee <walt@tilera.com>
3452 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3453 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3455 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3457 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3460 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3462 * print-rtl.c (in_call_function_usage): New var.
3463 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3464 EXPR_LIST mode as mode and not as reg note name.
3467 * cfgloopmanip.c (copy_loop_info): If
3468 loop->warned_aggressive_loop_optimizations, make sure
3469 the flag is set in target loop too.
3471 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3473 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3475 * builtins.def: Likewise.
3476 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3477 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3478 * ira.c (ira_setup_eliminable_regset): Likewise.
3479 * omp-low.c (gate_expand_omp): Likewise.
3480 (execute_lower_omp): Likewise.
3481 (diagnose_sb_0): Likewise.
3482 (gate_diagnose_omp_blocks): Likewise.
3483 (simd_clone_clauses_extract): Likewise.
3486 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3488 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3489 correction for little endian...
3490 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3493 2014-01-24 Jeff Law <law@redhat.com>
3495 PR tree-optimization/59919
3496 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3497 for non-returning calls.
3499 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3501 * common/config/aarch64/aarch64-common.c
3502 (aarch64_rewrite_mcpu): Handle multiple names.
3503 * config/aarch64/aarch64.h
3504 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3506 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3508 * input.c (add_file_to_cache_tab): Handle the case where fopen
3511 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3514 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3515 from push operand if code of push isn't PRE_DEC.
3517 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3520 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3521 -mquad-memory-atomic. Update -mquad-memory documentation to say
3522 it is only used for non-atomic loads/stores.
3524 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3525 -mquad-memory or -mquad-memory-atomic switches.
3527 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3528 -mquad-memory-atomic to ISA 2.07 support.
3530 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3531 to separate support of normal quad word memory operations (ldq, stq)
3532 from the atomic quad word memory operations.
3534 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3535 support to separate non-atomic quad word operations from atomic
3536 quad word operations. Disable non-atomic quad word operations in
3537 little endian mode so that we don't have to swap words after the
3538 load and before the store.
3539 (quad_load_store_p): Add comment about atomic quad word support.
3540 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3541 options printed with -mdebug=reg.
3543 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3544 -mquad-memory-atomic as the test for whether we have quad word
3545 atomic instructions.
3546 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3547 or -mp8-vector are used, allow byte/half-word atomic operations.
3549 * config/rs6000/sync.md (load_lockedti): Insure that the address
3550 is a proper indexed or indirect address for the lqarx instruction.
3551 On little endian systems, swap the hi/lo registers after the lqarx
3553 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3554 insure the address is valid for the lqarx instruction.
3555 (store_conditionalti): Insure that the address is a proper indexed
3556 or indirect address for the stqcrx. instruction. On little endian
3557 systems, swap the hi/lo registers before doing the stqcrx.
3559 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3560 insure the address is valid for the stqcrx. instruction.
3562 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3563 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3564 type of quad memory support is available.
3566 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3569 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3570 there is a danger of looping.
3572 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3575 force flag_ira_loop_pressure if set via command line.
3577 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3579 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3580 (ashr_simd): New builtin handling DI mode.
3581 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3582 (aarch64_sshr_simddi): New match pattern.
3583 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3584 (vshrd_n_s64): Likewise.
3585 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3587 2014-01-23 Nick Clifton <nickc@redhat.com>
3589 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3590 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3591 favour of mcu specific scripts.
3592 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3595 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3596 Alex Velenko <Alex.Velenko@arm.com>
3598 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3599 (vaddv_s16): Likewise.
3600 (vaddv_s32): Likewise.
3601 (vaddv_u8): Likewise.
3602 (vaddv_u16): Likewise.
3603 (vaddv_u32): Likewise.
3604 (vaddvq_s8): Likewise.
3605 (vaddvq_s16): Likewise.
3606 (vaddvq_s32): Likewise.
3607 (vaddvq_s64): Likewise.
3608 (vaddvq_u8): Likewise.
3609 (vaddvq_u16): Likewise.
3610 (vaddvq_u32): Likewise.
3611 (vaddvq_u64): Likewise.
3612 (vaddv_f32): Likewise.
3613 (vaddvq_f32): Likewise.
3614 (vaddvq_f64): Likewise.
3615 (vmaxv_f32): Likewise.
3616 (vmaxv_s8): Likewise.
3617 (vmaxv_s16): Likewise.
3618 (vmaxv_s32): Likewise.
3619 (vmaxv_u8): Likewise.
3620 (vmaxv_u16): Likewise.
3621 (vmaxv_u32): Likewise.
3622 (vmaxvq_f32): Likewise.
3623 (vmaxvq_f64): Likewise.
3624 (vmaxvq_s8): Likewise.
3625 (vmaxvq_s16): Likewise.
3626 (vmaxvq_s32): Likewise.
3627 (vmaxvq_u8): Likewise.
3628 (vmaxvq_u16): Likewise.
3629 (vmaxvq_u32): Likewise.
3630 (vmaxnmv_f32): Likewise.
3631 (vmaxnmvq_f32): Likewise.
3632 (vmaxnmvq_f64): Likewise.
3633 (vminv_f32): Likewise.
3634 (vminv_s8): Likewise.
3635 (vminv_s16): Likewise.
3636 (vminv_s32): Likewise.
3637 (vminv_u8): Likewise.
3638 (vminv_u16): Likewise.
3639 (vminv_u32): Likewise.
3640 (vminvq_f32): Likewise.
3641 (vminvq_f64): Likewise.
3642 (vminvq_s8): Likewise.
3643 (vminvq_s16): Likewise.
3644 (vminvq_s32): Likewise.
3645 (vminvq_u8): Likewise.
3646 (vminvq_u16): Likewise.
3647 (vminvq_u32): Likewise.
3648 (vminnmv_f32): Likewise.
3649 (vminnmvq_f32): Likewise.
3650 (vminnmvq_f64): Likewise.
3652 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3654 * config/aarch64/aarch64-simd.md
3655 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3656 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3657 (*aarch64_mul3_elt<mode>): Likewise.
3658 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3659 (*aarch64_mul3_elt_to_64v2df): Likewise.
3660 (*aarch64_mla_elt<mode>): Likewise.
3661 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3662 (*aarch64_mls_elt<mode>): Likewise.
3663 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3664 (*aarch64_fma4_elt<mode>): Likewise.
3665 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3666 (*aarch64_fma4_elt_to_64v2df): Likewise.
3667 (*aarch64_fnma4_elt<mode>): Likewise.
3668 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3669 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3670 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3671 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3672 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3673 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3674 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3675 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3676 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3678 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3680 * config/aarch64/aarch64-simd.md
3681 (aarch64_be_checked_get_lane<mode>): New define_expand.
3682 * config/aarch64/aarch64-simd-builtins.def
3683 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3684 New builtin definition.
3685 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3686 Use new safe be builtin.
3688 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3690 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3692 (aarch64_be_st1<mode>): Likewise.
3693 (aarch_ld1<VALL:mode>): Define_expand modified.
3694 (aarch_st1<VALL:mode>): Likewise.
3695 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3696 (UNSPEC_ST1): Likewise.
3698 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3700 * config/microblaze/microblaze.md: Add trap insn and attribute
3702 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3704 PR preprocessor/58580
3705 * input.h (location_get_source_line): Take an additional line_size
3707 (void diagnostics_file_cache_fini): Declare new function.
3708 * input.c (struct fcache): New type.
3709 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3710 New static constants.
3711 (diagnostic_file_cache_init, total_lines_num)
3712 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3713 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3714 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3715 (get_next_line, read_next_line, goto_next_line, read_line_num):
3716 New static function definitions.
3717 (diagnostic_file_cache_fini): New function.
3718 (location_get_source_line): Take an additional output line_len
3719 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3721 * diagnostic.c (diagnostic_finish): Call
3722 diagnostic_file_cache_fini.
3723 (adjust_line): Take an additional input parameter for the length
3724 of the line, rather than calculating it with strlen.
3725 (diagnostic_show_locus): Adjust the use of
3726 location_get_source_line and adjust_line with respect to their new
3727 signature. While displaying a line now, do not stop at the first
3728 null byte. Rather, display the zero byte as a space and keep
3729 going until we reach the size of the line.
3730 * Makefile.in: Add vec.o to OBJS-libcommon
3732 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3733 Ilya Tocar <ilya.tocar@intel.com>
3735 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3736 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3737 (__builtin_ia32_kmov16): Ditto.
3738 * config/i386/i386.md (UNSPEC_KMOV): New.
3741 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3743 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3744 (_mm512_storeu_si512): Ditto.
3746 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3749 * rtl.h (get_referenced_operands): Declare.
3750 * recog.c (get_referenced_operands): New function.
3751 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3752 operands have been referenced when recording LO_SUM references.
3754 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3756 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3758 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3760 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3761 Enable for generic and recent AMD targets.
3763 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3765 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3766 ARG_SIZE note when adjustment was eliminated.
3768 2014-01-22 Jeff Law <law@redhat.com>
3770 PR tree-optimization/59597
3771 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3772 in file. Accept new argument REGISTERING and use it to modify
3773 dump output appropriately.
3774 (register_jump_thread): Corresponding changes.
3775 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3776 thread paths involving joiner blocks. Add code to dump cancelled
3777 jump threading paths.
3779 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3781 PR rtl-optimization/59477
3782 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3783 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3785 2014-01-22 Tom Tromey <tromey@redhat.com>
3787 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3789 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3791 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3793 PR rtl-optimization/59896
3794 * lra-constraints.c (process_alt_operands): Check unused note for
3795 matched operands of insn with no output reloads.
3797 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3799 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3800 (mips_move_from_gpr_cost): Likewise.
3802 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3804 PR rtl-optimization/59858
3805 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3806 ira_class_hard_regs_num.
3807 (process_alt_operands): Increase reject for dying matched operand.
3809 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3812 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3813 smaller than size, perform several stores or loads and stores
3814 at dst + count - size to store or copy all of size bytes, rather
3815 than just last modesize bytes.
3817 2014-01-20 DJ Delorie <dj@redhat.com>
3819 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3820 that CLOBBERs are REGs before propogating their values.
3822 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3825 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3826 (cgraph_inline_failed_type): New function.
3827 * cgraph.h (DEFCIFCODE): Add type.
3828 (cgraph_inline_failed_type_t): New enum.
3829 (cgraph_inline_failed_type): New prototype.
3830 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3831 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3832 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3833 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3834 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3835 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3836 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3837 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3838 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3839 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3840 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3841 OPTIMIZATION_MISMATCH.
3842 * tree-inline.c (expand_call_inline): Emit errors during
3843 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3845 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3848 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3849 mode attribute in insn output.
3851 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3853 * output.h (output_constant): Delete.
3854 * varasm.c (output_constant): Make private.
3856 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3858 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3860 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3863 * tree.h (fold_builtin_strcat): New prototype.
3864 * builtins.c (fold_builtin_strcat): No longer static. Add len
3865 argument, if non-NULL, don't call c_strlen. Optimize
3866 directly into __builtin_memcpy instead of __builtin_strcpy.
3867 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3868 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3870 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3872 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3873 for SImode_address_operand operands, having only a REG argument.
3875 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3877 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3878 loader name using mbig-endian.
3879 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3881 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3883 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3887 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3889 * config/aarch64/aarch64-protos.h
3890 (aarch64_cannot_change_mode_class_ptr): Declare.
3891 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3892 aarch64_cannot_change_mode_class_ptr): New.
3893 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3894 backend hook aarch64_cannot_change_mode_class.
3896 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3898 * common/config/aarch64/aarch64-common.c
3899 (aarch64_handle_option): Don't handle any option order logic here.
3900 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3901 selected_cpu, warn on architecture version mismatch.
3902 (aarch64_override_options): Fix parsing order for option strings.
3904 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3905 Iain Sandoe <iain@codesourcery.com>
3908 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3909 warning. Amend comment to reflect current functionality.
3911 2014-01-20 Richard Biener <rguenther@suse.de>
3914 * builtins.c (fold_builtin_strcat): Remove case better handled
3915 by tree-ssa-strlen.c.
3917 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3919 * config/aarch64/aarch64.opt
3920 (mcpu, march, mtune): Make case-insensitive.
3922 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3925 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3926 if operands[1] is a REG or ZERO_EXTEND of a REG.
3928 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3930 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3932 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3934 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3935 long non-pic millicode calls.
3937 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3939 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3941 2014-01-19 Kito Cheng <kito@0xlab.org>
3943 * builtins.c (expand_movstr): Check movstr expand done or fail.
3945 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3946 H.J. Lu <hongjiu.lu@intel.com>
3949 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3950 to DImode for zero-extended addresses.
3952 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3954 PR rtl-optimization/57763
3955 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3956 on the new indirect jump_insn and increment LABEL_NUSES (label).
3958 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3962 * config.gcc (x86_archs): New variable.
3963 (x86_64_archs): Likewise.
3964 (x86_cpus): Likewise.
3965 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3966 --with-arch/--with-cpu= options.
3967 Support --with-arch=/--with-cpu={nehalem,westmere,
3968 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3970 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3972 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3973 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3975 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3977 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3979 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3982 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3983 clear cpp_get_options (parse_in)->warn_unused_macros for
3984 ix86_target_macros_internal with cpp_define.
3986 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3988 * jump.c (delete_related_insns): Keep (use (insn))s.
3989 * reorg.c (redundant_insn): Check for barriers too.
3991 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3993 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3995 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3997 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3998 call to $$dyncall when TARGET_LONG_CALLS is true.
4000 2014-01-17 Jeff Law <law@redhat.com>
4002 * ree.c (combine_set_extension): Temporarily disable test for
4003 changing number of hard registers.
4005 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4008 * ipa-inline-analysis.c (inline_free_summary):
4009 Do not free summary of aliases.
4011 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4014 * gimplify.c (gimplify_expr): Use create_tmp_var
4015 instead of create_tmp_var_raw. If cond doesn't have
4016 integral type, don't add the IFN_ANNOTATE builtin at all.
4018 2014-01-17 Martin Jambor <mjambor@suse.cz>
4021 * ipa-cp.c (prev_edge_clone): New variable.
4022 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4023 Also resize prev_edge_clone vector.
4024 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4025 (ipcp_edge_removal_hook): New function.
4026 (ipcp_driver): Register ipcp_edge_removal_hook.
4028 2014-01-17 Andrew Pinski <apinski@cavium.com>
4029 Steve Ellcey <sellcey@mips.com>
4032 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4035 2014-01-17 Jeff Law <law@redhat.com>
4038 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4039 so that pass_ccp runs first.
4041 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4043 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4044 (ix86_adjust_cost): Use !TARGET_XXX.
4045 (do_reorder_for_imul): Likewise.
4046 (swap_top_of_ready_list): Likewise.
4047 (ix86_sched_reorder): Likewise.
4049 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4051 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4052 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4053 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4054 (intel_memset): New. Duplicate slm_memset.
4055 (intel_cost): New. Duplicate slm_cost.
4056 (m_INTEL): New macro.
4057 (processor_target_table): Add "intel".
4058 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4059 with PROCESSOR_INTEL for "intel".
4060 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4061 PROCESSOR_SILVERMONT.
4062 (ix86_issue_rate): Likewise.
4063 (ix86_adjust_cost): Likewise.
4064 (ia32_multipass_dfa_lookahead): Likewise.
4065 (swap_top_of_ready_list): Likewise.
4066 (ix86_sched_reorder): Likewise.
4067 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4068 instead of TARGET_OPT_AGU.
4069 * config/i386/i386.h (TARGET_INTEL): New.
4070 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4071 (processor_type): Add PROCESSOR_INTEL.
4072 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4073 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4075 2014-01-17 Marek Polacek <polacek@redhat.com>
4078 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4081 2014-01-17 Richard Biener <rguenther@suse.de>
4083 PR tree-optimization/46590
4084 * opts.c (default_options_table): Add entries for
4085 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4086 all enabled at -O1 but not for -Og.
4087 * common.opt (fbranch-count-reg): Remove Init(1).
4088 (fmove-loop-invariants): Likewise.
4089 (ftree-pta): Likewise.
4091 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4093 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4094 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4095 decls to at least the GCC 4.8 used alignments.
4098 * tree-nested.c (convert_nonlocal_reference_stmt,
4099 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4100 of GIMPLE_BIND stmts, adjust associated decls.
4102 2014-01-17 Richard Biener <rguenther@suse.de>
4104 PR tree-optimization/46590
4105 * vec.h (vec<>::bseach): New member function implementing
4106 binary search according to C89 bsearch.
4107 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4108 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4109 bitmap pointer again. Make accesses_in_loop a flat array.
4110 (mem_ref_obstack): New global.
4111 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4112 (mark_ref_stored): Likewise.
4113 (ref_indep_loop_p_2): Likewise.
4114 (set_ref_stored_in_loop): New helper function.
4115 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4116 (memref_free): Adjust.
4117 (record_mem_ref_loc): Simplify.
4118 (gather_mem_refs_stmt): Adjust.
4119 (sort_locs_in_loop_postorder_cmp): New function.
4120 (analyze_memory_references): Sort accesses_in_loop after
4121 loop postorder number.
4122 (find_ref_loc_in_loop_cmp): New function.
4123 (for_all_locs_in_loop): Find relevant cluster of locs in
4124 accesses_in_loop and iterate without recursion.
4125 (execute_sm): Avoid uninit warning.
4126 (struct ref_always_accessed): Simplify.
4127 (ref_always_accessed::operator ()): Likewise.
4128 (ref_always_accessed_p): Likewise.
4129 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4130 loop postorder numbers here.
4131 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4134 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4137 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4138 on decls for which assemble_alias has been called.
4140 2014-01-17 Nick Clifton <nickc@redhat.com>
4142 * config/msp430/msp430.opt: (mcpu): New option.
4143 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4144 (msp430_option_override): Parse target_cpu. If the MCU name
4145 matches a generic string, clear target_mcu.
4146 (msp430_attr): Allow numeric interrupt values up to 63.
4147 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4148 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4150 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4152 * config/msp430/msp430.md (popm): Use %J rather than %I.
4153 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4154 (addhi_cy_i): Use immediate_operand for operand 2.
4155 * doc/invoke.texi: Document -mcpu option.
4157 2014-01-17 Richard Biener <rguenther@suse.de>
4159 PR rtl-optimization/38518
4160 * df.h (df_analyze_loop): Declare.
4161 * df-core.c: Include cfgloop.h.
4162 (df_analyze_1): Split out main part of df_analyze.
4163 (df_analyze): Adjust.
4164 (loop_inverted_post_order_compute): New function.
4165 (loop_post_order_compute): Likewise.
4166 (df_analyze_loop): New function avoiding whole-function
4168 * loop-invariant.c (find_defs): Use df_analyze_loop.
4169 (find_invariants): Adjust.
4170 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4172 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4174 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4175 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4177 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4179 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4180 traversal when removing references.
4182 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4185 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4187 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4190 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4191 pushing a reload for an autoinc when we had previously reloaded an
4192 inner part of the address.
4194 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4196 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4198 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4199 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4200 when not giving up or versioning for alias only because of
4202 (vect_analyze_data_ref_dependences): Set to true.
4203 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4205 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4206 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4210 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4213 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4214 operand 0 predicate for gathers, use a new pseudo as subtarget.
4216 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4219 * lra-constraints.c (process_alt_operands): Add printing debug info.
4220 Check absence of input/output reloads for matched operands too.
4222 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4224 PR rtl-optimization/59835
4225 * ira.c (ira_init_register_move_cost): Increase cost for
4228 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4230 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4232 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4235 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4236 non-register objects. Use gen_(high/low)part more consistently.
4239 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4242 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4243 endian support, remove tests for WORDS_BIG_ENDIAN.
4244 (p8_mfvsrd_3_<mode>): Likewise.
4245 (reload_gpr_from_vsx<mode>): Likewise.
4246 (reload_gpr_from_vsxsf): Likewise.
4247 (p8_mfvsrd_4_disf): Likewise.
4249 2014-01-16 Richard Biener <rguenther@suse.de>
4251 PR rtl-optimization/46590
4252 * lcm.c (compute_antinout_edge): Use postorder iteration.
4253 (compute_laterin): Use inverted postorder iteration.
4255 2014-01-16 Nick Clifton <nickc@redhat.com>
4258 * varasm.c (output_constant): Return the number of bytes actually
4260 (output_constructor_array_range): Update the field size with the
4261 number of bytes emitted by output_constant.
4262 (output_constructor_regular_field): Likewise. Also do not
4263 complain if the total number of bytes emitted is now greater
4264 than the expected fieldpos.
4265 * output.h (output_constant): Update prototype and descriptive comment.
4267 2014-01-16 Marek Polacek <polacek@redhat.com>
4270 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4271 it is error_mark_node.
4273 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4275 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4276 VALID_AVX256_REG_OR_OI_MODE.
4278 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4280 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4281 current procedure should be profiled.
4283 2014-01-15 Andrew Pinski <apinski@cavium.com>
4285 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4286 of moving from/to the STACK_REG register class.
4288 2014-01-15 Richard Henderson <rth@redhat.com>
4291 * reginfo.c (global_regs_decl): Globalize.
4292 * rtl.h (global_regs_decl): Declare.
4293 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4294 reserved via global_regs.
4296 2014-01-15 Teresa Johnson <tejohnson@google.com>
4298 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4300 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4302 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4303 and vmulosh rather than call gen_vec_widen_smult_*.
4304 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4305 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4306 (vec_widen_smult_even_v16qi): Likewise.
4307 (vec_widen_umult_even_v8hi): Likewise.
4308 (vec_widen_smult_even_v8hi): Likewise.
4309 (vec_widen_umult_odd_v16qi): Likewise.
4310 (vec_widen_smult_odd_v16qi): Likewise.
4311 (vec_widen_umult_odd_v8hi): Likewise.
4312 (vec_widen_smult_odd_v8hi): Likewise.
4313 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4314 vmuloub rather than call gen_vec_widen_umult_*.
4315 (vec_widen_umult_lo_v16qi): Likewise.
4316 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4317 vmulosb rather than call gen_vec_widen_smult_*.
4318 (vec_widen_smult_lo_v16qi): Likewise.
4319 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4320 rather than call gen_vec_widen_umult_*.
4321 (vec_widen_umult_lo_v8hi): Likewise.
4322 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4323 rather than call gen_vec_widen_smult_*.
4324 (vec_widen_smult_lo_v8hi): Likewise.
4326 2014-01-15 Jeff Law <law@redhat.com>
4328 PR tree-optimization/59747
4329 * ree.c (find_and_remove_re): Properly handle case where a second
4330 eliminated extension requires widening a copy created for elimination
4331 of a prior extension.
4332 (combine_set_extension): Ensure that the number of hard regs needed
4333 for a destination register does not change when we widen it.
4335 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4337 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4338 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4339 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4340 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4341 (avr-*-rtems*): Likewise.
4342 (bfin*-rtems*): Likewise.
4343 (moxie-*-rtems*): Likewise.
4344 (h8300-*-rtems*): Likewise.
4345 (i[34567]86-*-rtems*): Likewise.
4346 (lm32-*-rtems*): Likewise.
4347 (m32r-*-rtems*): Likewise.
4348 (m68k-*-rtems*): Likewise.
4349 (microblaze*-*-rtems*): Likewise.
4350 (mips*-*-rtems*): Likewise.
4351 (powerpc-*-rtems*): Likewise.
4352 (sh-*-rtems*): Likewise.
4353 (sparc-*-rtems*): Likewise.
4354 (sparc64-*-rtems*): Likewise.
4355 (v850-*-rtems*): Likewise.
4356 (m32c-*-rtems*): Likewise.
4358 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4360 PR rtl-optimization/59511
4361 * ira.c (ira_init_register_move_cost): Use memory costs for some
4362 cases of register move cost calculations.
4363 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4364 instead of BB frequency.
4365 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4366 * lra-assigns.c (find_hard_regno_for): Ditto.
4368 2014-01-15 Richard Biener <rguenther@suse.de>
4370 PR tree-optimization/59822
4371 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4372 (vectorizable_load): Use it to hoist defs of uses of invariant
4373 loads out of the loop.
4375 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4376 Kugan Vivekanandarajah <kuganv@linaro.org>
4379 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4382 2014-01-15 Richard Biener <rguenther@suse.de>
4384 PR rtl-optimization/59802
4385 * lcm.c (compute_available): Use inverted postorder to seed
4386 the initial worklist.
4388 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4391 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4392 ADDR_REGS for invalid symrefs in non-PIC code.
4394 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4397 * builtins.c (determine_block_size): Initialize *probable_max_size
4398 even if len_rtx is CONST_INT.
4400 2014-01-14 Andrew Pinski <apinski@cavium.com>
4402 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4403 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4404 (cortexa53_tunings): Likewise.
4405 (aarch64_sched_issue_rate): New function.
4406 (TARGET_SCHED_ISSUE_RATE): Define.
4408 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4410 * ira-costs.c (find_costs_and_classes): Add missed
4411 ira_init_register_move_cost_if_necessary.
4413 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4416 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4418 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4421 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4422 to indicate if type is used for function return value. Warn ABI
4423 change if the vector mode isn't available for function return value.
4424 (ix86_function_arg_advance): Pass false to type_natural_mode.
4425 (ix86_function_arg): Likewise.
4426 (ix86_gimplify_va_arg): Likewise.
4427 (function_arg_32): Don't warn ABI change.
4428 (ix86_function_value): Pass true to type_natural_mode.
4429 (ix86_return_in_memory): Likewise.
4430 (ix86_struct_value_rtx): Removed.
4431 (TARGET_STRUCT_VALUE_RTX): Likewise.
4433 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4435 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4436 converting a conditional jump into a conditional return.
4438 2014-01-14 Richard Biener <rguenther@suse.de>
4440 PR tree-optimization/58921
4441 PR tree-optimization/59006
4442 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4443 hoisting invariant stmts.
4444 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4445 invariant loads on the preheader edge if possible.
4447 2014-01-14 Joey Ye <joey.ye@arm.com>
4449 * doc/plugin.texi (Building GCC plugins): Update to C++.
4451 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4453 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4454 (_mm_rcp28_round_ss): Ditto.
4455 (_mm_rsqrt28_round_sd): Ditto.
4456 (_mm_rsqrt28_round_ss): Ditto.
4457 (_mm_rcp28_sd): Ditto.
4458 (_mm_rcp28_ss): Ditto.
4459 (_mm_rsqrt28_sd): Ditto.
4460 (_mm_rsqrt28_ss): Ditto.
4461 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4462 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4463 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4464 (IX86_BUILTIN_RCP28SD): Ditto.
4465 (IX86_BUILTIN_RCP28SS): Ditto.
4466 (IX86_BUILTIN_RSQRT28SD): Ditto.
4467 (IX86_BUILTIN_RSQRT28SS): Ditto.
4468 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4469 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4470 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4471 (ix86_expand_special_args_builtin): Expand new FTYPE.
4472 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4473 (srcp14<mode>): Make insn unary.
4474 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4475 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4476 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4477 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4478 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4479 Fix rounding: make it SAE only.
4480 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4482 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4484 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4485 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4486 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4487 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4488 (round_saeonly_mask_scalar_operand4): Ditto.
4489 (round_saeonly_mask_scalar_op3): Ditto.
4490 (round_saeonly_mask_scalar_op4): Ditto.
4492 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4495 Implement -maltivec=be for vec_insert and vec_extract.
4497 2014-01-10 DJ Delorie <dj@redhat.com>
4499 * config/msp430/msp430.md (call_internal): Don't allow memory
4500 references with SP as the base register.
4501 (call_value_internal): Likewise.
4502 * config/msp430/constraints.md (Yc): New. For memory references
4503 that don't use SP as a base register.
4505 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4506 "an integer without a # prefix"
4507 * config/msp430/msp430.md (epilogue_helper): Use it.
4509 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4512 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4513 AVX512F gather builtins.
4514 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4515 on gather decls with INTEGER_TYPE masktype.
4516 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4517 directly into the builtin rather than hoisting it before loop.
4519 PR tree-optimization/59387
4520 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4521 (scev_const_prop): If folded_casts and type has undefined overflow,
4522 use force_gimple_operand instead of force_gimple_operand_gsi and
4523 for each added stmt if it is assign with
4524 arith_code_with_undefined_signed_overflow, call
4525 rewrite_to_defined_overflow.
4526 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4527 gimple-fold.h instead.
4528 (arith_code_with_undefined_signed_overflow,
4529 rewrite_to_defined_overflow): Moved to ...
4530 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4531 rewrite_to_defined_overflow): ... here. No longer static.
4532 Include gimplify-me.h.
4533 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4534 rewrite_to_defined_overflow): New prototypes.
4536 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4538 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4540 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4542 * builtins.c (get_object_alignment_2): Minor tweak.
4543 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4545 2014-01-13 Christian Bruel <christian.bruel@st.com>
4547 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4548 optimized non constant lengths.
4550 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4553 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4554 load as __atomic_load_N if possible.
4556 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4558 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4560 (rs6000_expand_builtin): Adjust call.
4562 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4565 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4566 * config/rs6000/rs6000.c: Include target-globals.h.
4567 (rs6000_set_current_function): Instead of doing target_reinit
4568 unconditionally, use save_target_globals_default_opts and
4569 restore_target_globals.
4571 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4573 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4574 (rs6000_expand_builtin): Handle mffs and mtfsf.
4575 (rs6000_init_builtins): Define mffs and mtfsf.
4576 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4577 (rs6000_mffs): New pattern.
4578 (rs6000_mtfsf): New pattern.
4580 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4582 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4583 Start narrowing with START. Apply candidate-use pair
4584 and check overall cost in narrowing.
4585 (iv_ca_prune): Pass new argument.
4587 2014-01-10 Jeff Law <law@redhat.com>
4590 * ree.c (combine_reaching_defs): Ensure the defining statement
4591 occurs before the extension when optimizing extensions with
4592 different source and destination hard registers.
4594 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4597 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4598 vtables into the type inheritance graph.
4600 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4602 PR rtl-optimization/59754
4603 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4604 modes in the REGNO != REGNO case.
4606 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4608 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4610 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4612 PR tree-optimization/59745
4613 * tree-predcom.c (tree_predictive_commoning_loop): Call
4614 free_affine_expand_cache if giving up because components is NULL.
4616 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4617 GC in payload of target_globals struct instead of allocating them on
4618 the heap and the larger structs separately using GC.
4619 * target-globals.h (struct target_globals): Make regs, hard_regs,
4620 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4621 of GTY((skip)) and change type to void *.
4622 (reset_target_globals): Cast loads from those fields to corresponding
4625 2014-01-10 Steve Ellcey <sellcey@mips.com>
4628 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4629 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4630 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4632 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4635 * aarch64-modes.def (CC_Zmode): New flags mode.
4636 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4637 represents an equality.
4638 (aarch64_get_condition_code): Handle CC_Zmode.
4639 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4641 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4643 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4644 extraction in good case.
4646 2014-01-10 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/59374
4649 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4650 checking after SLP discovery. Mark stmts not participating
4651 in any SLP instance properly.
4653 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4655 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4656 when handling a SET rtx.
4658 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4660 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4661 (cortex-a57): Likewise.
4662 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4664 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4666 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4667 non-iwmmxt builtins.
4669 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4673 * ipa-devirt.c record_target_from_binfo): Take as argument
4674 stack of binfos and lookup matching one for virtual inheritance.
4675 (possible_polymorphic_call_targets_1): Update.
4677 2014-01-10 Huacai Chen <chenhc@lemote.com>
4679 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4680 kernel strings for Loongson-2E/2F/3A.
4682 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4685 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4686 is_gimple_call before calling gimple_call_internal_p.
4688 2014-01-09 Steve Ellcey <sellcey@mips.com>
4690 * Makefile.in (TREE_FLOW_H): Remove.
4691 (TREE_SSA_H): Add file names from tree-flow.h.
4692 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4693 * tree.h: Remove tree-flow.h reference.
4694 * hash-table.h: Remove tree-flow.h reference.
4695 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4696 reference with tree-ssa-loop.h.
4698 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4700 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4701 default element-order behavior for -maltivec.
4702 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4703 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4704 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4705 when targeting big endian, at least for now.
4706 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4708 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4711 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4712 var satisfies use_register_for_decl, just take into account type
4713 alignment, rather than decl alignment.
4715 PR tree-optimization/59622
4716 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4717 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4718 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4719 Don't devirtualize for inplace at all. For targets.length () == 1,
4720 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4722 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4724 * config/i386/i386.md (cpu): Remove the unused btver1.
4726 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4728 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4730 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4733 * tree-core.h (struct target_globals): New forward declaration.
4734 (struct tree_target_option): Add globals field.
4735 * tree.h (TREE_TARGET_GLOBALS): Define.
4736 (prepare_target_option_nodes_for_pch): New prototype.
4737 * target-globals.h (struct target_globals): Define even if
4739 * tree.c (prepare_target_option_node_for_pch,
4740 prepare_target_option_nodes_for_pch): New functions.
4741 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4742 * config/i386/i386.c: Include target-globals.h.
4743 (ix86_set_current_function): Instead of doing target_reinit
4744 unconditionally, use save_target_globals_default_opts and
4745 restore_target_globals.
4747 2014-01-09 Richard Biener <rguenther@suse.de>
4749 PR tree-optimization/59715
4750 * tree-cfg.h (split_critical_edges): Declare.
4751 * tree-cfg.c (split_critical_edges): Export.
4752 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4754 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4756 * cfgexpand.c (expand_stack_vars): Optionally disable
4757 asan stack protection.
4758 (expand_used_vars): Likewise.
4759 (partition_stack_vars): Likewise.
4760 * asan.c (asan_emit_stack_protection): Optionally disable
4761 after return stack usage.
4762 (instrument_derefs): Optionally disable memory access instrumentation.
4763 (instrument_builtin_call): Likewise.
4764 (instrument_strlen_call): Likewise.
4765 (asan_protect_global): Optionally disable global variables protection.
4766 * doc/invoke.texi: Added doc for new options.
4767 * params.def: Added new options.
4768 * params.h: Likewise.
4770 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4772 PR rtl-optimization/59724
4773 * ifcvt.c (cond_exec_process_if_block): Don't call
4774 flow_find_head_matching_sequence with 0 longest_match.
4775 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4776 non-active insns if !stop_after.
4777 (try_head_merge_bb): Revert 2014-01-07 changes.
4779 2014-01-08 Jeff Law <law@redhat.com>
4781 * ree.c (get_sub_rtx): New function, extracted from...
4782 (merge_def_and_ext): Here.
4783 (combine_reaching_defs): Use get_sub_rtx.
4785 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4787 * cgraph.h (varpool_variable_node): Do not choke on null node.
4789 2014-01-08 Catherine Moore <clm@codesourcery.com>
4791 * config/mips/mips.md (simple_return): Attempt to use JRC
4793 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4795 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4797 PR rtl-optimization/59137
4798 * reorg.c (steal_delay_list_from_target): Call update_block for
4800 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4802 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4804 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4805 two duplicate entries.
4807 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4810 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4812 * config/mips/mips.c (mips_truncated_op_cost): New function.
4813 (mips_rtx_costs): Adjust test for BADDU.
4814 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4816 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4818 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4819 (*baddu_si): ...this new pattern.
4821 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4824 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4826 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4829 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4831 (expand_expr, expand_normal): Adjust.
4832 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4833 inner_reference_p. Use inner_reference_p to expand inner references.
4834 (store_expr): Adjust.
4835 * cfgexpand.c (expand_call_stmt): Adjust.
4837 2014-01-08 Rong Xu <xur@google.com>
4839 * gcov-io.c (gcov_var): Move from gcov-io.h.
4840 (gcov_position): Ditto.
4841 (gcov_is_error): Ditto.
4842 (gcov_rewrite): Ditto.
4843 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4844 only part to libgcc/libgcov.h.
4846 2014-01-08 Marek Polacek <polacek@redhat.com>
4849 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4851 2014-01-08 Marek Polacek <polacek@redhat.com>
4854 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4856 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4858 PR rtl-optimization/59649
4859 * stor-layout.c (get_mode_bounds): For BImode return
4860 0 and STORE_FLAG_VALUE.
4862 2014-01-08 Richard Biener <rguenther@suse.de>
4865 * gimple.h (is_gimple_builtin_call): Remove.
4866 (gimple_builtin_call_types_compatible_p): New.
4867 (gimple_call_builtin_p): New overload.
4868 * gimple.c (is_gimple_builtin_call): Remove.
4869 (validate_call): Rename to ...
4870 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4872 (validate_type): New static function.
4873 (gimple_call_builtin_p): New overload and adjust.
4874 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4875 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4876 (gimple_fold_stmt_to_constant_1): Likewise.
4877 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4879 2014-01-08 Richard Biener <rguenther@suse.de>
4882 * gimplify.c (gimplify_expr): Gimplify register-register type
4883 VIEW_CONVERT_EXPRs to separate stmts.
4885 2014-01-07 Jeff Law <law@redhat.com>
4888 * ree.c (combine_set_extension): Handle case where source
4889 and destination registers in an extension insn are different.
4890 (combine_reaching_defs): Allow source and destination registers
4891 in extension to be different under limited circumstances.
4892 (add_removable_extension): Remove restriction that the
4893 source and destination registers in the extension are the same.
4894 (find_and_remove_re): Emit a copy from the extension's
4895 destination to its source after the defining insn if
4896 the source and destination registers are different.
4899 * ifcvt.c (merge_if_block): If we are merging a block with more than
4900 one successor with a block with no successors, remove any BARRIER
4901 after the second block.
4903 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4905 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4907 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4910 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4911 for 14-bit register offsets when INT14_OK_STRICT is false.
4913 2014-01-07 Roland Stigge <stigge@antcom.de>
4914 Michael Meissner <meissner@linux.vnet.ibm.com>
4917 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4918 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4920 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4922 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4925 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4927 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4928 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4929 rtx is const0_rtx or not.
4931 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4934 * target-globals.c (save_target_globals): Remove this_fn_optab
4936 * toplev.c: Include optabs.h.
4937 (target_reinit): Temporarily restore the global options if another
4938 set of options are in force.
4940 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4942 PR rtl-optimization/58668
4943 * cfgcleanup.c (flow_find_cross_jump): Don't count
4944 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4945 to determine what is counted.
4946 (flow_find_head_matching_sequence): Use active_insn_p to determine
4948 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4950 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4951 determine what is counted.
4953 PR tree-optimization/59643
4954 * tree-predcom.c (split_data_refs_to_components): If one dr is
4955 read and one write, determine_offset fails and the write isn't
4956 in the bad component, just put the read into the bad component.
4958 2014-01-07 Mike Stump <mikestump@comcast.net>
4959 Jakub Jelinek <jakub@redhat.com>
4962 * tree-core.h (struct tree_optimization_option): Change optabs
4963 type from unsigned char * to void *.
4964 * optabs.c (init_tree_optimization_optabs): Adjust
4965 TREE_OPTIMIZATION_OPTABS initialization.
4967 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4970 * config/i386/i386.h (struct machine_function): Add
4971 no_drap_save_restore field.
4972 * config/i386/i386.c (ix86_save_reg): Use
4973 !cfun->machine->no_drap_save_restore instead of
4974 crtl->stack_realign_needed.
4975 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4976 this function clears frame_pointer_needed. Set
4977 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4978 and DRAP reg is needed.
4980 2014-01-06 Marek Polacek <polacek@redhat.com>
4983 * doc/implement-c.texi: Mention that other integer types are
4984 permitted as bit-field types in strictly conforming mode.
4986 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4988 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4991 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4993 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4995 2014-01-06 Martin Jambor <mjambor@suse.cz>
4998 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5000 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5002 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5006 * var-tracking.c (add_stores): Preserve the value of the source even if
5007 we don't record the store.
5009 2014-01-06 Terry Guo <terry.guo@arm.com>
5011 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5013 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5016 * config/darwin.c (darwin_function_section): Adjust return values to
5017 correspond to optimisation changes made in r206070.
5019 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5021 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5022 from prefetch_block tune setting.
5023 (nocona_cost): Correct size of prefetch block to 64.
5025 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5027 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5028 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5029 used to save the static chain register in the computation of the offset
5030 from which the FP registers need to be restored.
5032 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5034 PR tree-optimization/59519
5035 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5036 ICE if get_current_def (current_new_name) is already non-NULL, as long
5037 as it is a phi result of some other phi in *new_exit_bb that has
5040 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5041 or vmovdqu* for misaligned_operand.
5042 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5043 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5044 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5045 aligned_mem for AVX512F masked aligned load and store builtins and for
5048 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5050 PR tree-optimization/59651
5051 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5052 Address range for negative step should be added by TYPE_SIZE_UNIT.
5054 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5056 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5057 overlapping registers also for registers other than the stack pointer.
5059 2014-01-03 Marek Polacek <polacek@redhat.com>
5062 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5065 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5068 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5071 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5072 (push splitter): Use <P:MODE_SIZE> instead of
5073 GET_MODE_SIZE (<P:MODE>mode).
5074 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5075 (mov -1, reg peephole2): Likewise.
5076 * config/i386/sse.md (*mov<mode>_internal,
5077 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5078 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5079 *<code><mode>3, *andnot<mode>3<mask_name>,
5080 <mask_codefor><code><mode>3<mask_name>): Likewise.
5081 * config/i386/subst.md (mask_mode512bit_condition,
5082 sd_mask_mode512bit_condition): Likewise.
5084 2014-01-02 Xinliang David Li <davidxl@google.com>
5086 PR tree-optimization/59303
5087 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5088 (dump_predicates): Better output format.
5089 (pred_equal_p): New function.
5090 (is_neq_relop_p): Ditto.
5091 (is_neq_zero_form_p): Ditto.
5092 (pred_expr_equal_p): Ditto.
5093 (pred_neg_p): Ditto.
5094 (simplify_pred): Ditto.
5095 (simplify_preds_2): Ditto.
5096 (simplify_preds_3): Ditto.
5097 (simplify_preds_4): Ditto.
5098 (simplify_preds): Ditto.
5100 (push_to_worklist): Ditto.
5101 (get_pred_info_from_cmp): Ditto.
5102 (is_degenerated_phi): Ditto.
5103 (normalize_one_pred_1): Ditto.
5104 (normalize_one_pred): Ditto.
5105 (normalize_one_pred_chain): Ditto.
5106 (normalize_preds): Ditto.
5107 (normalize_cond_1): Remove function.
5108 (normalize_cond): Ditto.
5109 (is_gcond_subset_of): Ditto.
5110 (is_subset_of_any): Ditto.
5111 (is_or_set_subset_of): Ditto.
5112 (is_and_set_subset_of): Ditto.
5113 (is_norm_cond_subset_of): Ditto.
5114 (pred_chain_length_cmp): Ditto.
5115 (convert_control_dep_chain_into_preds): Type change.
5116 (find_predicates): Ditto.
5117 (find_def_preds): Ditto.
5118 (destroy_predicates_vecs): Ditto.
5119 (find_matching_predicates_in_rest_chains): Ditto.
5120 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5121 (is_pred_expr_subset): Ditto.
5122 (is_pred_chain_subset_of): Ditto.
5123 (is_included_in): Ditto.
5124 (is_superset_of): Ditto.
5126 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5128 Update copyright years.
5130 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5132 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5133 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5134 config/arc/arc.md, config/arc/arc.opt,
5135 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5136 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5137 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5138 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5139 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5140 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5141 vtable-verify.c, vtable-verify.h: Use the standard form for the
5144 2014-01-02 Tobias Burnus <burnus@net-b.de>
5146 * gcc.c (process_command): Update copyright notice dates.
5147 * gcov-dump.c: Ditto.
5149 * doc/cpp.texi: Bump @copying's copyright year.
5150 * doc/cppinternals.texi: Ditto.
5151 * doc/gcc.texi: Ditto.
5152 * doc/gccint.texi: Ditto.
5153 * doc/gcov.texi: Ditto.
5154 * doc/install.texi: Ditto.
5155 * doc/invoke.texi: Ditto.
5157 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5159 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5161 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5163 * config/i386/sse.md (*mov<mode>_internal): Guard
5164 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5166 PR rtl-optimization/59647
5167 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5168 new_rtx into UNSIGNED_FLOAT rtxes.
5170 Copyright (C) 2014 Free Software Foundation, Inc.
5172 Copying and distribution of this file, with or without modification,
5173 are permitted in any medium without royalty provided the copyright
5174 notice and this notice are preserved.