* expr.c (expand_expr_real_1): Remove outdated comment.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3 * expr.c (expand_expr_real_1): Remove outdated comment.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