| blob | d629fa3eb8e347c892bd9b9a8753e549f5b906d4 |
1 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
3 PR rtl-optimization/60601
4 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
6 * gcc.c (eval_spec_function): Initialize save_growing_value.
8 2014-03-22 Jakub Jelinek <jakub@redhat.com>
10 PR sanitizer/60613
11 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
12 code == MINUS_EXPR, never swap op0 with op1.
14 * toplev.c (init_local_tick): Avoid signed integer multiplication
15 overflow.
16 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
17 shift by first operand's bitsize.
19 2014-03-21 Jakub Jelinek <jakub@redhat.com>
21 PR target/60610
22 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
23 redefine to 1 or 0.
24 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
25 TARGET_ISA_64BIT_P(x).
27 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
30 pattern for vector nor instead of subtract from splat(-1).
31 (altivec_expand_vec_perm_const_le): Likewise.
33 2014-03-21 Richard Henderson <rth@twiddle.net>
35 PR target/60598
36 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
37 related insns after epilogue_completed.
39 2014-03-21 Martin Jambor <mjambor@suse.cz>
41 PR ipa/59176
42 * cgraph.h (symtab_node): New flag body_removed.
43 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
44 when removing bodies.
45 * symtab.c (dump_symtab_base): Dump body_removed flag.
46 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
47 had their bodies removed.
49 2014-03-21 Martin Jambor <mjambor@suse.cz>
51 PR ipa/60419
52 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
53 in the border.
55 2014-03-21 Richard Biener <rguenther@suse.de>
57 PR tree-optimization/60577
58 * tree-core.h (struct tree_base): Document nothrow_flag use
59 in VAR_DECL_NONALIASED.
60 * tree.h (VAR_DECL_NONALIASED): New.
61 (may_be_aliased): Adjust.
62 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
64 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
66 * expr.c (expand_expr_real_1): Remove outdated comment.
68 2014-03-20 Jakub Jelinek <jakub@redhat.com>
70 PR middle-end/60597
71 * ira.c (adjust_cleared_regs): Call copy_rtx on
72 *reg_equiv[REGNO (loc)].src_p before passing it to
73 simplify_replace_fn_rtx.
75 PR target/60568
76 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
77 into CONST, put pic register as first operand of PLUS. Use
78 gen_const_mem for both 32-bit and 64-bit PIC got loads.
80 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
82 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
84 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
86 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
87 around for store forwarding issue in the FPU on the UT699.
88 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
89 loads and operations if -mfix-ut699 is specified.
90 (divtf3_hq): Tweak attribute.
91 (sqrttf2_hq): Likewise.
93 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
95 * calls.c (store_one_arg): Remove incorrect const qualification on the
96 type of the temporary.
97 * cfgexpand.c (expand_return): Likewise.
98 * expr.c (expand_constructor): Likewise.
99 (expand_expr_real_1): Likewise.
101 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
103 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
104 of parts.
106 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
108 PR target/60039
109 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
111 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
113 * config/arm/aarch-common-protos.h
114 (alu_cost_table): Fix spelling of "extend".
115 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
117 2014-03-19 Richard Biener <rguenther@suse.de>
119 PR middle-end/60553
120 * tree-core.h (tree_type_common): Re-order pointer members
121 to reduce recursion depth during GC walks.
123 2014-03-19 Marek Polacek <polacek@redhat.com>
125 PR sanitizer/60569
126 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
127 before accessing it.
129 2014-03-19 Richard Biener <rguenther@suse.de>
131 PR lto/59543
132 * lto-streamer-in.c (input_function): In WPA stage do not drop
133 debug stmts.
135 2014-03-19 Jakub Jelinek <jakub@redhat.com>
137 PR tree-optimization/60559
138 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
139 with build_zero_cst assignment.
141 2014-03-18 Kai Tietz <ktietz@redhat.com>
143 PR rtl-optimization/56356
144 * sdbout.c (sdbout_parms): Verify that parms'
145 incoming argument is valid.
146 (sdbout_reg_parms): Likewise.
148 2014-03-18 Richard Henderson <rth@redhat.com>
150 PR target/60562
151 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
152 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
153 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
155 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
157 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
158 * doc/plugins.texi (Plugin callbacks): Mention
159 PLUGIN_INCLUDE_FILE.
160 Italicize plugin event names in description. Explain that
161 PLUGIN_PRAGMAS has no sense for lto1. Explain
162 PLUGIN_INCLUDE_FILE.
163 Remind that no GCC functions should be called after
164 PLUGIN_FINISH.
165 Explain what pragmas with expansion are.
167 2014-03-18 Martin Liska <mliska@suse.cz>
169 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
170 gimple call statement is update.
171 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
172 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
174 2014-03-18 Jakub Jelinek <jakub@redhat.com>
176 PR sanitizer/60557
177 * ubsan.c (ubsan_instrument_unreachable): Call
178 initialize_sanitizer_builtins.
179 (ubsan_pass): Likewise.
181 PR sanitizer/60535
182 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
183 varpool_finalize_decl instead of rest_of_decl_compilation.
185 2014-03-18 Richard Biener <rguenther@suse.de>
187 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
188 by using bitmap_and_compl instead of bitmap_and_compl_into.
189 (df_rd_transfer_function): Likewise.
191 2014-03-18 Richard Biener <rguenther@suse.de>
193 * doc/lto.texi (fresolution): Fix typo.
195 2014-03-18 Richard Biener <rguenther@suse.de>
197 * doc/invoke.texi (flto): Update for changes in 4.9.
199 2014-03-18 Richard Biener <rguenther@suse.de>
201 * doc/loop.texi: Remove section on the removed lambda framework.
202 Update loop docs with recent changes in preserving loop structure.
204 2014-03-18 Richard Biener <rguenther@suse.de>
206 * doc/lto.texi (-fresolution): Document.
208 2014-03-18 Richard Biener <rguenther@suse.de>
210 * doc/contrib.texi: Adjust my name.
212 2014-03-18 Jakub Jelinek <jakub@redhat.com>
214 PR ipa/58721
215 * internal-fn.c: Include diagnostic-core.h.
216 (expand_BUILTIN_EXPECT): New function.
217 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
218 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
219 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
220 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
221 IFN_BUILTIN_EXPECT.
222 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
223 Revert 3 argument __builtin_expect code.
224 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
225 * gimple-fold.c (gimple_fold_call): Likewise.
226 * tree.h (fold_builtin_expect): New prototype.
227 * builtins.c (build_builtin_expect_predicate): Add predictor
228 argument, if non-NULL, create 3 argument __builtin_expect.
229 (fold_builtin_expect): No longer static. Add ARG2 argument,
230 pass it through to build_builtin_expect_predicate.
231 (fold_builtin_2): Adjust caller.
232 (fold_builtin_3): Handle BUILT_IN_EXPECT.
233 * internal-fn.def (BUILTIN_EXPECT): New.
235 2014-03-18 Tobias Burnus <burnus@net-b.de>
237 PR ipa/58721
238 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
239 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
240 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
242 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
244 PR ipa/58721
245 * predict.c (combine_predictions_for_bb): Fix up formatting.
246 (expr_expected_value_1, expr_expected_value): Add predictor argument,
247 fill what it points to if non-NULL.
248 (tree_predict_by_opcode): Adjust caller, use the predictor.
249 * predict.def (PRED_COMPARE_AND_SWAP): Add.
251 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
253 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
254 proper constant for the store mode.
256 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
258 * symtab.c (change_decl_assembler_name): Fix transparent alias
259 chain construction.
261 2014-03-16 Renlin Li <Renlin.Li@arm.com>
263 * config/aarch64/aarch64.c: Correct the comments about the
264 aarch64 stack layout.
266 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
268 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
269 check for GF_OMP_FOR_KIND_FOR.
271 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
273 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
274 ymm and zmm register names.
276 2014-03-17 Jakub Jelinek <jakub@redhat.com>
278 PR target/60516
279 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
280 note creation for the 2010-08-31 changes.
282 2014-03-17 Marek Polacek <polacek@redhat.com>
284 PR middle-end/60534
285 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
286 as -fno-tree-loop-vectorize.
287 (expand_omp_simd): Likewise.
289 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
291 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
292 (eligible_for_call_delay): New prototype.
293 * config/sparc/sparc.c (tls_call_delay): Rename into...
294 (eligible_for_call_delay): ...this. Return false if the instruction
295 cannot be put in the delay slot of a branch.
296 (eligible_for_restore_insn): Simplify.
297 (eligible_for_return_delay): Return false if the instruction cannot be
298 put in the delay slot of a branch and simplify.
299 (eligible_for_sibcall_delay): Return false if the instruction cannot be
300 put in the delay slot of a branch.
301 * config/sparc/sparc.md (fix_ut699): New attribute.
302 (tls_call_delay): Delete.
303 (in_call_delay): Reimplement.
304 (eligible_for_sibcall_delay): Rename into...
305 (in_sibcall_delay): ...this.
306 (eligible_for_return_delay): Rename into...
307 (in_return_delay): ...this.
308 (in_branch_delay): Reimplement.
309 (in_uncond_branch_delay): Delete.
310 (in_annul_branch_delay): Delete.
312 2014-03-14 Richard Henderson <rth@redhat.com>
314 PR target/60525
315 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
316 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
317 (*floathi<X87MODEF>2_i387_with_temp): Remove.
318 (floathi splitters): Remove.
319 (float<SWI48x>xf2): New pattern.
320 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
321 code that tried to handle DImode for 32-bit, but which was excluded
322 by the pattern's condition. Drop allocation of stack temporary.
323 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
324 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
325 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
326 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
327 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
328 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
329 (*float<SWI48><MODEF>2_sse_interunit): Remove.
330 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
331 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
332 (*float<SWI48x><X87MODEF>2_i387): Remove.
333 (all float _with_temp splitters): Remove.
334 (*float<SWI48x><MODEF>2_i387): New pattern.
335 (*float<SWI48><MODEF>2_sse): New pattern.
336 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
337 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
339 2014-03-14 Jakub Jelinek <jakub@redhat.com>
340 Marek Polacek <polacek@redhat.com>
342 PR middle-end/60484
343 * common.opt (dump_base_name_prefixed): New Variable.
344 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
345 if x_dump_base_name_prefixed is already set, set it at the end.
347 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
349 PR rtl-optimization/60508
350 * lra-constraints.c (get_reload_reg): Add new parameter
351 in_subreg_p.
352 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
353 Pass the new parameter values.
355 2014-03-14 Richard Biener <rguenther@suse.de>
357 * common.opt: Revert unintented changes from r205065.
358 * opts.c: Likewise.
360 2014-03-14 Richard Biener <rguenther@suse.de>
362 PR middle-end/60518
363 * cfghooks.c (split_block): Properly adjust all loops the
364 block was a latch of.
366 2014-03-14 Martin Jambor <mjambor@suse.cz>
368 PR lto/60461
369 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
370 and simplify it.
372 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
374 PR target/59396
375 * config/avr/avr.c (avr_set_current_function): Pass function name
376 through default_strip_name_encoding before sanity checking instead
377 of skipping the first char of the assembler name.
379 2014-03-13 Richard Henderson <rth@redhat.com>
381 PR debug/60438
382 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
383 (ix86_force_to_memory, ix86_free_from_memory): Remove.
384 * config/i386/i386-protos.h: Likewise.
385 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
386 in the expander instead of a splitter.
387 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
388 any possibility of requiring a memory.
389 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
390 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
391 (fp branch splitters): Update for ix86_split_fp_branch.
392 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
393 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
394 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
395 (*fop_<MODEF>_2_i387): Remove f/r alternative.
396 (*fop_<MODEF>_3_i387): Likewise.
397 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
398 (splitters for the fop_* register patterns): Remove.
399 (fscalexf4_i387): Rename from *fscalexf4_i387.
400 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
402 2014-03-13 Jakub Jelinek <jakub@redhat.com>
404 PR tree-optimization/59779
405 * tree-dfa.c (get_ref_base_and_extent): Use double_int
406 type for bitsize and maxsize instead of HOST_WIDE_INT.
408 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
410 PR rtl-optimization/57320
411 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
412 the CFG after thread_prologue_and_epilogue_insns.
414 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
416 PR rtl-optimization/57189
417 * lra-constraints.c (process_alt_operands): Disfavor spilling
418 vector pseudos.
420 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
422 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
423 messages.
425 2014-03-13 Jakub Jelinek <jakub@redhat.com>
427 PR tree-optimization/59025
428 PR middle-end/60418
429 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
430 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
432 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
434 PR target/60486
435 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
436 calls of avr_out_plus_1.
438 2014-03-13 Bin Cheng <bin.cheng@arm.com>
440 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
441 BB's single pred and update the father loop's latch info later.
443 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
445 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
446 (VEC_M): Likewise.
447 (VEC_N): Likewise.
448 (VEC_R): Likewise.
449 (VEC_base): Likewise.
450 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
451 registers, we need to swap double words in little endian mode.
453 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
454 to be a container mode for 128-bit integer operations added in ISA
455 2.07. Unlike TImode and PTImode, the preferred register set is
456 the Altivec/VMX registers for the 128-bit operations.
458 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
459 declarations.
460 (rs6000_split_128bit_ok_p): Likewise.
462 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
463 macros for creating ISA 2.07 normal and overloaded builtin
464 functions with 3 arguments.
465 (BU_P8V_OVERLOAD_3): Likewise.
466 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
467 for use as overloaded functions.
468 (VPERM_1TI_UNS): Likewise.
469 (VSEL_1TI): Likewise.
470 (VSEL_1TI_UNS): Likewise.
471 (ST_INTERNAL_1ti): Likewise.
472 (LD_INTERNAL_1ti): Likewise.
473 (XXSEL_1TI): Likewise.
474 (XXSEL_1TI_UNS): Likewise.
475 (VPERM_1TI): Likewise.
476 (VPERM_1TI_UNS): Likewise.
477 (XXPERMDI_1TI): Likewise.
478 (SET_1TI): Likewise.
479 (LXVD2X_V1TI): Likewise.
480 (STXVD2X_V1TI): Likewise.
481 (VEC_INIT_V1TI): Likewise.
482 (VEC_SET_V1TI): Likewise.
483 (VEC_EXT_V1TI): Likewise.
484 (EQV_V1TI): Likewise.
485 (NAND_V1TI): Likewise.
486 (ORC_V1TI): Likewise.
487 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
488 added in ISA 2.07. Add both normal 'altivec' builtins, and the
489 overloaded builtin.
490 (VADDUQM): Likewise.
491 (VSUBCUQ): Likewise.
492 (VADDEUQM): Likewise.
493 (VADDECUQ): Likewise.
494 (VSUBEUQM): Likewise.
495 (VSUBECUQ): Likewise.
497 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
498 __int128_t and __uint128_t types.
499 (__uint128_type): Likewise.
500 (altivec_categorize_keyword): Add support for vector __int128_t,
501 vector __uint128_t, vector __int128, and vector unsigned __int128
502 as a container type for TImode operations that need to be done in
503 VSX/Altivec registers.
504 (rs6000_macro_to_expand): Likewise.
505 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
506 to support 128-bit integer instructions vaddcuq, vadduqm,
507 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
508 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
510 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
511 for V1TImode, and set up preferences to use VSX/Altivec registers.
512 Setup VSX reload handlers.
513 (rs6000_debug_reg_global): Likewise.
514 (rs6000_init_hard_regno_mode_ok): Likewise.
515 (rs6000_preferred_simd_mode): Likewise.
516 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
517 (easy_altivec_constant): Likewise.
518 (output_vec_const_move): Likewise.
519 (rs6000_expand_vector_set): Convert V1TImode set and extract to
520 simple move.
521 (rs6000_expand_vector_extract): Likewise.
522 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
523 addressing.
524 (rs6000_const_vec): Add support for V1TImode.
525 (rs6000_emit_le_vsx_load): Swap double words when loading or
526 storing TImode/V1TImode.
527 (rs6000_emit_le_vsx_store): Likewise.
528 (rs6000_emit_le_vsx_move): Likewise.
529 (rs6000_emit_move): Add support for V1TImode.
530 (altivec_expand_ld_builtin): Likewise.
531 (altivec_expand_st_builtin): Likewise.
532 (altivec_expand_vec_init_builtin): Likewise.
533 (altivec_expand_builtin): Likewise.
534 (rs6000_init_builtins): Add support for V1TImode type. Add
535 support for ISA 2.07 128-bit integer builtins. Define type names
536 for the VSX/Altivec vector types.
537 (altivec_init_builtins): Add support for overloaded vector
538 functions with V1TImode type.
539 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
540 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
541 external function.
542 (rs6000_split_128bit_ok_p): Likewise.
543 (rs6000_handle_altivec_attribute): Create V1TImode from vector
544 __int128_t and vector __uint128_t.
546 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
547 and mode attributes.
548 (VSX_M): Likewise.
549 (VSX_M2): Likewise.
550 (VSm): Likewise.
551 (VSs): Likewise.
552 (VSr): Likewise.
553 (VSv): Likewise.
554 (VS_scalar): Likewise.
555 (VS_double): Likewise.
556 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
558 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
559 we support the ISA 2.07 128-bit integer arithmetic instructions.
560 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
561 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
562 and TImode types for use with the builtin functions.
563 (V1TI_type_node): Likewise.
564 (unsigned_V1TI_type_node): Likewise.
565 (intTI_type_internal_node): Likewise.
566 (uintTI_type_internal_node): Likewise.
568 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
569 128-bit builtin functions.
570 (UNSPEC_VADDEUQM): Likewise.
571 (UNSPEC_VADDECUQ): Likewise.
572 (UNSPEC_VSUBCUQ): Likewise.
573 (UNSPEC_VSUBEUQM): Likewise.
574 (UNSPEC_VSUBECUQ): Likewise.
575 (VM): Add V1TImode to vector mode iterators.
576 (VM2): Likewise.
577 (VI_unit): Likewise.
578 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
579 (altivec_vaddcuq): Likewise.
580 (altivec_vsubuqm): Likewise.
581 (altivec_vsubcuq): Likewise.
582 (altivec_vaddeuqm): Likewise.
583 (altivec_vaddecuq): Likewise.
584 (altivec_vsubeuqm): Likewise.
585 (altivec_vsubecuq): Likewise.
587 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
588 mode iterators.
589 (BOOL_128): Likewise.
590 (BOOL_REGS_OUTPUT): Likewise.
591 (BOOL_REGS_OP1): Likewise.
592 (BOOL_REGS_OP2): Likewise.
593 (BOOL_REGS_UNARY): Likewise.
594 (BOOL_REGS_AND_CR0): Likewise.
596 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
597 128-bit integer builtin support.
598 (vec_vadduqm): Likewise.
599 (vec_vaddecuq): Likewise.
600 (vec_vaddeuqm): Likewise.
601 (vec_vsubecuq): Likewise.
602 (vec_vsubeuqm): Likewise.
603 (vec_vsubcuq): Likewise.
604 (vec_vsubuqm): Likewise.
606 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
607 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
608 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
609 128-bit integer add/subtract to ISA 2.07.
611 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
613 * config/arc/arc.c (arc_predicate_delay_insns):
614 Fix third argument passed to conditionalize_nonjump.
616 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
618 * config/aarch64/aarch64-builtins.c
619 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
620 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
621 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
622 instead of __builtin_lfloor.
623 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
625 2014-03-12 Jakub Jelinek <jakub@redhat.com>
627 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
628 (tree_ssa_ifcombine_bb_1): New function.
629 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
630 is an empty forwarder block to then_bb or vice versa and then_bb
631 and else_bb are effectively swapped.
633 2014-03-12 Christian Bruel <christian.bruel@st.com>
635 PR target/60264
636 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
637 REG_CFA_DEF_CFA note.
638 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
639 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
641 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
643 PR tree-optimization/60454
644 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
646 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
648 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
649 Do not define target_cpu_default2 to generic.
650 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
651 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
652 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
654 2014-03-12 Jakub Jelinek <jakub@redhat.com>
655 Marc Glisse <marc.glisse@inria.fr>
657 PR tree-optimization/60502
658 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
659 instead of build_low_bits_mask.
661 2014-03-12 Jakub Jelinek <jakub@redhat.com>
663 PR middle-end/60482
664 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
665 if there are multiple uses, but op doesn't live on E edge.
666 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
667 clobber stmts before __builtin_unreachable.
669 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
671 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
672 hard_frame_pointer_rtx.
673 * cse.c (cse_insn): Remove volatile check.
674 * cselib.c (cselib_process_insn): Likewise.
675 * dse.c (scan_insn): Likewise.
677 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
679 * config/arc/arc.c (conditionalize_nonjump): New function,
680 broken out of ...
681 (arc_ifcvt): ... this.
682 (arc_predicate_delay_insns): Use it.
684 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
686 * config/arc/predicates.md (extend_operand): During/after reload,
687 allow const_int_operand.
688 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
689 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
690 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
691 to "i".
692 (umulsi3_highpart_i): Likewise.
694 2014-03-11 Richard Biener <rguenther@suse.de>
696 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
697 Add asserts to guard possible wrong-code bugs.
699 2014-03-11 Richard Biener <rguenther@suse.de>
701 PR tree-optimization/60429
702 PR tree-optimization/60485
703 * tree-ssa-structalias.c (set_union_with_increment): Properly
704 take into account all fields that overlap the shifted vars.
705 (do_sd_constraint): Likewise.
706 (do_ds_constraint): Likewise.
707 (get_constraint_for_ptr_offset): Likewise.
709 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
711 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
712 (nios2_compute_frame_layout):
713 Add calculation of cfun->machine->fp_save_offset.
714 (nios2_expand_prologue): Correct setting of frame pointer register
715 in prologue.
716 (nios2_expand_epilogue): Update recovery of stack pointer from
717 frame pointer accordingly.
718 (nios2_initial_elimination_offset): Update calculation of offset
719 for eliminating to HARD_FRAME_POINTER_REGNUM.
721 2014-03-10 Jakub Jelinek <jakub@redhat.com>
723 PR ipa/60457
724 * ipa.c (symtab_remove_unreachable_nodes): Don't call
725 cgraph_get_create_node on VAR_DECLs.
727 2014-03-10 Richard Biener <rguenther@suse.de>
729 PR middle-end/60474
730 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
732 2014-03-08 Douglas B Rupp <rupp@gnat.com>
734 * config/vms/vms.opt (vms_float_format): New variable.
736 2014-03-08 Tobias Burnus <burnus@net-b.de>
738 * doc/invoke.texi (-fcilkplus): Update implementation status.
740 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
741 Richard Biener <rguenther@suse.de>
743 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
744 consistently accross all TUs.
745 (run_gcc): Enable -fshort-double automatically at link at link-time
746 and disallow override.
748 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
750 PR target/58271
751 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
752 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
753 if they can't be used.
755 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
757 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
758 for Solaris 11/x86 ld.
759 * configure: Regenerate.
761 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
763 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
764 (LIB_TLS_SPEC): Save as ld_tls_libs.
765 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
766 (HAVE_AS_IX86_TLSLDM): New test.
767 * configure, config.in: Regenerate.
768 * config/i386/i386.c (legitimize_tls_address): Fall back to
769 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
770 cannot support TLS_MODEL_LOCAL_DYNAMIC.
771 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
772 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
774 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
776 * common.opt (fira-loop-pressure): Mark as optimization.
778 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
780 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
781 an OpenMP mappable type.
783 2014-03-06 Matthias Klose <doko@ubuntu.com>
785 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
786 MULTILIB_OSDIRNAMES is not defined.
788 2014-03-06 Jakub Jelinek <jakub@redhat.com>
789 Meador Inge <meadori@codesourcery.com>
791 PR target/58595
792 * config/arm/arm.c (arm_tls_symbol_p): Remove.
793 (arm_legitimize_address): Call legitimize_tls_address for any
794 arm_tls_referenced_p expression, handle constant addend. Call it
795 before testing for !TARGET_ARM.
796 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
798 2014-03-06 Richard Biener <rguenther@suse.de>
800 PR middle-end/60445
801 PR lto/60424
802 PR lto/60427
803 Revert
804 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
806 * tree-streamer.c (record_common_node): Assert we don't record
807 nodes with type double.
808 (preload_common_node): Skip type double, complex double and double
809 pointer since it is now frontend dependent due to fshort-double option.
811 2014-03-06 Richard Biener <rguenther@suse.de>
813 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
814 or -fno-lto is specified and the linker has full plugin support.
815 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
816 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
817 * lto-wrapper.c (merge_and_complain): Merge compile-time
818 optimization levels.
819 (run_gcc): And pass it through to the link options.
821 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
823 PR debug/60381
824 Revert:
825 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
826 PR debug/59992
827 * cselib.c (remove_useless_values): Skip to avoid quadratic
828 behavior if the condition moved from...
829 (cselib_process_insn): ... here holds.
831 2014-03-05 Jakub Jelinek <jakub@redhat.com>
833 PR plugins/59335
834 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
835 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
837 PR plugins/59335
838 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
839 (TM_H): Add x86-tune.def.
841 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
843 * config/aarch64/aarch64.c (generic_tunings):
844 Use cortexa57_extra_costs.
846 2014-03-05 Jakub Jelinek <jakub@redhat.com>
848 PR lto/60404
849 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
850 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
851 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
852 cost for in_lto_p.
854 2014-03-04 Heiher <r@hev.cc>
856 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
857 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
859 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
861 * config/i386/predicates.md (const2356_operand): Change to ...
862 (const2367_operand): ... this.
863 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
864 const2367_operand.
865 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
866 (*avx512pf_scatterpf<mode>sf): Ditto.
867 (avx512pf_scatterpf<mode>df): Ditto.
868 (*avx512pf_scatterpf<mode>df_mask): Ditto.
869 (*avx512pf_scatterpf<mode>df): Ditto.
870 * config/i386/i386.c (ix86_expand_builtin): Update
871 incorrect hint operand error message.
873 2014-03-04 Richard Biener <rguenther@suse.de>
875 * lto-section-in.c (lto_get_section_data): Fix const cast.
877 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
879 * tree-streamer.c (record_common_node): Assert we don't record
880 nodes with type double.
881 (preload_common_node): Skip type double, complex double and double
882 pointer since it is now frontend dependent due to fshort-double option.
884 2014-03-04 Richard Biener <rguenther@suse.de>
886 PR lto/60405
887 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
888 (lto_input_toplevel_asms): Likewise.
889 * lto-section-in.c (lto_get_section_data): Instead do it here
890 for every section.
892 2014-03-04 Richard Biener <rguenther@suse.de>
894 PR tree-optimization/60382
895 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
896 dead PHIs a reduction.
898 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
900 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
901 hint value.
902 (_mm_prefetch): Move out of GCC target("sse") pragma.
903 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
904 GCC target("prfchw") pragma.
905 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
906 for locality <= 2.
907 * config/i386/i386.c (ix86_option_override_internal): Enable
908 -mprfchw with -mprefetchwt1.
910 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
912 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
913 Mark as varying.
915 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
917 * opts.h (CL_PCH_IGNORE): Define.
918 * targhooks.c (option_affects_pch_p):
919 Return false for options that have CL_PCH_IGNORE set.
920 * opt-functions.awk: Process PchIgnore.
921 * doc/options.texi: Document PchIgnore.
923 * config/arc/arc.opt (misize): Add PchIgnore property.
925 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
927 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
928 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
929 constraint on constants to permit them being loaded into
930 GENERAL_REGS or BASE_REGS.
932 2014-03-03 Nick Clifton <nickc@redhat.com>
934 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
935 anti-cacnonical alternatives.
936 (negandhi3_real): New pattern.
937 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
939 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
941 * config/avr/avr-mcus.def: Remove atxmega16x1.
942 * config/avr/avr-tables.opt: Regenerate.
943 * config/avr/t-multilib: Regenerate.
944 * doc/avr-mmcu.texi: Regenerate.
946 2014-03-03 Tobias Grosser <tobias@grosser.es>
947 Mircea Namolaru <mircea.namolaru@inria.fr>
949 PR tree-optimization/58028
950 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
951 scalar dimensions.
953 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
955 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
956 not handled by recognizers.
958 2014-03-03 Jakub Jelinek <jakub@redhat.com>
960 PR middle-end/60175
961 * function.c (expand_function_end): Don't emit
962 clobber_return_register sequence if clobber_after is a BARRIER.
963 * cfgexpand.c (construct_exit_block): Append instructions before
964 return_label to prev_bb.
966 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
968 * config/rs6000/constraints.md: Document reserved use of "wc".
970 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
972 PR ipa/60150
973 * ipa.c (function_and_variable_visibility): When dissolving comdat
974 group, also set all symbols to local.
976 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
978 PR ipa/60306
980 Revert:
981 2013-12-14 Jan Hubicka <jh@suse.cz>
982 PR middle-end/58477
983 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
985 2014-03-02 Jon Beniston <jon@beniston.com>
987 PR bootstrap/48230
988 PR bootstrap/50927
989 PR bootstrap/52466
990 PR target/46898
991 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
992 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
993 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
994 (simple_return, *simple_return): New patterns
995 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
996 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
998 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1000 * dwarf2out.c (gen_subprogram_die): Tidy.
1002 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1004 PR target/60071
1005 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1006 (*mov_t_msb_neg_negc): ... this new insn.
1008 2014-02-28 Jason Merrill <jason@redhat.com>
1010 PR c++/58678
1011 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1012 function.
1014 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1016 PR c++/60314
1017 * dwarf2out.c (decltype_auto_die): New static.
1018 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1019 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1020 (is_cxx_auto): Likewise.
1022 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1024 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1025 we are not using general regs only.
1027 2014-02-28 Richard Biener <rguenther@suse.de>
1029 PR target/60280
1030 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1031 previous fix and only allow to remove trivial pre-headers
1032 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1033 (remove_forwarder_block): Properly update the latch of a loop.
1035 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1037 PR debug/59992
1038 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1039 (cselib_preserved_hash_table): New.
1040 (preserve_constants_and_equivs): Move preserved vals to it.
1041 (cselib_find_slot): Look it up first.
1042 (cselib_init): Initialize it.
1043 (cselib_finish): Release it.
1044 (dump_cselib_table): Dump it.
1046 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1048 PR debug/59992
1049 * cselib.c (remove_useless_values): Skip to avoid quadratic
1050 behavior if the condition moved from...
1051 (cselib_process_insn): ... here holds.
1053 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1055 PR debug/57232
1056 * var-tracking.c (vt_initialize): Apply the same condition to
1057 preserve the CFA base value.
1059 2014-02-28 Joey Ye <joey.ye@arm.com>
1061 PR target/PR60169
1062 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1063 if reload in progress or completed.
1065 2014-02-28 Tobias Burnus <burnus@net-b.de>
1067 PR middle-end/60147
1068 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1069 NAMELIST_DECL.
1071 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1073 * doc/tm.texi.in (Condition Code Status): Update documention for
1074 relative locations of cc0-setter and cc0-user.
1076 2014-02-27 Jeff Law <law@redhat.com>
1078 PR rtl-optimization/52714
1079 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1080 into two independent simple sets, if I3 is a jump, ensure the
1081 pattern we place into I3 is a (set (pc) ...).
1083 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1084 Jeff Law <law@redhat.com>
1086 PR rtl-optimization/49847
1087 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1088 are in different blocks.
1089 * doc/tm.texi (Condition Code Status): Update documention for
1090 relative locations of cc0-setter and cc0-user.
1092 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1094 PR target/59222
1095 * lra.c (lra_emit_add): Check SUBREG too.
1097 2014-02-27 Andreas Schwab <schwab@suse.de>
1099 * config/m68k/m68k.c (m68k_option_override): Disable
1100 -flive-range-shrinkage for classic m68k.
1101 (m68k_override_options_after_change): Likewise.
1103 2014-02-27 Marek Polacek <polacek@redhat.com>
1105 PR middle-end/59223
1106 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1107 -Wmaybe-uninitialized.
1109 2014-02-27 Alan Modra <amodra@gmail.com>
1111 PR target/57936
1112 * reload1.c (emit_input_reload_insns): When reload_override_in,
1113 set old to rl->in_reg when rl->in_reg is a subreg.
1115 2014-02-26 Richard Biener <rguenther@suse.de>
1117 PR bootstrap/60343
1118 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1120 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1122 * common/config/i386/predicates.md (const1256_operand): Remove.
1123 (const2356_operand): New.
1124 (const_1_to_2_operand): Remove.
1125 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1126 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1127 (*avx512pf_gatherpf<mode>sf): Ditto.
1128 (avx512pf_gatherpf<mode>df): Ditto.
1129 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1130 (*avx512pf_gatherpf<mode>df): Ditto.
1131 (avx512pf_scatterpf<mode>sf): Ditto.
1132 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1133 (*avx512pf_scatterpf<mode>sf): Ditto.
1134 (avx512pf_scatterpf<mode>df): Ditto.
1135 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1136 (*avx512pf_scatterpf<mode>df): Ditto.
1137 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1139 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1141 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1142 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1143 (_mm512_mask_testn_epi64_mask): Move to ...
1144 * config/i386/avx512cdintrin.h: Here.
1145 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1146 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1147 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1148 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1149 TARGET_AVX512F from TARGET_AVX512CD.
1151 2014-02-26 Richard Biener <rguenther@suse.de>
1153 PR ipa/60327
1154 * ipa.c (walk_polymorphic_call_targets): Properly guard
1155 call to inline_update_overall_summary.
1157 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1159 PR target/60280
1160 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1161 and latches only if requested. Fix latch if it is removed.
1162 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1163 LOOPS_HAVE_PREHEADERS.
1165 2014-02-25 Andrew Pinski <apinski@cavium.com>
1167 * builtins.c (expand_builtin_thread_pointer): Create a new target
1168 when the target is NULL.
1170 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1172 PR rtl-optimization/60317
1173 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1174 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1175 * lra-assigns.c: Include params.h.
1176 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1177 other reload pseudos considerations.
1179 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1181 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1182 to use canonical form for nor<mode>3.
1184 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1186 PR target/55426
1187 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1188 conversions.
1190 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1192 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1193 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1194 (ix86_handle_option): Handle OPT_mprefetchwt1.
1195 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1196 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1197 PREFETCHWT1 CPUID.
1198 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1199 OPTION_MASK_ISA_PREFETCHWT1.
1200 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1201 (PTA_PREFETCHWT1): New.
1202 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1203 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1204 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1205 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1206 (*prefetch_avx512pf_<mode>_: Change into ...
1207 (*prefetch_prefetchwt1_<mode>: This.
1208 * config/i386/i386.opt (mprefetchwt1): New.
1209 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1210 (_mm_prefetch): Handle intent to write.
1211 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1213 2014-02-25 Richard Biener <rguenther@suse.de>
1215 PR middle-end/60291
1216 * emit-rtl.c (mem_attrs_htab): Remove.
1217 (mem_attrs_htab_hash): Likewise.
1218 (mem_attrs_htab_eq): Likewise.
1219 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1220 (init_emit_once): Do not allocate mem_attrs_htab.
1222 2014-02-25 Richard Biener <rguenther@suse.de>
1224 PR lto/60319
1225 * lto-opts.c (lto_write_options): Output non-explicit conservative
1226 -fwrapv, -fno-trapv and -fno-strict-overflow.
1227 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1228 (run_gcc): And pass them through.
1230 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1232 * sel-sched.c (calculate_new_fences): New parameter ptime.
1233 Calculate it as a maximum over all fence cycles.
1234 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1235 Print the final schedule timing when sched_verbose.
1237 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1239 PR rtl-optimization/60292
1240 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1241 bit fot the fence instruction.
1243 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1245 * calls.h: Fix typo in comment.
1247 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1249 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1250 adjusting offsetable addresses.
1252 2014-02-24 Guozhi Wei <carrot@google.com>
1254 * sparseset.h (sparseset_pop): Fix the wrong index.
1256 2014-02-24 Walter Lee <walt@tilera.com>
1258 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1259 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1260 triplet.
1261 * common/config/tilegx/tilegx-common.c
1262 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1263 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1264 (LINK_SPEC): Ditto.
1265 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1266 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1267 (tilegx_gimplify_va_arg_expr): Handle big endian.
1268 (tilegx_expand_unaligned_load): Ditto.
1269 (tilegx_expand_unaligned_store): Ditto.
1270 (TARGET_RETURN_IN_MSB): New.
1271 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1272 (TARGET_ENDIAN_DEFAULT): New.
1273 (TARGET_BIG_ENDIAN): Handle big endian.
1274 (BYTES_BIG_ENDIAN): Ditto.
1275 (WORDS_BIG_ENDIAN): Ditto.
1276 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1277 (ENDIAN_SPEC): New.
1278 (EXTRA_SPECS): New.
1279 * config/tilegx/tilegx.md (extv): Handle big endian.
1280 (extzv): Ditto.
1281 (insn_st<n>): Ditto.
1282 (insn_st<n>_add<bitsuffix>): Ditto.
1283 (insn_stnt<n>): Ditto.
1284 (insn_stnt<n>_add<bitsuffix>):Ditto.
1285 (vec_interleave_highv8qi): Handle big endian.
1286 (vec_interleave_highv8qi_be): New.
1287 (vec_interleave_highv8qi_le): New.
1288 (insn_v1int_h): Handle big endian.
1289 (vec_interleave_lowv8qi): Handle big endian.
1290 (vec_interleave_lowv8qi_be): New.
1291 (vec_interleave_lowv8qi_le): New.
1292 (insn_v1int_l): Handle big endian.
1293 (vec_interleave_highv4hi): Handle big endian.
1294 (vec_interleave_highv4hi_be): New.
1295 (vec_interleave_highv4hi_le): New.
1296 (insn_v2int_h): Handle big endian.
1297 (vec_interleave_lowv4hi): Handle big endian.
1298 (vec_interleave_lowv4hi_be): New.
1299 (vec_interleave_lowv4hi_le): New.
1300 (insn_v2int_l): Handle big endian.
1301 (vec_interleave_highv2si): Handle big endian.
1302 (vec_interleave_highv2si_be): New.
1303 (vec_interleave_highv2si_le): New.
1304 (insn_v4int_h): Handle big endian.
1305 (vec_interleave_lowv2si): Handle big endian.
1306 (vec_interleave_lowv2si_be): New.
1307 (vec_interleave_lowv2si_le): New.
1308 (insn_v4int_l): Handle big endian.
1309 * config/tilegx/tilegx.opt (mbig-endian): New option.
1310 (mlittle-endian): New option.
1311 * doc/install.texi: Document tilegxbe-linux.
1312 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1314 2014-02-24 Martin Jambor <mjambor@suse.cz>
1316 PR ipa/60266
1317 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1318 there are no parameter descriptors.
1320 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1322 PR rtl-optimization/60268
1323 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1324 initialization to ...
1325 (sched_rgn_init): ... here.
1326 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1328 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1330 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1331 names.
1333 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1335 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1336 definition.
1338 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1340 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1341 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1343 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1345 * config/microblaze/predicates.md: Add cmp_op predicate.
1346 * config/microblaze/microblaze.md: Add branch_compare instruction
1347 which uses cmp_op predicate and emits cmp insn before branch.
1348 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1349 to microblaze_expand_conditional_branch and consolidate logic.
1350 (microblaze_expand_conditional_branch): emit branch_compare
1351 insn instead of handling cmp op separate from branch insn.
1353 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1355 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1356 to permit subregs.
1358 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1360 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1361 define_insn with define_expand and new define_insn
1362 *altivec_lve<VI_char>x_internal.
1363 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1364 and new define_insn *altivec_stve<VI_char>x_internal.
1365 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1366 prototype.
1367 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1368 lve*x built-ins.
1369 (altivec_expand_stvex_be): New function.
1371 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1373 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1374 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1375 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1376 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1378 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1380 PR target/60298
1381 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1382 instead of emit_move_insn.
1384 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1386 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1387 vspltw with vsldoi.
1388 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1389 gen_altivec_vsumsws.
1391 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1393 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1394 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1395 (altivec_lvxl_<mode>): New define_expand incorporating
1396 -maltivec=be semantics where needed.
1397 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1398 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1399 semantics where needed.
1400 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1401 (altivec_stvx_<mode>): New define_expand incorporating
1402 -maltivec=be semantics where needed.
1403 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1404 VM2 iterator instead of V4SI.
1405 (altivec_stvxl_<mode>): New define_expand incorporating
1406 -maltivec=be semantics where needed.
1407 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1408 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1409 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1410 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1411 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1412 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1413 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1414 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1415 ALTIVEC_BUILTIN_STVXL.
1416 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1417 (altivec_expand_stvx_be): Likewise.
1418 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1419 (altivec_expand_lvx_be): Likewise.
1420 (altivec_expand_stvx_be): Likewise.
1421 (altivec_expand_builtin): Add cases for
1422 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1423 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1424 (altivec_init_builtins): Add definitions for
1425 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1426 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1428 2014-02-21 Catherine Moore <clm@codesourcery.com>
1430 * doc/invoke.texi (mvirt, mno-virt): Document.
1431 * config/mips/mips.opt (mvirt): New option.
1432 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1434 2014-02-21 Richard Biener <rguenther@suse.de>
1436 PR tree-optimization/60276
1437 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1438 (STMT_VINFO_MIN_NEG_DIST): New macro.
1439 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1440 STMT_VINFO_MIN_NEG_DIST.
1441 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1442 made for negative dependence distances still hold.
1444 2014-02-21 Richard Biener <rguenther@suse.de>
1446 PR middle-end/60291
1447 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1448 DECL_INITIAL for globals not in the current function context.
1450 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1452 PR tree-optimization/56490
1453 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1454 * tree-ssa-uninit.c: Include params.h.
1455 (compute_control_dep_chain): Add num_calls argument, return false
1456 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1457 num_calls to recursive call.
1458 (find_predicates): Change dep_chain into normal array,
1459 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1460 variable and adjust compute_control_dep_chain caller.
1461 (find_def_preds): Likewise.
1463 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1465 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1466 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1468 2014-02-21 Nick Clifton <nickc@redhat.com>
1470 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1471 (pushhi1): Likewise.
1472 (popqi1): Add mode to pre_dec.
1473 (pophi1): Likewise.
1475 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1477 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1478 mode for mask of V8SFmode permutation.
1480 2014-02-20 Richard Henderson <rth@redhat.com>
1482 PR c++/60272
1483 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1484 a new pseudo for OLDVAL.
1486 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1488 PR target/57896
1489 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1490 gen_reg_rtx if d->testing_p.
1491 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1492 if d->testing_p and we will certainly return true.
1493 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1494 if d->testing_p.
1496 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1498 * emit-rtl.c (gen_reg_rtx): Assert that
1499 crtl->emit.regno_pointer_align_length is non-zero.
1501 2014-02-20 Richard Henderson <rth@redhat.com>
1503 PR c++/60272
1504 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1505 on failure the store back into EXPECT.
1507 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1508 Sandra Loosemore <sandra@codesourcery.com>
1510 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1511 * config/nios2/nios2.c (nios2_function_profiler): Add
1512 -fPIC (flag_pic == 2) support.
1513 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1514 (nios2_large_offset_p): New function.
1515 (nios2_unspec_reloc_p): Move up position, update to use
1516 nios2_large_offset_p.
1517 (nios2_unspec_address): Remove function.
1518 (nios2_unspec_offset): New function.
1519 (nios2_large_got_address): New function.
1520 (nios2_got_address): Add large offset support.
1521 (nios2_legitimize_tls_address): Update usage of removed and new
1522 functions.
1523 (nios2_symbol_binds_local_p): New function.
1524 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1525 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1526 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1527 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1528 processing for (const (unspec ...)).
1529 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1531 2014-02-20 Richard Biener <rguenther@suse.de>
1533 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1534 doing the substitution.
1535 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1537 2014-02-20 Martin Jambor <mjambor@suse.cz>
1539 PR ipa/55260
1540 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1541 info when checking whether lattices are bottom.
1543 2014-02-20 Richard Biener <rguenther@suse.de>
1545 PR middle-end/60221
1546 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1547 regions at -O0.
1549 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1551 PR ipa/58555
1552 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1553 parameter specifying the scaling.
1554 (inline_call): Update.
1555 (want_inline_recursively): Guard division by zero.
1556 (recursive_inlining): Update.
1557 * ipa-inline.h (clone_inlined_nodes): Update.
1559 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1561 PR target/60204
1562 * config/i386/i386.c (classify_argument): Pass structures of size
1563 64 bytes or less in register.
1565 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1566 Kirill Yukhin <kirill.yukhin@intel.com>
1568 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1569 (_mm_rcp28_round_ss): Ditto.
1570 (_mm_rsqrt28_round_sd): Ditto.
1571 (_mm_rsqrt28_round_ss): Ditto.
1572 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1573 (_mm_rcp14_round_ss): Ditto.
1574 (_mm_rsqrt14_round_sd): Ditto.
1575 (_mm_rsqrt14_round_ss): Ditto.
1576 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1577 the first input operand, get rid of match_dup.
1578 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1579 attribute to sse.
1580 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1581 Ditto.
1582 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1583 operand as the first input operand, set type attribute.
1584 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1585 Set type attribute.
1586 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1587 operand as the first input operand, set type attribute.
1589 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1591 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1592 bit of zero.
1594 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1596 PR target/60207
1597 * config/i386/i386.c (construct_container): Remove TFmode check
1598 for X86_64_INTEGER_CLASS.
1600 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1602 PR target/59794
1603 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1604 only when -Wpsabi is enabled.
1606 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1608 PR target/59799
1609 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1610 passing arrays in registers are the same as for structs, so remove the
1611 special case for them.
1613 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1615 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1616 destination type, extract only the valid bits if the source type is not
1617 integral and has a different mode.
1619 2014-02-19 Richard Biener <rguenther@suse.de>
1621 PR ipa/60243
1622 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1623 for all calls.
1625 2014-02-19 Richard Biener <rguenther@suse.de>
1627 PR ipa/60243
1628 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1629 (ipa_modify_call_arguments): Emit an argument load explicitely and
1630 preserve virtual SSA form there and for the replacement call.
1631 Do not update SSA form nor free dominance info.
1633 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1635 * ipa.c (function_and_variable_visibility): Also clear WEAK
1636 flag when disolving COMDAT_GROUP.
1638 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1640 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1641 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1642 not devirtualizing.
1643 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1644 do more sanity checks.
1645 (detect_type_change): Return true when giving up early.
1646 (compute_complex_assign_jump_func): Fix type parameter of
1647 ipa_set_ancestor_jf.
1648 (compute_complex_ancestor_jump_func): Likewise.
1649 (update_jump_functions_after_inlining): Fix updating of
1650 ancestor function.
1651 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1653 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1655 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1656 inline clones when edge disappears.
1658 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1660 PR target/60203
1661 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1662 Split 64-bit moves into 2 patterns. Do not allow the use of
1663 direct move for TDmode in little endian, since the decimal value
1664 has little endian bytes within a word, but the 64-bit pieces are
1665 ordered in a big endian fashion, and normal subreg's of TDmode are
1666 not allowed.
1667 (mov<mode>_64bit_dm): Likewise.
1668 (movtd_64bit_nodm): Likewise.
1670 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1672 PR tree-optimization/60174
1673 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1674 statement of an SSA_NAME that occurs in an abnormal PHI node.
1676 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1678 PR sanitizer/60142
1679 * final.c (SEEN_BB): Remove.
1680 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1681 (final_scan_insn): Don't force_source_line on second
1682 NOTE_INSN_BASIC_BLOCK.
1684 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1686 PR target/60205
1687 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1688 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1689 (type_natural_mode): Warn ABI change when %zmm register is not
1690 available for AVX512F vector value passing.
1692 2014-02-18 Kai Tietz <ktietz@redhat.com>
1694 PR target/60193
1695 * config/i386/i386.c (ix86_expand_prologue): Use value in
1696 rax register as displacement when restoring %r10 or %rax.
1697 Fix wrong offset when restoring both registers.
1699 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1701 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1702 assertion with conditional return.
1704 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1705 Uros Bizjak <ubizjak@gmail.com>
1707 PR driver/60233
1708 * config/i386/driver-i386.c (host_detect_local_cpu): If
1709 YMM state is not saved by the OS, also clear has_f16c. Move
1710 CPUID 0x80000001 handling before YMM state saving checking.
1712 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1714 PR rtl-optimization/58960
1715 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1716 factored out from ...
1717 (sched_init): ... here.
1718 (free_global_sched_pressure_data): New, factored out from ...
1719 (sched_finish): ... here.
1720 * sched-int.h (free_global_sched_pressure_data): Declare.
1721 * sched-rgn.c (nr_regions_initial): New static global.
1722 (haifa_find_rgns): Initialize it.
1723 (schedule_region): Disable sched-pressure for the newly
1724 generated regions.
1726 2014-02-17 Richard Biener <rguenther@suse.de>
1728 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1729 release SSA defs of pattern stmts.
1731 2014-02-17 Richard Biener <rguenther@suse.de>
1733 * tree-inline.c (expand_call_inline): Release the virtual
1734 operand defined by the call we are about to inline.
1736 2014-02-17 Richard Biener <rguenther@suse.de>
1738 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1740 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1741 Ilya Tocar <ilya.tocar@intel.com>
1743 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1744 arguments order in builtin.
1745 (_mm512_permutexvar_epi64): Ditto.
1746 (_mm512_mask_permutexvar_epi64): Ditto
1747 (_mm512_maskz_permutexvar_epi32): Ditto
1748 (_mm512_permutexvar_epi32): Ditto
1749 (_mm512_mask_permutexvar_epi32): Ditto
1751 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1753 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1754 (p8_vmrgow): Likewise.
1756 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1758 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1759 endian targets.
1761 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1763 PR target/60203
1764 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1765 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1766 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1767 using direct move instructions on ISA 2.07. Also adjust
1768 instruction length for 64-bit.
1769 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1770 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1772 2014-02-15 Alan Modra <amodra@gmail.com>
1774 PR target/58675
1775 PR target/57935
1776 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1777 find_replacement on parts of insn rtl that might be reloaded.
1779 2014-02-15 Richard Biener <rguenther@suse.de>
1781 PR tree-optimization/60183
1782 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1783 (tree_ssa_phiprop): Calculate and free post-dominators.
1785 2014-02-14 Jeff Law <law@redhat.com>
1787 PR rtl-optimization/60131
1788 * ree.c (get_extended_src_reg): New function.
1789 (combine_reaching_defs): Use it rather than assuming location of REG.
1790 (find_and_remove_re): Verify first operand of extension is
1791 a REG before adding the insns to the copy list.
1793 2014-02-14 Roland McGrath <mcgrathr@google.com>
1795 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1796 * configure: Regenerated.
1797 * config.in: Regenerated.
1798 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1799 instead of ASM_SHORT.
1801 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1802 Richard Earnshaw <rearnsha@arm.com>
1804 PR rtl-optimization/59535
1805 * lra-constraints.c (process_alt_operands): Encourage alternative
1806 when unassigned pseudo class is superset of the alternative class.
1807 (inherit_reload_reg): Don't inherit when optimizing for code size.
1808 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1809 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1810 modes not less than 4 for Thumb1.
1812 2014-02-14 Kyle McMartin <kyle@redhat.com>
1814 PR pch/60010
1815 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1817 2014-02-14 Richard Biener <rguenther@suse.de>
1819 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1820 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1821 Do not strip INDIRECT_REFs.
1823 2014-02-14 Richard Biener <rguenther@suse.de>
1825 PR lto/60179
1826 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1827 DECL_FUNCTION_SPECIFIC_TARGET.
1828 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1829 * tree-streamer-out.c (pack_ts_target_option): Remove.
1830 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1831 (write_ts_function_decl_tree_pointers): Do not stream
1832 DECL_FUNCTION_SPECIFIC_TARGET.
1833 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1834 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1835 (lto_input_ts_function_decl_tree_pointers): Do not stream
1836 DECL_FUNCTION_SPECIFIC_TARGET.
1838 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1840 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1841 (get_initial_def_for_induction, vectorizable_induction): Ignore
1842 debug stmts when looking for exit_phi.
1843 (vectorizable_live_operation): Fix up condition.
1845 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1847 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1848 nreverse() because it changes the content of original tree list.
1850 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1852 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1853 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1855 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1857 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1858 GNU coding standards.
1860 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1862 PR debug/60152
1863 * dwarf2out.c (gen_subprogram_die): Don't call
1864 add_calling_convention_attribute if subr_die is old_die.
1866 2014-02-13 Sharad Singhai <singhai@google.com>
1868 * doc/optinfo.texi: Fix order of nodes.
1870 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1872 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1873 operands[2], not operands[3].
1875 2014-02-13 Richard Biener <rguenther@suse.de>
1877 PR bootstrap/59878
1878 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1879 mention the possibility of an in-tree build.
1880 (CLooG): Update recommended version to 0.18.1, mention the
1881 possibility of an in-tree build and clarify that the ISL
1882 bundled with CLooG does not work.
1884 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1886 PR target/43546
1887 * expr.c (compress_float_constant): If x is a hard register,
1888 extend into a pseudo and then move to x.
1890 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1892 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1893 caused by bad second argument to warning_at() with -mhotpatch and
1894 nested functions (e.g. with gfortran).
1896 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1898 * opts.c (option_name): Remove "enabled by default" rider.
1900 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1902 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1904 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1905 Uros Bizjak <ubizjak@gmail.com>
1907 PR target/60151
1908 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1909 * configure: Regenerated.
1911 2014-02-12 Richard Biener <rguenther@suse.de>
1913 * vec.c (vec_prefix::calculate_allocation): Move as
1914 inline variant to vec.h.
1915 (vec_prefix::calculate_allocation_1): New out-of-line version.
1916 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1917 (vec_prefix::m_has_auto_buf): Rename to ...
1918 (vec_prefix::m_using_auto_storage): ... this.
1919 (vec_prefix::calculate_allocation): Inline the easy cases
1920 and dispatch to calculate_allocation_1 which doesn't need the
1921 prefix address.
1922 (va_heap::reserve): Use gcc_checking_assert.
1923 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1924 m_using_auto_storage.
1925 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1926 member and adjust.
1927 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1928 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1929 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1931 2014-02-12 Richard Biener <rguenther@suse.de>
1933 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1934 when we found a dependence.
1936 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1938 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1939 common code...
1940 (maybe_fold_stmt): ... into this new function.
1941 * omp-low.c (lower_omp): Update comment.
1943 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1944 last use.
1946 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1947 dereference.
1949 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1951 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1952 identifiers in comments.
1953 (cortexa53_extra_costs): Likewise.
1954 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1955 (cortexa7_extra_costs): Likewise.
1956 (cortexa12_extra_costs): Likewise.
1957 (cortexa15_extra_costs): Likewise.
1958 (v7m_extra_costs): Likewise.
1960 2014-02-12 Richard Biener <rguenther@suse.de>
1962 PR middle-end/60092
1963 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1964 of posix_memalign being successful.
1965 (lower_stmt): Restrict lowering of posix_memalign to when
1966 -ftree-bit-ccp is enabled.
1968 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1970 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1971 arg_loc.
1972 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1974 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1976 PR rtl-optimization/60116
1977 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1978 other_insn once the combination has been validated.
1980 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1982 PR lto/59468
1983 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1984 and wrapper.
1985 * ipa-devirt.c: Include demangle.h
1986 (odr_violation_reported): New static variable.
1987 (add_type_duplicate): Update odr_violations.
1988 (maybe_record_node): Add completep parameter; update it.
1989 (record_target_from_binfo): Add COMPLETEP parameter;
1990 update it as needed.
1991 (possible_polymorphic_call_targets_1): Likewise.
1992 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1993 rename FINAL to COMPLETE.
1994 (record_targets_from_bases): Sanity check we found the binfo;
1995 fix COMPLETEP updating.
1996 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1997 parameter, fix computing of COMPLETEP.
1998 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1999 at LTO time do demangling.
2000 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2001 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2002 parameter.
2003 (gimple_get_virt_method_for_binfo): Likewise.
2004 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2005 gimple_get_virt_method_for_vtable): Update prototypes.
2007 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2009 PR target/49008
2010 * genautomata.c (add_presence_absence): Fix typo with
2011 {final_}presence_list.
2013 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2015 PR target/60137
2016 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2017 for VSX/Altivec vectors that land in GPR registers.
2019 2014-02-11 Richard Henderson <rth@redhat.com>
2020 Jakub Jelinek <jakub@redhat.com>
2022 PR debug/59776
2023 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2024 around drhs if type conversion to lacc->type is not useless.
2026 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2028 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2029 tuning struct.
2030 (cortex-a57.cortex-a53): Likewise.
2031 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2033 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2035 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2036 arm_restrict_it.
2038 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2040 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2041 add_options_for_arm_vfp3.
2043 2014-02-11 Jeff Law <law@redhat.com>
2045 PR middle-end/54041
2046 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2047 object with an undesirable mode.
2049 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2051 PR libgomp/60107
2052 * config/i386/sol2-9.h: New file.
2053 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2054 *-*-solaris2.9*): Use it.
2056 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2058 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2059 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2061 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2063 * config/microblaze/microblaze.c: Extend mcpu version format
2065 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2067 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2069 2014-02-10 Richard Henderson <rth@redhat.com>
2071 PR target/59927
2072 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2073 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2074 ms-abi vs -mno-accumulate-outgoing-args.
2075 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2076 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2077 respect to ms-abi.
2079 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2081 PR middle-end/60080
2082 * cfgexpand.c (expand_asm_operands): Attach source location to
2083 ASM_INPUT rtx objects.
2084 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2086 2014-02-10 Nick Clifton <nickc@redhat.com>
2088 * config/mn10300/mn10300.c (popcount): New function.
2089 (mn10300_expand_prologue): Include saved registers in stack usage
2090 count.
2092 2014-02-10 Jeff Law <law@redhat.com>
2094 PR middle-end/52306
2095 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2096 when changing the SET_DEST of a prior insn to avoid an input reload.
2098 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2100 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2101 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2102 -mcall-openbsd, or -mcall-linux.
2103 (CC1_ENDIAN_BIG_SPEC): Remove.
2104 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2105 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2106 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2107 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2108 and %cc1_endian_default.
2109 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2111 2014-02-10 Richard Biener <rguenther@suse.de>
2113 PR tree-optimization/60115
2114 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2115 MEM_REF handling. Properly verify that the accesses are not
2116 out of the objects bound.
2118 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2120 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2121 coretex to cortex.
2123 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2125 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2126 proper constants and fix formatting.
2127 (possible_polymorphic_call_targets): Fix formatting.
2129 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2130 Ilya Tocar <ilya.tocar@intel.com>
2132 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2133 (_mm512_loadu_epi32): Renamed into...
2134 (_mm512_loadu_si512): This.
2135 (_mm512_storeu_epi32): Renamed into...
2136 (_mm512_storeu_si512): This.
2137 (_mm512_maskz_ceil_ps): Removed.
2138 (_mm512_maskz_ceil_pd): Ditto.
2139 (_mm512_maskz_floor_ps): Ditto.
2140 (_mm512_maskz_floor_pd): Ditto.
2141 (_mm512_floor_round_ps): Ditto.
2142 (_mm512_floor_round_pd): Ditto.
2143 (_mm512_ceil_round_ps): Ditto.
2144 (_mm512_ceil_round_pd): Ditto.
2145 (_mm512_mask_floor_round_ps): Ditto.
2146 (_mm512_mask_floor_round_pd): Ditto.
2147 (_mm512_mask_ceil_round_ps): Ditto.
2148 (_mm512_mask_ceil_round_pd): Ditto.
2149 (_mm512_maskz_floor_round_ps): Ditto.
2150 (_mm512_maskz_floor_round_pd): Ditto.
2151 (_mm512_maskz_ceil_round_ps): Ditto.
2152 (_mm512_maskz_ceil_round_pd): Ditto.
2153 (_mm512_expand_pd): Ditto.
2154 (_mm512_expand_ps): Ditto.
2155 * config/i386/i386.c (ix86_builtins): Remove
2156 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2157 (bdesc_args): Ditto.
2158 * config/i386/predicates.md (const1256_operand): New.
2159 (const_1_to_2_operand): Ditto.
2160 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2161 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2162 (*avx512pf_gatherpf<mode>sf): Ditto.
2163 (avx512pf_gatherpf<mode>df): Ditto.
2164 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2165 (*avx512pf_gatherpf<mode>df): Ditto.
2166 (avx512pf_scatterpf<mode>sf): Ditto.
2167 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2168 (*avx512pf_scatterpf<mode>sf): Ditto.
2169 (avx512pf_scatterpf<mode>df): Ditto.
2170 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2171 (*avx512pf_scatterpf<mode>df): Ditto.
2172 (avx512f_expand<mode>): Removed.
2173 (<shift_insn><mode>3<mask_name>): Change predicate type.
2175 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2177 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2178 not at the end of datarefs vector use ordered_remove to avoid
2179 reordering datarefs vector.
2181 PR c/59984
2182 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2183 mark local addressable non-static vars as GOVD_PRIVATE
2184 instead of GOVD_LOCAL.
2185 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2186 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2187 than copying them.
2189 PR middle-end/60092
2190 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2191 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2192 assume_aligned or alloc_align attributes.
2193 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2194 arguments. Handle also assume_aligned and alloc_align attributes.
2195 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2196 calls to functions with assume_aligned or alloc_align attributes.
2197 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2199 2014-02-08 Terry Guo <terry.guo@arm.com>
2201 * doc/invoke.texi: Document ARM -march=armv7e-m.
2203 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2205 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2206 flag on __cilkrts_rethrow builtin.
2208 PR ipa/60026
2209 * ipa-cp.c (determine_versionability): Fail at -O0
2210 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2211 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2213 Revert:
2214 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2216 PR ipa/60026
2217 * tree-inline.c (copy_forbidden): Fail for
2218 __attribute__((optimize (0))) functions.
2220 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2222 * varpool.c: Include pointer-set.h.
2223 (varpool_remove_unreferenced_decls): Variables in other partitions
2224 will not be output; be however careful to not lose information
2225 about partitioning.
2227 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2229 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2230 lookup in the vtable constructor.
2232 2014-02-07 Jeff Law <law@redhat.com>
2234 PR target/40977
2235 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2236 define_insn_and_split.
2238 * ipa-inline.c (inline_small_functions): Fix typos.
2240 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2242 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2243 (s390_can_use_return_insn): Declare.
2244 * config/s390/s390.h (EPILOGUE_USES): Define.
2245 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2246 instructions.
2247 (s390_chunkify_start): Handle return JUMP_LABELs.
2248 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2249 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2250 (s390_can_use_return_insn): New functions.
2251 (s390_fix_long_loop_prediction): Handle conditional returns.
2252 (TARGET_SET_UP_BY_PROLOGUE): Define.
2253 * config/s390/s390.md (ANY_RETURN): New code iterator.
2254 (*creturn, *csimple_return, return, simple_return): New patterns.
2256 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2258 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2259 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2260 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2261 REG_CFA_RESTORE list when deciding not to restore a register.
2263 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2265 * config/s390/s390.c: Include tree-pass.h and context.h.
2266 (s390_early_mach): New function, split out from...
2267 (s390_emit_prologue): ...here.
2268 (pass_data_s390_early_mach): New pass structure.
2269 (pass_s390_early_mach): New class.
2270 (s390_option_override): Create and register early_mach pass.
2271 Move to end of file.
2273 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2275 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2276 to match for the exit block.
2278 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2280 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2281 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2282 Reject misaligned operands.
2284 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2286 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2288 2014-02-07 Richard Biener <rguenther@suse.de>
2290 PR middle-end/60092
2291 * gimple-low.c (lower_builtin_posix_memalign): New function.
2292 (lower_stmt): Call it to lower posix_memalign in a way
2293 to make alignment info accessible.
2295 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2297 PR c++/60082
2298 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2299 __builtin_setjmp_receiver.
2301 2014-02-07 Richard Biener <rguenther@suse.de>
2303 PR middle-end/60092
2304 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2305 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2306 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2307 Handle BUILT_IN_POSIX_MEMALIGN.
2308 (find_func_clobbers): Likewise.
2309 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2310 (call_may_clobber_ref_p_1): Likewise.
2312 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2314 PR ipa/59918
2315 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2316 sanity check.
2318 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2320 PR ipa/59469
2321 * lto-cgraph.c (lto_output_node): Use
2322 symtab_get_symbol_partitioning_class.
2323 (lto_output_varpool_node): likewise.
2324 (symtab_get_symbol_partitioning_class): Move here from
2325 lto/lto-partition.c
2326 * cgraph.h (symbol_partitioning_class): Likewise.
2327 (symtab_get_symbol_partitioning_class): Declare.
2329 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2331 * ggc.h (ggc_internal_cleared_alloc): New macro.
2332 * vec.h (vec_safe_copy): Handle memory stats.
2333 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2334 * target-globals.c (save_target_globals): Likewise.
2336 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2338 PR target/60077
2339 * expr.c (emit_move_resolve_push): Export; be bit more selective
2340 on when to clear alias set.
2341 * expr.h (emit_move_resolve_push): Declare.
2342 * function.h (struct function): Add tail_call_marked.
2343 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2344 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2345 * config/i386/i386.md (TImode move expander): De not call
2346 ix86_expand_push.
2347 (FP push expanders): Preserve memory attributes.
2348 * config/i386/sse.md (push<mode>1): Remove.
2349 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2350 (ix86_expand_push): Remove.
2351 * config/i386/mmx.md (push<mode>1): Remove.
2353 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2355 PR rtl-optimization/60030
2356 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2357 lopart with paradoxical subreg before shifting it up by hprec.
2359 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2361 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2362 Remove extra newline at end of file.
2363 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2364 (arm_issue_rate): Handle cortexa57.
2365 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2366 (cortex-a57.cortex-a53): Likewise.
2368 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2370 PR target/59575
2371 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2372 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2373 bitmask.
2374 (arm_expand_prologue): Adjust all callers.
2375 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2376 info, registers also at the lowest numbered registers side. Use
2377 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2378 XEXP.
2380 PR debug/59992
2381 * var-tracking.c (adjust_mems): Before adding a SET to
2382 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2384 2014-02-06 Alan Modra <amodra@gmail.com>
2386 PR target/60032
2387 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2388 change SDmode to DDmode when lra_in_progress.
2390 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2392 PR middle-end/59150
2393 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2394 free_data_ref on the dr first, and before goto again also set dr
2395 to the next dr. For simd_lane_access, free old datarefs[i] before
2396 overwriting it. For get_vectype_for_scalar_type failure, don't
2397 free_data_ref if simd_lane_access.
2399 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2401 PR target/60062
2402 * tree.h (opts_for_fn): New inline function.
2403 (opt_for_fn): Define.
2404 * config/i386/i386.c (ix86_function_regparm): Use
2405 opt_for_fn (decl, optimize) instead of optimize.
2407 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2409 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2410 for SYMBOL_REF in large memory model.
2412 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2414 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2415 and crypto support.
2416 (cortex-a57): Likewise.
2417 (cortex-a57.cortex-a53): Likewise.
2419 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2420 Kugan Vivekanandarajah <kuganv@linaro.org>
2422 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2423 unaligned_access.
2424 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2426 2014-02-06 Richard Biener <rguenther@suse.de>
2428 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2429 set_loop_copy and initialize_original_copy_tables.
2431 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2433 * config/aarch64/aarch64-simd.md
2434 (aarch64_ashr_simddi): Change QI to SI.
2436 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2437 Jakub Jelinek <jakub@redhat.com>
2439 PR middle-end/60013
2440 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2441 of the dataflow.
2443 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2445 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2446 CODE_FOR_altivec_vpku[hw]um to
2447 CODE_FOR_altivec_vpku[hw]um_direct.
2448 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2449 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2450 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2451 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2453 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2455 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2456 generation for -maltivec=be.
2457 (altivec_vsumsws): Simplify redundant test.
2459 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2461 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2462 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2463 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2464 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2465 gen_altivec_vpkuwum.
2466 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2467 BYTES_BIG_ENDIAN.
2468 (altivec_vpks<VI_char>ss): Likewise.
2469 (altivec_vpks<VI_char>us): Likewise.
2470 (altivec_vpku<VI_char>us): Likewise.
2471 (altivec_vpku<VI_char>um): Likewise.
2472 (altivec_vpku<VI_char>um_direct): New (copy of
2473 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2474 internal use).
2475 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2476 target is little endian and -maltivec=be is not specified.
2477 (*altivec_vupkhs<VU_char>_direct): New (copy of
2478 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2479 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2480 target is little endian and -maltivec=be is not specified.
2481 (*altivec_vupkls<VU_char>_direct): New (copy of
2482 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2483 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2484 little endian and -maltivec=be is not specified.
2485 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2486 little endian and -maltivec=be is not specified.
2488 2014-02-05 Richard Henderson <rth@redhat.com>
2490 PR debug/52727
2491 * combine-stack-adj.c: Revert r206943.
2492 * sched-int.h (struct deps_desc): Add last_args_size.
2493 * sched-deps.c (init_deps): Initialize it.
2494 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2495 contain REG_ARGS_SIZE notes.
2497 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2499 * lto-cgraph.c (asm_nodes_output): Make global.
2500 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2501 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2502 (driver_handle_option): Handle OPT_fwpa.
2504 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2506 PR ipa/59947
2507 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2508 a comment typo and formatting issue. If odr_hash hasn't been
2509 created, return vNULL and set *completep to false.
2511 PR middle-end/57499
2512 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2513 bb with no successors.
2515 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2517 PR target/59718
2518 * doc/invoke.texi (-march): Clarify documentation for ARM.
2519 (-mtune): Likewise.
2520 (-mcpu): Likewise.
2522 2014-02-05 Richard Biener <rguenther@suse.de>
2524 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2525 when not vectorizing because of too many alias checks.
2526 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2527 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2529 2014-02-05 Nick Clifton <nickc@redhat.com>
2531 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2532 accept extended registers in any mode when compiling for the MN10300.
2534 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2536 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2537 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2538 sanitization attributes.
2539 (can_inline_edge_p): Likewise.
2540 (sanitize_attrs_match_for_inline_p): New function.
2542 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2544 * ipa-prop.c (detect_type_change): Shor circuit testing of
2545 type changes on THIS pointer.
2547 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2549 PR target/59777
2550 * config/pa/pa.c (legitimize_tls_address): Return original address
2551 if not passed a SYMBOL_REF rtx.
2552 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2553 addresses.
2554 (pa_emit_move_sequence): Simplify TLS source operands.
2555 (pa_legitimate_constant_p): Reject all TLS constants.
2556 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2557 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2559 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2561 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2562 groups when we know they are controlled by LTO.
2563 * varasm.c (default_binds_local_p_1): If object is in other partition,
2564 it will be resolved locally.
2566 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2568 * config/host-linux.c (linux_gt_pch_use_address): Don't
2569 use SSIZE_MAX because it is not always defined.
2571 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2573 PR bootstrap/59913
2574 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2575 threshold for pseudo splitting.
2576 (update_ebb_live_info): Process call argument hard registers and
2577 hard registers from insn definition too.
2578 (max_small_class_regs_num): New constant.
2579 (inherit_in_ebb): Update live hard regs through EBBs. Update
2580 reloads_num only for small register classes. Don't split for
2581 outputs of jumps.
2583 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2585 PR ipa/60058
2586 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2587 is non-null.
2589 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2591 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2592 visibility is safe.
2594 2014-02-04 Marek Polacek <polacek@redhat.com>
2596 * gdbinit.in (pel): Define.
2598 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2600 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2601 behavior.
2603 2014-02-04 Richard Biener <rguenther@suse.de>
2605 PR lto/59723
2606 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2607 in function context local.
2608 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2609 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2610 similar to LTO_imported_decl_ref.
2612 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2614 PR tree-optimization/60002
2615 * cgraphclones.c (build_function_decl_skip_args): Clear
2616 DECL_LANG_SPECIFIC.
2618 PR tree-optimization/60023
2619 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2620 false to gsi_replace.
2621 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2622 has been in some EH region and vec_stmt could throw, add
2623 vec_stmt into the same EH region.
2624 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2625 has no lhs, ignore it.
2626 * internal-fn.c (expand_MASK_LOAD): Likewise.
2628 PR ipa/60026
2629 * tree-inline.c (copy_forbidden): Fail for
2630 __attribute__((optimize (0))) functions.
2632 PR other/58712
2633 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2634 is set, copy one less argument.
2635 (expand_simd_clones): Don't subtract clone_info->inbranch
2636 from simd_clone_struct_alloc argument.
2638 PR rtl-optimization/57915
2639 * recog.c (simplify_while_replacing): If all unary/binary/relational
2640 operation arguments are constant, attempt to simplify those.
2642 PR middle-end/59261
2643 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2644 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2646 2014-02-04 Richard Biener <rguenther@suse.de>
2648 PR tree-optimization/60012
2649 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2650 TBAA disambiguation to all DDRs.
2652 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2654 PR target/59788
2655 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2656 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2658 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2660 PR ipa/59882
2661 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2663 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2665 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2666 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2668 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2670 PR ipa/59831
2671 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2672 to figure out targets of polymorphic calls with known decl.
2673 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2674 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2675 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2676 (get_polymorphic_call_info): ... here.
2677 (get_polymorphic_call_info_from_invariant): New function.
2679 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2681 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2682 lookup via vtable pointer; check for type consistency
2683 and turn inconsitent facts into UNREACHABLE.
2684 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2685 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2686 type inconsistent querries; return UNREACHABLE instead.
2688 2014-02-03 Richard Henderson <rth@twiddle.net>
2690 PR tree-opt/59924
2691 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2692 already processed this node.
2693 (normalize_one_pred_1): Pass along mark_set.
2694 (normalize_one_pred): Create and destroy a pointer_set_t.
2695 (normalize_one_pred_chain): Likewise.
2697 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2699 PR gcov-profile/58602
2700 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2702 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2704 PR ipa/59831
2705 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2706 -fno-devirtualize; try to devirtualize by the knowledge of
2707 virtual table pointer given by aggregate propagation.
2708 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2709 (ipa_print_node_jump_functions): Dump also offset that
2710 is relevant for polymorphic calls.
2711 (determine_known_aggregate_parts): Add arg_type parameter; use it
2712 instead of determining the type from pointer type.
2713 (ipa_compute_jump_functions_for_edge): Update call of
2714 determine_known_aggregate_parts.
2715 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2716 (gimple_get_virt_method_for_binfo): ... here; simplify using
2717 vtable_pointer_value_to_vtable.
2718 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2719 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2720 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2721 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2722 POINTER_PLUS_EXPR.
2723 (vtable_pointer_value_to_binfo): ... here.
2724 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2726 2014-02-03 Teresa Johnson <tejohnson@google.com>
2728 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2729 redef of outer loop index variable.
2731 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2733 PR c++/53017
2734 PR c++/59211
2735 * doc/extend.texi (Function Attributes): Typo.
2737 2014-02-03 Cong Hou <congh@google.com>
2739 PR tree-optimization/60000
2740 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2741 if the vectorized statement is a store. A store statement can only
2742 appear at the end of pattern statements.
2744 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2746 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2747 (ix86_option_override_internal): Default long double to 64-bit for
2748 32-bit Bionic and to 128-bit for 64-bit Bionic.
2750 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2751 TARGET_LONG_DOUBLE_128 is true.
2752 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2754 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2755 (mlong-double-64): Negate -mlong-double-128.
2756 (mlong-double-128): New option.
2758 * config/i386/i386-c.c (ix86_target_macros): Define
2759 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2761 * doc/invoke.texi: Document -mlong-double-128.
2763 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2765 PR rtl-optimization/60024
2766 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2768 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2770 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2772 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2774 PR rtl-optimization/57662
2775 * sel-sched.c (code_motion_path_driver): Do not mark already not
2776 existing blocks in the visiting bitmap.
2778 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2780 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2781 on the insn being emitted.
2783 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2784 Will Deacon <will.deacon@arm.com>
2786 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2788 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2790 * config/arm/arm-tables.opt: Regenerate.
2792 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2794 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2795 for vector types other than V16QImode.
2796 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2797 define_expand, and call altivec_expand_vec_perm_le when producing
2798 code with little endian element order.
2799 (*altivec_vperm_<mode>_internal): New insn having previous
2800 behavior of altivec_vperm_<mode>.
2801 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2802 altivec_expand_vec_perm_le when producing code with little endian
2803 element order.
2804 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2805 behavior of altivec_vperm_<mode>_uns.
2807 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2809 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2810 (altivec_vsumsws): Add handling for -maltivec=be with a little
2811 endian target.
2812 (altivec_vsumsws_direct): New.
2813 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2814 gen_altivec_vsumsws.
2816 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2818 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2819 vtable_pointer_value_to_binfo): New functions.
2820 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2821 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2823 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2825 * config/nios2/nios2.md (load_got_register): Initialize GOT
2826 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2827 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2829 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2831 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2832 preserverd by passthrough, do not propagate the type.
2834 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2836 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2837 (mips_atomic_assign_expand_fenv): New function.
2838 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2840 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2842 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2843 (__builtin_mips_set_fcsr): Likewise.
2844 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2845 MIPS_USI_FTYPE_VOID.
2846 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2847 (mips16_expand_set_fcsr): Likewise.
2848 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2849 (mips16_set_fcsr_stub): Likewise.
2850 (mips16_get_fcsr_one_only_stub): New class.
2851 (mips16_set_fcsr_one_only_stub): Likewise.
2852 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2853 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2854 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2855 (hard_float): New availability predicate.
2856 (mips_builtins): Add get_fcsr and set_fcsr.
2857 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2858 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2859 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2860 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2861 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2862 patterns.
2864 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2866 * config/mips/mips.c (mips_one_only_stub): New class.
2867 (mips_need_mips16_rdhwr_p): Replace with...
2868 (mips16_rdhwr_stub): ...this new variable.
2869 (mips16_stub_call_address): New function.
2870 (mips16_rdhwr_one_only_stub): New class.
2871 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2872 (mips_output_mips16_rdhwr): Delete.
2873 (mips_finish_stub): New function.
2874 (mips_code_end): Use it to handle rdhwr stubs.
2876 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2878 PR target/60017
2879 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2880 when calculating size of integer atomic types.
2882 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2884 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2886 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2888 PR tree-optimization/60003
2889 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2890 * profile.c (branch_prob): Use gimple_call_builtin_p
2891 to check for BUILT_IN_SETJMP_RECEIVER.
2892 * tree-inline.c (copy_bb): Call notice_special_calls.
2894 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2896 PR bootstrap/59985
2897 * lra-constraints.c (process_alt_operands): Update reload_sum only
2898 on the first pass.
2900 2014-01-31 Richard Henderson <rth@redhat.com>
2902 PR middle-end/60004
2903 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2904 until after else_eh is processed.
2906 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2908 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2909 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2910 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2911 in smmintrin.h, remove them.
2912 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2913 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2914 * config/i386/i386.md (ROUND_SAE): Fix value.
2915 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2916 (const48_operand): New.
2917 * config/i386/subst.md (round), (round_expand): Use
2918 const_4_or_8_to_11_operand.
2919 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2921 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2923 * config/i386/constraints.md (Yk): Swap meaning with k.
2924 * config/i386/i386.md (movhi_internal): Change Yk to k.
2925 (movqi_internal): Ditto.
2926 (*k<logic><mode>): Ditto.
2927 (*andhi_1): Ditto.
2928 (*andqi_1): Ditto.
2929 (kandn<mode>): Ditto.
2930 (*<code>hi_1): Ditto.
2931 (*<code>qi_1): Ditto.
2932 (kxnor<mode>): Ditto.
2933 (kortestzhi): Ditto.
2934 (kortestchi): Ditto.
2935 (kunpckhi): Ditto.
2936 (*one_cmplhi2_1): Ditto.
2937 (*one_cmplqi2_1): Ditto.
2938 * config/i386/sse.md (): Change k to Yk.
2939 (avx512f_load<mode>_mask): Ditto.
2940 (avx512f_blendm<mode>): Ditto.
2941 (avx512f_store<mode>_mask): Ditto.
2942 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2943 (avx512f_storedqu<mode>_mask): Ditto.
2944 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2945 Ditto.
2946 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2947 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2948 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2949 (avx512f_maskcmp<mode>3): Ditto.
2950 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2951 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2952 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2953 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2954 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2955 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2956 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2957 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2958 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2959 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2960 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2961 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2962 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2963 (vec_extract_lo_<mode>_maskm): Ditto.
2964 (vec_extract_hi_<mode>_maskm): Ditto.
2965 (avx512f_vternlog<mode>_mask): Ditto.
2966 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2967 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2968 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2969 (avx512f_<code>v8div16qi2_mask): Ditto.
2970 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2971 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2972 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2973 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2974 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2975 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2976 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2977 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2978 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2979 (avx512cd_maskb_vec_dupv8di): Ditto.
2980 (avx512cd_maskw_vec_dupv16si): Ditto.
2981 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2982 (avx512f_vpermi2var<mode>3_mask): Ditto.
2983 (avx512f_vpermi2var<mode>3_mask): Ditto.
2984 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2985 (*avx512f_gathersi<mode>): Ditto.
2986 (*avx512f_gathersi<mode>_2): Ditto.
2987 (*avx512f_gatherdi<mode>): Ditto.
2988 (*avx512f_gatherdi<mode>_2): Ditto.
2989 (*avx512f_scattersi<mode>): Ditto.
2990 (*avx512f_scatterdi<mode>): Ditto.
2991 (avx512f_compress<mode>_mask): Ditto.
2992 (avx512f_compressstore<mode>_mask): Ditto.
2993 (avx512f_expand<mode>_mask): Ditto.
2994 * config/i386/subst.md (mask): Change k to Yk.
2995 (mask_scalar_merge): Ditto.
2996 (sd): Ditto.
2998 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3000 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3002 2014-01-31 Richard Biener <rguenther@suse.de>
3004 PR middle-end/59990
3005 * builtins.c (fold_builtin_memory_op): Make sure to not
3006 use a floating-point mode or a boolean or enumeral type for
3007 the copy operation.
3009 2014-01-30 DJ Delorie <dj@redhat.com>
3011 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3012 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3013 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3014 whenever main() has an epilogue.
3016 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3018 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3019 unused variable "field".
3020 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3021 (vsx_mergeh_<mode>): Likewise.
3022 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3023 (altivec_vmrghh): Likewise.
3024 (altivec_vmrghw): Likewise.
3025 (altivec_vmrglb): Likewise.
3026 (altivec_vmrglh): Likewise.
3027 (altivec_vmrglw): Likewise.
3028 (altivec_vspltb): Add missing uses.
3029 (altivec_vsplth): Likewise.
3030 (altivec_vspltw): Likewise.
3031 (altivec_vspltsf): Likewise.
3033 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3035 PR target/59923
3036 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3037 frame related instructions.
3039 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3041 PR rtl-optimization/59959
3042 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3043 any reload of register whose subreg is invalid.
3045 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3047 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3048 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3049 Add missing return type - void.
3051 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3053 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3054 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3055 remove element index adjustment for endian (now handled in vsx.md
3056 and altivec.md).
3057 (altivec_expand_vec_perm_const): Use
3058 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3059 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3060 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3061 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3062 define_expand and a new define_insn *altivec_vspltb_internal;
3063 adjust for -maltivec=be on a little endian target.
3064 (altivec_vspltb_direct): New.
3065 (altivec_vsplth): Divide into a define_expand and a new
3066 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3067 little endian target.
3068 (altivec_vsplth_direct): New.
3069 (altivec_vspltw): Divide into a define_expand and a new
3070 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3071 little endian target.
3072 (altivec_vspltw_direct): New.
3073 (altivec_vspltsf): Divide into a define_expand and a new
3074 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3075 a little endian target.
3077 2014-01-30 Richard Biener <rguenther@suse.de>
3079 PR tree-optimization/59993
3080 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3081 can propagate form the earlier stmt and avoid the transform
3082 when the intermediate result is needed.
3084 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3086 * README.Portability: Fix typo.
3088 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3090 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3091 comparison_operator with ordered_comparison_operator.
3093 2014-01-30 Nick Clifton <nickc@redhat.com>
3095 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3096 Rename to mn10300_store_multiple_regs.
3097 * config/mn10300/mn10300.c: Likewise.
3098 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3099 store_multiple_regs.
3100 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3101 Call mn10300_store_multiple_regs.
3103 2014-01-30 Nick Clifton <nickc@redhat.com>
3104 DJ Delorie <dj@redhat.com>
3106 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3107 %fp 2 to keep registers after it properly word-aligned.
3108 (rl78_alloc_physical_registers_umul): Handle the case where both
3109 input operands are the same.
3111 2014-01-30 Richard Biener <rguenther@suse.de>
3113 PR tree-optimization/59903
3114 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3115 check properly.
3117 2014-01-30 Jason Merrill <jason@redhat.com>
3119 PR c++/59633
3120 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3122 PR c++/59645
3123 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3125 2014-01-30 Richard Biener <rguenther@suse.de>
3127 PR tree-optimization/59951
3128 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3130 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3132 PR target/59784
3133 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3134 SFmode to DFmode case.
3136 2014-01-29 DJ Delorie <dj@redhat.com>
3138 * config/msp430/msp430.opt (-minrt): New.
3139 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3140 if -minrt given.
3141 (ENDFILE_SPEC): Likewise.
3143 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3145 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3146 (estimate_function_body_sizes): Use it.
3148 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3150 PR c++/58561
3151 * dwarf2out.c (is_cxx_auto): New.
3152 (is_base_type): Use it.
3153 (gen_type_die_with_usage): Likewise.
3155 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3157 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3158 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3159 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3160 -maltivec=be with LE targets.
3161 (vsx_mergeh_<mode>): Likewise.
3162 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3163 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3164 (altivec_vmrghb): Replace with define_expand and new
3165 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3166 (altivec_vmrghb_direct): New define_insn.
3167 (altivec_vmrghh): Replace with define_expand and new
3168 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3169 (altivec_vmrghh_direct): New define_insn.
3170 (altivec_vmrghw): Replace with define_expand and new
3171 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3172 (altivec_vmrghw_direct): New define_insn.
3173 (*altivec_vmrghsf): Adjust for endianness.
3174 (altivec_vmrglb): Replace with define_expand and new
3175 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3176 (altivec_vmrglb_direct): New define_insn.
3177 (altivec_vmrglh): Replace with define_expand and new
3178 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3179 (altivec_vmrglh_direct): New define_insn.
3180 (altivec_vmrglw): Replace with define_expand and new
3181 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3182 (altivec_vmrglw_direct): New define_insn.
3183 (*altivec_vmrglsf): Adjust for endianness.
3184 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3185 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3186 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3187 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3188 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3189 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3190 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3191 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3193 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3195 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3196 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3197 whitespace.
3199 2014-01-29 Richard Biener <rguenther@suse.de>
3201 PR tree-optimization/58742
3202 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3203 associate_pointerplus_align.
3204 (associate_pointerplus_diff): New function.
3205 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3206 and associate_pointerplus_diff.
3208 2014-01-29 Richard Biener <rguenther@suse.de>
3210 * lto-streamer.h (LTO_major_version): Bump to 3.
3211 (LTO_minor_version): Reset to 0.
3213 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3215 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3216 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3217 (arm_file_start): Generate correct asm header for armv7ve.
3218 * config/arm/bpabi.h: Add multilib support for armv7ve.
3219 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3220 and cortex-a15 to armv7ve.
3221 * config/arm/t-aprofile: Add multilib support for armv7ve.
3222 * doc/invoke.texi: Document -march=armv7ve.
3224 2014-01-29 Richard Biener <rguenther@suse.de>
3226 PR tree-optimization/58742
3227 * tree-ssa-forwprop.c (associate_plusminus): Return true
3228 if we changed sth, defer EH cleanup to ...
3229 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3230 (simplify_mult): New function.
3232 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3234 PR middle-end/59917
3235 PR tree-optimization/59920
3236 * tree.c (build_common_builtin_nodes): Remove
3237 __builtin_setjmp_dispatcher initialization.
3238 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3239 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3240 instead of gsi_after_labels + manually skipping debug stmts.
3241 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3242 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3243 * tree-inline.c (copy_edges_for_bb): Remove
3244 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3245 argument. Ignore computed_goto_p stmts. Don't call
3246 make_abnormal_goto_edges. If a call might need abnormal edges
3247 for non-local gotos, see if it already has an edge to
3248 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3249 with true argument, don't do anything then, otherwise add
3250 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3251 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3252 caller.
3253 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3254 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3255 (lower_stmt): Don't set data->calls_builtin_setjmp.
3256 (lower_builtin_setjmp): Adjust comment.
3257 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3258 * tree-cfg.c (found_computed_goto): Remove.
3259 (factor_computed_gotos): Remove.
3260 (make_goto_expr_edges): Return bool, true for computed gotos.
3261 Don't call make_abnormal_goto_edges.
3262 (build_gimple_cfg): Don't set found_computed_goto, don't call
3263 factor_computed_gotos.
3264 (computed_goto_p): No longer static.
3265 (make_blocks): Don't set found_computed_goto.
3266 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3267 (make_edges): If make_goto_expr_edges returns true, push bb
3268 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3269 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3270 vector. Record mapping between bbs and OpenMP regions if there
3271 are any, adjust make_gimple_omp_edges caller. Call
3272 handle_abnormal_edges.
3273 (make_abnormal_goto_edges): Remove.
3274 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3275 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3276 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3277 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3278 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3279 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3280 filling *region also set *region_idx to (*region)->entry->index.
3282 PR other/58712
3283 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3284 For REGs set ORIGINAL_REGNO.
3286 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3288 * doc/md.texi: Mention that a target shouldn't implement
3289 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3290 than hi/lo pair.
3292 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3294 PR tree-optimization/59594
3295 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3296 a copy of the datarefs vector rather than the vector itself.
3298 2014-01-28 Jason Merrill <jason@redhat.com>
3300 PR c++/53756
3301 * dwarf2out.c (auto_die): New static.
3302 (gen_type_die_with_usage): Handle C++1y 'auto'.
3303 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3304 on definition.
3306 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3308 PR target/59672
3309 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3310 (SPEC_X32): Likewise.
3311 (SPEC_64): Likewise.
3312 * config/i386/i386.c (ix86_option_override_internal): Turn off
3313 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3314 for TARGET_16BIT.
3315 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3316 * config/i386/i386.h (TARGET_16BIT): New macro.
3317 (TARGET_16BIT_P): Likewise.
3318 * config/i386/i386.opt: Add m16.
3319 * doc/invoke.texi: Document -m16.
3321 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3323 PR preprocessor/59935
3324 * input.c (location_get_source_line): Bail out on when line number
3325 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3327 2014-01-28 Richard Biener <rguenther@suse.de>
3329 PR tree-optimization/58742
3330 * tree-ssa-forwprop.c (associate_plusminus): Handle
3331 pointer subtraction of the form (T)(P + A) - (T)P.
3333 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3335 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3336 at const_int_cost.
3338 2014-01-28 Richard Biener <rguenther@suse.de>
3340 Revert
3341 2014-01-28 Richard Biener <rguenther@suse.de>
3343 PR rtl-optimization/45364
3344 PR rtl-optimization/59890
3345 * var-tracking.c (local_get_addr_clear_given_value): Handle
3346 already cleared slot.
3347 (val_reset): Handle not allocated local_get_addr_cache.
3348 (vt_find_locations): Use post-order on the inverted CFG.
3350 2014-01-28 Richard Biener <rguenther@suse.de>
3352 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3354 2014-01-28 Richard Biener <rguenther@suse.de>
3356 PR rtl-optimization/45364
3357 PR rtl-optimization/59890
3358 * var-tracking.c (local_get_addr_clear_given_value): Handle
3359 already cleared slot.
3360 (val_reset): Handle not allocated local_get_addr_cache.
3361 (vt_find_locations): Use post-order on the inverted CFG.
3363 2014-01-28 Alan Modra <amodra@gmail.com>
3365 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3366 * configure.ac <recursive call for build != host>: Define
3367 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3368 and LD_FOR_BUILD too.
3369 * configure: Regenerate.
3371 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3373 * config/i386/i386.c (get_builtin_code_for_version): Separate
3374 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3375 Broadwell from Haswell.
3377 2014-01-27 Steve Ellcey <sellcey@mips.com>
3379 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3380 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3381 * config/mips/mips.c (mips_option_override): Change setting
3382 of TARGET_DSP.
3383 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3384 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3385 Change from Mask to Var.
3387 2014-01-27 Jeff Law <law@redhat.com>
3389 * ipa-inline.c (inline_small_functions): Fix typo.
3391 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3393 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3394 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3395 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3396 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3397 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3398 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3399 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3400 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3401 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3402 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3403 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3404 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3405 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3406 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3407 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3408 (_mm512_storeu_epi64): Ditto.
3409 (_mm512_cmpge_epi32_mask): Ditto.
3410 (_mm512_cmpge_epu32_mask): Ditto.
3411 (_mm512_cmpge_epi64_mask): Ditto.
3412 (_mm512_cmpge_epu64_mask): Ditto.
3413 (_mm512_cmple_epi32_mask): Ditto.
3414 (_mm512_cmple_epu32_mask): Ditto.
3415 (_mm512_cmple_epi64_mask): Ditto.
3416 (_mm512_cmple_epu64_mask): Ditto.
3417 (_mm512_cmplt_epi32_mask): Ditto.
3418 (_mm512_cmplt_epu32_mask): Ditto.
3419 (_mm512_cmplt_epi64_mask): Ditto.
3420 (_mm512_cmplt_epu64_mask): Ditto.
3421 (_mm512_cmpneq_epi32_mask): Ditto.
3422 (_mm512_cmpneq_epu32_mask): Ditto.
3423 (_mm512_cmpneq_epi64_mask): Ditto.
3424 (_mm512_cmpneq_epu64_mask): Ditto.
3425 (_mm512_expand_pd): Ditto.
3426 (_mm512_expand_ps): Ditto.
3427 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3428 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3429 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3430 * config/i386/i386.c (ix86_builtins): Add
3431 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3432 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3433 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3434 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3435 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3436 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3437 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3438 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3439 IX86_BUILTIN_PMOVUSQW512_MEM.
3440 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3441 __builtin_ia32_pmovsqd512mem_mask,
3442 __builtin_ia32_pmovqd512mem_mask,
3443 __builtin_ia32_pmovusqw512mem_mask,
3444 __builtin_ia32_pmovsqw512mem_mask,
3445 __builtin_ia32_pmovqw512mem_mask,
3446 __builtin_ia32_pmovusdw512mem_mask,
3447 __builtin_ia32_pmovsdw512mem_mask,
3448 __builtin_ia32_pmovdw512mem_mask,
3449 __builtin_ia32_pmovqb512mem_mask,
3450 __builtin_ia32_pmovusqb512mem_mask,
3451 __builtin_ia32_pmovsqb512mem_mask,
3452 __builtin_ia32_pmovusdb512mem_mask,
3453 __builtin_ia32_pmovsdb512mem_mask,
3454 __builtin_ia32_pmovdb512mem_mask.
3455 (bdesc_args): Add __builtin_ia32_expanddf512,
3456 __builtin_ia32_expandsf512.
3457 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3458 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3459 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3460 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3461 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3462 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3463 (avx512f_<code>v8div16qi2_mask_store): This.
3464 (avx512f_expand<mode>): New.
3466 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3468 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3469 New.
3470 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3471 (_mm512_prefetch_i32scatter_pd): Ditto.
3472 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3473 (_mm512_prefetch_i64scatter_pd): Ditto.
3474 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3475 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3476 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3477 (_mm512_prefetch_i32scatter_ps): Ditto.
3478 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3479 (_mm512_prefetch_i64scatter_ps): Ditto.
3480 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3481 * config/i386/i386-builtin-types.def: Define
3482 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3483 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3484 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3485 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3486 IX86_BUILTIN_SCATTERPFQPD.
3487 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3488 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3489 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3490 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3491 __builtin_ia32_scatterpfqps.
3492 (ix86_expand_builtin): Expand new built-ins.
3493 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3494 fix memory access data type.
3495 (*avx512pf_gatherpf<mode>_mask): Ditto.
3496 (*avx512pf_gatherpf<mode>): Ditto.
3497 (avx512pf_scatterpf<mode>): Ditto.
3498 (*avx512pf_scatterpf<mode>_mask): Ditto.
3499 (*avx512pf_scatterpf<mode>): Ditto.
3500 (GATHER_SCATTER_SF_MEM_MODE): New.
3501 (avx512pf_gatherpf<mode>df): Ditto.
3502 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3503 (*avx512pf_scatterpf<mode>df): Ditto.
3505 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3507 PR bootstrap/59934
3508 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3509 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3510 reached.
3512 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3514 * common/config/arm/arm-common.c
3515 (arm_rewrite_mcpu): Handle multiple names.
3516 * config/arm/arm.h
3517 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3519 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3521 * gimple-builder.h (create_gimple_tmp): Delete.
3523 2014-01-27 Christian Bruel <christian.bruel@st.com>
3525 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3526 words comparisons.
3528 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3530 * config/pa/pa.md (call): Generate indirect long calls to non-local
3531 functions when outputing 32-bit code.
3532 (call_value): Likewise except for special call to buggy powf function.
3534 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3535 portable runtime and PIC indirect calls.
3536 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3537 and PIC call sequences. Use ldo instead of blr to set return register
3538 in PIC call sequence.
3540 2014-01-25 Walter Lee <walt@tilera.com>
3542 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3543 avoid clobbering a live register.
3545 2014-01-25 Walter Lee <walt@tilera.com>
3547 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3548 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3549 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3550 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3552 2014-01-25 Walter Lee <walt@tilera.com>
3554 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3555 arguments on even registers.
3556 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3557 STACK_BOUNDARY.
3558 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3559 (BIGGEST_ALIGNMENT): Ditto.
3560 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3562 2014-01-25 Walter Lee <walt@tilera.com>
3564 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3565 insns before bundling.
3566 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3568 2014-01-25 Walter Lee <walt@tilera.com>
3570 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3571 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3572 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3574 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3576 * config/mips/constraints.md (kl): Delete.
3577 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3578 define expands, using...
3579 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3580 instructions for MIPS16.
3581 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3582 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3584 2014-01-25 Walter Lee <walt@tilera.com>
3586 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3587 (clzdi2): Ditto.
3588 (ffsdi2): Ditto.
3590 2014-01-25 Walter Lee <walt@tilera.com>
3592 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3593 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3595 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3597 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3598 Handle XOR.
3600 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3602 * print-rtl.c (in_call_function_usage): New var.
3603 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3604 EXPR_LIST mode as mode and not as reg note name.
3606 PR middle-end/59561
3607 * cfgloopmanip.c (copy_loop_info): If
3608 loop->warned_aggressive_loop_optimizations, make sure
3609 the flag is set in target loop too.
3611 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3613 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3614 flag_cilkplus.
3615 * builtins.def: Likewise.
3616 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3617 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3618 * ira.c (ira_setup_eliminable_regset): Likewise.
3619 * omp-low.c (gate_expand_omp): Likewise.
3620 (execute_lower_omp): Likewise.
3621 (diagnose_sb_0): Likewise.
3622 (gate_diagnose_omp_blocks): Likewise.
3623 (simd_clone_clauses_extract): Likewise.
3624 (gate): Likewise.
3626 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3628 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3629 correction for little endian...
3630 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3631 here.
3633 2014-01-24 Jeff Law <law@redhat.com>
3635 PR tree-optimization/59919
3636 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3637 for non-returning calls.
3639 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3641 * common/config/aarch64/aarch64-common.c
3642 (aarch64_rewrite_mcpu): Handle multiple names.
3643 * config/aarch64/aarch64.h
3644 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3646 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3648 * input.c (add_file_to_cache_tab): Handle the case where fopen
3649 returns NULL.
3651 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3653 PR target/59929
3654 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3655 from push operand if code of push isn't PRE_DEC.
3657 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3659 PR target/59909
3660 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3661 -mquad-memory-atomic. Update -mquad-memory documentation to say
3662 it is only used for non-atomic loads/stores.
3664 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3665 -mquad-memory or -mquad-memory-atomic switches.
3667 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3668 -mquad-memory-atomic to ISA 2.07 support.
3670 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3671 to separate support of normal quad word memory operations (ldq, stq)
3672 from the atomic quad word memory operations.
3674 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3675 support to separate non-atomic quad word operations from atomic
3676 quad word operations. Disable non-atomic quad word operations in
3677 little endian mode so that we don't have to swap words after the
3678 load and before the store.
3679 (quad_load_store_p): Add comment about atomic quad word support.
3680 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3681 options printed with -mdebug=reg.
3683 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3684 -mquad-memory-atomic as the test for whether we have quad word
3685 atomic instructions.
3686 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3687 or -mp8-vector are used, allow byte/half-word atomic operations.
3689 * config/rs6000/sync.md (load_lockedti): Insure that the address
3690 is a proper indexed or indirect address for the lqarx instruction.
3691 On little endian systems, swap the hi/lo registers after the lqarx
3692 instruction.
3693 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3694 insure the address is valid for the lqarx instruction.
3695 (store_conditionalti): Insure that the address is a proper indexed
3696 or indirect address for the stqcrx. instruction. On little endian
3697 systems, swap the hi/lo registers before doing the stqcrx.
3698 instruction.
3699 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3700 insure the address is valid for the stqcrx. instruction.
3702 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3703 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3704 type of quad memory support is available.
3706 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3708 PR regression/59915
3709 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3710 there is a danger of looping.
3712 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3715 force flag_ira_loop_pressure if set via command line.
3717 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3719 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3720 (ashr_simd): New builtin handling DI mode.
3721 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3722 (aarch64_sshr_simddi): New match pattern.
3723 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3724 (vshrd_n_s64): Likewise.
3725 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3727 2014-01-23 Nick Clifton <nickc@redhat.com>
3729 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3730 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3731 favour of mcu specific scripts.
3732 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3733 430x multilibs.
3735 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3736 Alex Velenko <Alex.Velenko@arm.com>
3738 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3739 (vaddv_s16): Likewise.
3740 (vaddv_s32): Likewise.
3741 (vaddv_u8): Likewise.
3742 (vaddv_u16): Likewise.
3743 (vaddv_u32): Likewise.
3744 (vaddvq_s8): Likewise.
3745 (vaddvq_s16): Likewise.
3746 (vaddvq_s32): Likewise.
3747 (vaddvq_s64): Likewise.
3748 (vaddvq_u8): Likewise.
3749 (vaddvq_u16): Likewise.
3750 (vaddvq_u32): Likewise.
3751 (vaddvq_u64): Likewise.
3752 (vaddv_f32): Likewise.
3753 (vaddvq_f32): Likewise.
3754 (vaddvq_f64): Likewise.
3755 (vmaxv_f32): Likewise.
3756 (vmaxv_s8): Likewise.
3757 (vmaxv_s16): Likewise.
3758 (vmaxv_s32): Likewise.
3759 (vmaxv_u8): Likewise.
3760 (vmaxv_u16): Likewise.
3761 (vmaxv_u32): Likewise.
3762 (vmaxvq_f32): Likewise.
3763 (vmaxvq_f64): Likewise.
3764 (vmaxvq_s8): Likewise.
3765 (vmaxvq_s16): Likewise.
3766 (vmaxvq_s32): Likewise.
3767 (vmaxvq_u8): Likewise.
3768 (vmaxvq_u16): Likewise.
3769 (vmaxvq_u32): Likewise.
3770 (vmaxnmv_f32): Likewise.
3771 (vmaxnmvq_f32): Likewise.
3772 (vmaxnmvq_f64): Likewise.
3773 (vminv_f32): Likewise.
3774 (vminv_s8): Likewise.
3775 (vminv_s16): Likewise.
3776 (vminv_s32): Likewise.
3777 (vminv_u8): Likewise.
3778 (vminv_u16): Likewise.
3779 (vminv_u32): Likewise.
3780 (vminvq_f32): Likewise.
3781 (vminvq_f64): Likewise.
3782 (vminvq_s8): Likewise.
3783 (vminvq_s16): Likewise.
3784 (vminvq_s32): Likewise.
3785 (vminvq_u8): Likewise.
3786 (vminvq_u16): Likewise.
3787 (vminvq_u32): Likewise.
3788 (vminnmv_f32): Likewise.
3789 (vminnmvq_f32): Likewise.
3790 (vminnmvq_f64): Likewise.
3792 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3794 * config/aarch64/aarch64-simd.md
3795 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3796 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3797 (*aarch64_mul3_elt<mode>): Likewise.
3798 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3799 (*aarch64_mul3_elt_to_64v2df): Likewise.
3800 (*aarch64_mla_elt<mode>): Likewise.
3801 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3802 (*aarch64_mls_elt<mode>): Likewise.
3803 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3804 (*aarch64_fma4_elt<mode>): Likewise.
3805 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3806 (*aarch64_fma4_elt_to_64v2df): Likewise.
3807 (*aarch64_fnma4_elt<mode>): Likewise.
3808 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3809 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3810 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3811 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3812 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3813 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3814 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3815 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3816 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3818 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3820 * config/aarch64/aarch64-simd.md
3821 (aarch64_be_checked_get_lane<mode>): New define_expand.
3822 * config/aarch64/aarch64-simd-builtins.def
3823 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3824 New builtin definition.
3825 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3826 Use new safe be builtin.
3828 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3830 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3831 New define_insn.
3832 (aarch64_be_st1<mode>): Likewise.
3833 (aarch_ld1<VALL:mode>): Define_expand modified.
3834 (aarch_st1<VALL:mode>): Likewise.
3835 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3836 (UNSPEC_ST1): Likewise.
3838 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3840 * config/microblaze/microblaze.md: Add trap insn and attribute
3842 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3844 PR preprocessor/58580
3845 * input.h (location_get_source_line): Take an additional line_size
3846 parameter.
3847 (void diagnostics_file_cache_fini): Declare new function.
3848 * input.c (struct fcache): New type.
3849 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3850 New static constants.
3851 (diagnostic_file_cache_init, total_lines_num)
3852 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3853 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3854 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3855 (get_next_line, read_next_line, goto_next_line, read_line_num):
3856 New static function definitions.
3857 (diagnostic_file_cache_fini): New function.
3858 (location_get_source_line): Take an additional output line_len
3859 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3860 read_line_num.
3861 * diagnostic.c (diagnostic_finish): Call
3862 diagnostic_file_cache_fini.
3863 (adjust_line): Take an additional input parameter for the length
3864 of the line, rather than calculating it with strlen.
3865 (diagnostic_show_locus): Adjust the use of
3866 location_get_source_line and adjust_line with respect to their new
3867 signature. While displaying a line now, do not stop at the first
3868 null byte. Rather, display the zero byte as a space and keep
3869 going until we reach the size of the line.
3870 * Makefile.in: Add vec.o to OBJS-libcommon
3872 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3873 Ilya Tocar <ilya.tocar@intel.com>
3875 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3876 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3877 (__builtin_ia32_kmov16): Ditto.
3878 * config/i386/i386.md (UNSPEC_KMOV): New.
3879 (kmovw): Ditto.
3881 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3883 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3884 (_mm512_storeu_si512): Ditto.
3886 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3888 PR target/52125
3889 * rtl.h (get_referenced_operands): Declare.
3890 * recog.c (get_referenced_operands): New function.
3891 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3892 operands have been referenced when recording LO_SUM references.
3894 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3896 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3898 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3900 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3901 Enable for generic and recent AMD targets.
3903 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3905 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3906 ARG_SIZE note when adjustment was eliminated.
3908 2014-01-22 Jeff Law <law@redhat.com>
3910 PR tree-optimization/59597
3911 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3912 in file. Accept new argument REGISTERING and use it to modify
3913 dump output appropriately.
3914 (register_jump_thread): Corresponding changes.
3915 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3916 thread paths involving joiner blocks. Add code to dump cancelled
3917 jump threading paths.
3919 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3921 PR rtl-optimization/59477
3922 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3923 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3925 2014-01-22 Tom Tromey <tromey@redhat.com>
3927 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3928 PARAMS.
3929 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3931 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3933 PR rtl-optimization/59896
3934 * lra-constraints.c (process_alt_operands): Check unused note for
3935 matched operands of insn with no output reloads.
3937 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3939 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3940 (mips_move_from_gpr_cost): Likewise.
3942 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3944 PR rtl-optimization/59858
3945 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3946 ira_class_hard_regs_num.
3947 (process_alt_operands): Increase reject for dying matched operand.
3949 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3951 PR target/59003
3952 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3953 smaller than size, perform several stores or loads and stores
3954 at dst + count - size to store or copy all of size bytes, rather
3955 than just last modesize bytes.
3957 2014-01-20 DJ Delorie <dj@redhat.com>
3959 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3960 that CLOBBERs are REGs before propogating their values.
3962 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3964 PR middle-end/59789
3965 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3966 (cgraph_inline_failed_type): New function.
3967 * cgraph.h (DEFCIFCODE): Add type.
3968 (cgraph_inline_failed_type_t): New enum.
3969 (cgraph_inline_failed_type): New prototype.
3970 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3971 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3972 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3973 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3974 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3975 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3976 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3977 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3978 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3979 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3980 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3981 OPTIMIZATION_MISMATCH.
3982 * tree-inline.c (expand_call_inline): Emit errors during
3983 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3985 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3987 PR target/59685
3988 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3989 mode attribute in insn output.
3991 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3993 * output.h (output_constant): Delete.
3994 * varasm.c (output_constant): Make private.
3996 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3998 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4000 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4002 PR middle-end/59860
4003 * tree.h (fold_builtin_strcat): New prototype.
4004 * builtins.c (fold_builtin_strcat): No longer static. Add len
4005 argument, if non-NULL, don't call c_strlen. Optimize
4006 directly into __builtin_memcpy instead of __builtin_strcpy.
4007 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4008 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4010 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4012 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4013 for SImode_address_operand operands, having only a REG argument.
4015 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4017 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4018 loader name using mbig-endian.
4019 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4021 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4023 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4024 (-mtune): Likewise.
4025 (-mcpu): Likewise.
4027 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4029 * config/aarch64/aarch64-protos.h
4030 (aarch64_cannot_change_mode_class_ptr): Declare.
4031 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4032 aarch64_cannot_change_mode_class_ptr): New.
4033 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4034 backend hook aarch64_cannot_change_mode_class.
4036 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4038 * common/config/aarch64/aarch64-common.c
4039 (aarch64_handle_option): Don't handle any option order logic here.
4040 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4041 selected_cpu, warn on architecture version mismatch.
4042 (aarch64_override_options): Fix parsing order for option strings.
4044 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4045 Iain Sandoe <iain@codesourcery.com>
4047 PR bootstrap/59496
4048 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4049 warning. Amend comment to reflect current functionality.
4051 2014-01-20 Richard Biener <rguenther@suse.de>
4053 PR middle-end/59860
4054 * builtins.c (fold_builtin_strcat): Remove case better handled
4055 by tree-ssa-strlen.c.
4057 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4059 * config/aarch64/aarch64.opt
4060 (mcpu, march, mtune): Make case-insensitive.
4062 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4064 PR target/59880
4065 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4066 if operands[1] is a REG or ZERO_EXTEND of a REG.
4068 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4070 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4072 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4074 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4075 long non-pic millicode calls.
4077 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4079 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4081 2014-01-19 Kito Cheng <kito@0xlab.org>
4083 * builtins.c (expand_movstr): Check movstr expand done or fail.
4085 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4086 H.J. Lu <hongjiu.lu@intel.com>
4088 PR target/59379
4089 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4090 to DImode for zero-extended addresses.
4092 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4094 PR rtl-optimization/57763
4095 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4096 on the new indirect jump_insn and increment LABEL_NUSES (label).
4098 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4100 PR bootstrap/59580
4101 PR bootstrap/59583
4102 * config.gcc (x86_archs): New variable.
4103 (x86_64_archs): Likewise.
4104 (x86_cpus): Likewise.
4105 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4106 --with-arch/--with-cpu= options.
4107 Support --with-arch=/--with-cpu={nehalem,westmere,
4108 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4110 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4112 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4113 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4115 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4117 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4119 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4121 PR target/58944
4122 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4123 clear cpp_get_options (parse_in)->warn_unused_macros for
4124 ix86_target_macros_internal with cpp_define.
4126 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4128 * jump.c (delete_related_insns): Keep (use (insn))s.
4129 * reorg.c (redundant_insn): Check for barriers too.
4131 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4133 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4135 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4137 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4138 call to $$dyncall when TARGET_LONG_CALLS is true.
4140 2014-01-17 Jeff Law <law@redhat.com>
4142 * ree.c (combine_set_extension): Temporarily disable test for
4143 changing number of hard registers.
4145 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4147 PR middle-end/58125
4148 * ipa-inline-analysis.c (inline_free_summary):
4149 Do not free summary of aliases.
4151 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4153 PR middle-end/59706
4154 * gimplify.c (gimplify_expr): Use create_tmp_var
4155 instead of create_tmp_var_raw. If cond doesn't have
4156 integral type, don't add the IFN_ANNOTATE builtin at all.
4158 2014-01-17 Martin Jambor <mjambor@suse.cz>
4160 PR ipa/59736
4161 * ipa-cp.c (prev_edge_clone): New variable.
4162 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4163 Also resize prev_edge_clone vector.
4164 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4165 (ipcp_edge_removal_hook): New function.
4166 (ipcp_driver): Register ipcp_edge_removal_hook.
4168 2014-01-17 Andrew Pinski <apinski@cavium.com>
4169 Steve Ellcey <sellcey@mips.com>
4171 PR target/59462
4172 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4173 of operator mode.
4175 2014-01-17 Jeff Law <law@redhat.com>
4177 PR middle-end/57904
4178 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4179 so that pass_ccp runs first.
4181 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4183 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4184 (ix86_adjust_cost): Use !TARGET_XXX.
4185 (do_reorder_for_imul): Likewise.
4186 (swap_top_of_ready_list): Likewise.
4187 (ix86_sched_reorder): Likewise.
4189 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4191 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4192 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4193 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4194 (intel_memset): New. Duplicate slm_memset.
4195 (intel_cost): New. Duplicate slm_cost.
4196 (m_INTEL): New macro.
4197 (processor_target_table): Add "intel".
4198 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4199 with PROCESSOR_INTEL for "intel".
4200 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4201 PROCESSOR_SILVERMONT.
4202 (ix86_issue_rate): Likewise.
4203 (ix86_adjust_cost): Likewise.
4204 (ia32_multipass_dfa_lookahead): Likewise.
4205 (swap_top_of_ready_list): Likewise.
4206 (ix86_sched_reorder): Likewise.
4207 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4208 instead of TARGET_OPT_AGU.
4209 * config/i386/i386.h (TARGET_INTEL): New.
4210 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4211 (processor_type): Add PROCESSOR_INTEL.
4212 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4213 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4215 2014-01-17 Marek Polacek <polacek@redhat.com>
4217 PR c/58346
4218 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4219 size is zero.
4221 2014-01-17 Richard Biener <rguenther@suse.de>
4223 PR tree-optimization/46590
4224 * opts.c (default_options_table): Add entries for
4225 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4226 all enabled at -O1 but not for -Og.
4227 * common.opt (fbranch-count-reg): Remove Init(1).
4228 (fmove-loop-invariants): Likewise.
4229 (ftree-pta): Likewise.
4231 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4233 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4234 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4235 decls to at least the GCC 4.8 used alignments.
4237 PR fortran/59440
4238 * tree-nested.c (convert_nonlocal_reference_stmt,
4239 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4240 of GIMPLE_BIND stmts, adjust associated decls.
4242 2014-01-17 Richard Biener <rguenther@suse.de>
4244 PR tree-optimization/46590
4245 * vec.h (vec<>::bseach): New member function implementing
4246 binary search according to C89 bsearch.
4247 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4248 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4249 bitmap pointer again. Make accesses_in_loop a flat array.
4250 (mem_ref_obstack): New global.
4251 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4252 (mark_ref_stored): Likewise.
4253 (ref_indep_loop_p_2): Likewise.
4254 (set_ref_stored_in_loop): New helper function.
4255 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4256 (memref_free): Adjust.
4257 (record_mem_ref_loc): Simplify.
4258 (gather_mem_refs_stmt): Adjust.
4259 (sort_locs_in_loop_postorder_cmp): New function.
4260 (analyze_memory_references): Sort accesses_in_loop after
4261 loop postorder number.
4262 (find_ref_loc_in_loop_cmp): New function.
4263 (for_all_locs_in_loop): Find relevant cluster of locs in
4264 accesses_in_loop and iterate without recursion.
4265 (execute_sm): Avoid uninit warning.
4266 (struct ref_always_accessed): Simplify.
4267 (ref_always_accessed::operator ()): Likewise.
4268 (ref_always_accessed_p): Likewise.
4269 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4270 loop postorder numbers here.
4271 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4272 numbers.
4274 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4276 PR c++/57945
4277 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4278 on decls for which assemble_alias has been called.
4280 2014-01-17 Nick Clifton <nickc@redhat.com>
4282 * config/msp430/msp430.opt: (mcpu): New option.
4283 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4284 (msp430_option_override): Parse target_cpu. If the MCU name
4285 matches a generic string, clear target_mcu.
4286 (msp430_attr): Allow numeric interrupt values up to 63.
4287 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4288 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4289 option.
4290 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4291 Add mcpu matches.
4292 * config/msp430/msp430.md (popm): Use %J rather than %I.
4293 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4294 (addhi_cy_i): Use immediate_operand for operand 2.
4295 * doc/invoke.texi: Document -mcpu option.
4297 2014-01-17 Richard Biener <rguenther@suse.de>
4299 PR rtl-optimization/38518
4300 * df.h (df_analyze_loop): Declare.
4301 * df-core.c: Include cfgloop.h.
4302 (df_analyze_1): Split out main part of df_analyze.
4303 (df_analyze): Adjust.
4304 (loop_inverted_post_order_compute): New function.
4305 (loop_post_order_compute): Likewise.
4306 (df_analyze_loop): New function avoiding whole-function
4307 postorder computes.
4308 * loop-invariant.c (find_defs): Use df_analyze_loop.
4309 (find_invariants): Adjust.
4310 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4312 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4314 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4315 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4317 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4319 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4320 traversal when removing references.
4322 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4324 PR ipa/59775
4325 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4327 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4329 PR middle-end/56791
4330 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4331 pushing a reload for an autoinc when we had previously reloaded an
4332 inner part of the address.
4334 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4336 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4337 field.
4338 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4339 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4340 when not giving up or versioning for alias only because of
4341 loop->safelen.
4342 (vect_analyze_data_ref_dependences): Set to true.
4343 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4344 is a GIMPLE_PHI.
4345 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4346 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4347 to the condition.
4349 PR middle-end/58344
4350 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4352 PR target/59839
4353 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4354 operand 0 predicate for gathers, use a new pseudo as subtarget.
4356 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4358 PR middle-end/59609
4359 * lra-constraints.c (process_alt_operands): Add printing debug info.
4360 Check absence of input/output reloads for matched operands too.
4362 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4364 PR rtl-optimization/59835
4365 * ira.c (ira_init_register_move_cost): Increase cost for
4366 impossible modes.
4368 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4370 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4372 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4374 PR target/59780
4375 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4376 non-register objects. Use gen_(high/low)part more consistently.
4377 Fix assertions.
4379 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4381 PR target/59844
4382 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4383 endian support, remove tests for WORDS_BIG_ENDIAN.
4384 (p8_mfvsrd_3_<mode>): Likewise.
4385 (reload_gpr_from_vsx<mode>): Likewise.
4386 (reload_gpr_from_vsxsf): Likewise.
4387 (p8_mfvsrd_4_disf): Likewise.
4389 2014-01-16 Richard Biener <rguenther@suse.de>
4391 PR rtl-optimization/46590
4392 * lcm.c (compute_antinout_edge): Use postorder iteration.
4393 (compute_laterin): Use inverted postorder iteration.
4395 2014-01-16 Nick Clifton <nickc@redhat.com>
4397 PR middle-end/28865
4398 * varasm.c (output_constant): Return the number of bytes actually
4399 emitted.
4400 (output_constructor_array_range): Update the field size with the
4401 number of bytes emitted by output_constant.
4402 (output_constructor_regular_field): Likewise. Also do not
4403 complain if the total number of bytes emitted is now greater
4404 than the expected fieldpos.
4405 * output.h (output_constant): Update prototype and descriptive comment.
4407 2014-01-16 Marek Polacek <polacek@redhat.com>
4409 PR middle-end/59827
4410 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4411 it is error_mark_node.
4413 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4415 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4416 VALID_AVX256_REG_OR_OI_MODE.
4418 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4420 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4421 current procedure should be profiled.
4423 2014-01-15 Andrew Pinski <apinski@cavium.com>
4425 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4426 of moving from/to the STACK_REG register class.
4428 2014-01-15 Richard Henderson <rth@redhat.com>
4430 PR debug/54694
4431 * reginfo.c (global_regs_decl): Globalize.
4432 * rtl.h (global_regs_decl): Declare.
4433 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4434 reserved via global_regs.
4436 2014-01-15 Teresa Johnson <tejohnson@google.com>
4438 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4440 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4442 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4443 and vmulosh rather than call gen_vec_widen_smult_*.
4444 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4445 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4446 (vec_widen_smult_even_v16qi): Likewise.
4447 (vec_widen_umult_even_v8hi): Likewise.
4448 (vec_widen_smult_even_v8hi): Likewise.
4449 (vec_widen_umult_odd_v16qi): Likewise.
4450 (vec_widen_smult_odd_v16qi): Likewise.
4451 (vec_widen_umult_odd_v8hi): Likewise.
4452 (vec_widen_smult_odd_v8hi): Likewise.
4453 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4454 vmuloub rather than call gen_vec_widen_umult_*.
4455 (vec_widen_umult_lo_v16qi): Likewise.
4456 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4457 vmulosb rather than call gen_vec_widen_smult_*.
4458 (vec_widen_smult_lo_v16qi): Likewise.
4459 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4460 rather than call gen_vec_widen_umult_*.
4461 (vec_widen_umult_lo_v8hi): Likewise.
4462 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4463 rather than call gen_vec_widen_smult_*.
4464 (vec_widen_smult_lo_v8hi): Likewise.
4466 2014-01-15 Jeff Law <law@redhat.com>
4468 PR tree-optimization/59747
4469 * ree.c (find_and_remove_re): Properly handle case where a second
4470 eliminated extension requires widening a copy created for elimination
4471 of a prior extension.
4472 (combine_set_extension): Ensure that the number of hard regs needed
4473 for a destination register does not change when we widen it.
4475 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4477 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4478 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4479 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4480 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4481 (avr-*-rtems*): Likewise.
4482 (bfin*-rtems*): Likewise.
4483 (moxie-*-rtems*): Likewise.
4484 (h8300-*-rtems*): Likewise.
4485 (i[34567]86-*-rtems*): Likewise.
4486 (lm32-*-rtems*): Likewise.
4487 (m32r-*-rtems*): Likewise.
4488 (m68k-*-rtems*): Likewise.
4489 (microblaze*-*-rtems*): Likewise.
4490 (mips*-*-rtems*): Likewise.
4491 (powerpc-*-rtems*): Likewise.
4492 (sh-*-rtems*): Likewise.
4493 (sparc-*-rtems*): Likewise.
4494 (sparc64-*-rtems*): Likewise.
4495 (v850-*-rtems*): Likewise.
4496 (m32c-*-rtems*): Likewise.
4498 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4500 PR rtl-optimization/59511
4501 * ira.c (ira_init_register_move_cost): Use memory costs for some
4502 cases of register move cost calculations.
4503 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4504 instead of BB frequency.
4505 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4506 * lra-assigns.c (find_hard_regno_for): Ditto.
4508 2014-01-15 Richard Biener <rguenther@suse.de>
4510 PR tree-optimization/59822
4511 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4512 (vectorizable_load): Use it to hoist defs of uses of invariant
4513 loads out of the loop.
4515 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4516 Kugan Vivekanandarajah <kuganv@linaro.org>
4518 PR target/59695
4519 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4520 truncation.
4522 2014-01-15 Richard Biener <rguenther@suse.de>
4524 PR rtl-optimization/59802
4525 * lcm.c (compute_available): Use inverted postorder to seed
4526 the initial worklist.
4528 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4530 PR target/59803
4531 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4532 ADDR_REGS for invalid symrefs in non-PIC code.
4534 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4536 PR other/58712
4537 * builtins.c (determine_block_size): Initialize *probable_max_size
4538 even if len_rtx is CONST_INT.
4540 2014-01-14 Andrew Pinski <apinski@cavium.com>
4542 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4543 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4544 (cortexa53_tunings): Likewise.
4545 (aarch64_sched_issue_rate): New function.
4546 (TARGET_SCHED_ISSUE_RATE): Define.
4548 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4550 * ira-costs.c (find_costs_and_classes): Add missed
4551 ira_init_register_move_cost_if_necessary.
4553 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4555 PR target/59787
4556 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4558 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4560 PR target/59794
4561 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4562 to indicate if type is used for function return value. Warn ABI
4563 change if the vector mode isn't available for function return value.
4564 (ix86_function_arg_advance): Pass false to type_natural_mode.
4565 (ix86_function_arg): Likewise.
4566 (ix86_gimplify_va_arg): Likewise.
4567 (function_arg_32): Don't warn ABI change.
4568 (ix86_function_value): Pass true to type_natural_mode.
4569 (ix86_return_in_memory): Likewise.
4570 (ix86_struct_value_rtx): Removed.
4571 (TARGET_STRUCT_VALUE_RTX): Likewise.
4573 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4575 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4576 converting a conditional jump into a conditional return.
4578 2014-01-14 Richard Biener <rguenther@suse.de>
4580 PR tree-optimization/58921
4581 PR tree-optimization/59006
4582 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4583 hoisting invariant stmts.
4584 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4585 invariant loads on the preheader edge if possible.
4587 2014-01-14 Joey Ye <joey.ye@arm.com>
4589 * doc/plugin.texi (Building GCC plugins): Update to C++.
4591 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4593 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4594 (_mm_rcp28_round_ss): Ditto.
4595 (_mm_rsqrt28_round_sd): Ditto.
4596 (_mm_rsqrt28_round_ss): Ditto.
4597 (_mm_rcp28_sd): Ditto.
4598 (_mm_rcp28_ss): Ditto.
4599 (_mm_rsqrt28_sd): Ditto.
4600 (_mm_rsqrt28_ss): Ditto.
4601 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4602 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4603 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4604 (IX86_BUILTIN_RCP28SD): Ditto.
4605 (IX86_BUILTIN_RCP28SS): Ditto.
4606 (IX86_BUILTIN_RSQRT28SD): Ditto.
4607 (IX86_BUILTIN_RSQRT28SS): Ditto.
4608 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4609 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4610 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4611 (ix86_expand_special_args_builtin): Expand new FTYPE.
4612 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4613 (srcp14<mode>): Make insn unary.
4614 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4615 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4616 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4617 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4618 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4619 Fix rounding: make it SAE only.
4620 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4621 Ditto.
4622 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4623 Ditto.
4624 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4625 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4626 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4627 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4628 (round_saeonly_mask_scalar_operand4): Ditto.
4629 (round_saeonly_mask_scalar_op3): Ditto.
4630 (round_saeonly_mask_scalar_op4): Ditto.
4632 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4634 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4635 Implement -maltivec=be for vec_insert and vec_extract.
4637 2014-01-10 DJ Delorie <dj@redhat.com>
4639 * config/msp430/msp430.md (call_internal): Don't allow memory
4640 references with SP as the base register.
4641 (call_value_internal): Likewise.
4642 * config/msp430/constraints.md (Yc): New. For memory references
4643 that don't use SP as a base register.
4645 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4646 "an integer without a # prefix"
4647 * config/msp430/msp430.md (epilogue_helper): Use it.
4649 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4651 PR target/59617
4652 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4653 AVX512F gather builtins.
4654 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4655 on gather decls with INTEGER_TYPE masktype.
4656 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4657 directly into the builtin rather than hoisting it before loop.
4659 PR tree-optimization/59387
4660 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4661 (scev_const_prop): If folded_casts and type has undefined overflow,
4662 use force_gimple_operand instead of force_gimple_operand_gsi and
4663 for each added stmt if it is assign with
4664 arith_code_with_undefined_signed_overflow, call
4665 rewrite_to_defined_overflow.
4666 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4667 gimple-fold.h instead.
4668 (arith_code_with_undefined_signed_overflow,
4669 rewrite_to_defined_overflow): Moved to ...
4670 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4671 rewrite_to_defined_overflow): ... here. No longer static.
4672 Include gimplify-me.h.
4673 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4674 rewrite_to_defined_overflow): New prototypes.
4676 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4678 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4680 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4682 * builtins.c (get_object_alignment_2): Minor tweak.
4683 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4685 2014-01-13 Christian Bruel <christian.bruel@st.com>
4687 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4688 optimized non constant lengths.
4690 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4692 PR libgomp/59194
4693 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4694 load as __atomic_load_N if possible.
4696 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4698 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4699 target parameter.
4700 (rs6000_expand_builtin): Adjust call.
4702 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4704 PR target/58115
4705 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4706 * config/rs6000/rs6000.c: Include target-globals.h.
4707 (rs6000_set_current_function): Instead of doing target_reinit
4708 unconditionally, use save_target_globals_default_opts and
4709 restore_target_globals.
4711 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4712 FPSCR.
4713 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4714 (rs6000_expand_builtin): Handle mffs and mtfsf.
4715 (rs6000_init_builtins): Define mffs and mtfsf.
4716 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4717 (rs6000_mffs): New pattern.
4718 (rs6000_mtfsf): New pattern.
4720 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4722 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4723 Start narrowing with START. Apply candidate-use pair
4724 and check overall cost in narrowing.
4725 (iv_ca_prune): Pass new argument.
4727 2014-01-10 Jeff Law <law@redhat.com>
4729 PR middle-end/59743
4730 * ree.c (combine_reaching_defs): Ensure the defining statement
4731 occurs before the extension when optimizing extensions with
4732 different source and destination hard registers.
4734 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4736 PR ipa/58585
4737 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4738 vtables into the type inheritance graph.
4740 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4742 PR rtl-optimization/59754
4743 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4744 modes in the REGNO != REGNO case.
4746 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4748 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4750 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4752 PR tree-optimization/59745
4753 * tree-predcom.c (tree_predictive_commoning_loop): Call
4754 free_affine_expand_cache if giving up because components is NULL.
4756 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4757 GC in payload of target_globals struct instead of allocating them on
4758 the heap and the larger structs separately using GC.
4759 * target-globals.h (struct target_globals): Make regs, hard_regs,
4760 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4761 of GTY((skip)) and change type to void *.
4762 (reset_target_globals): Cast loads from those fields to corresponding
4763 types.
4765 2014-01-10 Steve Ellcey <sellcey@mips.com>
4767 PR plugins/59335
4768 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4769 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4770 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4772 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4774 PR target/59744
4775 * aarch64-modes.def (CC_Zmode): New flags mode.
4776 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4777 represents an equality.
4778 (aarch64_get_condition_code): Handle CC_Zmode.
4779 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4781 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4783 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4784 extraction in good case.
4786 2014-01-10 Richard Biener <rguenther@suse.de>
4788 PR tree-optimization/59374
4789 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4790 checking after SLP discovery. Mark stmts not participating
4791 in any SLP instance properly.
4793 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4795 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4796 when handling a SET rtx.
4798 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4800 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4801 (cortex-a57): Likewise.
4802 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4804 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4806 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4807 non-iwmmxt builtins.
4809 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4811 PR ipa/58252
4812 PR ipa/59226
4813 * ipa-devirt.c record_target_from_binfo): Take as argument
4814 stack of binfos and lookup matching one for virtual inheritance.
4815 (possible_polymorphic_call_targets_1): Update.
4817 2014-01-10 Huacai Chen <chenhc@lemote.com>
4819 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4820 kernel strings for Loongson-2E/2F/3A.
4822 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4824 PR middle-end/59670
4825 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4826 is_gimple_call before calling gimple_call_internal_p.
4828 2014-01-09 Steve Ellcey <sellcey@mips.com>
4830 * Makefile.in (TREE_FLOW_H): Remove.
4831 (TREE_SSA_H): Add file names from tree-flow.h.
4832 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4833 * tree.h: Remove tree-flow.h reference.
4834 * hash-table.h: Remove tree-flow.h reference.
4835 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4836 reference with tree-ssa-loop.h.
4838 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4840 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4841 default element-order behavior for -maltivec.
4842 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4844 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4845 when targeting big endian, at least for now.
4846 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4848 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4850 PR middle-end/47735
4851 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4852 var satisfies use_register_for_decl, just take into account type
4853 alignment, rather than decl alignment.
4855 PR tree-optimization/59622
4856 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4857 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4858 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4859 Don't devirtualize for inplace at all. For targets.length () == 1,
4860 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4862 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4864 * config/i386/i386.md (cpu): Remove the unused btver1.
4866 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4868 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4870 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4872 PR target/58115
4873 * tree-core.h (struct target_globals): New forward declaration.
4874 (struct tree_target_option): Add globals field.
4875 * tree.h (TREE_TARGET_GLOBALS): Define.
4876 (prepare_target_option_nodes_for_pch): New prototype.
4877 * target-globals.h (struct target_globals): Define even if
4878 !SWITCHABLE_TARGET.
4879 * tree.c (prepare_target_option_node_for_pch,
4880 prepare_target_option_nodes_for_pch): New functions.
4881 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4882 * config/i386/i386.c: Include target-globals.h.
4883 (ix86_set_current_function): Instead of doing target_reinit
4884 unconditionally, use save_target_globals_default_opts and
4885 restore_target_globals.
4887 2014-01-09 Richard Biener <rguenther@suse.de>
4889 PR tree-optimization/59715
4890 * tree-cfg.h (split_critical_edges): Declare.
4891 * tree-cfg.c (split_critical_edges): Export.
4892 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4894 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4896 * cfgexpand.c (expand_stack_vars): Optionally disable
4897 asan stack protection.
4898 (expand_used_vars): Likewise.
4899 (partition_stack_vars): Likewise.
4900 * asan.c (asan_emit_stack_protection): Optionally disable
4901 after return stack usage.
4902 (instrument_derefs): Optionally disable memory access instrumentation.
4903 (instrument_builtin_call): Likewise.
4904 (instrument_strlen_call): Likewise.
4905 (asan_protect_global): Optionally disable global variables protection.
4906 * doc/invoke.texi: Added doc for new options.
4907 * params.def: Added new options.
4908 * params.h: Likewise.
4910 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4912 PR rtl-optimization/59724
4913 * ifcvt.c (cond_exec_process_if_block): Don't call
4914 flow_find_head_matching_sequence with 0 longest_match.
4915 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4916 non-active insns if !stop_after.
4917 (try_head_merge_bb): Revert 2014-01-07 changes.
4919 2014-01-08 Jeff Law <law@redhat.com>
4921 * ree.c (get_sub_rtx): New function, extracted from...
4922 (merge_def_and_ext): Here.
4923 (combine_reaching_defs): Use get_sub_rtx.
4925 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4927 * cgraph.h (varpool_variable_node): Do not choke on null node.
4929 2014-01-08 Catherine Moore <clm@codesourcery.com>
4931 * config/mips/mips.md (simple_return): Attempt to use JRC
4932 for microMIPS.
4933 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4935 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4937 PR rtl-optimization/59137
4938 * reorg.c (steal_delay_list_from_target): Call update_block for
4939 elided insns.
4940 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4942 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4944 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4945 two duplicate entries.
4947 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4949 Revert:
4950 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4952 * config/mips/mips.c (mips_truncated_op_cost): New function.
4953 (mips_rtx_costs): Adjust test for BADDU.
4954 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4956 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4958 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4959 (*baddu_si): ...this new pattern.
4961 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4963 PR ipa/59722
4964 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4966 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4968 PR middle-end/57748
4969 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4970 inner_reference_p.
4971 (expand_expr, expand_normal): Adjust.
4972 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4973 inner_reference_p. Use inner_reference_p to expand inner references.
4974 (store_expr): Adjust.
4975 * cfgexpand.c (expand_call_stmt): Adjust.
4977 2014-01-08 Rong Xu <xur@google.com>
4979 * gcov-io.c (gcov_var): Move from gcov-io.h.
4980 (gcov_position): Ditto.
4981 (gcov_is_error): Ditto.
4982 (gcov_rewrite): Ditto.
4983 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4984 only part to libgcc/libgcov.h.
4986 2014-01-08 Marek Polacek <polacek@redhat.com>
4988 PR middle-end/59669
4989 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4991 2014-01-08 Marek Polacek <polacek@redhat.com>
4993 PR sanitizer/59667
4994 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4996 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4998 PR rtl-optimization/59649
4999 * stor-layout.c (get_mode_bounds): For BImode return
5000 0 and STORE_FLAG_VALUE.
5002 2014-01-08 Richard Biener <rguenther@suse.de>
5004 PR middle-end/59630
5005 * gimple.h (is_gimple_builtin_call): Remove.
5006 (gimple_builtin_call_types_compatible_p): New.
5007 (gimple_call_builtin_p): New overload.
5008 * gimple.c (is_gimple_builtin_call): Remove.
5009 (validate_call): Rename to ...
5010 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5011 check return types.
5012 (validate_type): New static function.
5013 (gimple_call_builtin_p): New overload and adjust.
5014 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5015 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5016 (gimple_fold_stmt_to_constant_1): Likewise.
5017 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5019 2014-01-08 Richard Biener <rguenther@suse.de>
5021 PR middle-end/59471
5022 * gimplify.c (gimplify_expr): Gimplify register-register type
5023 VIEW_CONVERT_EXPRs to separate stmts.
5025 2014-01-07 Jeff Law <law@redhat.com>
5027 PR middle-end/53623
5028 * ree.c (combine_set_extension): Handle case where source
5029 and destination registers in an extension insn are different.
5030 (combine_reaching_defs): Allow source and destination registers
5031 in extension to be different under limited circumstances.
5032 (add_removable_extension): Remove restriction that the
5033 source and destination registers in the extension are the same.
5034 (find_and_remove_re): Emit a copy from the extension's
5035 destination to its source after the defining insn if
5036 the source and destination registers are different.
5038 PR middle-end/59285
5039 * ifcvt.c (merge_if_block): If we are merging a block with more than
5040 one successor with a block with no successors, remove any BARRIER
5041 after the second block.
5043 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5045 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5047 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5049 PR target/59652
5050 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5051 for 14-bit register offsets when INT14_OK_STRICT is false.
5053 2014-01-07 Roland Stigge <stigge@antcom.de>
5054 Michael Meissner <meissner@linux.vnet.ibm.com>
5056 PR 57386/target
5057 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5058 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5060 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5062 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5063 -mcpu.
5065 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5067 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5068 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5069 rtx is const0_rtx or not.
5071 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5073 PR target/58115
5074 * target-globals.c (save_target_globals): Remove this_fn_optab
5075 handling.
5076 * toplev.c: Include optabs.h.
5077 (target_reinit): Temporarily restore the global options if another
5078 set of options are in force.
5080 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5082 PR rtl-optimization/58668
5083 * cfgcleanup.c (flow_find_cross_jump): Don't count
5084 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5085 to determine what is counted.
5086 (flow_find_head_matching_sequence): Use active_insn_p to determine
5087 what is counted.
5088 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5089 counting change.
5090 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5091 determine what is counted.
5093 PR tree-optimization/59643
5094 * tree-predcom.c (split_data_refs_to_components): If one dr is
5095 read and one write, determine_offset fails and the write isn't
5096 in the bad component, just put the read into the bad component.
5098 2014-01-07 Mike Stump <mikestump@comcast.net>
5099 Jakub Jelinek <jakub@redhat.com>
5101 PR pch/59436
5102 * tree-core.h (struct tree_optimization_option): Change optabs
5103 type from unsigned char * to void *.
5104 * optabs.c (init_tree_optimization_optabs): Adjust
5105 TREE_OPTIMIZATION_OPTABS initialization.
5107 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5109 PR target/59644
5110 * config/i386/i386.h (struct machine_function): Add
5111 no_drap_save_restore field.
5112 * config/i386/i386.c (ix86_save_reg): Use
5113 !cfun->machine->no_drap_save_restore instead of
5114 crtl->stack_realign_needed.
5115 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5116 this function clears frame_pointer_needed. Set
5117 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5118 and DRAP reg is needed.
5120 2014-01-06 Marek Polacek <polacek@redhat.com>
5122 PR c/57773
5123 * doc/implement-c.texi: Mention that other integer types are
5124 permitted as bit-field types in strictly conforming mode.
5126 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5128 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5129 is newly allocated.
5131 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5133 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5135 2014-01-06 Martin Jambor <mjambor@suse.cz>
5137 PR ipa/59008
5138 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5139 to int.
5140 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5142 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5144 PR debug/59350
5145 PR debug/59510
5146 * var-tracking.c (add_stores): Preserve the value of the source even if
5147 we don't record the store.
5149 2014-01-06 Terry Guo <terry.guo@arm.com>
5151 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5153 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5155 PR bootstrap/59541
5156 * config/darwin.c (darwin_function_section): Adjust return values to
5157 correspond to optimisation changes made in r206070.
5159 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5161 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5162 from prefetch_block tune setting.
5163 (nocona_cost): Correct size of prefetch block to 64.
5165 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5167 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5168 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5169 used to save the static chain register in the computation of the offset
5170 from which the FP registers need to be restored.
5172 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5174 PR tree-optimization/59519
5175 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5176 ICE if get_current_def (current_new_name) is already non-NULL, as long
5177 as it is a phi result of some other phi in *new_exit_bb that has
5178 the same argument.
5180 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5181 or vmovdqu* for misaligned_operand.
5182 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5183 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5184 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5185 aligned_mem for AVX512F masked aligned load and store builtins and for
5186 non-temporal moves.
5188 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5190 PR tree-optimization/59651
5191 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5192 Address range for negative step should be added by TYPE_SIZE_UNIT.
5194 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5196 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5197 overlapping registers also for registers other than the stack pointer.
5199 2014-01-03 Marek Polacek <polacek@redhat.com>
5201 PR other/59661
5202 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5203 __builtin_FILE.
5205 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5207 PR target/59625
5208 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5209 asm goto as jump.
5211 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5212 (push splitter): Use <P:MODE_SIZE> instead of
5213 GET_MODE_SIZE (<P:MODE>mode).
5214 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5215 (mov -1, reg peephole2): Likewise.
5216 * config/i386/sse.md (*mov<mode>_internal,
5217 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5218 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5219 *<code><mode>3, *andnot<mode>3<mask_name>,
5220 <mask_codefor><code><mode>3<mask_name>): Likewise.
5221 * config/i386/subst.md (mask_mode512bit_condition,
5222 sd_mask_mode512bit_condition): Likewise.
5224 2014-01-02 Xinliang David Li <davidxl@google.com>
5226 PR tree-optimization/59303
5227 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5228 (dump_predicates): Better output format.
5229 (pred_equal_p): New function.
5230 (is_neq_relop_p): Ditto.
5231 (is_neq_zero_form_p): Ditto.
5232 (pred_expr_equal_p): Ditto.
5233 (pred_neg_p): Ditto.
5234 (simplify_pred): Ditto.
5235 (simplify_preds_2): Ditto.
5236 (simplify_preds_3): Ditto.
5237 (simplify_preds_4): Ditto.
5238 (simplify_preds): Ditto.
5239 (push_pred): Ditto.
5240 (push_to_worklist): Ditto.
5241 (get_pred_info_from_cmp): Ditto.
5242 (is_degenerated_phi): Ditto.
5243 (normalize_one_pred_1): Ditto.
5244 (normalize_one_pred): Ditto.
5245 (normalize_one_pred_chain): Ditto.
5246 (normalize_preds): Ditto.
5247 (normalize_cond_1): Remove function.
5248 (normalize_cond): Ditto.
5249 (is_gcond_subset_of): Ditto.
5250 (is_subset_of_any): Ditto.
5251 (is_or_set_subset_of): Ditto.
5252 (is_and_set_subset_of): Ditto.
5253 (is_norm_cond_subset_of): Ditto.
5254 (pred_chain_length_cmp): Ditto.
5255 (convert_control_dep_chain_into_preds): Type change.
5256 (find_predicates): Ditto.
5257 (find_def_preds): Ditto.
5258 (destroy_predicates_vecs): Ditto.
5259 (find_matching_predicates_in_rest_chains): Ditto.
5260 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5261 (is_pred_expr_subset): Ditto.
5262 (is_pred_chain_subset_of): Ditto.
5263 (is_included_in): Ditto.
5264 (is_superset_of): Ditto.
5266 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5268 Update copyright years.
5270 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5272 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5273 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5274 config/arc/arc.md, config/arc/arc.opt,
5275 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5276 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5277 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5278 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5279 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5280 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5281 vtable-verify.c, vtable-verify.h: Use the standard form for the
5282 copyright notice.
5284 2014-01-02 Tobias Burnus <burnus@net-b.de>
5286 * gcc.c (process_command): Update copyright notice dates.
5287 * gcov-dump.c: Ditto.
5288 * gcov.c: Ditto.
5289 * doc/cpp.texi: Bump @copying's copyright year.
5290 * doc/cppinternals.texi: Ditto.
5291 * doc/gcc.texi: Ditto.
5292 * doc/gccint.texi: Ditto.
5293 * doc/gcov.texi: Ditto.
5294 * doc/install.texi: Ditto.
5295 * doc/invoke.texi: Ditto.
5297 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5299 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5301 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5303 * config/i386/sse.md (*mov<mode>_internal): Guard
5304 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5306 PR rtl-optimization/59647
5307 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5308 new_rtx into UNSIGNED_FLOAT rtxes.
5309 \f
5310 Copyright (C) 2014 Free Software Foundation, Inc.
5312 Copying and distribution of this file, with or without modification,
5313 are permitted in any medium without royalty provided the copyright
5314 notice and this notice are preserved.