7cddeabb750bf95b5ea7d551354115f27b217358
[gcc.git] / gcc / ChangeLog
1 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2
3 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
4 pattern for vector nor instead of subtract from splat(-1).
5 (altivec_expand_vec_perm_const_le): Likewise.
6
7 2014-03-21 Richard Henderson <rth@twiddle.net>
8
9 PR target/60598
10 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
11 related insns after epilogue_completed.
12
13 2014-03-21 Martin Jambor <mjambor@suse.cz>
14
15 PR ipa/59176
16 * cgraph.h (symtab_node): New flag body_removed.
17 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
18 when removing bodies.
19 * symtab.c (dump_symtab_base): Dump body_removed flag.
20 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
21 had their bodies removed.
22
23 2014-03-21 Martin Jambor <mjambor@suse.cz>
24
25 PR ipa/60419
26 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
27 in the border.
28
29 2014-03-21 Richard Biener <rguenther@suse.de>
30
31 PR tree-optimization/60577
32 * tree-core.h (struct tree_base): Document nothrow_flag use
33 in VAR_DECL_NONALIASED.
34 * tree.h (VAR_DECL_NONALIASED): New.
35 (may_be_aliased): Adjust.
36 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
37
38 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
39
40 * expr.c (expand_expr_real_1): Remove outdated comment.
41
42 2014-03-20 Jakub Jelinek <jakub@redhat.com>
43
44 PR middle-end/60597
45 * ira.c (adjust_cleared_regs): Call copy_rtx on
46 *reg_equiv[REGNO (loc)].src_p before passing it to
47 simplify_replace_fn_rtx.
48
49 PR target/60568
50 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
51 into CONST, put pic register as first operand of PLUS. Use
52 gen_const_mem for both 32-bit and 64-bit PIC got loads.
53
54 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
55
56 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
57
58 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
59
60 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
61 around for store forwarding issue in the FPU on the UT699.
62 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
63 loads and operations if -mfix-ut699 is specified.
64 (divtf3_hq): Tweak attribute.
65 (sqrttf2_hq): Likewise.
66
67 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
68
69 * calls.c (store_one_arg): Remove incorrect const qualification on the
70 type of the temporary.
71 * cfgexpand.c (expand_return): Likewise.
72 * expr.c (expand_constructor): Likewise.
73 (expand_expr_real_1): Likewise.
74
75 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
76
77 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
78 of parts.
79
80 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
81
82 PR target/60039
83 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
84
85 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
86
87 * config/arm/aarch-common-protos.h
88 (alu_cost_table): Fix spelling of "extend".
89 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
90
91 2014-03-19 Richard Biener <rguenther@suse.de>
92
93 PR middle-end/60553
94 * tree-core.h (tree_type_common): Re-order pointer members
95 to reduce recursion depth during GC walks.
96
97 2014-03-19 Marek Polacek <polacek@redhat.com>
98
99 PR sanitizer/60569
100 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
101 before accessing it.
102
103 2014-03-19 Richard Biener <rguenther@suse.de>
104
105 PR lto/59543
106 * lto-streamer-in.c (input_function): In WPA stage do not drop
107 debug stmts.
108
109 2014-03-19 Jakub Jelinek <jakub@redhat.com>
110
111 PR tree-optimization/60559
112 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
113 with build_zero_cst assignment.
114
115 2014-03-18 Kai Tietz <ktietz@redhat.com>
116
117 PR rtl-optimization/56356
118 * sdbout.c (sdbout_parms): Verify that parms'
119 incoming argument is valid.
120 (sdbout_reg_parms): Likewise.
121
122 2014-03-18 Richard Henderson <rth@redhat.com>
123
124 PR target/60562
125 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
126 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
127 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
128
129 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
130
131 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
132 * doc/plugins.texi (Plugin callbacks): Mention
133 PLUGIN_INCLUDE_FILE.
134 Italicize plugin event names in description. Explain that
135 PLUGIN_PRAGMAS has no sense for lto1. Explain
136 PLUGIN_INCLUDE_FILE.
137 Remind that no GCC functions should be called after
138 PLUGIN_FINISH.
139 Explain what pragmas with expansion are.
140
141 2014-03-18 Martin Liska <mliska@suse.cz>
142
143 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
144 gimple call statement is update.
145 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
146 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
147
148 2014-03-18 Jakub Jelinek <jakub@redhat.com>
149
150 PR sanitizer/60557
151 * ubsan.c (ubsan_instrument_unreachable): Call
152 initialize_sanitizer_builtins.
153 (ubsan_pass): Likewise.
154
155 PR sanitizer/60535
156 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
157 varpool_finalize_decl instead of rest_of_decl_compilation.
158
159 2014-03-18 Richard Biener <rguenther@suse.de>
160
161 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
162 by using bitmap_and_compl instead of bitmap_and_compl_into.
163 (df_rd_transfer_function): Likewise.
164
165 2014-03-18 Richard Biener <rguenther@suse.de>
166
167 * doc/lto.texi (fresolution): Fix typo.
168
169 2014-03-18 Richard Biener <rguenther@suse.de>
170
171 * doc/invoke.texi (flto): Update for changes in 4.9.
172
173 2014-03-18 Richard Biener <rguenther@suse.de>
174
175 * doc/loop.texi: Remove section on the removed lambda framework.
176 Update loop docs with recent changes in preserving loop structure.
177
178 2014-03-18 Richard Biener <rguenther@suse.de>
179
180 * doc/lto.texi (-fresolution): Document.
181
182 2014-03-18 Richard Biener <rguenther@suse.de>
183
184 * doc/contrib.texi: Adjust my name.
185
186 2014-03-18 Jakub Jelinek <jakub@redhat.com>
187
188 PR ipa/58721
189 * internal-fn.c: Include diagnostic-core.h.
190 (expand_BUILTIN_EXPECT): New function.
191 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
192 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
193 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
194 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
195 IFN_BUILTIN_EXPECT.
196 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
197 Revert 3 argument __builtin_expect code.
198 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
199 * gimple-fold.c (gimple_fold_call): Likewise.
200 * tree.h (fold_builtin_expect): New prototype.
201 * builtins.c (build_builtin_expect_predicate): Add predictor
202 argument, if non-NULL, create 3 argument __builtin_expect.
203 (fold_builtin_expect): No longer static. Add ARG2 argument,
204 pass it through to build_builtin_expect_predicate.
205 (fold_builtin_2): Adjust caller.
206 (fold_builtin_3): Handle BUILT_IN_EXPECT.
207 * internal-fn.def (BUILTIN_EXPECT): New.
208
209 2014-03-18 Tobias Burnus <burnus@net-b.de>
210
211 PR ipa/58721
212 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
213 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
214 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
215
216 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
217
218 PR ipa/58721
219 * predict.c (combine_predictions_for_bb): Fix up formatting.
220 (expr_expected_value_1, expr_expected_value): Add predictor argument,
221 fill what it points to if non-NULL.
222 (tree_predict_by_opcode): Adjust caller, use the predictor.
223 * predict.def (PRED_COMPARE_AND_SWAP): Add.
224
225 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
226
227 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
228 proper constant for the store mode.
229
230 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
231
232 * symtab.c (change_decl_assembler_name): Fix transparent alias
233 chain construction.
234
235 2014-03-16 Renlin Li <Renlin.Li@arm.com>
236
237 * config/aarch64/aarch64.c: Correct the comments about the
238 aarch64 stack layout.
239
240 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
241
242 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
243 check for GF_OMP_FOR_KIND_FOR.
244
245 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
246
247 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
248 ymm and zmm register names.
249
250 2014-03-17 Jakub Jelinek <jakub@redhat.com>
251
252 PR target/60516
253 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
254 note creation for the 2010-08-31 changes.
255
256 2014-03-17 Marek Polacek <polacek@redhat.com>
257
258 PR middle-end/60534
259 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
260 as -fno-tree-loop-vectorize.
261 (expand_omp_simd): Likewise.
262
263 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
264
265 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
266 (eligible_for_call_delay): New prototype.
267 * config/sparc/sparc.c (tls_call_delay): Rename into...
268 (eligible_for_call_delay): ...this. Return false if the instruction
269 cannot be put in the delay slot of a branch.
270 (eligible_for_restore_insn): Simplify.
271 (eligible_for_return_delay): Return false if the instruction cannot be
272 put in the delay slot of a branch and simplify.
273 (eligible_for_sibcall_delay): Return false if the instruction cannot be
274 put in the delay slot of a branch.
275 * config/sparc/sparc.md (fix_ut699): New attribute.
276 (tls_call_delay): Delete.
277 (in_call_delay): Reimplement.
278 (eligible_for_sibcall_delay): Rename into...
279 (in_sibcall_delay): ...this.
280 (eligible_for_return_delay): Rename into...
281 (in_return_delay): ...this.
282 (in_branch_delay): Reimplement.
283 (in_uncond_branch_delay): Delete.
284 (in_annul_branch_delay): Delete.
285
286 2014-03-14 Richard Henderson <rth@redhat.com>
287
288 PR target/60525
289 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
290 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
291 (*floathi<X87MODEF>2_i387_with_temp): Remove.
292 (floathi splitters): Remove.
293 (float<SWI48x>xf2): New pattern.
294 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
295 code that tried to handle DImode for 32-bit, but which was excluded
296 by the pattern's condition. Drop allocation of stack temporary.
297 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
298 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
299 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
300 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
301 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
302 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
303 (*float<SWI48><MODEF>2_sse_interunit): Remove.
304 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
305 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
306 (*float<SWI48x><X87MODEF>2_i387): Remove.
307 (all float _with_temp splitters): Remove.
308 (*float<SWI48x><MODEF>2_i387): New pattern.
309 (*float<SWI48><MODEF>2_sse): New pattern.
310 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
311 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
312
313 2014-03-14 Jakub Jelinek <jakub@redhat.com>
314 Marek Polacek <polacek@redhat.com>
315
316 PR middle-end/60484
317 * common.opt (dump_base_name_prefixed): New Variable.
318 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
319 if x_dump_base_name_prefixed is already set, set it at the end.
320
321 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
322
323 PR rtl-optimization/60508
324 * lra-constraints.c (get_reload_reg): Add new parameter
325 in_subreg_p.
326 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
327 Pass the new parameter values.
328
329 2014-03-14 Richard Biener <rguenther@suse.de>
330
331 * common.opt: Revert unintented changes from r205065.
332 * opts.c: Likewise.
333
334 2014-03-14 Richard Biener <rguenther@suse.de>
335
336 PR middle-end/60518
337 * cfghooks.c (split_block): Properly adjust all loops the
338 block was a latch of.
339
340 2014-03-14 Martin Jambor <mjambor@suse.cz>
341
342 PR lto/60461
343 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
344 and simplify it.
345
346 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
347
348 PR target/59396
349 * config/avr/avr.c (avr_set_current_function): Pass function name
350 through default_strip_name_encoding before sanity checking instead
351 of skipping the first char of the assembler name.
352
353 2014-03-13 Richard Henderson <rth@redhat.com>
354
355 PR debug/60438
356 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
357 (ix86_force_to_memory, ix86_free_from_memory): Remove.
358 * config/i386/i386-protos.h: Likewise.
359 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
360 in the expander instead of a splitter.
361 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
362 any possibility of requiring a memory.
363 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
364 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
365 (fp branch splitters): Update for ix86_split_fp_branch.
366 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
367 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
368 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
369 (*fop_<MODEF>_2_i387): Remove f/r alternative.
370 (*fop_<MODEF>_3_i387): Likewise.
371 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
372 (splitters for the fop_* register patterns): Remove.
373 (fscalexf4_i387): Rename from *fscalexf4_i387.
374 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
375
376 2014-03-13 Jakub Jelinek <jakub@redhat.com>
377
378 PR tree-optimization/59779
379 * tree-dfa.c (get_ref_base_and_extent): Use double_int
380 type for bitsize and maxsize instead of HOST_WIDE_INT.
381
382 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
383
384 PR rtl-optimization/57320
385 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
386 the CFG after thread_prologue_and_epilogue_insns.
387
388 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
389
390 PR rtl-optimization/57189
391 * lra-constraints.c (process_alt_operands): Disfavor spilling
392 vector pseudos.
393
394 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
395
396 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
397 messages.
398
399 2014-03-13 Jakub Jelinek <jakub@redhat.com>
400
401 PR tree-optimization/59025
402 PR middle-end/60418
403 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
404 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
405
406 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
407
408 PR target/60486
409 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
410 calls of avr_out_plus_1.
411
412 2014-03-13 Bin Cheng <bin.cheng@arm.com>
413
414 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
415 BB's single pred and update the father loop's latch info later.
416
417 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
418
419 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
420 (VEC_M): Likewise.
421 (VEC_N): Likewise.
422 (VEC_R): Likewise.
423 (VEC_base): Likewise.
424 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
425 registers, we need to swap double words in little endian mode.
426
427 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
428 to be a container mode for 128-bit integer operations added in ISA
429 2.07. Unlike TImode and PTImode, the preferred register set is
430 the Altivec/VMX registers for the 128-bit operations.
431
432 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
433 declarations.
434 (rs6000_split_128bit_ok_p): Likewise.
435
436 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
437 macros for creating ISA 2.07 normal and overloaded builtin
438 functions with 3 arguments.
439 (BU_P8V_OVERLOAD_3): Likewise.
440 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
441 for use as overloaded functions.
442 (VPERM_1TI_UNS): Likewise.
443 (VSEL_1TI): Likewise.
444 (VSEL_1TI_UNS): Likewise.
445 (ST_INTERNAL_1ti): Likewise.
446 (LD_INTERNAL_1ti): Likewise.
447 (XXSEL_1TI): Likewise.
448 (XXSEL_1TI_UNS): Likewise.
449 (VPERM_1TI): Likewise.
450 (VPERM_1TI_UNS): Likewise.
451 (XXPERMDI_1TI): Likewise.
452 (SET_1TI): Likewise.
453 (LXVD2X_V1TI): Likewise.
454 (STXVD2X_V1TI): Likewise.
455 (VEC_INIT_V1TI): Likewise.
456 (VEC_SET_V1TI): Likewise.
457 (VEC_EXT_V1TI): Likewise.
458 (EQV_V1TI): Likewise.
459 (NAND_V1TI): Likewise.
460 (ORC_V1TI): Likewise.
461 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
462 added in ISA 2.07. Add both normal 'altivec' builtins, and the
463 overloaded builtin.
464 (VADDUQM): Likewise.
465 (VSUBCUQ): Likewise.
466 (VADDEUQM): Likewise.
467 (VADDECUQ): Likewise.
468 (VSUBEUQM): Likewise.
469 (VSUBECUQ): Likewise.
470
471 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
472 __int128_t and __uint128_t types.
473 (__uint128_type): Likewise.
474 (altivec_categorize_keyword): Add support for vector __int128_t,
475 vector __uint128_t, vector __int128, and vector unsigned __int128
476 as a container type for TImode operations that need to be done in
477 VSX/Altivec registers.
478 (rs6000_macro_to_expand): Likewise.
479 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
480 to support 128-bit integer instructions vaddcuq, vadduqm,
481 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
482 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
483
484 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
485 for V1TImode, and set up preferences to use VSX/Altivec registers.
486 Setup VSX reload handlers.
487 (rs6000_debug_reg_global): Likewise.
488 (rs6000_init_hard_regno_mode_ok): Likewise.
489 (rs6000_preferred_simd_mode): Likewise.
490 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
491 (easy_altivec_constant): Likewise.
492 (output_vec_const_move): Likewise.
493 (rs6000_expand_vector_set): Convert V1TImode set and extract to
494 simple move.
495 (rs6000_expand_vector_extract): Likewise.
496 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
497 addressing.
498 (rs6000_const_vec): Add support for V1TImode.
499 (rs6000_emit_le_vsx_load): Swap double words when loading or
500 storing TImode/V1TImode.
501 (rs6000_emit_le_vsx_store): Likewise.
502 (rs6000_emit_le_vsx_move): Likewise.
503 (rs6000_emit_move): Add support for V1TImode.
504 (altivec_expand_ld_builtin): Likewise.
505 (altivec_expand_st_builtin): Likewise.
506 (altivec_expand_vec_init_builtin): Likewise.
507 (altivec_expand_builtin): Likewise.
508 (rs6000_init_builtins): Add support for V1TImode type. Add
509 support for ISA 2.07 128-bit integer builtins. Define type names
510 for the VSX/Altivec vector types.
511 (altivec_init_builtins): Add support for overloaded vector
512 functions with V1TImode type.
513 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
514 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
515 external function.
516 (rs6000_split_128bit_ok_p): Likewise.
517 (rs6000_handle_altivec_attribute): Create V1TImode from vector
518 __int128_t and vector __uint128_t.
519
520 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
521 and mode attributes.
522 (VSX_M): Likewise.
523 (VSX_M2): Likewise.
524 (VSm): Likewise.
525 (VSs): Likewise.
526 (VSr): Likewise.
527 (VSv): Likewise.
528 (VS_scalar): Likewise.
529 (VS_double): Likewise.
530 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
531
532 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
533 we support the ISA 2.07 128-bit integer arithmetic instructions.
534 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
535 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
536 and TImode types for use with the builtin functions.
537 (V1TI_type_node): Likewise.
538 (unsigned_V1TI_type_node): Likewise.
539 (intTI_type_internal_node): Likewise.
540 (uintTI_type_internal_node): Likewise.
541
542 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
543 128-bit builtin functions.
544 (UNSPEC_VADDEUQM): Likewise.
545 (UNSPEC_VADDECUQ): Likewise.
546 (UNSPEC_VSUBCUQ): Likewise.
547 (UNSPEC_VSUBEUQM): Likewise.
548 (UNSPEC_VSUBECUQ): Likewise.
549 (VM): Add V1TImode to vector mode iterators.
550 (VM2): Likewise.
551 (VI_unit): Likewise.
552 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
553 (altivec_vaddcuq): Likewise.
554 (altivec_vsubuqm): Likewise.
555 (altivec_vsubcuq): Likewise.
556 (altivec_vaddeuqm): Likewise.
557 (altivec_vaddecuq): Likewise.
558 (altivec_vsubeuqm): Likewise.
559 (altivec_vsubecuq): Likewise.
560
561 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
562 mode iterators.
563 (BOOL_128): Likewise.
564 (BOOL_REGS_OUTPUT): Likewise.
565 (BOOL_REGS_OP1): Likewise.
566 (BOOL_REGS_OP2): Likewise.
567 (BOOL_REGS_UNARY): Likewise.
568 (BOOL_REGS_AND_CR0): Likewise.
569
570 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
571 128-bit integer builtin support.
572 (vec_vadduqm): Likewise.
573 (vec_vaddecuq): Likewise.
574 (vec_vaddeuqm): Likewise.
575 (vec_vsubecuq): Likewise.
576 (vec_vsubeuqm): Likewise.
577 (vec_vsubcuq): Likewise.
578 (vec_vsubuqm): Likewise.
579
580 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
581 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
582 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
583 128-bit integer add/subtract to ISA 2.07.
584
585 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
586
587 * config/arc/arc.c (arc_predicate_delay_insns):
588 Fix third argument passed to conditionalize_nonjump.
589
590 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
591
592 * config/aarch64/aarch64-builtins.c
593 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
594 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
595 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
596 instead of __builtin_lfloor.
597 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
598
599 2014-03-12 Jakub Jelinek <jakub@redhat.com>
600
601 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
602 (tree_ssa_ifcombine_bb_1): New function.
603 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
604 is an empty forwarder block to then_bb or vice versa and then_bb
605 and else_bb are effectively swapped.
606
607 2014-03-12 Christian Bruel <christian.bruel@st.com>
608
609 PR target/60264
610 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
611 REG_CFA_DEF_CFA note.
612 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
613 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
614
615 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
616
617 PR tree-optimization/60454
618 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
619
620 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
621
622 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
623 Do not define target_cpu_default2 to generic.
624 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
625 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
626 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
627
628 2014-03-12 Jakub Jelinek <jakub@redhat.com>
629 Marc Glisse <marc.glisse@inria.fr>
630
631 PR tree-optimization/60502
632 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
633 instead of build_low_bits_mask.
634
635 2014-03-12 Jakub Jelinek <jakub@redhat.com>
636
637 PR middle-end/60482
638 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
639 if there are multiple uses, but op doesn't live on E edge.
640 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
641 clobber stmts before __builtin_unreachable.
642
643 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
644
645 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
646 hard_frame_pointer_rtx.
647 * cse.c (cse_insn): Remove volatile check.
648 * cselib.c (cselib_process_insn): Likewise.
649 * dse.c (scan_insn): Likewise.
650
651 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
652
653 * config/arc/arc.c (conditionalize_nonjump): New function,
654 broken out of ...
655 (arc_ifcvt): ... this.
656 (arc_predicate_delay_insns): Use it.
657
658 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
659
660 * config/arc/predicates.md (extend_operand): During/after reload,
661 allow const_int_operand.
662 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
663 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
664 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
665 to "i".
666 (umulsi3_highpart_i): Likewise.
667
668 2014-03-11 Richard Biener <rguenther@suse.de>
669
670 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
671 Add asserts to guard possible wrong-code bugs.
672
673 2014-03-11 Richard Biener <rguenther@suse.de>
674
675 PR tree-optimization/60429
676 PR tree-optimization/60485
677 * tree-ssa-structalias.c (set_union_with_increment): Properly
678 take into account all fields that overlap the shifted vars.
679 (do_sd_constraint): Likewise.
680 (do_ds_constraint): Likewise.
681 (get_constraint_for_ptr_offset): Likewise.
682
683 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
684
685 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
686 (nios2_compute_frame_layout):
687 Add calculation of cfun->machine->fp_save_offset.
688 (nios2_expand_prologue): Correct setting of frame pointer register
689 in prologue.
690 (nios2_expand_epilogue): Update recovery of stack pointer from
691 frame pointer accordingly.
692 (nios2_initial_elimination_offset): Update calculation of offset
693 for eliminating to HARD_FRAME_POINTER_REGNUM.
694
695 2014-03-10 Jakub Jelinek <jakub@redhat.com>
696
697 PR ipa/60457
698 * ipa.c (symtab_remove_unreachable_nodes): Don't call
699 cgraph_get_create_node on VAR_DECLs.
700
701 2014-03-10 Richard Biener <rguenther@suse.de>
702
703 PR middle-end/60474
704 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
705
706 2014-03-08 Douglas B Rupp <rupp@gnat.com>
707
708 * config/vms/vms.opt (vms_float_format): New variable.
709
710 2014-03-08 Tobias Burnus <burnus@net-b.de>
711
712 * doc/invoke.texi (-fcilkplus): Update implementation status.
713
714 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
715 Richard Biener <rguenther@suse.de>
716
717 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
718 consistently accross all TUs.
719 (run_gcc): Enable -fshort-double automatically at link at link-time
720 and disallow override.
721
722 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
723
724 PR target/58271
725 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
726 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
727 if they can't be used.
728
729 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
730
731 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
732 for Solaris 11/x86 ld.
733 * configure: Regenerate.
734
735 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
736
737 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
738 (LIB_TLS_SPEC): Save as ld_tls_libs.
739 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
740 (HAVE_AS_IX86_TLSLDM): New test.
741 * configure, config.in: Regenerate.
742 * config/i386/i386.c (legitimize_tls_address): Fall back to
743 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
744 cannot support TLS_MODEL_LOCAL_DYNAMIC.
745 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
746 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
747
748 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
749
750 * common.opt (fira-loop-pressure): Mark as optimization.
751
752 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
753
754 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
755 an OpenMP mappable type.
756
757 2014-03-06 Matthias Klose <doko@ubuntu.com>
758
759 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
760 MULTILIB_OSDIRNAMES is not defined.
761
762 2014-03-06 Jakub Jelinek <jakub@redhat.com>
763 Meador Inge <meadori@codesourcery.com>
764
765 PR target/58595
766 * config/arm/arm.c (arm_tls_symbol_p): Remove.
767 (arm_legitimize_address): Call legitimize_tls_address for any
768 arm_tls_referenced_p expression, handle constant addend. Call it
769 before testing for !TARGET_ARM.
770 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
771
772 2014-03-06 Richard Biener <rguenther@suse.de>
773
774 PR middle-end/60445
775 PR lto/60424
776 PR lto/60427
777 Revert
778 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
779
780 * tree-streamer.c (record_common_node): Assert we don't record
781 nodes with type double.
782 (preload_common_node): Skip type double, complex double and double
783 pointer since it is now frontend dependent due to fshort-double option.
784
785 2014-03-06 Richard Biener <rguenther@suse.de>
786
787 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
788 or -fno-lto is specified and the linker has full plugin support.
789 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
790 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
791 * lto-wrapper.c (merge_and_complain): Merge compile-time
792 optimization levels.
793 (run_gcc): And pass it through to the link options.
794
795 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
796
797 PR debug/60381
798 Revert:
799 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
800 PR debug/59992
801 * cselib.c (remove_useless_values): Skip to avoid quadratic
802 behavior if the condition moved from...
803 (cselib_process_insn): ... here holds.
804
805 2014-03-05 Jakub Jelinek <jakub@redhat.com>
806
807 PR plugins/59335
808 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
809 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
810
811 PR plugins/59335
812 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
813 (TM_H): Add x86-tune.def.
814
815 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
816
817 * config/aarch64/aarch64.c (generic_tunings):
818 Use cortexa57_extra_costs.
819
820 2014-03-05 Jakub Jelinek <jakub@redhat.com>
821
822 PR lto/60404
823 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
824 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
825 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
826 cost for in_lto_p.
827
828 2014-03-04 Heiher <r@hev.cc>
829
830 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
831 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
832
833 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
834
835 * config/i386/predicates.md (const2356_operand): Change to ...
836 (const2367_operand): ... this.
837 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
838 const2367_operand.
839 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
840 (*avx512pf_scatterpf<mode>sf): Ditto.
841 (avx512pf_scatterpf<mode>df): Ditto.
842 (*avx512pf_scatterpf<mode>df_mask): Ditto.
843 (*avx512pf_scatterpf<mode>df): Ditto.
844 * config/i386/i386.c (ix86_expand_builtin): Update
845 incorrect hint operand error message.
846
847 2014-03-04 Richard Biener <rguenther@suse.de>
848
849 * lto-section-in.c (lto_get_section_data): Fix const cast.
850
851 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
852
853 * tree-streamer.c (record_common_node): Assert we don't record
854 nodes with type double.
855 (preload_common_node): Skip type double, complex double and double
856 pointer since it is now frontend dependent due to fshort-double option.
857
858 2014-03-04 Richard Biener <rguenther@suse.de>
859
860 PR lto/60405
861 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
862 (lto_input_toplevel_asms): Likewise.
863 * lto-section-in.c (lto_get_section_data): Instead do it here
864 for every section.
865
866 2014-03-04 Richard Biener <rguenther@suse.de>
867
868 PR tree-optimization/60382
869 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
870 dead PHIs a reduction.
871
872 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
873
874 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
875 hint value.
876 (_mm_prefetch): Move out of GCC target("sse") pragma.
877 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
878 GCC target("prfchw") pragma.
879 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
880 for locality <= 2.
881 * config/i386/i386.c (ix86_option_override_internal): Enable
882 -mprfchw with -mprefetchwt1.
883
884 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
885
886 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
887 Mark as varying.
888
889 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
890
891 * opts.h (CL_PCH_IGNORE): Define.
892 * targhooks.c (option_affects_pch_p):
893 Return false for options that have CL_PCH_IGNORE set.
894 * opt-functions.awk: Process PchIgnore.
895 * doc/options.texi: Document PchIgnore.
896
897 * config/arc/arc.opt (misize): Add PchIgnore property.
898
899 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
900
901 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
902 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
903 constraint on constants to permit them being loaded into
904 GENERAL_REGS or BASE_REGS.
905
906 2014-03-03 Nick Clifton <nickc@redhat.com>
907
908 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
909 anti-cacnonical alternatives.
910 (negandhi3_real): New pattern.
911 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
912
913 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
914
915 * config/avr/avr-mcus.def: Remove atxmega16x1.
916 * config/avr/avr-tables.opt: Regenerate.
917 * config/avr/t-multilib: Regenerate.
918 * doc/avr-mmcu.texi: Regenerate.
919
920 2014-03-03 Tobias Grosser <tobias@grosser.es>
921 Mircea Namolaru <mircea.namolaru@inria.fr>
922
923 PR tree-optimization/58028
924 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
925 scalar dimensions.
926
927 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
928
929 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
930 not handled by recognizers.
931
932 2014-03-03 Jakub Jelinek <jakub@redhat.com>
933
934 PR middle-end/60175
935 * function.c (expand_function_end): Don't emit
936 clobber_return_register sequence if clobber_after is a BARRIER.
937 * cfgexpand.c (construct_exit_block): Append instructions before
938 return_label to prev_bb.
939
940 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
941
942 * config/rs6000/constraints.md: Document reserved use of "wc".
943
944 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
945
946 PR ipa/60150
947 * ipa.c (function_and_variable_visibility): When dissolving comdat
948 group, also set all symbols to local.
949
950 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
951
952 PR ipa/60306
953
954 Revert:
955 2013-12-14 Jan Hubicka <jh@suse.cz>
956 PR middle-end/58477
957 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
958
959 2014-03-02 Jon Beniston <jon@beniston.com>
960
961 PR bootstrap/48230
962 PR bootstrap/50927
963 PR bootstrap/52466
964 PR target/46898
965 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
966 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
967 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
968 (simple_return, *simple_return): New patterns
969 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
970 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
971
972 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
973
974 * dwarf2out.c (gen_subprogram_die): Tidy.
975
976 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
977
978 PR target/60071
979 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
980 (*mov_t_msb_neg_negc): ... this new insn.
981
982 2014-02-28 Jason Merrill <jason@redhat.com>
983
984 PR c++/58678
985 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
986 function.
987
988 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
989
990 PR c++/60314
991 * dwarf2out.c (decltype_auto_die): New static.
992 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
993 (gen_type_die_with_usage): Handle 'decltype(auto)'.
994 (is_cxx_auto): Likewise.
995
996 2014-02-28 Ian Bolton <ian.bolton@arm.com>
997
998 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
999 we are not using general regs only.
1000
1001 2014-02-28 Richard Biener <rguenther@suse.de>
1002
1003 PR target/60280
1004 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1005 previous fix and only allow to remove trivial pre-headers
1006 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1007 (remove_forwarder_block): Properly update the latch of a loop.
1008
1009 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1010
1011 PR debug/59992
1012 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1013 (cselib_preserved_hash_table): New.
1014 (preserve_constants_and_equivs): Move preserved vals to it.
1015 (cselib_find_slot): Look it up first.
1016 (cselib_init): Initialize it.
1017 (cselib_finish): Release it.
1018 (dump_cselib_table): Dump it.
1019
1020 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1021
1022 PR debug/59992
1023 * cselib.c (remove_useless_values): Skip to avoid quadratic
1024 behavior if the condition moved from...
1025 (cselib_process_insn): ... here holds.
1026
1027 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1028
1029 PR debug/57232
1030 * var-tracking.c (vt_initialize): Apply the same condition to
1031 preserve the CFA base value.
1032
1033 2014-02-28 Joey Ye <joey.ye@arm.com>
1034
1035 PR target/PR60169
1036 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1037 if reload in progress or completed.
1038
1039 2014-02-28 Tobias Burnus <burnus@net-b.de>
1040
1041 PR middle-end/60147
1042 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1043 NAMELIST_DECL.
1044
1045 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1046
1047 * doc/tm.texi.in (Condition Code Status): Update documention for
1048 relative locations of cc0-setter and cc0-user.
1049
1050 2014-02-27 Jeff Law <law@redhat.com>
1051
1052 PR rtl-optimization/52714
1053 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1054 into two independent simple sets, if I3 is a jump, ensure the
1055 pattern we place into I3 is a (set (pc) ...).
1056
1057 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1058 Jeff Law <law@redhat.com>
1059
1060 PR rtl-optimization/49847
1061 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1062 are in different blocks.
1063 * doc/tm.texi (Condition Code Status): Update documention for
1064 relative locations of cc0-setter and cc0-user.
1065
1066 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1067
1068 PR target/59222
1069 * lra.c (lra_emit_add): Check SUBREG too.
1070
1071 2014-02-27 Andreas Schwab <schwab@suse.de>
1072
1073 * config/m68k/m68k.c (m68k_option_override): Disable
1074 -flive-range-shrinkage for classic m68k.
1075 (m68k_override_options_after_change): Likewise.
1076
1077 2014-02-27 Marek Polacek <polacek@redhat.com>
1078
1079 PR middle-end/59223
1080 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1081 -Wmaybe-uninitialized.
1082
1083 2014-02-27 Alan Modra <amodra@gmail.com>
1084
1085 PR target/57936
1086 * reload1.c (emit_input_reload_insns): When reload_override_in,
1087 set old to rl->in_reg when rl->in_reg is a subreg.
1088
1089 2014-02-26 Richard Biener <rguenther@suse.de>
1090
1091 PR bootstrap/60343
1092 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1093
1094 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1095
1096 * common/config/i386/predicates.md (const1256_operand): Remove.
1097 (const2356_operand): New.
1098 (const_1_to_2_operand): Remove.
1099 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1100 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1101 (*avx512pf_gatherpf<mode>sf): Ditto.
1102 (avx512pf_gatherpf<mode>df): Ditto.
1103 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1104 (*avx512pf_gatherpf<mode>df): Ditto.
1105 (avx512pf_scatterpf<mode>sf): Ditto.
1106 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1107 (*avx512pf_scatterpf<mode>sf): Ditto.
1108 (avx512pf_scatterpf<mode>df): Ditto.
1109 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1110 (*avx512pf_scatterpf<mode>df): Ditto.
1111 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1112
1113 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1114
1115 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1116 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1117 (_mm512_mask_testn_epi64_mask): Move to ...
1118 * config/i386/avx512cdintrin.h: Here.
1119 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1120 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1121 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1122 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1123 TARGET_AVX512F from TARGET_AVX512CD.
1124
1125 2014-02-26 Richard Biener <rguenther@suse.de>
1126
1127 PR ipa/60327
1128 * ipa.c (walk_polymorphic_call_targets): Properly guard
1129 call to inline_update_overall_summary.
1130
1131 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1132
1133 PR target/60280
1134 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1135 and latches only if requested. Fix latch if it is removed.
1136 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1137 LOOPS_HAVE_PREHEADERS.
1138
1139 2014-02-25 Andrew Pinski <apinski@cavium.com>
1140
1141 * builtins.c (expand_builtin_thread_pointer): Create a new target
1142 when the target is NULL.
1143
1144 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1145
1146 PR rtl-optimization/60317
1147 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1148 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1149 * lra-assigns.c: Include params.h.
1150 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1151 other reload pseudos considerations.
1152
1153 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1154
1155 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1156 to use canonical form for nor<mode>3.
1157
1158 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1159
1160 PR target/55426
1161 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1162 conversions.
1163
1164 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1165
1166 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1167 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1168 (ix86_handle_option): Handle OPT_mprefetchwt1.
1169 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1170 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1171 PREFETCHWT1 CPUID.
1172 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1173 OPTION_MASK_ISA_PREFETCHWT1.
1174 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1175 (PTA_PREFETCHWT1): New.
1176 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1177 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1178 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1179 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1180 (*prefetch_avx512pf_<mode>_: Change into ...
1181 (*prefetch_prefetchwt1_<mode>: This.
1182 * config/i386/i386.opt (mprefetchwt1): New.
1183 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1184 (_mm_prefetch): Handle intent to write.
1185 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1186
1187 2014-02-25 Richard Biener <rguenther@suse.de>
1188
1189 PR middle-end/60291
1190 * emit-rtl.c (mem_attrs_htab): Remove.
1191 (mem_attrs_htab_hash): Likewise.
1192 (mem_attrs_htab_eq): Likewise.
1193 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1194 (init_emit_once): Do not allocate mem_attrs_htab.
1195
1196 2014-02-25 Richard Biener <rguenther@suse.de>
1197
1198 PR lto/60319
1199 * lto-opts.c (lto_write_options): Output non-explicit conservative
1200 -fwrapv, -fno-trapv and -fno-strict-overflow.
1201 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1202 (run_gcc): And pass them through.
1203
1204 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1205
1206 * sel-sched.c (calculate_new_fences): New parameter ptime.
1207 Calculate it as a maximum over all fence cycles.
1208 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1209 Print the final schedule timing when sched_verbose.
1210
1211 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1212
1213 PR rtl-optimization/60292
1214 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1215 bit fot the fence instruction.
1216
1217 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1218
1219 * calls.h: Fix typo in comment.
1220
1221 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1222
1223 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1224 adjusting offsetable addresses.
1225
1226 2014-02-24 Guozhi Wei <carrot@google.com>
1227
1228 * sparseset.h (sparseset_pop): Fix the wrong index.
1229
1230 2014-02-24 Walter Lee <walt@tilera.com>
1231
1232 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1233 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1234 triplet.
1235 * common/config/tilegx/tilegx-common.c
1236 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1237 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1238 (LINK_SPEC): Ditto.
1239 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1240 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1241 (tilegx_gimplify_va_arg_expr): Handle big endian.
1242 (tilegx_expand_unaligned_load): Ditto.
1243 (tilegx_expand_unaligned_store): Ditto.
1244 (TARGET_RETURN_IN_MSB): New.
1245 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1246 (TARGET_ENDIAN_DEFAULT): New.
1247 (TARGET_BIG_ENDIAN): Handle big endian.
1248 (BYTES_BIG_ENDIAN): Ditto.
1249 (WORDS_BIG_ENDIAN): Ditto.
1250 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1251 (ENDIAN_SPEC): New.
1252 (EXTRA_SPECS): New.
1253 * config/tilegx/tilegx.md (extv): Handle big endian.
1254 (extzv): Ditto.
1255 (insn_st<n>): Ditto.
1256 (insn_st<n>_add<bitsuffix>): Ditto.
1257 (insn_stnt<n>): Ditto.
1258 (insn_stnt<n>_add<bitsuffix>):Ditto.
1259 (vec_interleave_highv8qi): Handle big endian.
1260 (vec_interleave_highv8qi_be): New.
1261 (vec_interleave_highv8qi_le): New.
1262 (insn_v1int_h): Handle big endian.
1263 (vec_interleave_lowv8qi): Handle big endian.
1264 (vec_interleave_lowv8qi_be): New.
1265 (vec_interleave_lowv8qi_le): New.
1266 (insn_v1int_l): Handle big endian.
1267 (vec_interleave_highv4hi): Handle big endian.
1268 (vec_interleave_highv4hi_be): New.
1269 (vec_interleave_highv4hi_le): New.
1270 (insn_v2int_h): Handle big endian.
1271 (vec_interleave_lowv4hi): Handle big endian.
1272 (vec_interleave_lowv4hi_be): New.
1273 (vec_interleave_lowv4hi_le): New.
1274 (insn_v2int_l): Handle big endian.
1275 (vec_interleave_highv2si): Handle big endian.
1276 (vec_interleave_highv2si_be): New.
1277 (vec_interleave_highv2si_le): New.
1278 (insn_v4int_h): Handle big endian.
1279 (vec_interleave_lowv2si): Handle big endian.
1280 (vec_interleave_lowv2si_be): New.
1281 (vec_interleave_lowv2si_le): New.
1282 (insn_v4int_l): Handle big endian.
1283 * config/tilegx/tilegx.opt (mbig-endian): New option.
1284 (mlittle-endian): New option.
1285 * doc/install.texi: Document tilegxbe-linux.
1286 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1287
1288 2014-02-24 Martin Jambor <mjambor@suse.cz>
1289
1290 PR ipa/60266
1291 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1292 there are no parameter descriptors.
1293
1294 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1295
1296 PR rtl-optimization/60268
1297 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1298 initialization to ...
1299 (sched_rgn_init): ... here.
1300 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1301
1302 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1303
1304 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1305 names.
1306
1307 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1308
1309 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1310 definition.
1311
1312 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1313
1314 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1315 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1316
1317 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1318
1319 * config/microblaze/predicates.md: Add cmp_op predicate.
1320 * config/microblaze/microblaze.md: Add branch_compare instruction
1321 which uses cmp_op predicate and emits cmp insn before branch.
1322 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1323 to microblaze_expand_conditional_branch and consolidate logic.
1324 (microblaze_expand_conditional_branch): emit branch_compare
1325 insn instead of handling cmp op separate from branch insn.
1326
1327 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1328
1329 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1330 to permit subregs.
1331
1332 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1333
1334 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1335 define_insn with define_expand and new define_insn
1336 *altivec_lve<VI_char>x_internal.
1337 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1338 and new define_insn *altivec_stve<VI_char>x_internal.
1339 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1340 prototype.
1341 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1342 lve*x built-ins.
1343 (altivec_expand_stvex_be): New function.
1344
1345 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1346
1347 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1348 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1349 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1350 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1351
1352 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1353
1354 PR target/60298
1355 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1356 instead of emit_move_insn.
1357
1358 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1359
1360 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1361 vspltw with vsldoi.
1362 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1363 gen_altivec_vsumsws.
1364
1365 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1366
1367 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1368 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1369 (altivec_lvxl_<mode>): New define_expand incorporating
1370 -maltivec=be semantics where needed.
1371 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1372 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1373 semantics where needed.
1374 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1375 (altivec_stvx_<mode>): New define_expand incorporating
1376 -maltivec=be semantics where needed.
1377 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1378 VM2 iterator instead of V4SI.
1379 (altivec_stvxl_<mode>): New define_expand incorporating
1380 -maltivec=be semantics where needed.
1381 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1382 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1383 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1384 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1385 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1386 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1387 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1388 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1389 ALTIVEC_BUILTIN_STVXL.
1390 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1391 (altivec_expand_stvx_be): Likewise.
1392 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1393 (altivec_expand_lvx_be): Likewise.
1394 (altivec_expand_stvx_be): Likewise.
1395 (altivec_expand_builtin): Add cases for
1396 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1397 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1398 (altivec_init_builtins): Add definitions for
1399 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1400 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1401
1402 2014-02-21 Catherine Moore <clm@codesourcery.com>
1403
1404 * doc/invoke.texi (mvirt, mno-virt): Document.
1405 * config/mips/mips.opt (mvirt): New option.
1406 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1407
1408 2014-02-21 Richard Biener <rguenther@suse.de>
1409
1410 PR tree-optimization/60276
1411 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1412 (STMT_VINFO_MIN_NEG_DIST): New macro.
1413 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1414 STMT_VINFO_MIN_NEG_DIST.
1415 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1416 made for negative dependence distances still hold.
1417
1418 2014-02-21 Richard Biener <rguenther@suse.de>
1419
1420 PR middle-end/60291
1421 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1422 DECL_INITIAL for globals not in the current function context.
1423
1424 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1425
1426 PR tree-optimization/56490
1427 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1428 * tree-ssa-uninit.c: Include params.h.
1429 (compute_control_dep_chain): Add num_calls argument, return false
1430 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1431 num_calls to recursive call.
1432 (find_predicates): Change dep_chain into normal array,
1433 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1434 variable and adjust compute_control_dep_chain caller.
1435 (find_def_preds): Likewise.
1436
1437 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1438
1439 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1440 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1441
1442 2014-02-21 Nick Clifton <nickc@redhat.com>
1443
1444 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1445 (pushhi1): Likewise.
1446 (popqi1): Add mode to pre_dec.
1447 (pophi1): Likewise.
1448
1449 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1450
1451 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1452 mode for mask of V8SFmode permutation.
1453
1454 2014-02-20 Richard Henderson <rth@redhat.com>
1455
1456 PR c++/60272
1457 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1458 a new pseudo for OLDVAL.
1459
1460 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1461
1462 PR target/57896
1463 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1464 gen_reg_rtx if d->testing_p.
1465 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1466 if d->testing_p and we will certainly return true.
1467 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1468 if d->testing_p.
1469
1470 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1471
1472 * emit-rtl.c (gen_reg_rtx): Assert that
1473 crtl->emit.regno_pointer_align_length is non-zero.
1474
1475 2014-02-20 Richard Henderson <rth@redhat.com>
1476
1477 PR c++/60272
1478 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1479 on failure the store back into EXPECT.
1480
1481 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1482 Sandra Loosemore <sandra@codesourcery.com>
1483
1484 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1485 * config/nios2/nios2.c (nios2_function_profiler): Add
1486 -fPIC (flag_pic == 2) support.
1487 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1488 (nios2_large_offset_p): New function.
1489 (nios2_unspec_reloc_p): Move up position, update to use
1490 nios2_large_offset_p.
1491 (nios2_unspec_address): Remove function.
1492 (nios2_unspec_offset): New function.
1493 (nios2_large_got_address): New function.
1494 (nios2_got_address): Add large offset support.
1495 (nios2_legitimize_tls_address): Update usage of removed and new
1496 functions.
1497 (nios2_symbol_binds_local_p): New function.
1498 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1499 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1500 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1501 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1502 processing for (const (unspec ...)).
1503 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1504
1505 2014-02-20 Richard Biener <rguenther@suse.de>
1506
1507 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1508 doing the substitution.
1509 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1510
1511 2014-02-20 Martin Jambor <mjambor@suse.cz>
1512
1513 PR ipa/55260
1514 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1515 info when checking whether lattices are bottom.
1516
1517 2014-02-20 Richard Biener <rguenther@suse.de>
1518
1519 PR middle-end/60221
1520 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1521 regions at -O0.
1522
1523 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1524
1525 PR ipa/58555
1526 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1527 parameter specifying the scaling.
1528 (inline_call): Update.
1529 (want_inline_recursively): Guard division by zero.
1530 (recursive_inlining): Update.
1531 * ipa-inline.h (clone_inlined_nodes): Update.
1532
1533 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1534
1535 PR target/60204
1536 * config/i386/i386.c (classify_argument): Pass structures of size
1537 64 bytes or less in register.
1538
1539 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1540 Kirill Yukhin <kirill.yukhin@intel.com>
1541
1542 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1543 (_mm_rcp28_round_ss): Ditto.
1544 (_mm_rsqrt28_round_sd): Ditto.
1545 (_mm_rsqrt28_round_ss): Ditto.
1546 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1547 (_mm_rcp14_round_ss): Ditto.
1548 (_mm_rsqrt14_round_sd): Ditto.
1549 (_mm_rsqrt14_round_ss): Ditto.
1550 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1551 the first input operand, get rid of match_dup.
1552 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1553 attribute to sse.
1554 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1555 Ditto.
1556 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1557 operand as the first input operand, set type attribute.
1558 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1559 Set type attribute.
1560 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1561 operand as the first input operand, set type attribute.
1562
1563 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1564
1565 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1566 bit of zero.
1567
1568 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1569
1570 PR target/60207
1571 * config/i386/i386.c (construct_container): Remove TFmode check
1572 for X86_64_INTEGER_CLASS.
1573
1574 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1575
1576 PR target/59794
1577 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1578 only when -Wpsabi is enabled.
1579
1580 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1581
1582 PR target/59799
1583 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1584 passing arrays in registers are the same as for structs, so remove the
1585 special case for them.
1586
1587 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1588
1589 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1590 destination type, extract only the valid bits if the source type is not
1591 integral and has a different mode.
1592
1593 2014-02-19 Richard Biener <rguenther@suse.de>
1594
1595 PR ipa/60243
1596 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1597 for all calls.
1598
1599 2014-02-19 Richard Biener <rguenther@suse.de>
1600
1601 PR ipa/60243
1602 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1603 (ipa_modify_call_arguments): Emit an argument load explicitely and
1604 preserve virtual SSA form there and for the replacement call.
1605 Do not update SSA form nor free dominance info.
1606
1607 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1608
1609 * ipa.c (function_and_variable_visibility): Also clear WEAK
1610 flag when disolving COMDAT_GROUP.
1611
1612 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1613
1614 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1615 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1616 not devirtualizing.
1617 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1618 do more sanity checks.
1619 (detect_type_change): Return true when giving up early.
1620 (compute_complex_assign_jump_func): Fix type parameter of
1621 ipa_set_ancestor_jf.
1622 (compute_complex_ancestor_jump_func): Likewise.
1623 (update_jump_functions_after_inlining): Fix updating of
1624 ancestor function.
1625 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1626
1627 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1628
1629 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1630 inline clones when edge disappears.
1631
1632 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1633
1634 PR target/60203
1635 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1636 Split 64-bit moves into 2 patterns. Do not allow the use of
1637 direct move for TDmode in little endian, since the decimal value
1638 has little endian bytes within a word, but the 64-bit pieces are
1639 ordered in a big endian fashion, and normal subreg's of TDmode are
1640 not allowed.
1641 (mov<mode>_64bit_dm): Likewise.
1642 (movtd_64bit_nodm): Likewise.
1643
1644 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1645
1646 PR tree-optimization/60174
1647 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1648 statement of an SSA_NAME that occurs in an abnormal PHI node.
1649
1650 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1651
1652 PR sanitizer/60142
1653 * final.c (SEEN_BB): Remove.
1654 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1655 (final_scan_insn): Don't force_source_line on second
1656 NOTE_INSN_BASIC_BLOCK.
1657
1658 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1659
1660 PR target/60205
1661 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1662 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1663 (type_natural_mode): Warn ABI change when %zmm register is not
1664 available for AVX512F vector value passing.
1665
1666 2014-02-18 Kai Tietz <ktietz@redhat.com>
1667
1668 PR target/60193
1669 * config/i386/i386.c (ix86_expand_prologue): Use value in
1670 rax register as displacement when restoring %r10 or %rax.
1671 Fix wrong offset when restoring both registers.
1672
1673 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1674
1675 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1676 assertion with conditional return.
1677
1678 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1679 Uros Bizjak <ubizjak@gmail.com>
1680
1681 PR driver/60233
1682 * config/i386/driver-i386.c (host_detect_local_cpu): If
1683 YMM state is not saved by the OS, also clear has_f16c. Move
1684 CPUID 0x80000001 handling before YMM state saving checking.
1685
1686 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1687
1688 PR rtl-optimization/58960
1689 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1690 factored out from ...
1691 (sched_init): ... here.
1692 (free_global_sched_pressure_data): New, factored out from ...
1693 (sched_finish): ... here.
1694 * sched-int.h (free_global_sched_pressure_data): Declare.
1695 * sched-rgn.c (nr_regions_initial): New static global.
1696 (haifa_find_rgns): Initialize it.
1697 (schedule_region): Disable sched-pressure for the newly
1698 generated regions.
1699
1700 2014-02-17 Richard Biener <rguenther@suse.de>
1701
1702 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1703 release SSA defs of pattern stmts.
1704
1705 2014-02-17 Richard Biener <rguenther@suse.de>
1706
1707 * tree-inline.c (expand_call_inline): Release the virtual
1708 operand defined by the call we are about to inline.
1709
1710 2014-02-17 Richard Biener <rguenther@suse.de>
1711
1712 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1713
1714 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1715 Ilya Tocar <ilya.tocar@intel.com>
1716
1717 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1718 arguments order in builtin.
1719 (_mm512_permutexvar_epi64): Ditto.
1720 (_mm512_mask_permutexvar_epi64): Ditto
1721 (_mm512_maskz_permutexvar_epi32): Ditto
1722 (_mm512_permutexvar_epi32): Ditto
1723 (_mm512_mask_permutexvar_epi32): Ditto
1724
1725 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1726
1727 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1728 (p8_vmrgow): Likewise.
1729
1730 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1731
1732 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1733 endian targets.
1734
1735 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1736
1737 PR target/60203
1738 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1739 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1740 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1741 using direct move instructions on ISA 2.07. Also adjust
1742 instruction length for 64-bit.
1743 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1744 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1745
1746 2014-02-15 Alan Modra <amodra@gmail.com>
1747
1748 PR target/58675
1749 PR target/57935
1750 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1751 find_replacement on parts of insn rtl that might be reloaded.
1752
1753 2014-02-15 Richard Biener <rguenther@suse.de>
1754
1755 PR tree-optimization/60183
1756 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1757 (tree_ssa_phiprop): Calculate and free post-dominators.
1758
1759 2014-02-14 Jeff Law <law@redhat.com>
1760
1761 PR rtl-optimization/60131
1762 * ree.c (get_extended_src_reg): New function.
1763 (combine_reaching_defs): Use it rather than assuming location of REG.
1764 (find_and_remove_re): Verify first operand of extension is
1765 a REG before adding the insns to the copy list.
1766
1767 2014-02-14 Roland McGrath <mcgrathr@google.com>
1768
1769 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1770 * configure: Regenerated.
1771 * config.in: Regenerated.
1772 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1773 instead of ASM_SHORT.
1774
1775 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1776 Richard Earnshaw <rearnsha@arm.com>
1777
1778 PR rtl-optimization/59535
1779 * lra-constraints.c (process_alt_operands): Encourage alternative
1780 when unassigned pseudo class is superset of the alternative class.
1781 (inherit_reload_reg): Don't inherit when optimizing for code size.
1782 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1783 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1784 modes not less than 4 for Thumb1.
1785
1786 2014-02-14 Kyle McMartin <kyle@redhat.com>
1787
1788 PR pch/60010
1789 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1790
1791 2014-02-14 Richard Biener <rguenther@suse.de>
1792
1793 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1794 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1795 Do not strip INDIRECT_REFs.
1796
1797 2014-02-14 Richard Biener <rguenther@suse.de>
1798
1799 PR lto/60179
1800 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1801 DECL_FUNCTION_SPECIFIC_TARGET.
1802 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1803 * tree-streamer-out.c (pack_ts_target_option): Remove.
1804 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1805 (write_ts_function_decl_tree_pointers): Do not stream
1806 DECL_FUNCTION_SPECIFIC_TARGET.
1807 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1808 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1809 (lto_input_ts_function_decl_tree_pointers): Do not stream
1810 DECL_FUNCTION_SPECIFIC_TARGET.
1811
1812 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1813
1814 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1815 (get_initial_def_for_induction, vectorizable_induction): Ignore
1816 debug stmts when looking for exit_phi.
1817 (vectorizable_live_operation): Fix up condition.
1818
1819 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1820
1821 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1822 nreverse() because it changes the content of original tree list.
1823
1824 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1825
1826 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1827 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1828
1829 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1830
1831 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1832 GNU coding standards.
1833
1834 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1835
1836 PR debug/60152
1837 * dwarf2out.c (gen_subprogram_die): Don't call
1838 add_calling_convention_attribute if subr_die is old_die.
1839
1840 2014-02-13 Sharad Singhai <singhai@google.com>
1841
1842 * doc/optinfo.texi: Fix order of nodes.
1843
1844 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1845
1846 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1847 operands[2], not operands[3].
1848
1849 2014-02-13 Richard Biener <rguenther@suse.de>
1850
1851 PR bootstrap/59878
1852 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1853 mention the possibility of an in-tree build.
1854 (CLooG): Update recommended version to 0.18.1, mention the
1855 possibility of an in-tree build and clarify that the ISL
1856 bundled with CLooG does not work.
1857
1858 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1859
1860 PR target/43546
1861 * expr.c (compress_float_constant): If x is a hard register,
1862 extend into a pseudo and then move to x.
1863
1864 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1865
1866 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1867 caused by bad second argument to warning_at() with -mhotpatch and
1868 nested functions (e.g. with gfortran).
1869
1870 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1871
1872 * opts.c (option_name): Remove "enabled by default" rider.
1873
1874 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1875
1876 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1877
1878 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1879 Uros Bizjak <ubizjak@gmail.com>
1880
1881 PR target/60151
1882 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1883 * configure: Regenerated.
1884
1885 2014-02-12 Richard Biener <rguenther@suse.de>
1886
1887 * vec.c (vec_prefix::calculate_allocation): Move as
1888 inline variant to vec.h.
1889 (vec_prefix::calculate_allocation_1): New out-of-line version.
1890 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1891 (vec_prefix::m_has_auto_buf): Rename to ...
1892 (vec_prefix::m_using_auto_storage): ... this.
1893 (vec_prefix::calculate_allocation): Inline the easy cases
1894 and dispatch to calculate_allocation_1 which doesn't need the
1895 prefix address.
1896 (va_heap::reserve): Use gcc_checking_assert.
1897 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1898 m_using_auto_storage.
1899 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1900 member and adjust.
1901 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1902 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1903 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1904
1905 2014-02-12 Richard Biener <rguenther@suse.de>
1906
1907 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1908 when we found a dependence.
1909
1910 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1911
1912 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1913 common code...
1914 (maybe_fold_stmt): ... into this new function.
1915 * omp-low.c (lower_omp): Update comment.
1916
1917 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1918 last use.
1919
1920 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1921 dereference.
1922
1923 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1924
1925 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1926 identifiers in comments.
1927 (cortexa53_extra_costs): Likewise.
1928 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1929 (cortexa7_extra_costs): Likewise.
1930 (cortexa12_extra_costs): Likewise.
1931 (cortexa15_extra_costs): Likewise.
1932 (v7m_extra_costs): Likewise.
1933
1934 2014-02-12 Richard Biener <rguenther@suse.de>
1935
1936 PR middle-end/60092
1937 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1938 of posix_memalign being successful.
1939 (lower_stmt): Restrict lowering of posix_memalign to when
1940 -ftree-bit-ccp is enabled.
1941
1942 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1943
1944 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1945 arg_loc.
1946 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1947
1948 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1949
1950 PR rtl-optimization/60116
1951 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1952 other_insn once the combination has been validated.
1953
1954 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1955
1956 PR lto/59468
1957 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1958 and wrapper.
1959 * ipa-devirt.c: Include demangle.h
1960 (odr_violation_reported): New static variable.
1961 (add_type_duplicate): Update odr_violations.
1962 (maybe_record_node): Add completep parameter; update it.
1963 (record_target_from_binfo): Add COMPLETEP parameter;
1964 update it as needed.
1965 (possible_polymorphic_call_targets_1): Likewise.
1966 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1967 rename FINAL to COMPLETE.
1968 (record_targets_from_bases): Sanity check we found the binfo;
1969 fix COMPLETEP updating.
1970 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1971 parameter, fix computing of COMPLETEP.
1972 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1973 at LTO time do demangling.
1974 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1975 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1976 parameter.
1977 (gimple_get_virt_method_for_binfo): Likewise.
1978 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1979 gimple_get_virt_method_for_vtable): Update prototypes.
1980
1981 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1982
1983 PR target/49008
1984 * genautomata.c (add_presence_absence): Fix typo with
1985 {final_}presence_list.
1986
1987 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1988
1989 PR target/60137
1990 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1991 for VSX/Altivec vectors that land in GPR registers.
1992
1993 2014-02-11 Richard Henderson <rth@redhat.com>
1994 Jakub Jelinek <jakub@redhat.com>
1995
1996 PR debug/59776
1997 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1998 around drhs if type conversion to lacc->type is not useless.
1999
2000 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2001
2002 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2003 tuning struct.
2004 (cortex-a57.cortex-a53): Likewise.
2005 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2006
2007 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2008
2009 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2010 arm_restrict_it.
2011
2012 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2013
2014 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2015 add_options_for_arm_vfp3.
2016
2017 2014-02-11 Jeff Law <law@redhat.com>
2018
2019 PR middle-end/54041
2020 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2021 object with an undesirable mode.
2022
2023 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2024
2025 PR libgomp/60107
2026 * config/i386/sol2-9.h: New file.
2027 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2028 *-*-solaris2.9*): Use it.
2029
2030 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2031
2032 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2033 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2034
2035 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2036
2037 * config/microblaze/microblaze.c: Extend mcpu version format
2038
2039 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2040
2041 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2042
2043 2014-02-10 Richard Henderson <rth@redhat.com>
2044
2045 PR target/59927
2046 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2047 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2048 ms-abi vs -mno-accumulate-outgoing-args.
2049 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2050 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2051 respect to ms-abi.
2052
2053 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2054
2055 PR middle-end/60080
2056 * cfgexpand.c (expand_asm_operands): Attach source location to
2057 ASM_INPUT rtx objects.
2058 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2059
2060 2014-02-10 Nick Clifton <nickc@redhat.com>
2061
2062 * config/mn10300/mn10300.c (popcount): New function.
2063 (mn10300_expand_prologue): Include saved registers in stack usage
2064 count.
2065
2066 2014-02-10 Jeff Law <law@redhat.com>
2067
2068 PR middle-end/52306
2069 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2070 when changing the SET_DEST of a prior insn to avoid an input reload.
2071
2072 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2073
2074 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2075 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2076 -mcall-openbsd, or -mcall-linux.
2077 (CC1_ENDIAN_BIG_SPEC): Remove.
2078 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2079 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2080 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2081 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2082 and %cc1_endian_default.
2083 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2084
2085 2014-02-10 Richard Biener <rguenther@suse.de>
2086
2087 PR tree-optimization/60115
2088 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2089 MEM_REF handling. Properly verify that the accesses are not
2090 out of the objects bound.
2091
2092 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2093
2094 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2095 coretex to cortex.
2096
2097 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2098
2099 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2100 proper constants and fix formatting.
2101 (possible_polymorphic_call_targets): Fix formatting.
2102
2103 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2104 Ilya Tocar <ilya.tocar@intel.com>
2105
2106 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2107 (_mm512_loadu_epi32): Renamed into...
2108 (_mm512_loadu_si512): This.
2109 (_mm512_storeu_epi32): Renamed into...
2110 (_mm512_storeu_si512): This.
2111 (_mm512_maskz_ceil_ps): Removed.
2112 (_mm512_maskz_ceil_pd): Ditto.
2113 (_mm512_maskz_floor_ps): Ditto.
2114 (_mm512_maskz_floor_pd): Ditto.
2115 (_mm512_floor_round_ps): Ditto.
2116 (_mm512_floor_round_pd): Ditto.
2117 (_mm512_ceil_round_ps): Ditto.
2118 (_mm512_ceil_round_pd): Ditto.
2119 (_mm512_mask_floor_round_ps): Ditto.
2120 (_mm512_mask_floor_round_pd): Ditto.
2121 (_mm512_mask_ceil_round_ps): Ditto.
2122 (_mm512_mask_ceil_round_pd): Ditto.
2123 (_mm512_maskz_floor_round_ps): Ditto.
2124 (_mm512_maskz_floor_round_pd): Ditto.
2125 (_mm512_maskz_ceil_round_ps): Ditto.
2126 (_mm512_maskz_ceil_round_pd): Ditto.
2127 (_mm512_expand_pd): Ditto.
2128 (_mm512_expand_ps): Ditto.
2129 * config/i386/i386.c (ix86_builtins): Remove
2130 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2131 (bdesc_args): Ditto.
2132 * config/i386/predicates.md (const1256_operand): New.
2133 (const_1_to_2_operand): Ditto.
2134 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2135 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2136 (*avx512pf_gatherpf<mode>sf): Ditto.
2137 (avx512pf_gatherpf<mode>df): Ditto.
2138 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2139 (*avx512pf_gatherpf<mode>df): Ditto.
2140 (avx512pf_scatterpf<mode>sf): Ditto.
2141 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2142 (*avx512pf_scatterpf<mode>sf): Ditto.
2143 (avx512pf_scatterpf<mode>df): Ditto.
2144 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2145 (*avx512pf_scatterpf<mode>df): Ditto.
2146 (avx512f_expand<mode>): Removed.
2147 (<shift_insn><mode>3<mask_name>): Change predicate type.
2148
2149 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2150
2151 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2152 not at the end of datarefs vector use ordered_remove to avoid
2153 reordering datarefs vector.
2154
2155 PR c/59984
2156 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2157 mark local addressable non-static vars as GOVD_PRIVATE
2158 instead of GOVD_LOCAL.
2159 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2160 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2161 than copying them.
2162
2163 PR middle-end/60092
2164 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2165 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2166 assume_aligned or alloc_align attributes.
2167 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2168 arguments. Handle also assume_aligned and alloc_align attributes.
2169 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2170 calls to functions with assume_aligned or alloc_align attributes.
2171 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2172
2173 2014-02-08 Terry Guo <terry.guo@arm.com>
2174
2175 * doc/invoke.texi: Document ARM -march=armv7e-m.
2176
2177 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2178
2179 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2180 flag on __cilkrts_rethrow builtin.
2181
2182 PR ipa/60026
2183 * ipa-cp.c (determine_versionability): Fail at -O0
2184 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2185 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2186
2187 Revert:
2188 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2189
2190 PR ipa/60026
2191 * tree-inline.c (copy_forbidden): Fail for
2192 __attribute__((optimize (0))) functions.
2193
2194 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2195
2196 * varpool.c: Include pointer-set.h.
2197 (varpool_remove_unreferenced_decls): Variables in other partitions
2198 will not be output; be however careful to not lose information
2199 about partitioning.
2200
2201 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2202
2203 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2204 lookup in the vtable constructor.
2205
2206 2014-02-07 Jeff Law <law@redhat.com>
2207
2208 PR target/40977
2209 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2210 define_insn_and_split.
2211
2212 * ipa-inline.c (inline_small_functions): Fix typos.
2213
2214 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2215
2216 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2217 (s390_can_use_return_insn): Declare.
2218 * config/s390/s390.h (EPILOGUE_USES): Define.
2219 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2220 instructions.
2221 (s390_chunkify_start): Handle return JUMP_LABELs.
2222 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2223 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2224 (s390_can_use_return_insn): New functions.
2225 (s390_fix_long_loop_prediction): Handle conditional returns.
2226 (TARGET_SET_UP_BY_PROLOGUE): Define.
2227 * config/s390/s390.md (ANY_RETURN): New code iterator.
2228 (*creturn, *csimple_return, return, simple_return): New patterns.
2229
2230 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2231
2232 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2233 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2234 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2235 REG_CFA_RESTORE list when deciding not to restore a register.
2236
2237 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2238
2239 * config/s390/s390.c: Include tree-pass.h and context.h.
2240 (s390_early_mach): New function, split out from...
2241 (s390_emit_prologue): ...here.
2242 (pass_data_s390_early_mach): New pass structure.
2243 (pass_s390_early_mach): New class.
2244 (s390_option_override): Create and register early_mach pass.
2245 Move to end of file.
2246
2247 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2248
2249 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2250 to match for the exit block.
2251
2252 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2253
2254 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2255 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2256 Reject misaligned operands.
2257
2258 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2259
2260 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2261
2262 2014-02-07 Richard Biener <rguenther@suse.de>
2263
2264 PR middle-end/60092
2265 * gimple-low.c (lower_builtin_posix_memalign): New function.
2266 (lower_stmt): Call it to lower posix_memalign in a way
2267 to make alignment info accessible.
2268
2269 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2270
2271 PR c++/60082
2272 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2273 __builtin_setjmp_receiver.
2274
2275 2014-02-07 Richard Biener <rguenther@suse.de>
2276
2277 PR middle-end/60092
2278 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2279 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2280 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2281 Handle BUILT_IN_POSIX_MEMALIGN.
2282 (find_func_clobbers): Likewise.
2283 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2284 (call_may_clobber_ref_p_1): Likewise.
2285
2286 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2287
2288 PR ipa/59918
2289 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2290 sanity check.
2291
2292 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2293
2294 PR ipa/59469
2295 * lto-cgraph.c (lto_output_node): Use
2296 symtab_get_symbol_partitioning_class.
2297 (lto_output_varpool_node): likewise.
2298 (symtab_get_symbol_partitioning_class): Move here from
2299 lto/lto-partition.c
2300 * cgraph.h (symbol_partitioning_class): Likewise.
2301 (symtab_get_symbol_partitioning_class): Declare.
2302
2303 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2304
2305 * ggc.h (ggc_internal_cleared_alloc): New macro.
2306 * vec.h (vec_safe_copy): Handle memory stats.
2307 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2308 * target-globals.c (save_target_globals): Likewise.
2309
2310 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2311
2312 PR target/60077
2313 * expr.c (emit_move_resolve_push): Export; be bit more selective
2314 on when to clear alias set.
2315 * expr.h (emit_move_resolve_push): Declare.
2316 * function.h (struct function): Add tail_call_marked.
2317 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2318 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2319 * config/i386/i386.md (TImode move expander): De not call
2320 ix86_expand_push.
2321 (FP push expanders): Preserve memory attributes.
2322 * config/i386/sse.md (push<mode>1): Remove.
2323 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2324 (ix86_expand_push): Remove.
2325 * config/i386/mmx.md (push<mode>1): Remove.
2326
2327 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2328
2329 PR rtl-optimization/60030
2330 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2331 lopart with paradoxical subreg before shifting it up by hprec.
2332
2333 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2334
2335 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2336 Remove extra newline at end of file.
2337 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2338 (arm_issue_rate): Handle cortexa57.
2339 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2340 (cortex-a57.cortex-a53): Likewise.
2341
2342 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2343
2344 PR target/59575
2345 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2346 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2347 bitmask.
2348 (arm_expand_prologue): Adjust all callers.
2349 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2350 info, registers also at the lowest numbered registers side. Use
2351 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2352 XEXP.
2353
2354 PR debug/59992
2355 * var-tracking.c (adjust_mems): Before adding a SET to
2356 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2357
2358 2014-02-06 Alan Modra <amodra@gmail.com>
2359
2360 PR target/60032
2361 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2362 change SDmode to DDmode when lra_in_progress.
2363
2364 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2365
2366 PR middle-end/59150
2367 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2368 free_data_ref on the dr first, and before goto again also set dr
2369 to the next dr. For simd_lane_access, free old datarefs[i] before
2370 overwriting it. For get_vectype_for_scalar_type failure, don't
2371 free_data_ref if simd_lane_access.
2372
2373 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2374
2375 PR target/60062
2376 * tree.h (opts_for_fn): New inline function.
2377 (opt_for_fn): Define.
2378 * config/i386/i386.c (ix86_function_regparm): Use
2379 opt_for_fn (decl, optimize) instead of optimize.
2380
2381 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2382
2383 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2384 for SYMBOL_REF in large memory model.
2385
2386 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2387
2388 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2389 and crypto support.
2390 (cortex-a57): Likewise.
2391 (cortex-a57.cortex-a53): Likewise.
2392
2393 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2394 Kugan Vivekanandarajah <kuganv@linaro.org>
2395
2396 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2397 unaligned_access.
2398 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2399
2400 2014-02-06 Richard Biener <rguenther@suse.de>
2401
2402 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2403 set_loop_copy and initialize_original_copy_tables.
2404
2405 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2406
2407 * config/aarch64/aarch64-simd.md
2408 (aarch64_ashr_simddi): Change QI to SI.
2409
2410 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2411 Jakub Jelinek <jakub@redhat.com>
2412
2413 PR middle-end/60013
2414 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2415 of the dataflow.
2416
2417 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2418
2419 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2420 CODE_FOR_altivec_vpku[hw]um to
2421 CODE_FOR_altivec_vpku[hw]um_direct.
2422 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2423 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2424 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2425 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2426
2427 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2428
2429 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2430 generation for -maltivec=be.
2431 (altivec_vsumsws): Simplify redundant test.
2432
2433 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2434
2435 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2436 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2437 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2438 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2439 gen_altivec_vpkuwum.
2440 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2441 BYTES_BIG_ENDIAN.
2442 (altivec_vpks<VI_char>ss): Likewise.
2443 (altivec_vpks<VI_char>us): Likewise.
2444 (altivec_vpku<VI_char>us): Likewise.
2445 (altivec_vpku<VI_char>um): Likewise.
2446 (altivec_vpku<VI_char>um_direct): New (copy of
2447 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2448 internal use).
2449 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2450 target is little endian and -maltivec=be is not specified.
2451 (*altivec_vupkhs<VU_char>_direct): New (copy of
2452 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2453 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2454 target is little endian and -maltivec=be is not specified.
2455 (*altivec_vupkls<VU_char>_direct): New (copy of
2456 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2457 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2458 little endian and -maltivec=be is not specified.
2459 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2460 little endian and -maltivec=be is not specified.
2461
2462 2014-02-05 Richard Henderson <rth@redhat.com>
2463
2464 PR debug/52727
2465 * combine-stack-adj.c: Revert r206943.
2466 * sched-int.h (struct deps_desc): Add last_args_size.
2467 * sched-deps.c (init_deps): Initialize it.
2468 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2469 contain REG_ARGS_SIZE notes.
2470
2471 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2472
2473 * lto-cgraph.c (asm_nodes_output): Make global.
2474 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2475 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2476 (driver_handle_option): Handle OPT_fwpa.
2477
2478 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2479
2480 PR ipa/59947
2481 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2482 a comment typo and formatting issue. If odr_hash hasn't been
2483 created, return vNULL and set *completep to false.
2484
2485 PR middle-end/57499
2486 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2487 bb with no successors.
2488
2489 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2490
2491 PR target/59718
2492 * doc/invoke.texi (-march): Clarify documentation for ARM.
2493 (-mtune): Likewise.
2494 (-mcpu): Likewise.
2495
2496 2014-02-05 Richard Biener <rguenther@suse.de>
2497
2498 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2499 when not vectorizing because of too many alias checks.
2500 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2501 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2502
2503 2014-02-05 Nick Clifton <nickc@redhat.com>
2504
2505 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2506 accept extended registers in any mode when compiling for the MN10300.
2507
2508 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2509
2510 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2511 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2512 sanitization attributes.
2513 (can_inline_edge_p): Likewise.
2514 (sanitize_attrs_match_for_inline_p): New function.
2515
2516 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2517
2518 * ipa-prop.c (detect_type_change): Shor circuit testing of
2519 type changes on THIS pointer.
2520
2521 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2522
2523 PR target/59777
2524 * config/pa/pa.c (legitimize_tls_address): Return original address
2525 if not passed a SYMBOL_REF rtx.
2526 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2527 addresses.
2528 (pa_emit_move_sequence): Simplify TLS source operands.
2529 (pa_legitimate_constant_p): Reject all TLS constants.
2530 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2531 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2532
2533 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2534
2535 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2536 groups when we know they are controlled by LTO.
2537 * varasm.c (default_binds_local_p_1): If object is in other partition,
2538 it will be resolved locally.
2539
2540 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2541
2542 * config/host-linux.c (linux_gt_pch_use_address): Don't
2543 use SSIZE_MAX because it is not always defined.
2544
2545 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2546
2547 PR bootstrap/59913
2548 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2549 threshold for pseudo splitting.
2550 (update_ebb_live_info): Process call argument hard registers and
2551 hard registers from insn definition too.
2552 (max_small_class_regs_num): New constant.
2553 (inherit_in_ebb): Update live hard regs through EBBs. Update
2554 reloads_num only for small register classes. Don't split for
2555 outputs of jumps.
2556
2557 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2558
2559 PR ipa/60058
2560 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2561 is non-null.
2562
2563 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2564
2565 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2566 visibility is safe.
2567
2568 2014-02-04 Marek Polacek <polacek@redhat.com>
2569
2570 * gdbinit.in (pel): Define.
2571
2572 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2573
2574 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2575 behavior.
2576
2577 2014-02-04 Richard Biener <rguenther@suse.de>
2578
2579 PR lto/59723
2580 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2581 in function context local.
2582 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2583 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2584 similar to LTO_imported_decl_ref.
2585
2586 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2587
2588 PR tree-optimization/60002
2589 * cgraphclones.c (build_function_decl_skip_args): Clear
2590 DECL_LANG_SPECIFIC.
2591
2592 PR tree-optimization/60023
2593 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2594 false to gsi_replace.
2595 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2596 has been in some EH region and vec_stmt could throw, add
2597 vec_stmt into the same EH region.
2598 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2599 has no lhs, ignore it.
2600 * internal-fn.c (expand_MASK_LOAD): Likewise.
2601
2602 PR ipa/60026
2603 * tree-inline.c (copy_forbidden): Fail for
2604 __attribute__((optimize (0))) functions.
2605
2606 PR other/58712
2607 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2608 is set, copy one less argument.
2609 (expand_simd_clones): Don't subtract clone_info->inbranch
2610 from simd_clone_struct_alloc argument.
2611
2612 PR rtl-optimization/57915
2613 * recog.c (simplify_while_replacing): If all unary/binary/relational
2614 operation arguments are constant, attempt to simplify those.
2615
2616 PR middle-end/59261
2617 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2618 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2619
2620 2014-02-04 Richard Biener <rguenther@suse.de>
2621
2622 PR tree-optimization/60012
2623 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2624 TBAA disambiguation to all DDRs.
2625
2626 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2627
2628 PR target/59788
2629 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2630 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2631
2632 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2633
2634 PR ipa/59882
2635 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2636
2637 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2638
2639 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2640 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2641
2642 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2643
2644 PR ipa/59831
2645 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2646 to figure out targets of polymorphic calls with known decl.
2647 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2648 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2649 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2650 (get_polymorphic_call_info): ... here.
2651 (get_polymorphic_call_info_from_invariant): New function.
2652
2653 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2654
2655 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2656 lookup via vtable pointer; check for type consistency
2657 and turn inconsitent facts into UNREACHABLE.
2658 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2659 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2660 type inconsistent querries; return UNREACHABLE instead.
2661
2662 2014-02-03 Richard Henderson <rth@twiddle.net>
2663
2664 PR tree-opt/59924
2665 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2666 already processed this node.
2667 (normalize_one_pred_1): Pass along mark_set.
2668 (normalize_one_pred): Create and destroy a pointer_set_t.
2669 (normalize_one_pred_chain): Likewise.
2670
2671 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2672
2673 PR gcov-profile/58602
2674 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2675
2676 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2677
2678 PR ipa/59831
2679 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2680 -fno-devirtualize; try to devirtualize by the knowledge of
2681 virtual table pointer given by aggregate propagation.
2682 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2683 (ipa_print_node_jump_functions): Dump also offset that
2684 is relevant for polymorphic calls.
2685 (determine_known_aggregate_parts): Add arg_type parameter; use it
2686 instead of determining the type from pointer type.
2687 (ipa_compute_jump_functions_for_edge): Update call of
2688 determine_known_aggregate_parts.
2689 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2690 (gimple_get_virt_method_for_binfo): ... here; simplify using
2691 vtable_pointer_value_to_vtable.
2692 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2693 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2694 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2695 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2696 POINTER_PLUS_EXPR.
2697 (vtable_pointer_value_to_binfo): ... here.
2698 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2699
2700 2014-02-03 Teresa Johnson <tejohnson@google.com>
2701
2702 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2703 redef of outer loop index variable.
2704
2705 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2706
2707 PR c++/53017
2708 PR c++/59211
2709 * doc/extend.texi (Function Attributes): Typo.
2710
2711 2014-02-03 Cong Hou <congh@google.com>
2712
2713 PR tree-optimization/60000
2714 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2715 if the vectorized statement is a store. A store statement can only
2716 appear at the end of pattern statements.
2717
2718 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2719
2720 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2721 (ix86_option_override_internal): Default long double to 64-bit for
2722 32-bit Bionic and to 128-bit for 64-bit Bionic.
2723
2724 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2725 TARGET_LONG_DOUBLE_128 is true.
2726 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2727
2728 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2729 (mlong-double-64): Negate -mlong-double-128.
2730 (mlong-double-128): New option.
2731
2732 * config/i386/i386-c.c (ix86_target_macros): Define
2733 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2734
2735 * doc/invoke.texi: Document -mlong-double-128.
2736
2737 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2738
2739 PR rtl-optimization/60024
2740 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2741
2742 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2743
2744 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2745
2746 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2747
2748 PR rtl-optimization/57662
2749 * sel-sched.c (code_motion_path_driver): Do not mark already not
2750 existing blocks in the visiting bitmap.
2751
2752 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2753
2754 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2755 on the insn being emitted.
2756
2757 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2758 Will Deacon <will.deacon@arm.com>
2759
2760 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2761
2762 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2763
2764 * config/arm/arm-tables.opt: Regenerate.
2765
2766 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2767
2768 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2769 for vector types other than V16QImode.
2770 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2771 define_expand, and call altivec_expand_vec_perm_le when producing
2772 code with little endian element order.
2773 (*altivec_vperm_<mode>_internal): New insn having previous
2774 behavior of altivec_vperm_<mode>.
2775 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2776 altivec_expand_vec_perm_le when producing code with little endian
2777 element order.
2778 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2779 behavior of altivec_vperm_<mode>_uns.
2780
2781 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2782
2783 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2784 (altivec_vsumsws): Add handling for -maltivec=be with a little
2785 endian target.
2786 (altivec_vsumsws_direct): New.
2787 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2788 gen_altivec_vsumsws.
2789
2790 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2791
2792 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2793 vtable_pointer_value_to_binfo): New functions.
2794 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2795 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2796
2797 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2798
2799 * config/nios2/nios2.md (load_got_register): Initialize GOT
2800 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2801 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2802
2803 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2804
2805 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2806 preserverd by passthrough, do not propagate the type.
2807
2808 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2809
2810 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2811 (mips_atomic_assign_expand_fenv): New function.
2812 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2813
2814 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2815
2816 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2817 (__builtin_mips_set_fcsr): Likewise.
2818 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2819 MIPS_USI_FTYPE_VOID.
2820 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2821 (mips16_expand_set_fcsr): Likewise.
2822 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2823 (mips16_set_fcsr_stub): Likewise.
2824 (mips16_get_fcsr_one_only_stub): New class.
2825 (mips16_set_fcsr_one_only_stub): Likewise.
2826 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2827 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2828 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2829 (hard_float): New availability predicate.
2830 (mips_builtins): Add get_fcsr and set_fcsr.
2831 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2832 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2833 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2834 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2835 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2836 patterns.
2837
2838 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2839
2840 * config/mips/mips.c (mips_one_only_stub): New class.
2841 (mips_need_mips16_rdhwr_p): Replace with...
2842 (mips16_rdhwr_stub): ...this new variable.
2843 (mips16_stub_call_address): New function.
2844 (mips16_rdhwr_one_only_stub): New class.
2845 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2846 (mips_output_mips16_rdhwr): Delete.
2847 (mips_finish_stub): New function.
2848 (mips_code_end): Use it to handle rdhwr stubs.
2849
2850 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2851
2852 PR target/60017
2853 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2854 when calculating size of integer atomic types.
2855
2856 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2857
2858 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2859
2860 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2861
2862 PR tree-optimization/60003
2863 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2864 * profile.c (branch_prob): Use gimple_call_builtin_p
2865 to check for BUILT_IN_SETJMP_RECEIVER.
2866 * tree-inline.c (copy_bb): Call notice_special_calls.
2867
2868 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2869
2870 PR bootstrap/59985
2871 * lra-constraints.c (process_alt_operands): Update reload_sum only
2872 on the first pass.
2873
2874 2014-01-31 Richard Henderson <rth@redhat.com>
2875
2876 PR middle-end/60004
2877 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2878 until after else_eh is processed.
2879
2880 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2881
2882 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2883 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2884 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2885 in smmintrin.h, remove them.
2886 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2887 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2888 * config/i386/i386.md (ROUND_SAE): Fix value.
2889 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2890 (const48_operand): New.
2891 * config/i386/subst.md (round), (round_expand): Use
2892 const_4_or_8_to_11_operand.
2893 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2894
2895 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2896
2897 * config/i386/constraints.md (Yk): Swap meaning with k.
2898 * config/i386/i386.md (movhi_internal): Change Yk to k.
2899 (movqi_internal): Ditto.
2900 (*k<logic><mode>): Ditto.
2901 (*andhi_1): Ditto.
2902 (*andqi_1): Ditto.
2903 (kandn<mode>): Ditto.
2904 (*<code>hi_1): Ditto.
2905 (*<code>qi_1): Ditto.
2906 (kxnor<mode>): Ditto.
2907 (kortestzhi): Ditto.
2908 (kortestchi): Ditto.
2909 (kunpckhi): Ditto.
2910 (*one_cmplhi2_1): Ditto.
2911 (*one_cmplqi2_1): Ditto.
2912 * config/i386/sse.md (): Change k to Yk.
2913 (avx512f_load<mode>_mask): Ditto.
2914 (avx512f_blendm<mode>): Ditto.
2915 (avx512f_store<mode>_mask): Ditto.
2916 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2917 (avx512f_storedqu<mode>_mask): Ditto.
2918 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2919 Ditto.
2920 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2921 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2922 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2923 (avx512f_maskcmp<mode>3): Ditto.
2924 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2925 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2926 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2927 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2928 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2929 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2930 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2931 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2932 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2933 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2934 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2935 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2936 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2937 (vec_extract_lo_<mode>_maskm): Ditto.
2938 (vec_extract_hi_<mode>_maskm): Ditto.
2939 (avx512f_vternlog<mode>_mask): Ditto.
2940 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2941 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2942 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2943 (avx512f_<code>v8div16qi2_mask): Ditto.
2944 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2945 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2946 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2947 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2948 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2949 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2950 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2951 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2952 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2953 (avx512cd_maskb_vec_dupv8di): Ditto.
2954 (avx512cd_maskw_vec_dupv16si): Ditto.
2955 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2956 (avx512f_vpermi2var<mode>3_mask): Ditto.
2957 (avx512f_vpermi2var<mode>3_mask): Ditto.
2958 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2959 (*avx512f_gathersi<mode>): Ditto.
2960 (*avx512f_gathersi<mode>_2): Ditto.
2961 (*avx512f_gatherdi<mode>): Ditto.
2962 (*avx512f_gatherdi<mode>_2): Ditto.
2963 (*avx512f_scattersi<mode>): Ditto.
2964 (*avx512f_scatterdi<mode>): Ditto.
2965 (avx512f_compress<mode>_mask): Ditto.
2966 (avx512f_compressstore<mode>_mask): Ditto.
2967 (avx512f_expand<mode>_mask): Ditto.
2968 * config/i386/subst.md (mask): Change k to Yk.
2969 (mask_scalar_merge): Ditto.
2970 (sd): Ditto.
2971
2972 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2973
2974 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2975
2976 2014-01-31 Richard Biener <rguenther@suse.de>
2977
2978 PR middle-end/59990
2979 * builtins.c (fold_builtin_memory_op): Make sure to not
2980 use a floating-point mode or a boolean or enumeral type for
2981 the copy operation.
2982
2983 2014-01-30 DJ Delorie <dj@redhat.com>
2984
2985 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2986 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2987 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2988 whenever main() has an epilogue.
2989
2990 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2991
2992 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2993 unused variable "field".
2994 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2995 (vsx_mergeh_<mode>): Likewise.
2996 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2997 (altivec_vmrghh): Likewise.
2998 (altivec_vmrghw): Likewise.
2999 (altivec_vmrglb): Likewise.
3000 (altivec_vmrglh): Likewise.
3001 (altivec_vmrglw): Likewise.
3002 (altivec_vspltb): Add missing uses.
3003 (altivec_vsplth): Likewise.
3004 (altivec_vspltw): Likewise.
3005 (altivec_vspltsf): Likewise.
3006
3007 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3008
3009 PR target/59923
3010 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3011 frame related instructions.
3012
3013 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3014
3015 PR rtl-optimization/59959
3016 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3017 any reload of register whose subreg is invalid.
3018
3019 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3020
3021 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3022 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3023 Add missing return type - void.
3024
3025 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3026
3027 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3028 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3029 remove element index adjustment for endian (now handled in vsx.md
3030 and altivec.md).
3031 (altivec_expand_vec_perm_const): Use
3032 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3033 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3034 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3035 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3036 define_expand and a new define_insn *altivec_vspltb_internal;
3037 adjust for -maltivec=be on a little endian target.
3038 (altivec_vspltb_direct): New.
3039 (altivec_vsplth): Divide into a define_expand and a new
3040 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3041 little endian target.
3042 (altivec_vsplth_direct): New.
3043 (altivec_vspltw): Divide into a define_expand and a new
3044 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3045 little endian target.
3046 (altivec_vspltw_direct): New.
3047 (altivec_vspltsf): Divide into a define_expand and a new
3048 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3049 a little endian target.
3050
3051 2014-01-30 Richard Biener <rguenther@suse.de>
3052
3053 PR tree-optimization/59993
3054 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3055 can propagate form the earlier stmt and avoid the transform
3056 when the intermediate result is needed.
3057
3058 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3059
3060 * README.Portability: Fix typo.
3061
3062 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3063
3064 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3065 comparison_operator with ordered_comparison_operator.
3066
3067 2014-01-30 Nick Clifton <nickc@redhat.com>
3068
3069 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3070 Rename to mn10300_store_multiple_regs.
3071 * config/mn10300/mn10300.c: Likewise.
3072 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3073 store_multiple_regs.
3074 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3075 Call mn10300_store_multiple_regs.
3076
3077 2014-01-30 Nick Clifton <nickc@redhat.com>
3078 DJ Delorie <dj@redhat.com>
3079
3080 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3081 %fp 2 to keep registers after it properly word-aligned.
3082 (rl78_alloc_physical_registers_umul): Handle the case where both
3083 input operands are the same.
3084
3085 2014-01-30 Richard Biener <rguenther@suse.de>
3086
3087 PR tree-optimization/59903
3088 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3089 check properly.
3090
3091 2014-01-30 Jason Merrill <jason@redhat.com>
3092
3093 PR c++/59633
3094 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3095
3096 PR c++/59645
3097 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3098
3099 2014-01-30 Richard Biener <rguenther@suse.de>
3100
3101 PR tree-optimization/59951
3102 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3103
3104 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3105
3106 PR target/59784
3107 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3108 SFmode to DFmode case.
3109
3110 2014-01-29 DJ Delorie <dj@redhat.com>
3111
3112 * config/msp430/msp430.opt (-minrt): New.
3113 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3114 if -minrt given.
3115 (ENDFILE_SPEC): Likewise.
3116
3117 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3118
3119 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3120 (estimate_function_body_sizes): Use it.
3121
3122 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3123
3124 PR c++/58561
3125 * dwarf2out.c (is_cxx_auto): New.
3126 (is_base_type): Use it.
3127 (gen_type_die_with_usage): Likewise.
3128
3129 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3130
3131 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3132 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3133 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3134 -maltivec=be with LE targets.
3135 (vsx_mergeh_<mode>): Likewise.
3136 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3137 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3138 (altivec_vmrghb): Replace with define_expand and new
3139 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3140 (altivec_vmrghb_direct): New define_insn.
3141 (altivec_vmrghh): Replace with define_expand and new
3142 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3143 (altivec_vmrghh_direct): New define_insn.
3144 (altivec_vmrghw): Replace with define_expand and new
3145 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3146 (altivec_vmrghw_direct): New define_insn.
3147 (*altivec_vmrghsf): Adjust for endianness.
3148 (altivec_vmrglb): Replace with define_expand and new
3149 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3150 (altivec_vmrglb_direct): New define_insn.
3151 (altivec_vmrglh): Replace with define_expand and new
3152 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3153 (altivec_vmrglh_direct): New define_insn.
3154 (altivec_vmrglw): Replace with define_expand and new
3155 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3156 (altivec_vmrglw_direct): New define_insn.
3157 (*altivec_vmrglsf): Adjust for endianness.
3158 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3159 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3160 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3161 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3162 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3163 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3164 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3165 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3166
3167 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3168
3169 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3170 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3171 whitespace.
3172
3173 2014-01-29 Richard Biener <rguenther@suse.de>
3174
3175 PR tree-optimization/58742
3176 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3177 associate_pointerplus_align.
3178 (associate_pointerplus_diff): New function.
3179 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3180 and associate_pointerplus_diff.
3181
3182 2014-01-29 Richard Biener <rguenther@suse.de>
3183
3184 * lto-streamer.h (LTO_major_version): Bump to 3.
3185 (LTO_minor_version): Reset to 0.
3186
3187 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3188
3189 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3190 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3191 (arm_file_start): Generate correct asm header for armv7ve.
3192 * config/arm/bpabi.h: Add multilib support for armv7ve.
3193 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3194 and cortex-a15 to armv7ve.
3195 * config/arm/t-aprofile: Add multilib support for armv7ve.
3196 * doc/invoke.texi: Document -march=armv7ve.
3197
3198 2014-01-29 Richard Biener <rguenther@suse.de>
3199
3200 PR tree-optimization/58742
3201 * tree-ssa-forwprop.c (associate_plusminus): Return true
3202 if we changed sth, defer EH cleanup to ...
3203 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3204 (simplify_mult): New function.
3205
3206 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3207
3208 PR middle-end/59917
3209 PR tree-optimization/59920
3210 * tree.c (build_common_builtin_nodes): Remove
3211 __builtin_setjmp_dispatcher initialization.
3212 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3213 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3214 instead of gsi_after_labels + manually skipping debug stmts.
3215 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3216 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3217 * tree-inline.c (copy_edges_for_bb): Remove
3218 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3219 argument. Ignore computed_goto_p stmts. Don't call
3220 make_abnormal_goto_edges. If a call might need abnormal edges
3221 for non-local gotos, see if it already has an edge to
3222 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3223 with true argument, don't do anything then, otherwise add
3224 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3225 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3226 caller.
3227 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3228 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3229 (lower_stmt): Don't set data->calls_builtin_setjmp.
3230 (lower_builtin_setjmp): Adjust comment.
3231 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3232 * tree-cfg.c (found_computed_goto): Remove.
3233 (factor_computed_gotos): Remove.
3234 (make_goto_expr_edges): Return bool, true for computed gotos.
3235 Don't call make_abnormal_goto_edges.
3236 (build_gimple_cfg): Don't set found_computed_goto, don't call
3237 factor_computed_gotos.
3238 (computed_goto_p): No longer static.
3239 (make_blocks): Don't set found_computed_goto.
3240 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3241 (make_edges): If make_goto_expr_edges returns true, push bb
3242 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3243 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3244 vector. Record mapping between bbs and OpenMP regions if there
3245 are any, adjust make_gimple_omp_edges caller. Call
3246 handle_abnormal_edges.
3247 (make_abnormal_goto_edges): Remove.
3248 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3249 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3250 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3251 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3252 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3253 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3254 filling *region also set *region_idx to (*region)->entry->index.
3255
3256 PR other/58712
3257 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3258 For REGs set ORIGINAL_REGNO.
3259
3260 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3261
3262 * doc/md.texi: Mention that a target shouldn't implement
3263 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3264 than hi/lo pair.
3265
3266 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3267
3268 PR tree-optimization/59594
3269 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3270 a copy of the datarefs vector rather than the vector itself.
3271
3272 2014-01-28 Jason Merrill <jason@redhat.com>
3273
3274 PR c++/53756
3275 * dwarf2out.c (auto_die): New static.
3276 (gen_type_die_with_usage): Handle C++1y 'auto'.
3277 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3278 on definition.
3279
3280 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3281
3282 PR target/59672
3283 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3284 (SPEC_X32): Likewise.
3285 (SPEC_64): Likewise.
3286 * config/i386/i386.c (ix86_option_override_internal): Turn off
3287 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3288 for TARGET_16BIT.
3289 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3290 * config/i386/i386.h (TARGET_16BIT): New macro.
3291 (TARGET_16BIT_P): Likewise.
3292 * config/i386/i386.opt: Add m16.
3293 * doc/invoke.texi: Document -m16.
3294
3295 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3296
3297 PR preprocessor/59935
3298 * input.c (location_get_source_line): Bail out on when line number
3299 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3300
3301 2014-01-28 Richard Biener <rguenther@suse.de>
3302
3303 PR tree-optimization/58742
3304 * tree-ssa-forwprop.c (associate_plusminus): Handle
3305 pointer subtraction of the form (T)(P + A) - (T)P.
3306
3307 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3308
3309 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3310 at const_int_cost.
3311
3312 2014-01-28 Richard Biener <rguenther@suse.de>
3313
3314 Revert
3315 2014-01-28 Richard Biener <rguenther@suse.de>
3316
3317 PR rtl-optimization/45364
3318 PR rtl-optimization/59890
3319 * var-tracking.c (local_get_addr_clear_given_value): Handle
3320 already cleared slot.
3321 (val_reset): Handle not allocated local_get_addr_cache.
3322 (vt_find_locations): Use post-order on the inverted CFG.
3323
3324 2014-01-28 Richard Biener <rguenther@suse.de>
3325
3326 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3327
3328 2014-01-28 Richard Biener <rguenther@suse.de>
3329
3330 PR rtl-optimization/45364
3331 PR rtl-optimization/59890
3332 * var-tracking.c (local_get_addr_clear_given_value): Handle
3333 already cleared slot.
3334 (val_reset): Handle not allocated local_get_addr_cache.
3335 (vt_find_locations): Use post-order on the inverted CFG.
3336
3337 2014-01-28 Alan Modra <amodra@gmail.com>
3338
3339 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3340 * configure.ac <recursive call for build != host>: Define
3341 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3342 and LD_FOR_BUILD too.
3343 * configure: Regenerate.
3344
3345 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3346
3347 * config/i386/i386.c (get_builtin_code_for_version): Separate
3348 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3349 Broadwell from Haswell.
3350
3351 2014-01-27 Steve Ellcey <sellcey@mips.com>
3352
3353 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3354 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3355 * config/mips/mips.c (mips_option_override): Change setting
3356 of TARGET_DSP.
3357 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3358 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3359 Change from Mask to Var.
3360
3361 2014-01-27 Jeff Law <law@redhat.com>
3362
3363 * ipa-inline.c (inline_small_functions): Fix typo.
3364
3365 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3366
3367 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3368 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3369 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3370 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3371 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3372 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3373 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3374 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3375 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3376 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3377 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3378 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3379 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3380 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3381 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3382 (_mm512_storeu_epi64): Ditto.
3383 (_mm512_cmpge_epi32_mask): Ditto.
3384 (_mm512_cmpge_epu32_mask): Ditto.
3385 (_mm512_cmpge_epi64_mask): Ditto.
3386 (_mm512_cmpge_epu64_mask): Ditto.
3387 (_mm512_cmple_epi32_mask): Ditto.
3388 (_mm512_cmple_epu32_mask): Ditto.
3389 (_mm512_cmple_epi64_mask): Ditto.
3390 (_mm512_cmple_epu64_mask): Ditto.
3391 (_mm512_cmplt_epi32_mask): Ditto.
3392 (_mm512_cmplt_epu32_mask): Ditto.
3393 (_mm512_cmplt_epi64_mask): Ditto.
3394 (_mm512_cmplt_epu64_mask): Ditto.
3395 (_mm512_cmpneq_epi32_mask): Ditto.
3396 (_mm512_cmpneq_epu32_mask): Ditto.
3397 (_mm512_cmpneq_epi64_mask): Ditto.
3398 (_mm512_cmpneq_epu64_mask): Ditto.
3399 (_mm512_expand_pd): Ditto.
3400 (_mm512_expand_ps): Ditto.
3401 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3402 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3403 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3404 * config/i386/i386.c (ix86_builtins): Add
3405 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3406 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3407 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3408 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3409 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3410 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3411 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3412 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3413 IX86_BUILTIN_PMOVUSQW512_MEM.
3414 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3415 __builtin_ia32_pmovsqd512mem_mask,
3416 __builtin_ia32_pmovqd512mem_mask,
3417 __builtin_ia32_pmovusqw512mem_mask,
3418 __builtin_ia32_pmovsqw512mem_mask,
3419 __builtin_ia32_pmovqw512mem_mask,
3420 __builtin_ia32_pmovusdw512mem_mask,
3421 __builtin_ia32_pmovsdw512mem_mask,
3422 __builtin_ia32_pmovdw512mem_mask,
3423 __builtin_ia32_pmovqb512mem_mask,
3424 __builtin_ia32_pmovusqb512mem_mask,
3425 __builtin_ia32_pmovsqb512mem_mask,
3426 __builtin_ia32_pmovusdb512mem_mask,
3427 __builtin_ia32_pmovsdb512mem_mask,
3428 __builtin_ia32_pmovdb512mem_mask.
3429 (bdesc_args): Add __builtin_ia32_expanddf512,
3430 __builtin_ia32_expandsf512.
3431 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3432 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3433 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3434 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3435 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3436 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3437 (avx512f_<code>v8div16qi2_mask_store): This.
3438 (avx512f_expand<mode>): New.
3439
3440 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3441
3442 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3443 New.
3444 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3445 (_mm512_prefetch_i32scatter_pd): Ditto.
3446 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3447 (_mm512_prefetch_i64scatter_pd): Ditto.
3448 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3449 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3450 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3451 (_mm512_prefetch_i32scatter_ps): Ditto.
3452 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3453 (_mm512_prefetch_i64scatter_ps): Ditto.
3454 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3455 * config/i386/i386-builtin-types.def: Define
3456 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3457 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3458 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3459 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3460 IX86_BUILTIN_SCATTERPFQPD.
3461 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3462 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3463 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3464 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3465 __builtin_ia32_scatterpfqps.
3466 (ix86_expand_builtin): Expand new built-ins.
3467 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3468 fix memory access data type.
3469 (*avx512pf_gatherpf<mode>_mask): Ditto.
3470 (*avx512pf_gatherpf<mode>): Ditto.
3471 (avx512pf_scatterpf<mode>): Ditto.
3472 (*avx512pf_scatterpf<mode>_mask): Ditto.
3473 (*avx512pf_scatterpf<mode>): Ditto.
3474 (GATHER_SCATTER_SF_MEM_MODE): New.
3475 (avx512pf_gatherpf<mode>df): Ditto.
3476 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3477 (*avx512pf_scatterpf<mode>df): Ditto.
3478
3479 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3480
3481 PR bootstrap/59934
3482 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3483 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3484 reached.
3485
3486 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3487
3488 * common/config/arm/arm-common.c
3489 (arm_rewrite_mcpu): Handle multiple names.
3490 * config/arm/arm.h
3491 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3492
3493 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3494
3495 * gimple-builder.h (create_gimple_tmp): Delete.
3496
3497 2014-01-27 Christian Bruel <christian.bruel@st.com>
3498
3499 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3500 words comparisons.
3501
3502 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3503
3504 * config/pa/pa.md (call): Generate indirect long calls to non-local
3505 functions when outputing 32-bit code.
3506 (call_value): Likewise except for special call to buggy powf function.
3507
3508 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3509 portable runtime and PIC indirect calls.
3510 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3511 and PIC call sequences. Use ldo instead of blr to set return register
3512 in PIC call sequence.
3513
3514 2014-01-25 Walter Lee <walt@tilera.com>
3515
3516 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3517 avoid clobbering a live register.
3518
3519 2014-01-25 Walter Lee <walt@tilera.com>
3520
3521 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3522 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3523 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3524 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3525
3526 2014-01-25 Walter Lee <walt@tilera.com>
3527
3528 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3529 arguments on even registers.
3530 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3531 STACK_BOUNDARY.
3532 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3533 (BIGGEST_ALIGNMENT): Ditto.
3534 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3535
3536 2014-01-25 Walter Lee <walt@tilera.com>
3537
3538 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3539 insns before bundling.
3540 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3541
3542 2014-01-25 Walter Lee <walt@tilera.com>
3543
3544 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3545 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3546 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3547
3548 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3549
3550 * config/mips/constraints.md (kl): Delete.
3551 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3552 define expands, using...
3553 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3554 instructions for MIPS16.
3555 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3556 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3557
3558 2014-01-25 Walter Lee <walt@tilera.com>
3559
3560 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3561 (clzdi2): Ditto.
3562 (ffsdi2): Ditto.
3563
3564 2014-01-25 Walter Lee <walt@tilera.com>
3565
3566 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3567 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3568
3569 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3570
3571 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3572 Handle XOR.
3573
3574 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3575
3576 * print-rtl.c (in_call_function_usage): New var.
3577 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3578 EXPR_LIST mode as mode and not as reg note name.
3579
3580 PR middle-end/59561
3581 * cfgloopmanip.c (copy_loop_info): If
3582 loop->warned_aggressive_loop_optimizations, make sure
3583 the flag is set in target loop too.
3584
3585 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3586
3587 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3588 flag_cilkplus.
3589 * builtins.def: Likewise.
3590 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3591 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3592 * ira.c (ira_setup_eliminable_regset): Likewise.
3593 * omp-low.c (gate_expand_omp): Likewise.
3594 (execute_lower_omp): Likewise.
3595 (diagnose_sb_0): Likewise.
3596 (gate_diagnose_omp_blocks): Likewise.
3597 (simd_clone_clauses_extract): Likewise.
3598 (gate): Likewise.
3599
3600 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3601
3602 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3603 correction for little endian...
3604 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3605 here.
3606
3607 2014-01-24 Jeff Law <law@redhat.com>
3608
3609 PR tree-optimization/59919
3610 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3611 for non-returning calls.
3612
3613 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3614
3615 * common/config/aarch64/aarch64-common.c
3616 (aarch64_rewrite_mcpu): Handle multiple names.
3617 * config/aarch64/aarch64.h
3618 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3619
3620 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3621
3622 * input.c (add_file_to_cache_tab): Handle the case where fopen
3623 returns NULL.
3624
3625 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3626
3627 PR target/59929
3628 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3629 from push operand if code of push isn't PRE_DEC.
3630
3631 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3632
3633 PR target/59909
3634 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3635 -mquad-memory-atomic. Update -mquad-memory documentation to say
3636 it is only used for non-atomic loads/stores.
3637
3638 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3639 -mquad-memory or -mquad-memory-atomic switches.
3640
3641 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3642 -mquad-memory-atomic to ISA 2.07 support.
3643
3644 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3645 to separate support of normal quad word memory operations (ldq, stq)
3646 from the atomic quad word memory operations.
3647
3648 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3649 support to separate non-atomic quad word operations from atomic
3650 quad word operations. Disable non-atomic quad word operations in
3651 little endian mode so that we don't have to swap words after the
3652 load and before the store.
3653 (quad_load_store_p): Add comment about atomic quad word support.
3654 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3655 options printed with -mdebug=reg.
3656
3657 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3658 -mquad-memory-atomic as the test for whether we have quad word
3659 atomic instructions.
3660 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3661 or -mp8-vector are used, allow byte/half-word atomic operations.
3662
3663 * config/rs6000/sync.md (load_lockedti): Insure that the address
3664 is a proper indexed or indirect address for the lqarx instruction.
3665 On little endian systems, swap the hi/lo registers after the lqarx
3666 instruction.
3667 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3668 insure the address is valid for the lqarx instruction.
3669 (store_conditionalti): Insure that the address is a proper indexed
3670 or indirect address for the stqcrx. instruction. On little endian
3671 systems, swap the hi/lo registers before doing the stqcrx.
3672 instruction.
3673 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3674 insure the address is valid for the stqcrx. instruction.
3675
3676 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3677 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3678 type of quad memory support is available.
3679
3680 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3681
3682 PR regression/59915
3683 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3684 there is a danger of looping.
3685
3686 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3687
3688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3689 force flag_ira_loop_pressure if set via command line.
3690
3691 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3692
3693 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3694 (ashr_simd): New builtin handling DI mode.
3695 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3696 (aarch64_sshr_simddi): New match pattern.
3697 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3698 (vshrd_n_s64): Likewise.
3699 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3700
3701 2014-01-23 Nick Clifton <nickc@redhat.com>
3702
3703 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3704 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3705 favour of mcu specific scripts.
3706 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3707 430x multilibs.
3708
3709 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3710 Alex Velenko <Alex.Velenko@arm.com>
3711
3712 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3713 (vaddv_s16): Likewise.
3714 (vaddv_s32): Likewise.
3715 (vaddv_u8): Likewise.
3716 (vaddv_u16): Likewise.
3717 (vaddv_u32): Likewise.
3718 (vaddvq_s8): Likewise.
3719 (vaddvq_s16): Likewise.
3720 (vaddvq_s32): Likewise.
3721 (vaddvq_s64): Likewise.
3722 (vaddvq_u8): Likewise.
3723 (vaddvq_u16): Likewise.
3724 (vaddvq_u32): Likewise.
3725 (vaddvq_u64): Likewise.
3726 (vaddv_f32): Likewise.
3727 (vaddvq_f32): Likewise.
3728 (vaddvq_f64): Likewise.
3729 (vmaxv_f32): Likewise.
3730 (vmaxv_s8): Likewise.
3731 (vmaxv_s16): Likewise.
3732 (vmaxv_s32): Likewise.
3733 (vmaxv_u8): Likewise.
3734 (vmaxv_u16): Likewise.
3735 (vmaxv_u32): Likewise.
3736 (vmaxvq_f32): Likewise.
3737 (vmaxvq_f64): Likewise.
3738 (vmaxvq_s8): Likewise.
3739 (vmaxvq_s16): Likewise.
3740 (vmaxvq_s32): Likewise.
3741 (vmaxvq_u8): Likewise.
3742 (vmaxvq_u16): Likewise.
3743 (vmaxvq_u32): Likewise.
3744 (vmaxnmv_f32): Likewise.
3745 (vmaxnmvq_f32): Likewise.
3746 (vmaxnmvq_f64): Likewise.
3747 (vminv_f32): Likewise.
3748 (vminv_s8): Likewise.
3749 (vminv_s16): Likewise.
3750 (vminv_s32): Likewise.
3751 (vminv_u8): Likewise.
3752 (vminv_u16): Likewise.
3753 (vminv_u32): Likewise.
3754 (vminvq_f32): Likewise.
3755 (vminvq_f64): Likewise.
3756 (vminvq_s8): Likewise.
3757 (vminvq_s16): Likewise.
3758 (vminvq_s32): Likewise.
3759 (vminvq_u8): Likewise.
3760 (vminvq_u16): Likewise.
3761 (vminvq_u32): Likewise.
3762 (vminnmv_f32): Likewise.
3763 (vminnmvq_f32): Likewise.
3764 (vminnmvq_f64): Likewise.
3765
3766 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3767
3768 * config/aarch64/aarch64-simd.md
3769 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3770 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3771 (*aarch64_mul3_elt<mode>): Likewise.
3772 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3773 (*aarch64_mul3_elt_to_64v2df): Likewise.
3774 (*aarch64_mla_elt<mode>): Likewise.
3775 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3776 (*aarch64_mls_elt<mode>): Likewise.
3777 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3778 (*aarch64_fma4_elt<mode>): Likewise.
3779 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3780 (*aarch64_fma4_elt_to_64v2df): Likewise.
3781 (*aarch64_fnma4_elt<mode>): Likewise.
3782 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3783 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3784 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3785 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3786 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3787 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3788 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3789 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3790 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3791
3792 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3793
3794 * config/aarch64/aarch64-simd.md
3795 (aarch64_be_checked_get_lane<mode>): New define_expand.
3796 * config/aarch64/aarch64-simd-builtins.def
3797 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3798 New builtin definition.
3799 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3800 Use new safe be builtin.
3801
3802 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3803
3804 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3805 New define_insn.
3806 (aarch64_be_st1<mode>): Likewise.
3807 (aarch_ld1<VALL:mode>): Define_expand modified.
3808 (aarch_st1<VALL:mode>): Likewise.
3809 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3810 (UNSPEC_ST1): Likewise.
3811
3812 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3813
3814 * config/microblaze/microblaze.md: Add trap insn and attribute
3815
3816 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3817
3818 PR preprocessor/58580
3819 * input.h (location_get_source_line): Take an additional line_size
3820 parameter.
3821 (void diagnostics_file_cache_fini): Declare new function.
3822 * input.c (struct fcache): New type.
3823 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3824 New static constants.
3825 (diagnostic_file_cache_init, total_lines_num)
3826 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3827 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3828 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3829 (get_next_line, read_next_line, goto_next_line, read_line_num):
3830 New static function definitions.
3831 (diagnostic_file_cache_fini): New function.
3832 (location_get_source_line): Take an additional output line_len
3833 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3834 read_line_num.
3835 * diagnostic.c (diagnostic_finish): Call
3836 diagnostic_file_cache_fini.
3837 (adjust_line): Take an additional input parameter for the length
3838 of the line, rather than calculating it with strlen.
3839 (diagnostic_show_locus): Adjust the use of
3840 location_get_source_line and adjust_line with respect to their new
3841 signature. While displaying a line now, do not stop at the first
3842 null byte. Rather, display the zero byte as a space and keep
3843 going until we reach the size of the line.
3844 * Makefile.in: Add vec.o to OBJS-libcommon
3845
3846 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3847 Ilya Tocar <ilya.tocar@intel.com>
3848
3849 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3850 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3851 (__builtin_ia32_kmov16): Ditto.
3852 * config/i386/i386.md (UNSPEC_KMOV): New.
3853 (kmovw): Ditto.
3854
3855 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3856
3857 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3858 (_mm512_storeu_si512): Ditto.
3859
3860 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3861
3862 PR target/52125
3863 * rtl.h (get_referenced_operands): Declare.
3864 * recog.c (get_referenced_operands): New function.
3865 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3866 operands have been referenced when recording LO_SUM references.
3867
3868 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3869
3870 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3871
3872 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3873
3874 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3875 Enable for generic and recent AMD targets.
3876
3877 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3878
3879 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3880 ARG_SIZE note when adjustment was eliminated.
3881
3882 2014-01-22 Jeff Law <law@redhat.com>
3883
3884 PR tree-optimization/59597
3885 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3886 in file. Accept new argument REGISTERING and use it to modify
3887 dump output appropriately.
3888 (register_jump_thread): Corresponding changes.
3889 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3890 thread paths involving joiner blocks. Add code to dump cancelled
3891 jump threading paths.
3892
3893 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3894
3895 PR rtl-optimization/59477
3896 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3897 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3898
3899 2014-01-22 Tom Tromey <tromey@redhat.com>
3900
3901 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3902 PARAMS.
3903 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3904
3905 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3906
3907 PR rtl-optimization/59896
3908 * lra-constraints.c (process_alt_operands): Check unused note for
3909 matched operands of insn with no output reloads.
3910
3911 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3912
3913 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3914 (mips_move_from_gpr_cost): Likewise.
3915
3916 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3917
3918 PR rtl-optimization/59858
3919 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3920 ira_class_hard_regs_num.
3921 (process_alt_operands): Increase reject for dying matched operand.
3922
3923 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3924
3925 PR target/59003
3926 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3927 smaller than size, perform several stores or loads and stores
3928 at dst + count - size to store or copy all of size bytes, rather
3929 than just last modesize bytes.
3930
3931 2014-01-20 DJ Delorie <dj@redhat.com>
3932
3933 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3934 that CLOBBERs are REGs before propogating their values.
3935
3936 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3937
3938 PR middle-end/59789
3939 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3940 (cgraph_inline_failed_type): New function.
3941 * cgraph.h (DEFCIFCODE): Add type.
3942 (cgraph_inline_failed_type_t): New enum.
3943 (cgraph_inline_failed_type): New prototype.
3944 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3945 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3946 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3947 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3948 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3949 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3950 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3951 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3952 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3953 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3954 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3955 OPTIMIZATION_MISMATCH.
3956 * tree-inline.c (expand_call_inline): Emit errors during
3957 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3958
3959 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3960
3961 PR target/59685
3962 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3963 mode attribute in insn output.
3964
3965 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3966
3967 * output.h (output_constant): Delete.
3968 * varasm.c (output_constant): Make private.
3969
3970 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3971
3972 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3973
3974 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3975
3976 PR middle-end/59860
3977 * tree.h (fold_builtin_strcat): New prototype.
3978 * builtins.c (fold_builtin_strcat): No longer static. Add len
3979 argument, if non-NULL, don't call c_strlen. Optimize
3980 directly into __builtin_memcpy instead of __builtin_strcpy.
3981 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3982 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3983
3984 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3985
3986 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3987 for SImode_address_operand operands, having only a REG argument.
3988
3989 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3990
3991 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3992 loader name using mbig-endian.
3993 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3994
3995 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3996
3997 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3998 (-mtune): Likewise.
3999 (-mcpu): Likewise.
4000
4001 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4002
4003 * config/aarch64/aarch64-protos.h
4004 (aarch64_cannot_change_mode_class_ptr): Declare.
4005 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4006 aarch64_cannot_change_mode_class_ptr): New.
4007 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4008 backend hook aarch64_cannot_change_mode_class.
4009
4010 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4011
4012 * common/config/aarch64/aarch64-common.c
4013 (aarch64_handle_option): Don't handle any option order logic here.
4014 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4015 selected_cpu, warn on architecture version mismatch.
4016 (aarch64_override_options): Fix parsing order for option strings.
4017
4018 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4019 Iain Sandoe <iain@codesourcery.com>
4020
4021 PR bootstrap/59496
4022 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4023 warning. Amend comment to reflect current functionality.
4024
4025 2014-01-20 Richard Biener <rguenther@suse.de>
4026
4027 PR middle-end/59860
4028 * builtins.c (fold_builtin_strcat): Remove case better handled
4029 by tree-ssa-strlen.c.
4030
4031 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4032
4033 * config/aarch64/aarch64.opt
4034 (mcpu, march, mtune): Make case-insensitive.
4035
4036 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4037
4038 PR target/59880
4039 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4040 if operands[1] is a REG or ZERO_EXTEND of a REG.
4041
4042 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4043
4044 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4045
4046 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4047
4048 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4049 long non-pic millicode calls.
4050
4051 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4052
4053 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4054
4055 2014-01-19 Kito Cheng <kito@0xlab.org>
4056
4057 * builtins.c (expand_movstr): Check movstr expand done or fail.
4058
4059 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4060 H.J. Lu <hongjiu.lu@intel.com>
4061
4062 PR target/59379
4063 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4064 to DImode for zero-extended addresses.
4065
4066 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4067
4068 PR rtl-optimization/57763
4069 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4070 on the new indirect jump_insn and increment LABEL_NUSES (label).
4071
4072 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4073
4074 PR bootstrap/59580
4075 PR bootstrap/59583
4076 * config.gcc (x86_archs): New variable.
4077 (x86_64_archs): Likewise.
4078 (x86_cpus): Likewise.
4079 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4080 --with-arch/--with-cpu= options.
4081 Support --with-arch=/--with-cpu={nehalem,westmere,
4082 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4083
4084 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4085
4086 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4087 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4088
4089 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4090
4091 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4092
4093 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4094
4095 PR target/58944
4096 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4097 clear cpp_get_options (parse_in)->warn_unused_macros for
4098 ix86_target_macros_internal with cpp_define.
4099
4100 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4101
4102 * jump.c (delete_related_insns): Keep (use (insn))s.
4103 * reorg.c (redundant_insn): Check for barriers too.
4104
4105 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4106
4107 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4108
4109 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4110
4111 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4112 call to $$dyncall when TARGET_LONG_CALLS is true.
4113
4114 2014-01-17 Jeff Law <law@redhat.com>
4115
4116 * ree.c (combine_set_extension): Temporarily disable test for
4117 changing number of hard registers.
4118
4119 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4120
4121 PR middle-end/58125
4122 * ipa-inline-analysis.c (inline_free_summary):
4123 Do not free summary of aliases.
4124
4125 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4126
4127 PR middle-end/59706
4128 * gimplify.c (gimplify_expr): Use create_tmp_var
4129 instead of create_tmp_var_raw. If cond doesn't have
4130 integral type, don't add the IFN_ANNOTATE builtin at all.
4131
4132 2014-01-17 Martin Jambor <mjambor@suse.cz>
4133
4134 PR ipa/59736
4135 * ipa-cp.c (prev_edge_clone): New variable.
4136 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4137 Also resize prev_edge_clone vector.
4138 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4139 (ipcp_edge_removal_hook): New function.
4140 (ipcp_driver): Register ipcp_edge_removal_hook.
4141
4142 2014-01-17 Andrew Pinski <apinski@cavium.com>
4143 Steve Ellcey <sellcey@mips.com>
4144
4145 PR target/59462
4146 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4147 of operator mode.
4148
4149 2014-01-17 Jeff Law <law@redhat.com>
4150
4151 PR middle-end/57904
4152 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4153 so that pass_ccp runs first.
4154
4155 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4156
4157 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4158 (ix86_adjust_cost): Use !TARGET_XXX.
4159 (do_reorder_for_imul): Likewise.
4160 (swap_top_of_ready_list): Likewise.
4161 (ix86_sched_reorder): Likewise.
4162
4163 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4164
4165 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4166 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4167 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4168 (intel_memset): New. Duplicate slm_memset.
4169 (intel_cost): New. Duplicate slm_cost.
4170 (m_INTEL): New macro.
4171 (processor_target_table): Add "intel".
4172 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4173 with PROCESSOR_INTEL for "intel".
4174 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4175 PROCESSOR_SILVERMONT.
4176 (ix86_issue_rate): Likewise.
4177 (ix86_adjust_cost): Likewise.
4178 (ia32_multipass_dfa_lookahead): Likewise.
4179 (swap_top_of_ready_list): Likewise.
4180 (ix86_sched_reorder): Likewise.
4181 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4182 instead of TARGET_OPT_AGU.
4183 * config/i386/i386.h (TARGET_INTEL): New.
4184 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4185 (processor_type): Add PROCESSOR_INTEL.
4186 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4187 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4188
4189 2014-01-17 Marek Polacek <polacek@redhat.com>
4190
4191 PR c/58346
4192 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4193 size is zero.
4194
4195 2014-01-17 Richard Biener <rguenther@suse.de>
4196
4197 PR tree-optimization/46590
4198 * opts.c (default_options_table): Add entries for
4199 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4200 all enabled at -O1 but not for -Og.
4201 * common.opt (fbranch-count-reg): Remove Init(1).
4202 (fmove-loop-invariants): Likewise.
4203 (ftree-pta): Likewise.
4204
4205 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4206
4207 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4208 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4209 decls to at least the GCC 4.8 used alignments.
4210
4211 PR fortran/59440
4212 * tree-nested.c (convert_nonlocal_reference_stmt,
4213 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4214 of GIMPLE_BIND stmts, adjust associated decls.
4215
4216 2014-01-17 Richard Biener <rguenther@suse.de>
4217
4218 PR tree-optimization/46590
4219 * vec.h (vec<>::bseach): New member function implementing
4220 binary search according to C89 bsearch.
4221 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4222 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4223 bitmap pointer again. Make accesses_in_loop a flat array.
4224 (mem_ref_obstack): New global.
4225 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4226 (mark_ref_stored): Likewise.
4227 (ref_indep_loop_p_2): Likewise.
4228 (set_ref_stored_in_loop): New helper function.
4229 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4230 (memref_free): Adjust.
4231 (record_mem_ref_loc): Simplify.
4232 (gather_mem_refs_stmt): Adjust.
4233 (sort_locs_in_loop_postorder_cmp): New function.
4234 (analyze_memory_references): Sort accesses_in_loop after
4235 loop postorder number.
4236 (find_ref_loc_in_loop_cmp): New function.
4237 (for_all_locs_in_loop): Find relevant cluster of locs in
4238 accesses_in_loop and iterate without recursion.
4239 (execute_sm): Avoid uninit warning.
4240 (struct ref_always_accessed): Simplify.
4241 (ref_always_accessed::operator ()): Likewise.
4242 (ref_always_accessed_p): Likewise.
4243 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4244 loop postorder numbers here.
4245 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4246 numbers.
4247
4248 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4249
4250 PR c++/57945
4251 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4252 on decls for which assemble_alias has been called.
4253
4254 2014-01-17 Nick Clifton <nickc@redhat.com>
4255
4256 * config/msp430/msp430.opt: (mcpu): New option.
4257 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4258 (msp430_option_override): Parse target_cpu. If the MCU name
4259 matches a generic string, clear target_mcu.
4260 (msp430_attr): Allow numeric interrupt values up to 63.
4261 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4262 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4263 option.
4264 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4265 Add mcpu matches.
4266 * config/msp430/msp430.md (popm): Use %J rather than %I.
4267 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4268 (addhi_cy_i): Use immediate_operand for operand 2.
4269 * doc/invoke.texi: Document -mcpu option.
4270
4271 2014-01-17 Richard Biener <rguenther@suse.de>
4272
4273 PR rtl-optimization/38518
4274 * df.h (df_analyze_loop): Declare.
4275 * df-core.c: Include cfgloop.h.
4276 (df_analyze_1): Split out main part of df_analyze.
4277 (df_analyze): Adjust.
4278 (loop_inverted_post_order_compute): New function.
4279 (loop_post_order_compute): Likewise.
4280 (df_analyze_loop): New function avoiding whole-function
4281 postorder computes.
4282 * loop-invariant.c (find_defs): Use df_analyze_loop.
4283 (find_invariants): Adjust.
4284 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4285
4286 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4287
4288 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4289 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4290
4291 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4292
4293 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4294 traversal when removing references.
4295
4296 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4297
4298 PR ipa/59775
4299 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4300
4301 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4302
4303 PR middle-end/56791
4304 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4305 pushing a reload for an autoinc when we had previously reloaded an
4306 inner part of the address.
4307
4308 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4309
4310 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4311 field.
4312 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4313 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4314 when not giving up or versioning for alias only because of
4315 loop->safelen.
4316 (vect_analyze_data_ref_dependences): Set to true.
4317 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4318 is a GIMPLE_PHI.
4319 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4320 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4321 to the condition.
4322
4323 PR middle-end/58344
4324 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4325
4326 PR target/59839
4327 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4328 operand 0 predicate for gathers, use a new pseudo as subtarget.
4329
4330 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4331
4332 PR middle-end/59609
4333 * lra-constraints.c (process_alt_operands): Add printing debug info.
4334 Check absence of input/output reloads for matched operands too.
4335
4336 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4337
4338 PR rtl-optimization/59835
4339 * ira.c (ira_init_register_move_cost): Increase cost for
4340 impossible modes.
4341
4342 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4343
4344 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4345
4346 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4347
4348 PR target/59780
4349 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4350 non-register objects. Use gen_(high/low)part more consistently.
4351 Fix assertions.
4352
4353 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4354
4355 PR target/59844
4356 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4357 endian support, remove tests for WORDS_BIG_ENDIAN.
4358 (p8_mfvsrd_3_<mode>): Likewise.
4359 (reload_gpr_from_vsx<mode>): Likewise.
4360 (reload_gpr_from_vsxsf): Likewise.
4361 (p8_mfvsrd_4_disf): Likewise.
4362
4363 2014-01-16 Richard Biener <rguenther@suse.de>
4364
4365 PR rtl-optimization/46590
4366 * lcm.c (compute_antinout_edge): Use postorder iteration.
4367 (compute_laterin): Use inverted postorder iteration.
4368
4369 2014-01-16 Nick Clifton <nickc@redhat.com>
4370
4371 PR middle-end/28865
4372 * varasm.c (output_constant): Return the number of bytes actually
4373 emitted.
4374 (output_constructor_array_range): Update the field size with the
4375 number of bytes emitted by output_constant.
4376 (output_constructor_regular_field): Likewise. Also do not
4377 complain if the total number of bytes emitted is now greater
4378 than the expected fieldpos.
4379 * output.h (output_constant): Update prototype and descriptive comment.
4380
4381 2014-01-16 Marek Polacek <polacek@redhat.com>
4382
4383 PR middle-end/59827
4384 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4385 it is error_mark_node.
4386
4387 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4388
4389 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4390 VALID_AVX256_REG_OR_OI_MODE.
4391
4392 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4393
4394 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4395 current procedure should be profiled.
4396
4397 2014-01-15 Andrew Pinski <apinski@cavium.com>
4398
4399 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4400 of moving from/to the STACK_REG register class.
4401
4402 2014-01-15 Richard Henderson <rth@redhat.com>
4403
4404 PR debug/54694
4405 * reginfo.c (global_regs_decl): Globalize.
4406 * rtl.h (global_regs_decl): Declare.
4407 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4408 reserved via global_regs.
4409
4410 2014-01-15 Teresa Johnson <tejohnson@google.com>
4411
4412 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4413
4414 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4415
4416 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4417 and vmulosh rather than call gen_vec_widen_smult_*.
4418 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4419 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4420 (vec_widen_smult_even_v16qi): Likewise.
4421 (vec_widen_umult_even_v8hi): Likewise.
4422 (vec_widen_smult_even_v8hi): Likewise.
4423 (vec_widen_umult_odd_v16qi): Likewise.
4424 (vec_widen_smult_odd_v16qi): Likewise.
4425 (vec_widen_umult_odd_v8hi): Likewise.
4426 (vec_widen_smult_odd_v8hi): Likewise.
4427 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4428 vmuloub rather than call gen_vec_widen_umult_*.
4429 (vec_widen_umult_lo_v16qi): Likewise.
4430 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4431 vmulosb rather than call gen_vec_widen_smult_*.
4432 (vec_widen_smult_lo_v16qi): Likewise.
4433 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4434 rather than call gen_vec_widen_umult_*.
4435 (vec_widen_umult_lo_v8hi): Likewise.
4436 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4437 rather than call gen_vec_widen_smult_*.
4438 (vec_widen_smult_lo_v8hi): Likewise.
4439
4440 2014-01-15 Jeff Law <law@redhat.com>
4441
4442 PR tree-optimization/59747
4443 * ree.c (find_and_remove_re): Properly handle case where a second
4444 eliminated extension requires widening a copy created for elimination
4445 of a prior extension.
4446 (combine_set_extension): Ensure that the number of hard regs needed
4447 for a destination register does not change when we widen it.
4448
4449 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4450
4451 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4452 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4453 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4454 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4455 (avr-*-rtems*): Likewise.
4456 (bfin*-rtems*): Likewise.
4457 (moxie-*-rtems*): Likewise.
4458 (h8300-*-rtems*): Likewise.
4459 (i[34567]86-*-rtems*): Likewise.
4460 (lm32-*-rtems*): Likewise.
4461 (m32r-*-rtems*): Likewise.
4462 (m68k-*-rtems*): Likewise.
4463 (microblaze*-*-rtems*): Likewise.
4464 (mips*-*-rtems*): Likewise.
4465 (powerpc-*-rtems*): Likewise.
4466 (sh-*-rtems*): Likewise.
4467 (sparc-*-rtems*): Likewise.
4468 (sparc64-*-rtems*): Likewise.
4469 (v850-*-rtems*): Likewise.
4470 (m32c-*-rtems*): Likewise.
4471
4472 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4473
4474 PR rtl-optimization/59511
4475 * ira.c (ira_init_register_move_cost): Use memory costs for some
4476 cases of register move cost calculations.
4477 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4478 instead of BB frequency.
4479 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4480 * lra-assigns.c (find_hard_regno_for): Ditto.
4481
4482 2014-01-15 Richard Biener <rguenther@suse.de>
4483
4484 PR tree-optimization/59822
4485 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4486 (vectorizable_load): Use it to hoist defs of uses of invariant
4487 loads out of the loop.
4488
4489 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4490 Kugan Vivekanandarajah <kuganv@linaro.org>
4491
4492 PR target/59695
4493 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4494 truncation.
4495
4496 2014-01-15 Richard Biener <rguenther@suse.de>
4497
4498 PR rtl-optimization/59802
4499 * lcm.c (compute_available): Use inverted postorder to seed
4500 the initial worklist.
4501
4502 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4503
4504 PR target/59803
4505 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4506 ADDR_REGS for invalid symrefs in non-PIC code.
4507
4508 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4509
4510 PR other/58712
4511 * builtins.c (determine_block_size): Initialize *probable_max_size
4512 even if len_rtx is CONST_INT.
4513
4514 2014-01-14 Andrew Pinski <apinski@cavium.com>
4515
4516 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4517 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4518 (cortexa53_tunings): Likewise.
4519 (aarch64_sched_issue_rate): New function.
4520 (TARGET_SCHED_ISSUE_RATE): Define.
4521
4522 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4523
4524 * ira-costs.c (find_costs_and_classes): Add missed
4525 ira_init_register_move_cost_if_necessary.
4526
4527 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4528
4529 PR target/59787
4530 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4531
4532 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4533
4534 PR target/59794
4535 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4536 to indicate if type is used for function return value. Warn ABI
4537 change if the vector mode isn't available for function return value.
4538 (ix86_function_arg_advance): Pass false to type_natural_mode.
4539 (ix86_function_arg): Likewise.
4540 (ix86_gimplify_va_arg): Likewise.
4541 (function_arg_32): Don't warn ABI change.
4542 (ix86_function_value): Pass true to type_natural_mode.
4543 (ix86_return_in_memory): Likewise.
4544 (ix86_struct_value_rtx): Removed.
4545 (TARGET_STRUCT_VALUE_RTX): Likewise.
4546
4547 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4548
4549 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4550 converting a conditional jump into a conditional return.
4551
4552 2014-01-14 Richard Biener <rguenther@suse.de>
4553
4554 PR tree-optimization/58921
4555 PR tree-optimization/59006
4556 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4557 hoisting invariant stmts.
4558 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4559 invariant loads on the preheader edge if possible.
4560
4561 2014-01-14 Joey Ye <joey.ye@arm.com>
4562
4563 * doc/plugin.texi (Building GCC plugins): Update to C++.
4564
4565 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4566
4567 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4568 (_mm_rcp28_round_ss): Ditto.
4569 (_mm_rsqrt28_round_sd): Ditto.
4570 (_mm_rsqrt28_round_ss): Ditto.
4571 (_mm_rcp28_sd): Ditto.
4572 (_mm_rcp28_ss): Ditto.
4573 (_mm_rsqrt28_sd): Ditto.
4574 (_mm_rsqrt28_ss): Ditto.
4575 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4576 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4577 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4578 (IX86_BUILTIN_RCP28SD): Ditto.
4579 (IX86_BUILTIN_RCP28SS): Ditto.
4580 (IX86_BUILTIN_RSQRT28SD): Ditto.
4581 (IX86_BUILTIN_RSQRT28SS): Ditto.
4582 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4583 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4584 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4585 (ix86_expand_special_args_builtin): Expand new FTYPE.
4586 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4587 (srcp14<mode>): Make insn unary.
4588 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4589 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4590 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4591 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4592 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4593 Fix rounding: make it SAE only.
4594 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4595 Ditto.
4596 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4597 Ditto.
4598 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4599 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4600 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4601 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4602 (round_saeonly_mask_scalar_operand4): Ditto.
4603 (round_saeonly_mask_scalar_op3): Ditto.
4604 (round_saeonly_mask_scalar_op4): Ditto.
4605
4606 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4607
4608 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4609 Implement -maltivec=be for vec_insert and vec_extract.
4610
4611 2014-01-10 DJ Delorie <dj@redhat.com>
4612
4613 * config/msp430/msp430.md (call_internal): Don't allow memory
4614 references with SP as the base register.
4615 (call_value_internal): Likewise.
4616 * config/msp430/constraints.md (Yc): New. For memory references
4617 that don't use SP as a base register.
4618
4619 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4620 "an integer without a # prefix"
4621 * config/msp430/msp430.md (epilogue_helper): Use it.
4622
4623 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4624
4625 PR target/59617
4626 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4627 AVX512F gather builtins.
4628 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4629 on gather decls with INTEGER_TYPE masktype.
4630 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4631 directly into the builtin rather than hoisting it before loop.
4632
4633 PR tree-optimization/59387
4634 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4635 (scev_const_prop): If folded_casts and type has undefined overflow,
4636 use force_gimple_operand instead of force_gimple_operand_gsi and
4637 for each added stmt if it is assign with
4638 arith_code_with_undefined_signed_overflow, call
4639 rewrite_to_defined_overflow.
4640 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4641 gimple-fold.h instead.
4642 (arith_code_with_undefined_signed_overflow,
4643 rewrite_to_defined_overflow): Moved to ...
4644 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4645 rewrite_to_defined_overflow): ... here. No longer static.
4646 Include gimplify-me.h.
4647 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4648 rewrite_to_defined_overflow): New prototypes.
4649
4650 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4651
4652 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4653
4654 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4655
4656 * builtins.c (get_object_alignment_2): Minor tweak.
4657 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4658
4659 2014-01-13 Christian Bruel <christian.bruel@st.com>
4660
4661 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4662 optimized non constant lengths.
4663
4664 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4665
4666 PR libgomp/59194
4667 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4668 load as __atomic_load_N if possible.
4669
4670 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4671
4672 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4673 target parameter.
4674 (rs6000_expand_builtin): Adjust call.
4675
4676 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4677
4678 PR target/58115
4679 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4680 * config/rs6000/rs6000.c: Include target-globals.h.
4681 (rs6000_set_current_function): Instead of doing target_reinit
4682 unconditionally, use save_target_globals_default_opts and
4683 restore_target_globals.
4684
4685 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4686 FPSCR.
4687 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4688 (rs6000_expand_builtin): Handle mffs and mtfsf.
4689 (rs6000_init_builtins): Define mffs and mtfsf.
4690 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4691 (rs6000_mffs): New pattern.
4692 (rs6000_mtfsf): New pattern.
4693
4694 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4695
4696 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4697 Start narrowing with START. Apply candidate-use pair
4698 and check overall cost in narrowing.
4699 (iv_ca_prune): Pass new argument.
4700
4701 2014-01-10 Jeff Law <law@redhat.com>
4702
4703 PR middle-end/59743
4704 * ree.c (combine_reaching_defs): Ensure the defining statement
4705 occurs before the extension when optimizing extensions with
4706 different source and destination hard registers.
4707
4708 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4709
4710 PR ipa/58585
4711 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4712 vtables into the type inheritance graph.
4713
4714 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4715
4716 PR rtl-optimization/59754
4717 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4718 modes in the REGNO != REGNO case.
4719
4720 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4721
4722 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4723
4724 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4725
4726 PR tree-optimization/59745
4727 * tree-predcom.c (tree_predictive_commoning_loop): Call
4728 free_affine_expand_cache if giving up because components is NULL.
4729
4730 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4731 GC in payload of target_globals struct instead of allocating them on
4732 the heap and the larger structs separately using GC.
4733 * target-globals.h (struct target_globals): Make regs, hard_regs,
4734 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4735 of GTY((skip)) and change type to void *.
4736 (reset_target_globals): Cast loads from those fields to corresponding
4737 types.
4738
4739 2014-01-10 Steve Ellcey <sellcey@mips.com>
4740
4741 PR plugins/59335
4742 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4743 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4744 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4745
4746 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4747
4748 PR target/59744
4749 * aarch64-modes.def (CC_Zmode): New flags mode.
4750 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4751 represents an equality.
4752 (aarch64_get_condition_code): Handle CC_Zmode.
4753 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4754
4755 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4756
4757 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4758 extraction in good case.
4759
4760 2014-01-10 Richard Biener <rguenther@suse.de>
4761
4762 PR tree-optimization/59374
4763 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4764 checking after SLP discovery. Mark stmts not participating
4765 in any SLP instance properly.
4766
4767 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4768
4769 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4770 when handling a SET rtx.
4771
4772 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4773
4774 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4775 (cortex-a57): Likewise.
4776 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4777
4778 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4779
4780 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4781 non-iwmmxt builtins.
4782
4783 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4784
4785 PR ipa/58252
4786 PR ipa/59226
4787 * ipa-devirt.c record_target_from_binfo): Take as argument
4788 stack of binfos and lookup matching one for virtual inheritance.
4789 (possible_polymorphic_call_targets_1): Update.
4790
4791 2014-01-10 Huacai Chen <chenhc@lemote.com>
4792
4793 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4794 kernel strings for Loongson-2E/2F/3A.
4795
4796 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4797
4798 PR middle-end/59670
4799 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4800 is_gimple_call before calling gimple_call_internal_p.
4801
4802 2014-01-09 Steve Ellcey <sellcey@mips.com>
4803
4804 * Makefile.in (TREE_FLOW_H): Remove.
4805 (TREE_SSA_H): Add file names from tree-flow.h.
4806 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4807 * tree.h: Remove tree-flow.h reference.
4808 * hash-table.h: Remove tree-flow.h reference.
4809 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4810 reference with tree-ssa-loop.h.
4811
4812 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4813
4814 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4815 default element-order behavior for -maltivec.
4816 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4817 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4818 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4819 when targeting big endian, at least for now.
4820 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4821
4822 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4823
4824 PR middle-end/47735
4825 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4826 var satisfies use_register_for_decl, just take into account type
4827 alignment, rather than decl alignment.
4828
4829 PR tree-optimization/59622
4830 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4831 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4832 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4833 Don't devirtualize for inplace at all. For targets.length () == 1,
4834 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4835
4836 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4837
4838 * config/i386/i386.md (cpu): Remove the unused btver1.
4839
4840 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4841
4842 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4843
4844 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4845
4846 PR target/58115
4847 * tree-core.h (struct target_globals): New forward declaration.
4848 (struct tree_target_option): Add globals field.
4849 * tree.h (TREE_TARGET_GLOBALS): Define.
4850 (prepare_target_option_nodes_for_pch): New prototype.
4851 * target-globals.h (struct target_globals): Define even if
4852 !SWITCHABLE_TARGET.
4853 * tree.c (prepare_target_option_node_for_pch,
4854 prepare_target_option_nodes_for_pch): New functions.
4855 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4856 * config/i386/i386.c: Include target-globals.h.
4857 (ix86_set_current_function): Instead of doing target_reinit
4858 unconditionally, use save_target_globals_default_opts and
4859 restore_target_globals.
4860
4861 2014-01-09 Richard Biener <rguenther@suse.de>
4862
4863 PR tree-optimization/59715
4864 * tree-cfg.h (split_critical_edges): Declare.
4865 * tree-cfg.c (split_critical_edges): Export.
4866 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4867
4868 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4869
4870 * cfgexpand.c (expand_stack_vars): Optionally disable
4871 asan stack protection.
4872 (expand_used_vars): Likewise.
4873 (partition_stack_vars): Likewise.
4874 * asan.c (asan_emit_stack_protection): Optionally disable
4875 after return stack usage.
4876 (instrument_derefs): Optionally disable memory access instrumentation.
4877 (instrument_builtin_call): Likewise.
4878 (instrument_strlen_call): Likewise.
4879 (asan_protect_global): Optionally disable global variables protection.
4880 * doc/invoke.texi: Added doc for new options.
4881 * params.def: Added new options.
4882 * params.h: Likewise.
4883
4884 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4885
4886 PR rtl-optimization/59724
4887 * ifcvt.c (cond_exec_process_if_block): Don't call
4888 flow_find_head_matching_sequence with 0 longest_match.
4889 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4890 non-active insns if !stop_after.
4891 (try_head_merge_bb): Revert 2014-01-07 changes.
4892
4893 2014-01-08 Jeff Law <law@redhat.com>
4894
4895 * ree.c (get_sub_rtx): New function, extracted from...
4896 (merge_def_and_ext): Here.
4897 (combine_reaching_defs): Use get_sub_rtx.
4898
4899 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4900
4901 * cgraph.h (varpool_variable_node): Do not choke on null node.
4902
4903 2014-01-08 Catherine Moore <clm@codesourcery.com>
4904
4905 * config/mips/mips.md (simple_return): Attempt to use JRC
4906 for microMIPS.
4907 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4908
4909 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4910
4911 PR rtl-optimization/59137
4912 * reorg.c (steal_delay_list_from_target): Call update_block for
4913 elided insns.
4914 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4915
4916 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4917
4918 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4919 two duplicate entries.
4920
4921 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4922
4923 Revert:
4924 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4925
4926 * config/mips/mips.c (mips_truncated_op_cost): New function.
4927 (mips_rtx_costs): Adjust test for BADDU.
4928 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4929
4930 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4931
4932 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4933 (*baddu_si): ...this new pattern.
4934
4935 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4936
4937 PR ipa/59722
4938 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4939
4940 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4941
4942 PR middle-end/57748
4943 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4944 inner_reference_p.
4945 (expand_expr, expand_normal): Adjust.
4946 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4947 inner_reference_p. Use inner_reference_p to expand inner references.
4948 (store_expr): Adjust.
4949 * cfgexpand.c (expand_call_stmt): Adjust.
4950
4951 2014-01-08 Rong Xu <xur@google.com>
4952
4953 * gcov-io.c (gcov_var): Move from gcov-io.h.
4954 (gcov_position): Ditto.
4955 (gcov_is_error): Ditto.
4956 (gcov_rewrite): Ditto.
4957 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4958 only part to libgcc/libgcov.h.
4959
4960 2014-01-08 Marek Polacek <polacek@redhat.com>
4961
4962 PR middle-end/59669
4963 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4964
4965 2014-01-08 Marek Polacek <polacek@redhat.com>
4966
4967 PR sanitizer/59667
4968 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4969
4970 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4971
4972 PR rtl-optimization/59649
4973 * stor-layout.c (get_mode_bounds): For BImode return
4974 0 and STORE_FLAG_VALUE.
4975
4976 2014-01-08 Richard Biener <rguenther@suse.de>
4977
4978 PR middle-end/59630
4979 * gimple.h (is_gimple_builtin_call): Remove.
4980 (gimple_builtin_call_types_compatible_p): New.
4981 (gimple_call_builtin_p): New overload.
4982 * gimple.c (is_gimple_builtin_call): Remove.
4983 (validate_call): Rename to ...
4984 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4985 check return types.
4986 (validate_type): New static function.
4987 (gimple_call_builtin_p): New overload and adjust.
4988 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4989 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4990 (gimple_fold_stmt_to_constant_1): Likewise.
4991 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4992
4993 2014-01-08 Richard Biener <rguenther@suse.de>
4994
4995 PR middle-end/59471
4996 * gimplify.c (gimplify_expr): Gimplify register-register type
4997 VIEW_CONVERT_EXPRs to separate stmts.
4998
4999 2014-01-07 Jeff Law <law@redhat.com>
5000
5001 PR middle-end/53623
5002 * ree.c (combine_set_extension): Handle case where source
5003 and destination registers in an extension insn are different.
5004 (combine_reaching_defs): Allow source and destination registers
5005 in extension to be different under limited circumstances.
5006 (add_removable_extension): Remove restriction that the
5007 source and destination registers in the extension are the same.
5008 (find_and_remove_re): Emit a copy from the extension's
5009 destination to its source after the defining insn if
5010 the source and destination registers are different.
5011
5012 PR middle-end/59285
5013 * ifcvt.c (merge_if_block): If we are merging a block with more than
5014 one successor with a block with no successors, remove any BARRIER
5015 after the second block.
5016
5017 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5018
5019 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5020
5021 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5022
5023 PR target/59652
5024 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5025 for 14-bit register offsets when INT14_OK_STRICT is false.
5026
5027 2014-01-07 Roland Stigge <stigge@antcom.de>
5028 Michael Meissner <meissner@linux.vnet.ibm.com>
5029
5030 PR 57386/target
5031 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5032 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5033
5034 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5035
5036 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5037 -mcpu.
5038
5039 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5040
5041 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5042 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5043 rtx is const0_rtx or not.
5044
5045 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5046
5047 PR target/58115
5048 * target-globals.c (save_target_globals): Remove this_fn_optab
5049 handling.
5050 * toplev.c: Include optabs.h.
5051 (target_reinit): Temporarily restore the global options if another
5052 set of options are in force.
5053
5054 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5055
5056 PR rtl-optimization/58668
5057 * cfgcleanup.c (flow_find_cross_jump): Don't count
5058 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5059 to determine what is counted.
5060 (flow_find_head_matching_sequence): Use active_insn_p to determine
5061 what is counted.
5062 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5063 counting change.
5064 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5065 determine what is counted.
5066
5067 PR tree-optimization/59643
5068 * tree-predcom.c (split_data_refs_to_components): If one dr is
5069 read and one write, determine_offset fails and the write isn't
5070 in the bad component, just put the read into the bad component.
5071
5072 2014-01-07 Mike Stump <mikestump@comcast.net>
5073 Jakub Jelinek <jakub@redhat.com>
5074
5075 PR pch/59436
5076 * tree-core.h (struct tree_optimization_option): Change optabs
5077 type from unsigned char * to void *.
5078 * optabs.c (init_tree_optimization_optabs): Adjust
5079 TREE_OPTIMIZATION_OPTABS initialization.
5080
5081 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5082
5083 PR target/59644
5084 * config/i386/i386.h (struct machine_function): Add
5085 no_drap_save_restore field.
5086 * config/i386/i386.c (ix86_save_reg): Use
5087 !cfun->machine->no_drap_save_restore instead of
5088 crtl->stack_realign_needed.
5089 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5090 this function clears frame_pointer_needed. Set
5091 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5092 and DRAP reg is needed.
5093
5094 2014-01-06 Marek Polacek <polacek@redhat.com>
5095
5096 PR c/57773
5097 * doc/implement-c.texi: Mention that other integer types are
5098 permitted as bit-field types in strictly conforming mode.
5099
5100 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5101
5102 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5103 is newly allocated.
5104
5105 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5106
5107 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5108
5109 2014-01-06 Martin Jambor <mjambor@suse.cz>
5110
5111 PR ipa/59008
5112 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5113 to int.
5114 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5115
5116 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5117
5118 PR debug/59350
5119 PR debug/59510
5120 * var-tracking.c (add_stores): Preserve the value of the source even if
5121 we don't record the store.
5122
5123 2014-01-06 Terry Guo <terry.guo@arm.com>
5124
5125 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5126
5127 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5128
5129 PR bootstrap/59541
5130 * config/darwin.c (darwin_function_section): Adjust return values to
5131 correspond to optimisation changes made in r206070.
5132
5133 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5134
5135 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5136 from prefetch_block tune setting.
5137 (nocona_cost): Correct size of prefetch block to 64.
5138
5139 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5140
5141 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5142 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5143 used to save the static chain register in the computation of the offset
5144 from which the FP registers need to be restored.
5145
5146 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5147
5148 PR tree-optimization/59519
5149 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5150 ICE if get_current_def (current_new_name) is already non-NULL, as long
5151 as it is a phi result of some other phi in *new_exit_bb that has
5152 the same argument.
5153
5154 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5155 or vmovdqu* for misaligned_operand.
5156 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5157 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5158 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5159 aligned_mem for AVX512F masked aligned load and store builtins and for
5160 non-temporal moves.
5161
5162 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5163
5164 PR tree-optimization/59651
5165 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5166 Address range for negative step should be added by TYPE_SIZE_UNIT.
5167
5168 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5169
5170 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5171 overlapping registers also for registers other than the stack pointer.
5172
5173 2014-01-03 Marek Polacek <polacek@redhat.com>
5174
5175 PR other/59661
5176 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5177 __builtin_FILE.
5178
5179 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5180
5181 PR target/59625
5182 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5183 asm goto as jump.
5184
5185 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5186 (push splitter): Use <P:MODE_SIZE> instead of
5187 GET_MODE_SIZE (<P:MODE>mode).
5188 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5189 (mov -1, reg peephole2): Likewise.
5190 * config/i386/sse.md (*mov<mode>_internal,
5191 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5192 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5193 *<code><mode>3, *andnot<mode>3<mask_name>,
5194 <mask_codefor><code><mode>3<mask_name>): Likewise.
5195 * config/i386/subst.md (mask_mode512bit_condition,
5196 sd_mask_mode512bit_condition): Likewise.
5197
5198 2014-01-02 Xinliang David Li <davidxl@google.com>
5199
5200 PR tree-optimization/59303
5201 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5202 (dump_predicates): Better output format.
5203 (pred_equal_p): New function.
5204 (is_neq_relop_p): Ditto.
5205 (is_neq_zero_form_p): Ditto.
5206 (pred_expr_equal_p): Ditto.
5207 (pred_neg_p): Ditto.
5208 (simplify_pred): Ditto.
5209 (simplify_preds_2): Ditto.
5210 (simplify_preds_3): Ditto.
5211 (simplify_preds_4): Ditto.
5212 (simplify_preds): Ditto.
5213 (push_pred): Ditto.
5214 (push_to_worklist): Ditto.
5215 (get_pred_info_from_cmp): Ditto.
5216 (is_degenerated_phi): Ditto.
5217 (normalize_one_pred_1): Ditto.
5218 (normalize_one_pred): Ditto.
5219 (normalize_one_pred_chain): Ditto.
5220 (normalize_preds): Ditto.
5221 (normalize_cond_1): Remove function.
5222 (normalize_cond): Ditto.
5223 (is_gcond_subset_of): Ditto.
5224 (is_subset_of_any): Ditto.
5225 (is_or_set_subset_of): Ditto.
5226 (is_and_set_subset_of): Ditto.
5227 (is_norm_cond_subset_of): Ditto.
5228 (pred_chain_length_cmp): Ditto.
5229 (convert_control_dep_chain_into_preds): Type change.
5230 (find_predicates): Ditto.
5231 (find_def_preds): Ditto.
5232 (destroy_predicates_vecs): Ditto.
5233 (find_matching_predicates_in_rest_chains): Ditto.
5234 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5235 (is_pred_expr_subset): Ditto.
5236 (is_pred_chain_subset_of): Ditto.
5237 (is_included_in): Ditto.
5238 (is_superset_of): Ditto.
5239
5240 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5241
5242 Update copyright years.
5243
5244 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5245
5246 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5247 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5248 config/arc/arc.md, config/arc/arc.opt,
5249 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5250 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5251 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5252 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5253 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5254 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5255 vtable-verify.c, vtable-verify.h: Use the standard form for the
5256 copyright notice.
5257
5258 2014-01-02 Tobias Burnus <burnus@net-b.de>
5259
5260 * gcc.c (process_command): Update copyright notice dates.
5261 * gcov-dump.c: Ditto.
5262 * gcov.c: Ditto.
5263 * doc/cpp.texi: Bump @copying's copyright year.
5264 * doc/cppinternals.texi: Ditto.
5265 * doc/gcc.texi: Ditto.
5266 * doc/gccint.texi: Ditto.
5267 * doc/gcov.texi: Ditto.
5268 * doc/install.texi: Ditto.
5269 * doc/invoke.texi: Ditto.
5270
5271 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5272
5273 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5274
5275 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5276
5277 * config/i386/sse.md (*mov<mode>_internal): Guard
5278 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5279
5280 PR rtl-optimization/59647
5281 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5282 new_rtx into UNSIGNED_FLOAT rtxes.
5283 \f
5284 Copyright (C) 2014 Free Software Foundation, Inc.
5285
5286 Copying and distribution of this file, with or without modification,
5287 are permitted in any medium without royalty provided the copyright
5288 notice and this notice are preserved.