1 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
4 * config/i386/i386.c (ix86_expand_builtin)
5 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
6 register for target RTX.
7 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
9 2014-04-18 Cong Hou <congh@google.com>
11 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
12 the widen-mult pattern by handling two operands with different sizes,
13 and operands whose size is smaller than half of the result type.
15 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
17 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
18 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
19 (do_estimate_edge_time): Compute it.
20 * ipa-inline.c (want_inline_small_function_p): Bypass
21 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
23 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
25 * ipa-inline.c (spec_rem): New static variable.
26 (dump_overall_stats): New function.
27 (dump_inline_stats): New function.
29 2014-04-18 Richard Henderson <rth@redhat.com>
31 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
32 to GET_MODE_SIZE, not a reg_class_t.
34 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
37 (vsx_xxmrglw_<mode>): Likewise.
39 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
42 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
43 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
44 (rs6000_init_hard_regno_mode_ok): Likewise.
46 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
48 * ipa-inline.c (inline_small_functions): Account only non-cold
50 * doc/invoke.texi (inline-unit-growth): Update documentation.
52 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
54 * config/rs6000/rs6000.md (addti3, subti3): New.
56 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
59 * config/i386/i386.c (ix86_expand_clear): Remove outdated
60 comment. Check optimize_insn_for_size_p instead of
61 optimize_insn_for_speed_p.
63 2014-04-17 Martin Jambor <mjambor@suse.cz>
65 * gimple-iterator.c (gsi_start_edge): New function.
66 * gimple-iterator.h (gsi_start_edge): Declare.
67 * tree-sra.c (single_non_eh_succ): New function.
68 (disqualify_ops_if_throwing_stmt): Renamed to
69 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
70 having one non-EH successor BB.
71 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
72 generate loads into replacements.
73 (sra_modify_assign): Likewise and and also use the simple path for
75 (sra_modify_function_body): Commit statements on edges.
77 2014-04-17 Richard Biener <rguenther@suse.de>
80 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
81 comparison results and add clarifying comment.
83 2014-04-17 Jakub Jelinek <jakub@redhat.com>
85 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
86 (blank_mode): Initialize it.
87 (emit_mode_size_inline, emit_mode_nunits_inline,
88 emit_mode_inner_inline): New functions.
89 (emit_insn_modes_h): Call them and surround their output with
90 #if GCC_VERSION >= 4001 ... #endif.
91 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
92 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
93 mode_* arrays if the argument is __builtin_constant_p.
94 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
97 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
99 * passes.c (opt_pass::execute): Adjust.
100 (pass_manager::execute_pass_mode_switching): Likewise.
101 (early_local_passes::execute): Likewise.
102 (execute_one_pass): Pass cfun to the pass's execute method.
103 * tree-pass.h (opt_pass::execute): Add function * argument.
104 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
105 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
106 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
107 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
108 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
109 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
110 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
111 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
112 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
113 ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
114 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
115 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
116 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
117 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
118 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
119 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
120 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
121 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
122 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
123 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
124 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
125 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
126 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
127 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
128 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
129 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
130 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
131 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
132 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
135 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
137 * passes.c (opt_pass::gate): Take function * argument.
138 (gate_all_early_local_passes): Merge into
139 (early_local_passes::gate): this.
140 (gate_all_early_optimizations): Merge into
141 (all_early_optimizations::gate): this.
142 (gate_all_optimizations): Mege into
143 (all_optimizations::gate): this.
144 (gate_all_optimizations_g): Merge into
145 (all_optimizations_g::gate): this.
146 (gate_rest_of_compilation): Mege into
147 (rest_of_compilation::gate): this.
148 (gate_postreload): Merge into
149 (postreload::gate): this.
150 (dump_one_pass): Pass cfun to the pass's gate method.
151 (execute_ipa_summary_passes): Likewise.
152 (execute_one_pass): Likewise.
153 (ipa_write_summaries_2): Likewise.
154 (ipa_write_optimization_summaries_1): Likewise.
155 (ipa_read_summaries_1): Likewise.
156 (ipa_read_optimization_summaries_1): Likewise.
157 (execute_ipa_stmt_fixups): Likewise.
158 * tree-pass.h (opt_pass::gate): Add function * argument.
159 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
160 combine-stack-adj.c, combine.c, compare-elim.c,
161 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
162 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
163 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
164 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
165 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
166 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
167 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
168 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
169 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
170 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
171 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
172 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
173 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
174 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
175 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
176 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
177 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
178 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
179 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
180 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
181 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
182 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
183 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
184 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
185 var-tracking.c, vtable-verify.c, web.c: Adjust.
187 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
189 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
190 * configure: Regenerate.
192 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
194 * passes.c (dump_one_pass): don't check pass->has_gate.
195 (execute_ipa_summary_passes): Likewise.
196 (execute_one_pass): Likewise.
197 (ipa_write_summaries_2): Likewise.
198 (ipa_write_optimization_summaries_1): Likewise.
199 (ipa_read_optimization_summaries_1): Likewise.
200 (execute_ipa_stmt_fixups): Likewise.
201 * tree-pass.h (pass_data::has_gate): Remove.
202 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
203 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
204 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
205 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
206 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
207 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
208 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
209 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
210 gimple-low.c, gimple-ssa-isolate-paths.c,
211 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
212 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
213 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
214 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
215 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
216 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
217 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
218 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
219 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
220 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
221 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
222 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
223 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
224 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
225 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
226 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
227 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
228 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
229 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
230 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
231 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
232 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
235 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
237 * pass_manager.h (pass_manager::register_dump_files_1): Remove
239 * passes.c (pass_manager::register_dump_files_1): Merge into
240 (pass_manager::register_dump_files): this, and remove its handling of
241 properties since the pass always has the properties anyway.
242 (pass_manager::pass_manager): Adjust.
244 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
246 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
247 * passes.c (pass_manager::register_dump_files_1): Remove dead code
248 dealing with properties.
249 (pass_manager::register_dump_files): Adjust.
251 2014-03-20 Mark Wielaard <mjw@redhat.com>
253 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
254 then represent the bound as normal constant value.
256 2014-04-17 Jakub Jelinek <jakub@redhat.com>
259 Forward port from 4.8 branch
260 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
262 * config/i386/bmiintrin.h (_blsi_u32): New.
266 (_blsmsk_u32): Ditto.
267 (_blsmsk_u64): Ditto.
271 2014-04-17 Kito Cheng <kito@0xlab.org>
273 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
275 2014-04-17 Richard Biener <rguenther@suse.de>
278 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
279 boolean results for comparisons.
281 2014-04-17 Richard Biener <rguenther@suse.de>
283 PR tree-optimization/60836
284 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
285 initial PHI args to be gimple values.
287 2014-04-17 Richard Biener <rguenther@suse.de>
289 PR tree-optimization/60841
290 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
291 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
292 of stmts to SLP build.
293 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
294 (vect_analyze_slp): Likewise.
295 (vect_analyze_slp_instance): Likewise.
296 (vect_build_slp_tree): Limit overall SLP tree growth.
297 * tree-vectorizer.h (vect_analyze_data_refs,
298 vect_analyze_slp): Adjust prototypes.
300 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
302 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
305 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
307 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
308 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
309 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
310 for TARGET_SLOW_PSHUFB
312 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
314 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
315 * config/i386/i386.c (intel_cost): Ditto.
317 2014-04-17 Joey Ye <joey.ye@arm.com>
319 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
321 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
323 * opts.c (common_handle_option): Disable -fipa-reference coorectly
326 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
328 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
329 (type_all_derivations_known_p): New predicate.
330 (type_all_ctors_visible_p): New predicate.
331 (type_possibly_instantiated_p): New predicate.
332 (get_odr_type): Compute all_derivations_known.
333 (dump_odr_type): Dump the flag.
334 (maybe_record_type): Cleanup.
335 (record_target_from_binfo): Add bases_to_consider array;
336 record bases for types w/o instances and skip CXX destructor.
337 (possible_polymorphic_call_targets_1): Add bases_to_consider
338 and consider_construction parameters; check if type may have instance.
339 (get_polymorphic_call_info): Set maybe_in_construction to true
340 when we know nothing.
341 (record_targets_from_bases): Skip CXX destructors; they are
342 never called for types in construction.
343 (possible_polymorphic_call_targets): Do not record target when
344 type may not have instance.
346 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
349 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
350 external aliases alive, too.
352 2014-04-16 Andrew Pinski <apinski@cavium.com>
354 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
357 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
359 * final.c (compute_alignments): Do not apply loop alignment to a block
360 falling through to the exit.
362 2014-04-16 Catherine Moore <clm@codesourcery.com>
364 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
365 Adjust constraints for microMIPS store patterns.
367 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
369 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
371 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
373 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
374 (append_use): Run at -O0.
375 (append_vdef): Likewise.
376 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
377 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
379 2014-04-16 Jakub Jelinek <jakub@redhat.com>
381 PR tree-optimization/60844
382 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
383 (propagate_op_to_single_use, remove_visited_stmt_chain,
384 linearize_expr, repropagate_negates, reassociate_bb): Use it
385 instead of gsi_remove.
387 2014-04-16 Martin Jambor <mjambor@suse.cz>
389 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
390 ipa_transforms_to_apply.
391 (cgraph_function_versioning): Assert that old_node has empty
392 ipa_transforms_to_apply.
393 * trans-mem.c (ipa_tm_create_version): Likewise.
394 * tree-inline.c (tree_function_versioning): Do not duplicate
395 ipa_transforms_to_apply.
397 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
400 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
402 Pass necessary as flags on 64-bit Solaris/x86.
403 Use lowercase relocs for x86_64-*-*.
404 * configure: Regenerate.
406 2014-04-15 Jan Hubicka <jh@suse.cz>
408 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
409 (maybe_record_node, likely_target_p): Use it.
411 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
414 Revert following patch
416 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
419 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
420 software floating point or no floating point registers, do not
421 allow any type in the FPRs. Eliminate a test for SPE SIMD types
422 in GPRs that occurs after we tested for GPRs that would never be
425 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
426 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
427 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
428 specifically allow DDmode, since that does not use the SPE SIMD
431 2014-03-21 Mark Wielaard <mjw@redhat.com>
433 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
434 as unsigned or int depending on type and value used.
436 2014-04-15 Richard Biener <rguenther@suse.de>
438 PR rtl-optimization/56965
439 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
440 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
442 * alias.c (true_dependence_1): Do not call
443 nonoverlapping_component_refs_p.
444 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
445 nonoverlapping_component_refs_p.
446 (indirect_refs_may_alias_p): Likewise.
448 2014-04-15 Teresa Johnson <tejohnson@google.com>
450 * cfg.c (dump_bb_info): Fix flags check.
451 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
453 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
455 PR rtl-optimization/60663
456 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
459 2014-04-15 Richard Biener <rguenther@suse.de>
461 * lto-streamer.h (LTO_major_version): Bump to 4.
463 2014-04-15 Richard Biener <rguenther@suse.de>
465 * common.opt (lto_partition_model): New enum.
466 (flto-partition=): Merge separate options with a single with argument,
467 add -flto-partition=one support.
468 * flag-types.h (enum lto_partition_model): Declare.
469 * opts.c (finish_options): Remove duplicate -flto-partition=
471 * lto-wrapper.c (run_gcc): Adjust.
473 2014-04-15 Richard Biener <rguenther@suse.de>
475 * alias.c (ncr_compar): New function.
476 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
478 2014-04-15 Richard Biener <rguenther@suse.de>
480 * alias.c (record_component_aliases): Do not walk BINFOs.
482 2014-04-15 Richard Biener <rguenther@suse.de>
484 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
485 Add struct function argument and adjust.
486 (find_func_aliases_for_call): Likewise.
487 (find_func_aliases): Likewise.
488 (find_func_clobbers): Likewise.
489 (intra_create_variable_infos): Likewise.
490 (compute_points_to_sets): Likewise.
491 (ipa_pta_execute): Adjust. Do not push/pop cfun.
493 2014-04-15 Richard Biener <rguenther@suse.de>
495 * tree.c (iterative_hash_expr): Use enum tree_code_class
496 to store TREE_CODE_CLASS.
497 (tree_block): Likewise.
498 (tree_set_block): Likewise.
499 * tree.h (fold_build_pointer_plus_loc): Use
500 convert_to_ptrofftype_loc.
502 2014-04-15 Jakub Jelinek <jakub@redhat.com>
505 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
508 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
510 * cfgloop.h (struct loop): Move force_vectorize down.
511 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
512 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
513 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
514 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
515 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
516 * tree-core.h (enum annot_expr_kind): Add new kind values.
517 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
518 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
520 * tree.def (ANNOTATE_EXPR): Tweak comment.
522 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
524 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
527 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
529 * tree.h (TYPE_IDENTIFIER): Declare.
530 * tree.c (subrange_type_for_debug_p): Use it.
531 * godump.c (go_format_type): Likewise.
532 * dwarf2out.c (is_cxx_auto, modified_type_die,
533 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
534 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
536 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
539 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
541 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
543 * config/i386/i386.c (examine_argument): Return bool. Return true if
544 parameter should be passed in memory.
545 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
546 (construct_container): Update calls to examine_argument.
547 (function_arg_advance_64): Ditto.
548 (return_in_memory_32): Merge with ix86_return_in_memory.
549 (return_in_memory_64): Ditto.
550 (return_in_memory_ms_64): Ditto.
552 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
554 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
555 * coverage.c (coverage_compute_profile_id): Handle externally visible
558 2014-04-14 Martin Jambor <mjambor@suse.cz>
560 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
561 DECL_DISREGARD_INLINE_LIMITS functions.
563 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
566 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
568 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
571 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
572 optimize_insn_for_speed_p instead of
573 optimize_function_for_speed_p.
575 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
577 * doc/invoke.texi (free): Document AArch64.
579 2014-04-14 Richard Biener <rguenther@suse.de>
581 PR tree-optimization/60042
582 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
583 (insert_into_preds_of_block): Do not prevent PHI insertion
584 for REFERENCE exprs here ...
585 (eliminate_dom_walker::before_dom_children): ... but prevent
586 their use here under similar conditions when applied to the
587 IL after PRE optimizations.
589 2014-04-14 Richard Biener <rguenther@suse.de>
591 * passes.def: Move early points-to after early SRA.
593 2014-04-14 Richard Biener <rguenther@suse.de>
595 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
596 check for which sign-changes we allow when forwarding
597 a converted value into a switch.
599 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
601 * stor-layout.c (place_field): Finalize non-constant offset for the
604 2014-04-14 Richard Biener <rguenther@suse.de>
606 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
608 (expand_switch_using_bit_tests_p): Likewise.
609 (process_switch): Compute and pass on speed_p based on the
611 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
612 optimize_bb_for_speed_p.
614 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
616 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
617 * function.h (struct function): Rename has_force_vect_loops into
618 has_force_vectorize_loops.
619 * lto-streamer-in.c (input_cfg): Adjust for renaming.
620 (input_struct_function_base): Likewise.
621 * lto-streamer-out.c (output_cfg): Likewise.
622 (output_struct_function_base): Likewise.
623 * omp-low.c (expand_omp_simd): Likewise.
624 * tree-cfg.c (move_sese_region_to_fn): Likewise.
625 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
626 (version_loop_for_if_conversion): Likewise.
627 (tree_if_conversion): Likewise.
628 (main_tree_if_conversion): Likewise.
629 (gate_tree_if_conversion): Likewise.
630 * tree-inline.c (copy_loops): Likewise.
631 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
632 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
633 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
634 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
635 * tree-vectorizer.c (vectorize_loops): Likewise.
636 * tree-vectorizer.h (unlimited_cost_model): Likewise.
638 2014-04-14 Richard Biener <rguenther@suse.de>
641 * lto-streamer-out.c (wrap_refs): New function.
642 (lto_output): Wrap symbol references in global initializes in
643 type-preserving MEM_REFs.
645 2014-04-14 Christian Bruel <christian.bruel@st.com>
647 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
649 2014-04-14 Christian Bruel <christian.bruel@st.com>
651 * config/sh/sh.md (setmemqi): New expand pattern.
652 * config/sh/sh.h (CLEAR_RATIO): Define.
653 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
654 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
656 2014-04-14 Richard Biener <rguenther@suse.de>
659 * fold-const.c (negate_expr_p): Don't negate directional rounding
661 (fold_negate_expr): Likewise.
663 2014-04-14 Richard Biener <rguenther@suse.de>
665 PR tree-optimization/59817
666 PR tree-optimization/60453
667 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
668 recursion to catch all CHRECs in the scalar evolution and restrict
669 the predicate for the remains appropriately.
671 2014-04-12 Catherine Moore <clm@codesourcery.com>
673 * config/mips/constraints.md: Add new register constraint "kb".
674 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
675 (*movhi_internal): Likewise.
676 (*movqi_internal): Likewise.
677 * config/mips/mips.h (M16_STORE_REGS): New register class.
678 (REG_CLASS_NAMES): Add M16_STORE_REGS.
679 (REG_CLASS_CONTENTS): Likewise.
680 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
682 2014-04-11 Tobias Burnus <burnus@net-b.de>
685 * doc/invoke.texi (-Wformat-signedness): Document it.
686 (Wformat=2): Mention that this enables -Wformat-signedness.
688 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
690 * common/config/epiphany/epiphany-common.c
691 (epiphany_option_optimization_table): Enable section anchors by
692 default at -O1 or higher.
693 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
694 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
695 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
696 carries no extra cost.
697 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
698 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
699 * config/epiphany/predicates.md (memclob_operand): New predicate.
700 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
701 Use memclob_operand predicate and X constraint for operand 3.
703 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
705 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
706 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
709 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
711 PR rtl-optimization/60651
712 * mode-switching.c (optimize_mode_switching): Make sure to emit
713 sets of a lower numbered entity before sets of a higher numbered
714 entity to a mode of the same or lower priority.
715 When creating a seginfo for a basic block that starts with a code
716 label, move the insertion point past the code label.
717 (new_seginfo): Document and enforce requirement that
718 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
719 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
720 * doc/tm.texi: Regenerate.
722 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
725 * config/arc/arc.c (arc_save_restore): Fix assert typo.
727 2013-04-11 Jakub Jelinek <jakub@redhat.com>
729 * BASE-VER: Set to 4.10.0.
731 2014-04-11 Tobias Burnus <burnus@net-b.de>
734 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
735 * doc/gcc.texi (Service): Update description in the @menu
736 * doc/invoke.texi (Option Summary): Remove misplaced and
739 2014-04-11 Steve Ellcey <sellcey@mips.com>
740 Jakub Jelinek <jakub@redhat.com>
743 * expr.c (convert_move): Use emit_store_flag_force instead of
744 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
747 2014-04-11 Richard Biener <rguenther@suse.de>
750 * varasm.c (assemble_alias): Avoid endless error reporting
751 recursion by setting TREE_ASM_WRITTEN.
753 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
755 * config/s390/s390.md: Add a splitter for NOT rtx.
757 2014-04-11 Jakub Jelinek <jakub@redhat.com>
759 PR rtl-optimization/60663
760 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
762 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
763 Jakub Jelinek <jakub@redhat.com>
766 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
767 flag from decl_node to node.
769 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
772 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
773 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
774 ameliorating the cases where it can be.
776 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
779 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
781 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
782 (loadsync_<mode>): Change mode.
783 (load_quadpti, store_quadpti): New.
784 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
785 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
786 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
788 2014-04-09 Cong Hou <congh@google.com>
791 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
794 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
796 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
797 instead of vnor to exploit possible fusion opportunity in the
799 (altivec_expand_vec_perm_const_le): Likewise.
801 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
803 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
804 (loadsync_<mode>): Change mode.
805 (load_quadpti, store_quadpti): New.
806 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
807 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
809 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
812 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
813 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
814 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
816 2014-04-08 Richard Biener <rguenther@suse.de>
819 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
820 a 64bit widest int print double-int similar to on HWI64 hosts.
822 2014-04-08 Richard Biener <rguenther@suse.de>
824 PR tree-optimization/60785
825 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
826 default defs properly.
828 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
830 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
833 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
835 * ipa-devirt.c (maybe_record_node): When node is not recorded,
836 set completep to false rather than true.
838 2014-04-07 Douglas B Rupp <rupp@adacore.com>
841 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
842 ARM_TARGET2_DWARF_FORMAT.
844 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
847 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
848 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
851 2014-04-07 Richard Biener <rguenther@suse.de>
853 PR tree-optimization/60766
854 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
855 (may_eliminate_iv): Convert cand_value_at result to desired type.
857 2014-04-07 Jason Merrill <jason@redhat.com>
860 * common.opt (-fno-gnu-unique): Add.
861 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
863 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
865 * haifa-sched.c: Fix outdated function reference and minor
866 grammar errors in introductory comment.
868 2014-04-07 Richard Biener <rguenther@suse.de>
871 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
873 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
875 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
878 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
879 size accounting for thunks.
880 (pa_asm_output_mi_thunk): Use final_start_function() and
881 final_end_function() to output function start and end directives.
883 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
885 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
886 device specific ISA/ feature information. Remove short_sp and
887 errata_skip ds. Add avr_device_specific_features enum to have device
889 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
890 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
891 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
892 updated device specific info.
893 * config/avr/avr-mcus.def: Merge device specific details to
895 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
897 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
898 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
899 assembler if RMW isa supported by current device.
900 * config/avr/genmultilib.awk: Update as device info structure changed.
901 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
903 2014-04-04 Cong Hou <congh@google.com>
905 PR tree-optimization/60656
906 * tree-vect-stmts.c (supportable_widening_operation):
907 Fix a bug that elements in a vector with vect_used_by_reduction
908 property are incorrectly reordered when the operation on it is not
909 consistant with the one in reduction operation.
911 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
913 PR rtl-optimization/60155
914 * gcse.c (record_set_data): New function.
915 (single_set_gcse): New function.
916 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
917 (hoist_code): Likewise.
918 (get_pressure_class_and_nregs): Likewise.
920 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
922 * explow.c (probe_stack_range): Emit a final optimization blockage.
924 2014-04-04 Anthony Green <green@moxielogic.com>
926 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
929 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
932 * lto-cgraph.c (input_overwrite_node): Check that partitioning
933 flags are set only during streaming.
934 * ipa.c (process_references, walk_polymorphic_call_targets,
935 symtab_remove_unreachable_nodes): Drop bodies of always inline
936 after early inlining.
937 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
939 2014-04-04 Jakub Jelinek <jakub@redhat.com>
940 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
943 * dwarf2out.c (const_ok_for_output_1): Reject expressions
946 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
949 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
951 (cortex_a53_fdivd): Likewise.
953 2014-04-04 Martin Jambor <mjambor@suse.cz>
956 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
958 * cgraph.c (clone_of_p): Also return true if thunks match.
959 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
960 cgraph_function_or_thunk_node and an obsolete comment.
961 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
963 (build_function_decl_skip_args): Likewise.
964 (set_new_clone_decl_and_node_flags): New function.
965 (duplicate_thunk_for_node): Likewise.
966 (redirect_edge_duplicating_thunks): Likewise.
967 (cgraph_clone_node): New parameter args_to_skip, pass it to
968 redirect_edge_duplicating_thunks which is called instead of
969 cgraph_redirect_edge_callee.
970 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
971 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
973 2014-04-04 Jeff Law <law@redhat.com>
976 * config/arm/predicates.md (const_int_I_operand): New predicate.
977 (const_int_M_operand): Similarly.
978 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
980 (insv_t2, extv_reg, extzv_t2): Likewise.
981 (load_multiple_with_writeback): Similarly for const_int_I_operand.
982 (pop_multiple_with_writeback_and_return): Likewise.
983 (vfp_pop_multiple_with_writeback): Likewise
985 2014-04-04 Richard Biener <rguenther@suse.de>
988 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
989 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
991 * gimplify.h (gimple_add_tmp_var_fn): Declare.
992 * gimplify.c (gimple_add_tmp_var_fn): New function.
993 * gimple-expr.h (create_tmp_reg_fn): Declare.
994 * gimple-expr.c (create_tmp_reg_fn): New function.
995 * gimple-low.c (record_vars_into): Don't change cfun.
996 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
997 code generation without cfun.
999 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
1002 * Makefile.in (install-driver): Fix shell scripting.
1004 2014-04-03 Cong Hou <congh@google.com>
1006 PR tree-optimization/60505
1007 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
1008 threshold of number of iterations below which no vectorization
1010 * tree-vect-loop.c (new_loop_vec_info):
1011 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
1012 * tree-vect-loop.c (vect_analyze_loop_operations):
1013 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
1014 * tree-vect-loop.c (vect_transform_loop):
1015 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
1016 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
1017 of iterations of the loop and see if we should build the epilogue.
1019 2014-04-03 Richard Biener <rguenther@suse.de>
1021 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
1022 (streamer_tree_cache_create): Adjust.
1023 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
1024 to allow optional nodes array.
1025 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
1026 (streamer_tree_cache_append): Likewise.
1027 (streamer_tree_cache_create): Create nodes array optionally
1028 as specified by parameter.
1029 * lto-streamer-out.c (create_output_block): Avoid maintaining
1030 the node array in the writer cache.
1031 (DFS_write_tree): Remove assertion.
1032 (produce_asm_for_decls): Free the out decl state hash table early.
1033 * lto-streamer-in.c (lto_data_in_create): Adjust for
1034 streamer_tree_cache_create prototype change.
1036 2014-04-03 Richard Biener <rguenther@suse.de>
1038 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
1039 set TREE_CHAIN to NULL_TREE.
1041 2014-04-03 Richard Biener <rguenther@suse.de>
1043 PR tree-optimization/60740
1044 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
1045 over all GIMPLE_COND operands.
1047 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
1049 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
1050 (Weffc++): Remove Scott's numbering, merge lists and reference
1053 2014-04-03 Nick Clifton <nickc@redhat.com>
1055 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
1058 2014-04-03 Martin Jambor <mjambor@suse.cz>
1060 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
1061 mention gcc_unreachable before failing.
1062 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
1065 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
1068 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
1069 inconsistent code and instead mark the context inconsistent.
1070 (possible_polymorphic_call_targets): For inconsistent contexts
1071 return empty complete list.
1073 2014-04-02 Anthony Green <green@moxielogic.com>
1075 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
1076 (extendqisi2, extendhisi2): Define.
1077 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
1078 (WCHAR_TYPE): Change to unsigned int.
1080 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1082 PR tree-optimization/60733
1083 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
1084 insertion point for PHI candidates to be the end of the feeding
1085 block for the PHI argument.
1087 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
1089 PR rtl-optimization/60650
1090 * lra-constraints.c (process_alt_operands): Decrease reject for
1091 earlyclobber matching.
1093 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1095 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
1097 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1099 * config/spu/spu.c (pad_bb): Do not crash when the last
1100 insn is CODE_FOR_blockage.
1102 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1104 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
1105 lies outside the target mode.
1107 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1110 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1111 software floating point or no floating point registers, do not
1112 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1113 in GPRs that occurs after we tested for GPRs that would never be
1116 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1117 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1118 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1119 specifically allow DDmode, since that does not use the SPE SIMD
1122 2014-04-02 Richard Biener <rguenther@suse.de>
1125 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
1126 MODE_INTs. Properly use negv_optab.
1127 (expand_abs): Likewise.
1129 2014-04-02 Richard Biener <rguenther@suse.de>
1132 * Makefile.in (install-driver): Guard extra installs with special
1135 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1137 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1140 2014-04-01 Richard Henderson <rth@redhat.com>
1143 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
1144 alternative enabled before register allocation.
1146 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
1148 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
1149 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
1151 (nios2_large_got_address): Remove unneeded 'sym' parameter.
1152 (nios2_got_address): Update nios2_large_got_address call site.
1153 (nios2_delegitimize_address): New function.
1154 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
1155 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
1156 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
1158 2014-04-01 Martin Husemann <martin@duskware.de>
1160 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
1163 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
1165 PR rtl-optimization/60604
1166 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
1167 check from register_operand.
1168 (register_operand): Redefine in terms of general_operand.
1169 (nonmemory_operand): Use register_operand for the non-constant cases.
1171 2014-04-01 Richard Biener <rguenther@suse.de>
1173 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
1175 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
1177 * doc/invoke.texi (mapp-regs): Clarify.
1179 2014-03-31 Ulrich Drepper <drepper@gmail.com>
1181 * config/i386/avx512fintrin.h (__v32hi): Define type.
1182 (__v64qi): Likewise.
1183 (_mm512_set1_epi8): Define.
1184 (_mm512_set1_epi16): Define.
1185 (_mm512_set4_epi32): Define.
1186 (_mm512_set4_epi64): Define.
1187 (_mm512_set4_pd): Define.
1188 (_mm512_set4_ps): Define.
1189 (_mm512_setr4_epi64): Define.
1190 (_mm512_setr4_epi32): Define.
1191 (_mm512_setr4_pd): Define.
1192 (_mm512_setr4_ps): Define.
1193 (_mm512_setzero_epi32): Define.
1195 2014-03-31 Martin Jambor <mjambor@suse.cz>
1198 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
1199 callsite_arguments_match_p. Updated all callers. Also check types of
1200 corresponding formal parameters and actual arguments.
1201 (not_all_callers_have_enough_arguments_p) Renamed to
1202 some_callers_have_mismatched_arguments_p.
1204 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
1206 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
1208 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
1211 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
1214 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
1216 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
1217 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
1219 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
1220 Use FMAMODE_NOVF512 mode iterator.
1221 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
1222 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
1223 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
1224 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
1226 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
1227 Use VF_128_256 mode iterator.
1228 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
1231 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1233 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
1234 static chain if needed.
1236 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1239 * lra-constraints.c (index_part_to_reg): New.
1240 (process_address): Use it.
1242 2014-03-27 Jeff Law <law@redhat.com>
1243 Jakub Jelinek <jakub@redhat.com>
1246 * expr.c (do_tablejump): Use simplify_gen_binary rather than
1247 gen_rtx_{PLUS,MULT} to build up the address expression.
1249 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
1250 creating non-canonical RTL.
1252 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1255 * ipa-inline.c (want_inline_small_function_p): Short circuit large
1256 functions; reorganize to make cheap checks first.
1257 (inline_small_functions): Do not estimate growth when dumping;
1259 * ipa-inline.h (inline_summary): Add min_size.
1260 (growth_likely_positive): New function.
1261 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
1262 (set_cond_stmt_execution_predicate): Cleanup.
1263 (estimate_edge_size_and_time): Compute min_size.
1264 (estimate_calls_size_and_time): Likewise.
1265 (estimate_node_size_and_time): Likewise.
1266 (inline_update_overall_summary): Update min_size.
1267 (do_estimate_edge_time): Likewise.
1268 (do_estimate_edge_size): Update.
1269 (do_estimate_edge_hints): Update.
1270 (growth_likely_positive): New function.
1272 2014-03-28 Jakub Jelinek <jakub@redhat.com>
1275 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
1276 also if addr has VOIDmode.
1278 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1280 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
1281 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
1283 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
1284 instructions as well as AdvancedSIMD loads.
1286 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1288 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1289 Use crypto_aese type.
1290 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
1291 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
1292 crypto_aese, crypto_aesmc. Move to types.md.
1293 * config/arm/types.md (crypto_aes): Split into crypto_aese,
1295 * config/arm/iterators.md (crypto_type): Likewise.
1297 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1299 * cgraph.c: Include expr.h and tree-dfa.h.
1300 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
1303 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1306 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
1307 regs from checking multi-reg pseudos.
1309 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1311 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
1313 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1315 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
1316 if it would clobber the stack pointer, even temporarily.
1318 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
1320 * mode-switching.c: Make small adjustments to the top comment.
1322 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1324 * config/rs6000/constraints.md (wD constraint): New constraint to
1325 match the constant integer to get the top DImode/DFmode out of a
1326 vector in a VSX register.
1328 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
1329 match the constant integer to get the top DImode/DFmode out of a
1330 vector in a VSX register.
1332 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
1335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1338 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
1339 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
1341 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
1342 Optimize vec_extract of 64-bit values, where the value being
1343 extracted is in the top word, where we can use scalar
1344 instructions. Add direct move and store support. Combine the big
1345 endian/little endian vector select load support into a single insn.
1346 (vsx_extract_<mode>_internal1): Likewise.
1347 (vsx_extract_<mode>_internal2): Likewise.
1348 (vsx_extract_<mode>_load): Likewise.
1349 (vsx_extract_<mode>_store): Likewise.
1350 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
1351 combined into vsx_extract_<mode>_load.
1352 (vsx_extract_<mode>_one_le): Likewise.
1354 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
1355 define the top 64-bit vector element.
1357 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
1360 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1361 Document vec_vbpermq builtin.
1364 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
1365 enable use of xxsldwi and xxpermdi builtin functions.
1366 (vec_xxpermdi): Likewise.
1368 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1369 Document use of vec_xxsldwi and vec_xxpermdi builtins.
1371 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
1373 PR rtl-optimization/60650
1374 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
1376 (find_spills_for): New.
1377 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
1378 Spill all pseudos on the second iteration.
1380 2014-03-27 Marek Polacek <polacek@redhat.com>
1383 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
1386 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1388 * config/s390/s390.c (s390_can_use_return_insn): Check for
1389 call-saved FPRs on 31 bit.
1391 2014-03-27 Jakub Jelinek <jakub@redhat.com>
1394 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
1395 if they need regimplification, just drop them instead of
1396 calling gimple_regimplify_operands on them.
1398 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
1401 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
1402 (aarch64_frame_pointer_required): Adjust logic.
1403 (aarch64_can_eliminate): Adjust logic.
1404 (aarch64_override_options_after_change): Adjust logic.
1406 2014-03-27 Dehao Chen <dehao@google.com>
1408 * ipa-inline.c (early_inliner): Update node's inline info.
1410 2014-03-26 Dehao Chen <dehao@google.com>
1412 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
1413 compiler inserted conditional jumps for NAN float check.
1415 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1417 * ubsan.h (ubsan_create_data): Change second argument's type
1418 to const location_t *.
1419 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
1421 (ubsan_create_data): Change second argument to const location_t *PLOC.
1422 Create Loc field whenever PLOC is non-NULL.
1423 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
1424 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
1428 * real.c (real_to_integer2): Change type of low to UHWI.
1430 2014-03-26 Tobias Burnus <burnus@net-b.de>
1432 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
1433 (CILK_SELF_SPECS): New define.
1434 (driver_self_specs): Use it.
1436 2014-03-26 Richard Biener <rguenther@suse.de>
1438 * tree-pretty-print.c (percent_K_format): Implement special
1439 case for LTO and its stripped down BLOCK tree.
1441 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1444 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
1446 * tree-vrp.c (simplify_internal_call_using_ranges): If only
1447 one range is range_int_cst_p, but not both, at least optimize
1448 addition/subtraction of 0 and multiplication by 0 or 1.
1449 * gimple-fold.c (gimple_fold_call): Fold
1450 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
1451 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
1452 INTEGER_CSTs, try to fold at least x * 0 and y - y.
1454 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
1456 PR rtl-optimization/60452
1457 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
1458 <case REG>: Return 1 for invalid offsets from the frame pointer.
1460 2014-03-26 Marek Polacek <polacek@redhat.com>
1463 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1466 2014-03-26 Marek Polacek <polacek@redhat.com>
1469 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1472 2014-03-26 Marek Polacek <polacek@redhat.com>
1475 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1476 multiplication in unsigned type.
1478 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1480 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1482 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1484 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1486 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1489 * cif-code.def (UNREACHABLE) New code.
1490 * ipa-inline.c (inline_small_functions): Skip edges to
1491 __builtlin_unreachable.
1492 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1493 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1494 predicate to __bulitin_unreachable.
1495 (set_cond_stmt_execution_predicate): Fix issue when
1496 invert_tree_comparison returns ERROR_MARK.
1497 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1498 propagate to inline clones.
1499 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1501 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1502 * cgraphclones.c (cgraph_clone_node): If call destination is already
1503 ureachable, do not redirect it back.
1504 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1507 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1509 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1510 Do not modify inline clones.
1512 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1514 * config/i386/i386.md (general_sext_operand): New mode attr.
1515 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1516 don't generate (sign_extend (const_int)).
1517 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1518 operands[2]. Use We constraint instead of <i> and
1519 <general_sext_operand> predicate instead of <general_operand>.
1520 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1521 * config/i386/constraints.md (We): New constraint.
1522 * config/i386/predicates.md (x86_64_sext_operand,
1523 sext_operand): New predicates.
1525 2014-03-25 Martin Jambor <mjambor@suse.cz>
1528 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1529 inconsistent devirtualizations to __builtin_unreachable.
1531 2014-03-25 Marek Polacek <polacek@redhat.com>
1534 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1536 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1538 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1539 order of elements for big-endian.
1541 2014-03-25 Richard Biener <rguenther@suse.de>
1544 * gimplify-me.c (gimple_regimplify_operands): Update the
1547 2014-03-25 Martin Jambor <mjambor@suse.cz>
1550 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1551 (lto_output_varpool_node): Likewise.
1552 (input_overwrite_node): Likewise.
1553 (input_varpool_node): Likewise.
1555 2014-03-25 Richard Biener <rguenther@suse.de>
1557 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1558 (run_gcc): Likewise.
1560 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1562 * combine.c (simplify_compare_const): Add MODE argument.
1563 Handle mode_width 0 as very large mode_width.
1564 (try_combine, simplify_comparison): Adjust callers.
1566 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1567 type to avoid signed integer overflow.
1568 * explow.c (plus_constant): Likewise.
1570 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1572 * doc/generic.texi: Correct typos.
1574 2014-03-24 Tobias Burnus <burnus@net-b.de>
1576 * doc/invoke.texi (-flto): Expand section about
1577 using static libraries with LTO.
1579 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1581 PR rtl-optimization/60501
1582 * optabs.def (addptr3_optab): New optab.
1583 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1584 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1585 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1587 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1589 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1591 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1593 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1596 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1597 (_mm256_undefined_ps): Define.
1598 (_mm256_undefined_pd): Define.
1599 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1600 (_mm_undefined_pd): Define.
1601 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1602 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1603 (_mm512_undefined_ps): Define.
1604 (_mm512_undefined_pd): Define.
1605 Use _mm*_undefined_*.
1606 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1608 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1610 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1611 (lshr_simd): DI mode added.
1612 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1613 (aarch64_ushr_simddi): Likewise.
1614 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1615 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1616 (vshrd_n_u64): Likewise.
1618 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1620 * Makefile.in (s-macro_list): Depend on cc1.
1622 2014-03-23 Teresa Johnson <tejohnson@google.com>
1624 * ipa-utils.c (ipa_print_order): Use specified dump file.
1626 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1628 PR rtl-optimization/60601
1629 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1631 * gcc.c (eval_spec_function): Initialize save_growing_value.
1633 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1636 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1637 code == MINUS_EXPR, never swap op0 with op1.
1639 * toplev.c (init_local_tick): Avoid signed integer multiplication
1641 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1642 shift by first operand's bitsize.
1644 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1647 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1649 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1650 TARGET_ISA_64BIT_P(x).
1652 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1654 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1655 pattern for vector nor instead of subtract from splat(-1).
1656 (altivec_expand_vec_perm_const_le): Likewise.
1658 2014-03-21 Richard Henderson <rth@twiddle.net>
1661 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1662 related insns after epilogue_completed.
1664 2014-03-21 Martin Jambor <mjambor@suse.cz>
1667 * cgraph.h (symtab_node): New flag body_removed.
1668 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1669 when removing bodies.
1670 * symtab.c (dump_symtab_base): Dump body_removed flag.
1671 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1672 had their bodies removed.
1674 2014-03-21 Martin Jambor <mjambor@suse.cz>
1677 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1680 2014-03-21 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/60577
1683 * tree-core.h (struct tree_base): Document nothrow_flag use
1684 in VAR_DECL_NONALIASED.
1685 * tree.h (VAR_DECL_NONALIASED): New.
1686 (may_be_aliased): Adjust.
1687 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1689 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1691 * expr.c (expand_expr_real_1): Remove outdated comment.
1693 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1696 * ira.c (adjust_cleared_regs): Call copy_rtx on
1697 *reg_equiv[REGNO (loc)].src_p before passing it to
1698 simplify_replace_fn_rtx.
1701 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1702 into CONST, put pic register as first operand of PLUS. Use
1703 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1705 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1707 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1709 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1711 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1712 around for store forwarding issue in the FPU on the UT699.
1713 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1714 loads and operations if -mfix-ut699 is specified.
1715 (divtf3_hq): Tweak attribute.
1716 (sqrttf2_hq): Likewise.
1718 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1720 * calls.c (store_one_arg): Remove incorrect const qualification on the
1721 type of the temporary.
1722 * cfgexpand.c (expand_return): Likewise.
1723 * expr.c (expand_constructor): Likewise.
1724 (expand_expr_real_1): Likewise.
1726 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1728 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1731 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1734 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1736 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1738 * config/arm/aarch-common-protos.h
1739 (alu_cost_table): Fix spelling of "extend".
1740 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1742 2014-03-19 Richard Biener <rguenther@suse.de>
1745 * tree-core.h (tree_type_common): Re-order pointer members
1746 to reduce recursion depth during GC walks.
1748 2014-03-19 Marek Polacek <polacek@redhat.com>
1751 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1752 before accessing it.
1754 2014-03-19 Richard Biener <rguenther@suse.de>
1757 * lto-streamer-in.c (input_function): In WPA stage do not drop
1760 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1762 PR tree-optimization/60559
1763 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1764 with build_zero_cst assignment.
1766 2014-03-18 Kai Tietz <ktietz@redhat.com>
1768 PR rtl-optimization/56356
1769 * sdbout.c (sdbout_parms): Verify that parms'
1770 incoming argument is valid.
1771 (sdbout_reg_parms): Likewise.
1773 2014-03-18 Richard Henderson <rth@redhat.com>
1776 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1777 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1778 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1780 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1782 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1783 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1784 Italicize plugin event names in description. Explain that
1785 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1786 Remind that no GCC functions should be called after PLUGIN_FINISH.
1787 Explain what pragmas with expansion are.
1789 2014-03-18 Martin Liska <mliska@suse.cz>
1791 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1792 gimple call statement is update.
1793 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1794 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1796 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1799 * ubsan.c (ubsan_instrument_unreachable): Call
1800 initialize_sanitizer_builtins.
1801 (ubsan_pass): Likewise.
1804 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1805 varpool_finalize_decl instead of rest_of_decl_compilation.
1807 2014-03-18 Richard Biener <rguenther@suse.de>
1809 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1810 by using bitmap_and_compl instead of bitmap_and_compl_into.
1811 (df_rd_transfer_function): Likewise.
1813 2014-03-18 Richard Biener <rguenther@suse.de>
1815 * doc/lto.texi (fresolution): Fix typo.
1817 2014-03-18 Richard Biener <rguenther@suse.de>
1819 * doc/invoke.texi (flto): Update for changes in 4.9.
1821 2014-03-18 Richard Biener <rguenther@suse.de>
1823 * doc/loop.texi: Remove section on the removed lambda framework.
1824 Update loop docs with recent changes in preserving loop structure.
1826 2014-03-18 Richard Biener <rguenther@suse.de>
1828 * doc/lto.texi (-fresolution): Document.
1830 2014-03-18 Richard Biener <rguenther@suse.de>
1832 * doc/contrib.texi: Adjust my name.
1834 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1837 * internal-fn.c: Include diagnostic-core.h.
1838 (expand_BUILTIN_EXPECT): New function.
1839 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1840 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1841 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1842 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1844 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1845 Revert 3 argument __builtin_expect code.
1846 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1847 * gimple-fold.c (gimple_fold_call): Likewise.
1848 * tree.h (fold_builtin_expect): New prototype.
1849 * builtins.c (build_builtin_expect_predicate): Add predictor
1850 argument, if non-NULL, create 3 argument __builtin_expect.
1851 (fold_builtin_expect): No longer static. Add ARG2 argument,
1852 pass it through to build_builtin_expect_predicate.
1853 (fold_builtin_2): Adjust caller.
1854 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1855 * internal-fn.def (BUILTIN_EXPECT): New.
1857 2014-03-18 Tobias Burnus <burnus@net-b.de>
1860 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1861 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1862 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1864 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1867 * predict.c (combine_predictions_for_bb): Fix up formatting.
1868 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1869 fill what it points to if non-NULL.
1870 (tree_predict_by_opcode): Adjust caller, use the predictor.
1871 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1873 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1875 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1876 proper constant for the store mode.
1878 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1880 * symtab.c (change_decl_assembler_name): Fix transparent alias
1883 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1885 * config/aarch64/aarch64.c: Correct the comments about the
1886 aarch64 stack layout.
1888 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1890 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1891 check for GF_OMP_FOR_KIND_FOR.
1893 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1895 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1896 ymm and zmm register names.
1898 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1901 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1902 note creation for the 2010-08-31 changes.
1904 2014-03-17 Marek Polacek <polacek@redhat.com>
1907 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1908 as -fno-tree-loop-vectorize.
1909 (expand_omp_simd): Likewise.
1911 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1913 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1914 (eligible_for_call_delay): New prototype.
1915 * config/sparc/sparc.c (tls_call_delay): Rename into...
1916 (eligible_for_call_delay): ...this. Return false if the instruction
1917 cannot be put in the delay slot of a branch.
1918 (eligible_for_restore_insn): Simplify.
1919 (eligible_for_return_delay): Return false if the instruction cannot be
1920 put in the delay slot of a branch and simplify.
1921 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1922 put in the delay slot of a branch.
1923 * config/sparc/sparc.md (fix_ut699): New attribute.
1924 (tls_call_delay): Delete.
1925 (in_call_delay): Reimplement.
1926 (eligible_for_sibcall_delay): Rename into...
1927 (in_sibcall_delay): ...this.
1928 (eligible_for_return_delay): Rename into...
1929 (in_return_delay): ...this.
1930 (in_branch_delay): Reimplement.
1931 (in_uncond_branch_delay): Delete.
1932 (in_annul_branch_delay): Delete.
1934 2014-03-14 Richard Henderson <rth@redhat.com>
1937 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1938 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1939 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1940 (floathi splitters): Remove.
1941 (float<SWI48x>xf2): New pattern.
1942 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1943 code that tried to handle DImode for 32-bit, but which was excluded
1944 by the pattern's condition. Drop allocation of stack temporary.
1945 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1946 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1947 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1948 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1949 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1950 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1951 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1952 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1953 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1954 (*float<SWI48x><X87MODEF>2_i387): Remove.
1955 (all float _with_temp splitters): Remove.
1956 (*float<SWI48x><MODEF>2_i387): New pattern.
1957 (*float<SWI48><MODEF>2_sse): New pattern.
1958 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1959 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1961 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1962 Marek Polacek <polacek@redhat.com>
1965 * common.opt (dump_base_name_prefixed): New Variable.
1966 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1967 if x_dump_base_name_prefixed is already set, set it at the end.
1969 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1971 PR rtl-optimization/60508
1972 * lra-constraints.c (get_reload_reg): Add new parameter
1974 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1975 Pass the new parameter values.
1977 2014-03-14 Richard Biener <rguenther@suse.de>
1979 * common.opt: Revert unintented changes from r205065.
1982 2014-03-14 Richard Biener <rguenther@suse.de>
1985 * cfghooks.c (split_block): Properly adjust all loops the
1986 block was a latch of.
1988 2014-03-14 Martin Jambor <mjambor@suse.cz>
1991 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1994 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1997 * config/avr/avr.c (avr_set_current_function): Pass function name
1998 through default_strip_name_encoding before sanity checking instead
1999 of skipping the first char of the assembler name.
2001 2014-03-13 Richard Henderson <rth@redhat.com>
2004 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
2005 (ix86_force_to_memory, ix86_free_from_memory): Remove.
2006 * config/i386/i386-protos.h: Likewise.
2007 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
2008 in the expander instead of a splitter.
2009 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
2010 any possibility of requiring a memory.
2011 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
2012 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
2013 (fp branch splitters): Update for ix86_split_fp_branch.
2014 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
2015 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
2016 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
2017 (*fop_<MODEF>_2_i387): Remove f/r alternative.
2018 (*fop_<MODEF>_3_i387): Likewise.
2019 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
2020 (splitters for the fop_* register patterns): Remove.
2021 (fscalexf4_i387): Rename from *fscalexf4_i387.
2022 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
2024 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2026 PR tree-optimization/59779
2027 * tree-dfa.c (get_ref_base_and_extent): Use double_int
2028 type for bitsize and maxsize instead of HOST_WIDE_INT.
2030 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
2032 PR rtl-optimization/57320
2033 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
2034 the CFG after thread_prologue_and_epilogue_insns.
2036 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
2038 PR rtl-optimization/57189
2039 * lra-constraints.c (process_alt_operands): Disfavor spilling
2042 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
2044 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
2046 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2048 PR tree-optimization/59025
2050 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
2051 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
2053 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
2056 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
2057 calls of avr_out_plus_1.
2059 2014-03-13 Bin Cheng <bin.cheng@arm.com>
2061 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
2062 BB's single pred and update the father loop's latch info later.
2064 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2066 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
2070 (VEC_base): Likewise.
2071 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
2072 registers, we need to swap double words in little endian mode.
2074 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
2075 to be a container mode for 128-bit integer operations added in ISA
2076 2.07. Unlike TImode and PTImode, the preferred register set is
2077 the Altivec/VMX registers for the 128-bit operations.
2079 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
2081 (rs6000_split_128bit_ok_p): Likewise.
2083 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
2084 macros for creating ISA 2.07 normal and overloaded builtin
2085 functions with 3 arguments.
2086 (BU_P8V_OVERLOAD_3): Likewise.
2087 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
2088 for use as overloaded functions.
2089 (VPERM_1TI_UNS): Likewise.
2090 (VSEL_1TI): Likewise.
2091 (VSEL_1TI_UNS): Likewise.
2092 (ST_INTERNAL_1ti): Likewise.
2093 (LD_INTERNAL_1ti): Likewise.
2094 (XXSEL_1TI): Likewise.
2095 (XXSEL_1TI_UNS): Likewise.
2096 (VPERM_1TI): Likewise.
2097 (VPERM_1TI_UNS): Likewise.
2098 (XXPERMDI_1TI): Likewise.
2099 (SET_1TI): Likewise.
2100 (LXVD2X_V1TI): Likewise.
2101 (STXVD2X_V1TI): Likewise.
2102 (VEC_INIT_V1TI): Likewise.
2103 (VEC_SET_V1TI): Likewise.
2104 (VEC_EXT_V1TI): Likewise.
2105 (EQV_V1TI): Likewise.
2106 (NAND_V1TI): Likewise.
2107 (ORC_V1TI): Likewise.
2108 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
2109 added in ISA 2.07. Add both normal 'altivec' builtins, and the
2111 (VADDUQM): Likewise.
2112 (VSUBCUQ): Likewise.
2113 (VADDEUQM): Likewise.
2114 (VADDECUQ): Likewise.
2115 (VSUBEUQM): Likewise.
2116 (VSUBECUQ): Likewise.
2118 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
2119 __int128_t and __uint128_t types.
2120 (__uint128_type): Likewise.
2121 (altivec_categorize_keyword): Add support for vector __int128_t,
2122 vector __uint128_t, vector __int128, and vector unsigned __int128
2123 as a container type for TImode operations that need to be done in
2124 VSX/Altivec registers.
2125 (rs6000_macro_to_expand): Likewise.
2126 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
2127 to support 128-bit integer instructions vaddcuq, vadduqm,
2128 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
2129 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
2131 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
2132 for V1TImode, and set up preferences to use VSX/Altivec registers.
2133 Setup VSX reload handlers.
2134 (rs6000_debug_reg_global): Likewise.
2135 (rs6000_init_hard_regno_mode_ok): Likewise.
2136 (rs6000_preferred_simd_mode): Likewise.
2137 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
2138 (easy_altivec_constant): Likewise.
2139 (output_vec_const_move): Likewise.
2140 (rs6000_expand_vector_set): Convert V1TImode set and extract to
2142 (rs6000_expand_vector_extract): Likewise.
2143 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
2145 (rs6000_const_vec): Add support for V1TImode.
2146 (rs6000_emit_le_vsx_load): Swap double words when loading or
2147 storing TImode/V1TImode.
2148 (rs6000_emit_le_vsx_store): Likewise.
2149 (rs6000_emit_le_vsx_move): Likewise.
2150 (rs6000_emit_move): Add support for V1TImode.
2151 (altivec_expand_ld_builtin): Likewise.
2152 (altivec_expand_st_builtin): Likewise.
2153 (altivec_expand_vec_init_builtin): Likewise.
2154 (altivec_expand_builtin): Likewise.
2155 (rs6000_init_builtins): Add support for V1TImode type. Add
2156 support for ISA 2.07 128-bit integer builtins. Define type names
2157 for the VSX/Altivec vector types.
2158 (altivec_init_builtins): Add support for overloaded vector
2159 functions with V1TImode type.
2160 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
2161 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
2163 (rs6000_split_128bit_ok_p): Likewise.
2164 (rs6000_handle_altivec_attribute): Create V1TImode from vector
2165 __int128_t and vector __uint128_t.
2167 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
2168 and mode attributes.
2175 (VS_scalar): Likewise.
2176 (VS_double): Likewise.
2177 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
2179 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
2180 we support the ISA 2.07 128-bit integer arithmetic instructions.
2181 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
2182 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
2183 and TImode types for use with the builtin functions.
2184 (V1TI_type_node): Likewise.
2185 (unsigned_V1TI_type_node): Likewise.
2186 (intTI_type_internal_node): Likewise.
2187 (uintTI_type_internal_node): Likewise.
2189 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
2190 128-bit builtin functions.
2191 (UNSPEC_VADDEUQM): Likewise.
2192 (UNSPEC_VADDECUQ): Likewise.
2193 (UNSPEC_VSUBCUQ): Likewise.
2194 (UNSPEC_VSUBEUQM): Likewise.
2195 (UNSPEC_VSUBECUQ): Likewise.
2196 (VM): Add V1TImode to vector mode iterators.
2198 (VI_unit): Likewise.
2199 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
2200 (altivec_vaddcuq): Likewise.
2201 (altivec_vsubuqm): Likewise.
2202 (altivec_vsubcuq): Likewise.
2203 (altivec_vaddeuqm): Likewise.
2204 (altivec_vaddecuq): Likewise.
2205 (altivec_vsubeuqm): Likewise.
2206 (altivec_vsubecuq): Likewise.
2208 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
2210 (BOOL_128): Likewise.
2211 (BOOL_REGS_OUTPUT): Likewise.
2212 (BOOL_REGS_OP1): Likewise.
2213 (BOOL_REGS_OP2): Likewise.
2214 (BOOL_REGS_UNARY): Likewise.
2215 (BOOL_REGS_AND_CR0): Likewise.
2217 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
2218 128-bit integer builtin support.
2219 (vec_vadduqm): Likewise.
2220 (vec_vaddecuq): Likewise.
2221 (vec_vaddeuqm): Likewise.
2222 (vec_vsubecuq): Likewise.
2223 (vec_vsubeuqm): Likewise.
2224 (vec_vsubcuq): Likewise.
2225 (vec_vsubuqm): Likewise.
2227 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2228 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
2229 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
2230 128-bit integer add/subtract to ISA 2.07.
2232 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
2234 * config/arc/arc.c (arc_predicate_delay_insns):
2235 Fix third argument passed to conditionalize_nonjump.
2237 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
2239 * config/aarch64/aarch64-builtins.c
2240 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
2241 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
2242 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
2243 instead of __builtin_lfloor.
2244 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
2246 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2248 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
2249 (tree_ssa_ifcombine_bb_1): New function.
2250 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
2251 is an empty forwarder block to then_bb or vice versa and then_bb
2252 and else_bb are effectively swapped.
2254 2014-03-12 Christian Bruel <christian.bruel@st.com>
2257 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
2258 REG_CFA_DEF_CFA note.
2259 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
2260 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
2262 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
2264 PR tree-optimization/60454
2265 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
2267 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2269 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
2270 Do not define target_cpu_default2 to generic.
2271 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
2272 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
2273 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
2275 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2276 Marc Glisse <marc.glisse@inria.fr>
2278 PR tree-optimization/60502
2279 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
2280 instead of build_low_bits_mask.
2282 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2285 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
2286 if there are multiple uses, but op doesn't live on E edge.
2287 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
2288 clobber stmts before __builtin_unreachable.
2290 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
2292 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
2293 hard_frame_pointer_rtx.
2294 * cse.c (cse_insn): Remove volatile check.
2295 * cselib.c (cselib_process_insn): Likewise.
2296 * dse.c (scan_insn): Likewise.
2298 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2300 * config/arc/arc.c (conditionalize_nonjump): New function,
2302 (arc_ifcvt): ... this.
2303 (arc_predicate_delay_insns): Use it.
2305 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2307 * config/arc/predicates.md (extend_operand): During/after reload,
2308 allow const_int_operand.
2309 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
2310 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
2311 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
2313 (umulsi3_highpart_i): Likewise.
2315 2014-03-11 Richard Biener <rguenther@suse.de>
2317 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
2318 Add asserts to guard possible wrong-code bugs.
2320 2014-03-11 Richard Biener <rguenther@suse.de>
2322 PR tree-optimization/60429
2323 PR tree-optimization/60485
2324 * tree-ssa-structalias.c (set_union_with_increment): Properly
2325 take into account all fields that overlap the shifted vars.
2326 (do_sd_constraint): Likewise.
2327 (do_ds_constraint): Likewise.
2328 (get_constraint_for_ptr_offset): Likewise.
2330 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
2332 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
2333 (nios2_compute_frame_layout):
2334 Add calculation of cfun->machine->fp_save_offset.
2335 (nios2_expand_prologue): Correct setting of frame pointer register
2337 (nios2_expand_epilogue): Update recovery of stack pointer from
2338 frame pointer accordingly.
2339 (nios2_initial_elimination_offset): Update calculation of offset
2340 for eliminating to HARD_FRAME_POINTER_REGNUM.
2342 2014-03-10 Jakub Jelinek <jakub@redhat.com>
2345 * ipa.c (symtab_remove_unreachable_nodes): Don't call
2346 cgraph_get_create_node on VAR_DECLs.
2348 2014-03-10 Richard Biener <rguenther@suse.de>
2351 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
2353 2014-03-08 Douglas B Rupp <rupp@gnat.com>
2355 * config/vms/vms.opt (vms_float_format): New variable.
2357 2014-03-08 Tobias Burnus <burnus@net-b.de>
2359 * doc/invoke.texi (-fcilkplus): Update implementation status.
2361 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
2362 Richard Biener <rguenther@suse.de>
2364 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
2365 consistently accross all TUs.
2366 (run_gcc): Enable -fshort-double automatically at link at link-time
2367 and disallow override.
2369 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
2372 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
2373 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
2374 if they can't be used.
2376 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2378 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
2379 for Solaris 11/x86 ld.
2380 * configure: Regenerate.
2382 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2384 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
2385 (LIB_TLS_SPEC): Save as ld_tls_libs.
2386 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
2387 (HAVE_AS_IX86_TLSLDM): New test.
2388 * configure, config.in: Regenerate.
2389 * config/i386/i386.c (legitimize_tls_address): Fall back to
2390 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
2391 cannot support TLS_MODEL_LOCAL_DYNAMIC.
2392 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
2393 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
2395 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
2397 * common.opt (fira-loop-pressure): Mark as optimization.
2399 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
2401 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
2402 an OpenMP mappable type.
2404 2014-03-06 Matthias Klose <doko@ubuntu.com>
2406 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
2407 MULTILIB_OSDIRNAMES is not defined.
2409 2014-03-06 Jakub Jelinek <jakub@redhat.com>
2410 Meador Inge <meadori@codesourcery.com>
2413 * config/arm/arm.c (arm_tls_symbol_p): Remove.
2414 (arm_legitimize_address): Call legitimize_tls_address for any
2415 arm_tls_referenced_p expression, handle constant addend. Call it
2416 before testing for !TARGET_ARM.
2417 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
2419 2014-03-06 Richard Biener <rguenther@suse.de>
2425 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2427 * tree-streamer.c (record_common_node): Assert we don't record
2428 nodes with type double.
2429 (preload_common_node): Skip type double, complex double and double
2430 pointer since it is now frontend dependent due to fshort-double option.
2432 2014-03-06 Richard Biener <rguenther@suse.de>
2434 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
2435 or -fno-lto is specified and the linker has full plugin support.
2436 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
2437 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
2438 * lto-wrapper.c (merge_and_complain): Merge compile-time
2439 optimization levels.
2440 (run_gcc): And pass it through to the link options.
2442 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
2446 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2448 * cselib.c (remove_useless_values): Skip to avoid quadratic
2449 behavior if the condition moved from...
2450 (cselib_process_insn): ... here holds.
2452 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2455 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
2456 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2459 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2460 (TM_H): Add x86-tune.def.
2462 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2464 * config/aarch64/aarch64.c (generic_tunings):
2465 Use cortexa57_extra_costs.
2467 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2470 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2471 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2472 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2475 2014-03-04 Heiher <r@hev.cc>
2477 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2478 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2480 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2482 * config/i386/predicates.md (const2356_operand): Change to ...
2483 (const2367_operand): ... this.
2484 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2486 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2487 (*avx512pf_scatterpf<mode>sf): Ditto.
2488 (avx512pf_scatterpf<mode>df): Ditto.
2489 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2490 (*avx512pf_scatterpf<mode>df): Ditto.
2491 * config/i386/i386.c (ix86_expand_builtin): Update
2492 incorrect hint operand error message.
2494 2014-03-04 Richard Biener <rguenther@suse.de>
2496 * lto-section-in.c (lto_get_section_data): Fix const cast.
2498 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2500 * tree-streamer.c (record_common_node): Assert we don't record
2501 nodes with type double.
2502 (preload_common_node): Skip type double, complex double and double
2503 pointer since it is now frontend dependent due to fshort-double option.
2505 2014-03-04 Richard Biener <rguenther@suse.de>
2508 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2509 (lto_input_toplevel_asms): Likewise.
2510 * lto-section-in.c (lto_get_section_data): Instead do it here
2513 2014-03-04 Richard Biener <rguenther@suse.de>
2515 PR tree-optimization/60382
2516 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2517 dead PHIs a reduction.
2519 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2521 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2523 (_mm_prefetch): Move out of GCC target("sse") pragma.
2524 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2525 GCC target("prfchw") pragma.
2526 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2528 * config/i386/i386.c (ix86_option_override_internal): Enable
2529 -mprfchw with -mprefetchwt1.
2531 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2533 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2536 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2538 * opts.h (CL_PCH_IGNORE): Define.
2539 * targhooks.c (option_affects_pch_p):
2540 Return false for options that have CL_PCH_IGNORE set.
2541 * opt-functions.awk: Process PchIgnore.
2542 * doc/options.texi: Document PchIgnore.
2544 * config/arc/arc.opt (misize): Add PchIgnore property.
2546 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2548 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2549 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2550 constraint on constants to permit them being loaded into
2551 GENERAL_REGS or BASE_REGS.
2553 2014-03-03 Nick Clifton <nickc@redhat.com>
2555 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2556 anti-cacnonical alternatives.
2557 (negandhi3_real): New pattern.
2558 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2560 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2562 * config/avr/avr-mcus.def: Remove atxmega16x1.
2563 * config/avr/avr-tables.opt: Regenerate.
2564 * config/avr/t-multilib: Regenerate.
2565 * doc/avr-mmcu.texi: Regenerate.
2567 2014-03-03 Tobias Grosser <tobias@grosser.es>
2568 Mircea Namolaru <mircea.namolaru@inria.fr>
2570 PR tree-optimization/58028
2571 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2574 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2576 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2577 not handled by recognizers.
2579 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2582 * function.c (expand_function_end): Don't emit
2583 clobber_return_register sequence if clobber_after is a BARRIER.
2584 * cfgexpand.c (construct_exit_block): Append instructions before
2585 return_label to prev_bb.
2587 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2589 * config/rs6000/constraints.md: Document reserved use of "wc".
2591 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2594 * ipa.c (function_and_variable_visibility): When dissolving comdat
2595 group, also set all symbols to local.
2597 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2602 2013-12-14 Jan Hubicka <jh@suse.cz>
2604 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2606 2014-03-02 Jon Beniston <jon@beniston.com>
2612 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2613 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2614 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2615 (simple_return, *simple_return): New patterns
2616 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2617 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2619 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2621 * dwarf2out.c (gen_subprogram_die): Tidy.
2623 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2626 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2627 (*mov_t_msb_neg_negc): ... this new insn.
2629 2014-02-28 Jason Merrill <jason@redhat.com>
2632 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2635 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2638 * dwarf2out.c (decltype_auto_die): New static.
2639 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2640 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2641 (is_cxx_auto): Likewise.
2643 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2645 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2646 we are not using general regs only.
2648 2014-02-28 Richard Biener <rguenther@suse.de>
2651 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2652 previous fix and only allow to remove trivial pre-headers
2653 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2654 (remove_forwarder_block): Properly update the latch of a loop.
2656 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2659 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2660 (cselib_preserved_hash_table): New.
2661 (preserve_constants_and_equivs): Move preserved vals to it.
2662 (cselib_find_slot): Look it up first.
2663 (cselib_init): Initialize it.
2664 (cselib_finish): Release it.
2665 (dump_cselib_table): Dump it.
2667 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2670 * cselib.c (remove_useless_values): Skip to avoid quadratic
2671 behavior if the condition moved from...
2672 (cselib_process_insn): ... here holds.
2674 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2677 * var-tracking.c (vt_initialize): Apply the same condition to
2678 preserve the CFA base value.
2680 2014-02-28 Joey Ye <joey.ye@arm.com>
2683 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2684 if reload in progress or completed.
2686 2014-02-28 Tobias Burnus <burnus@net-b.de>
2689 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2692 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2694 * doc/tm.texi.in (Condition Code Status): Update documention for
2695 relative locations of cc0-setter and cc0-user.
2697 2014-02-27 Jeff Law <law@redhat.com>
2699 PR rtl-optimization/52714
2700 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2701 into two independent simple sets, if I3 is a jump, ensure the
2702 pattern we place into I3 is a (set (pc) ...).
2704 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2705 Jeff Law <law@redhat.com>
2707 PR rtl-optimization/49847
2708 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2709 are in different blocks.
2710 * doc/tm.texi (Condition Code Status): Update documention for
2711 relative locations of cc0-setter and cc0-user.
2713 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2716 * lra.c (lra_emit_add): Check SUBREG too.
2718 2014-02-27 Andreas Schwab <schwab@suse.de>
2720 * config/m68k/m68k.c (m68k_option_override): Disable
2721 -flive-range-shrinkage for classic m68k.
2722 (m68k_override_options_after_change): Likewise.
2724 2014-02-27 Marek Polacek <polacek@redhat.com>
2727 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2728 -Wmaybe-uninitialized.
2730 2014-02-27 Alan Modra <amodra@gmail.com>
2733 * reload1.c (emit_input_reload_insns): When reload_override_in,
2734 set old to rl->in_reg when rl->in_reg is a subreg.
2736 2014-02-26 Richard Biener <rguenther@suse.de>
2739 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2741 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2743 * common/config/i386/predicates.md (const1256_operand): Remove.
2744 (const2356_operand): New.
2745 (const_1_to_2_operand): Remove.
2746 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2747 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2748 (*avx512pf_gatherpf<mode>sf): Ditto.
2749 (avx512pf_gatherpf<mode>df): Ditto.
2750 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2751 (*avx512pf_gatherpf<mode>df): Ditto.
2752 (avx512pf_scatterpf<mode>sf): Ditto.
2753 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2754 (*avx512pf_scatterpf<mode>sf): Ditto.
2755 (avx512pf_scatterpf<mode>df): Ditto.
2756 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2757 (*avx512pf_scatterpf<mode>df): Ditto.
2758 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2760 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2762 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2763 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2764 (_mm512_mask_testn_epi64_mask): Move to ...
2765 * config/i386/avx512cdintrin.h: Here.
2766 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2767 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2768 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2769 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2770 TARGET_AVX512F from TARGET_AVX512CD.
2772 2014-02-26 Richard Biener <rguenther@suse.de>
2775 * ipa.c (walk_polymorphic_call_targets): Properly guard
2776 call to inline_update_overall_summary.
2778 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2781 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2782 and latches only if requested. Fix latch if it is removed.
2783 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2784 LOOPS_HAVE_PREHEADERS.
2786 2014-02-25 Andrew Pinski <apinski@cavium.com>
2788 * builtins.c (expand_builtin_thread_pointer): Create a new target
2789 when the target is NULL.
2791 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2793 PR rtl-optimization/60317
2794 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2795 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2796 * lra-assigns.c: Include params.h.
2797 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2798 other reload pseudos considerations.
2800 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2802 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2803 to use canonical form for nor<mode>3.
2805 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2808 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2811 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2813 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2814 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2815 (ix86_handle_option): Handle OPT_mprefetchwt1.
2816 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2817 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2819 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2820 OPTION_MASK_ISA_PREFETCHWT1.
2821 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2822 (PTA_PREFETCHWT1): New.
2823 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2824 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2825 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2826 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2827 (*prefetch_avx512pf_<mode>_: Change into ...
2828 (*prefetch_prefetchwt1_<mode>: This.
2829 * config/i386/i386.opt (mprefetchwt1): New.
2830 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2831 (_mm_prefetch): Handle intent to write.
2832 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2834 2014-02-25 Richard Biener <rguenther@suse.de>
2837 * emit-rtl.c (mem_attrs_htab): Remove.
2838 (mem_attrs_htab_hash): Likewise.
2839 (mem_attrs_htab_eq): Likewise.
2840 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2841 (init_emit_once): Do not allocate mem_attrs_htab.
2843 2014-02-25 Richard Biener <rguenther@suse.de>
2846 * lto-opts.c (lto_write_options): Output non-explicit conservative
2847 -fwrapv, -fno-trapv and -fno-strict-overflow.
2848 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2849 (run_gcc): And pass them through.
2851 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2853 * sel-sched.c (calculate_new_fences): New parameter ptime.
2854 Calculate it as a maximum over all fence cycles.
2855 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2856 Print the final schedule timing when sched_verbose.
2858 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2860 PR rtl-optimization/60292
2861 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2862 bit fot the fence instruction.
2864 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2866 * calls.h: Fix typo in comment.
2868 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2870 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2871 adjusting offsetable addresses.
2873 2014-02-24 Guozhi Wei <carrot@google.com>
2875 * sparseset.h (sparseset_pop): Fix the wrong index.
2877 2014-02-24 Walter Lee <walt@tilera.com>
2879 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2880 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2882 * common/config/tilegx/tilegx-common.c
2883 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2884 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2886 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2887 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2888 (tilegx_gimplify_va_arg_expr): Handle big endian.
2889 (tilegx_expand_unaligned_load): Ditto.
2890 (tilegx_expand_unaligned_store): Ditto.
2891 (TARGET_RETURN_IN_MSB): New.
2892 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2893 (TARGET_ENDIAN_DEFAULT): New.
2894 (TARGET_BIG_ENDIAN): Handle big endian.
2895 (BYTES_BIG_ENDIAN): Ditto.
2896 (WORDS_BIG_ENDIAN): Ditto.
2897 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2900 * config/tilegx/tilegx.md (extv): Handle big endian.
2902 (insn_st<n>): Ditto.
2903 (insn_st<n>_add<bitsuffix>): Ditto.
2904 (insn_stnt<n>): Ditto.
2905 (insn_stnt<n>_add<bitsuffix>):Ditto.
2906 (vec_interleave_highv8qi): Handle big endian.
2907 (vec_interleave_highv8qi_be): New.
2908 (vec_interleave_highv8qi_le): New.
2909 (insn_v1int_h): Handle big endian.
2910 (vec_interleave_lowv8qi): Handle big endian.
2911 (vec_interleave_lowv8qi_be): New.
2912 (vec_interleave_lowv8qi_le): New.
2913 (insn_v1int_l): Handle big endian.
2914 (vec_interleave_highv4hi): Handle big endian.
2915 (vec_interleave_highv4hi_be): New.
2916 (vec_interleave_highv4hi_le): New.
2917 (insn_v2int_h): Handle big endian.
2918 (vec_interleave_lowv4hi): Handle big endian.
2919 (vec_interleave_lowv4hi_be): New.
2920 (vec_interleave_lowv4hi_le): New.
2921 (insn_v2int_l): Handle big endian.
2922 (vec_interleave_highv2si): Handle big endian.
2923 (vec_interleave_highv2si_be): New.
2924 (vec_interleave_highv2si_le): New.
2925 (insn_v4int_h): Handle big endian.
2926 (vec_interleave_lowv2si): Handle big endian.
2927 (vec_interleave_lowv2si_be): New.
2928 (vec_interleave_lowv2si_le): New.
2929 (insn_v4int_l): Handle big endian.
2930 * config/tilegx/tilegx.opt (mbig-endian): New option.
2931 (mlittle-endian): New option.
2932 * doc/install.texi: Document tilegxbe-linux.
2933 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2935 2014-02-24 Martin Jambor <mjambor@suse.cz>
2938 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2939 there are no parameter descriptors.
2941 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2943 PR rtl-optimization/60268
2944 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2945 initialization to ...
2946 (sched_rgn_init): ... here.
2947 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2949 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2951 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2954 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2956 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2959 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2961 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2962 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2964 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2966 * config/microblaze/predicates.md: Add cmp_op predicate.
2967 * config/microblaze/microblaze.md: Add branch_compare instruction
2968 which uses cmp_op predicate and emits cmp insn before branch.
2969 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2970 to microblaze_expand_conditional_branch and consolidate logic.
2971 (microblaze_expand_conditional_branch): emit branch_compare
2972 insn instead of handling cmp op separate from branch insn.
2974 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2976 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2979 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2981 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2982 define_insn with define_expand and new define_insn
2983 *altivec_lve<VI_char>x_internal.
2984 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2985 and new define_insn *altivec_stve<VI_char>x_internal.
2986 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2988 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2990 (altivec_expand_stvex_be): New function.
2992 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2994 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2995 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2996 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2997 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2999 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
3002 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
3003 instead of emit_move_insn.
3005 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3007 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
3009 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
3010 gen_altivec_vsumsws.
3012 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3014 * config/rs6000/altivec.md (altivec_lvxl): Rename as
3015 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
3016 (altivec_lvxl_<mode>): New define_expand incorporating
3017 -maltivec=be semantics where needed.
3018 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
3019 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
3020 semantics where needed.
3021 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
3022 (altivec_stvx_<mode>): New define_expand incorporating
3023 -maltivec=be semantics where needed.
3024 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
3025 VM2 iterator instead of V4SI.
3026 (altivec_stvxl_<mode>): New define_expand incorporating
3027 -maltivec=be semantics where needed.
3028 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
3029 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
3030 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
3031 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
3032 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
3033 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
3034 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
3035 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
3036 ALTIVEC_BUILTIN_STVXL.
3037 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
3038 (altivec_expand_stvx_be): Likewise.
3039 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
3040 (altivec_expand_lvx_be): Likewise.
3041 (altivec_expand_stvx_be): Likewise.
3042 (altivec_expand_builtin): Add cases for
3043 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
3044 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
3045 (altivec_init_builtins): Add definitions for
3046 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
3047 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
3049 2014-02-21 Catherine Moore <clm@codesourcery.com>
3051 * doc/invoke.texi (mvirt, mno-virt): Document.
3052 * config/mips/mips.opt (mvirt): New option.
3053 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
3055 2014-02-21 Richard Biener <rguenther@suse.de>
3057 PR tree-optimization/60276
3058 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
3059 (STMT_VINFO_MIN_NEG_DIST): New macro.
3060 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
3061 STMT_VINFO_MIN_NEG_DIST.
3062 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
3063 made for negative dependence distances still hold.
3065 2014-02-21 Richard Biener <rguenther@suse.de>
3068 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
3069 DECL_INITIAL for globals not in the current function context.
3071 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3073 PR tree-optimization/56490
3074 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
3075 * tree-ssa-uninit.c: Include params.h.
3076 (compute_control_dep_chain): Add num_calls argument, return false
3077 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
3078 num_calls to recursive call.
3079 (find_predicates): Change dep_chain into normal array,
3080 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
3081 variable and adjust compute_control_dep_chain caller.
3082 (find_def_preds): Likewise.
3084 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3086 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
3087 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
3089 2014-02-21 Nick Clifton <nickc@redhat.com>
3091 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
3092 (pushhi1): Likewise.
3093 (popqi1): Add mode to pre_dec.
3096 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3098 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
3099 mode for mask of V8SFmode permutation.
3101 2014-02-20 Richard Henderson <rth@redhat.com>
3104 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
3105 a new pseudo for OLDVAL.
3107 2014-02-20 Jakub Jelinek <jakub@redhat.com>
3110 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
3111 gen_reg_rtx if d->testing_p.
3112 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
3113 if d->testing_p and we will certainly return true.
3114 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
3117 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3119 * emit-rtl.c (gen_reg_rtx): Assert that
3120 crtl->emit.regno_pointer_align_length is non-zero.
3122 2014-02-20 Richard Henderson <rth@redhat.com>
3125 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
3126 on failure the store back into EXPECT.
3128 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
3129 Sandra Loosemore <sandra@codesourcery.com>
3131 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
3132 * config/nios2/nios2.c (nios2_function_profiler): Add
3133 -fPIC (flag_pic == 2) support.
3134 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
3135 (nios2_large_offset_p): New function.
3136 (nios2_unspec_reloc_p): Move up position, update to use
3137 nios2_large_offset_p.
3138 (nios2_unspec_address): Remove function.
3139 (nios2_unspec_offset): New function.
3140 (nios2_large_got_address): New function.
3141 (nios2_got_address): Add large offset support.
3142 (nios2_legitimize_tls_address): Update usage of removed and new
3144 (nios2_symbol_binds_local_p): New function.
3145 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
3146 (nios2_legitimize_address): Update to use nios2_large_offset_p.
3147 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
3148 (nios2_print_operand): Merge H/L processing, add hiadj/lo
3149 processing for (const (unspec ...)).
3150 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
3152 2014-02-20 Richard Biener <rguenther@suse.de>
3154 * tree-cfg.c (replace_uses_by): Mark altered BBs before
3155 doing the substitution.
3156 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
3158 2014-02-20 Martin Jambor <mjambor@suse.cz>
3161 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
3162 info when checking whether lattices are bottom.
3164 2014-02-20 Richard Biener <rguenther@suse.de>
3167 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
3170 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
3173 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
3174 parameter specifying the scaling.
3175 (inline_call): Update.
3176 (want_inline_recursively): Guard division by zero.
3177 (recursive_inlining): Update.
3178 * ipa-inline.h (clone_inlined_nodes): Update.
3180 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3183 * config/i386/i386.c (classify_argument): Pass structures of size
3184 64 bytes or less in register.
3186 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3187 Kirill Yukhin <kirill.yukhin@intel.com>
3189 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
3190 (_mm_rcp28_round_ss): Ditto.
3191 (_mm_rsqrt28_round_sd): Ditto.
3192 (_mm_rsqrt28_round_ss): Ditto.
3193 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
3194 (_mm_rcp14_round_ss): Ditto.
3195 (_mm_rsqrt14_round_sd): Ditto.
3196 (_mm_rsqrt14_round_ss): Ditto.
3197 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
3198 the first input operand, get rid of match_dup.
3199 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
3201 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3203 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
3204 operand as the first input operand, set type attribute.
3205 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3207 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
3208 operand as the first input operand, set type attribute.
3210 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3212 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
3215 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
3218 * config/i386/i386.c (construct_container): Remove TFmode check
3219 for X86_64_INTEGER_CLASS.
3221 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
3224 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
3225 only when -Wpsabi is enabled.
3227 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
3230 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
3231 passing arrays in registers are the same as for structs, so remove the
3232 special case for them.
3234 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
3236 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
3237 destination type, extract only the valid bits if the source type is not
3238 integral and has a different mode.
3240 2014-02-19 Richard Biener <rguenther@suse.de>
3243 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
3246 2014-02-19 Richard Biener <rguenther@suse.de>
3249 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
3250 (ipa_modify_call_arguments): Emit an argument load explicitely and
3251 preserve virtual SSA form there and for the replacement call.
3252 Do not update SSA form nor free dominance info.
3254 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3256 * ipa.c (function_and_variable_visibility): Also clear WEAK
3257 flag when disolving COMDAT_GROUP.
3259 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3261 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
3262 * ipa-prop.c (ipa_set_jf_known_type): Return early when
3264 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
3265 do more sanity checks.
3266 (detect_type_change): Return true when giving up early.
3267 (compute_complex_assign_jump_func): Fix type parameter of
3268 ipa_set_ancestor_jf.
3269 (compute_complex_ancestor_jump_func): Likewise.
3270 (update_jump_functions_after_inlining): Fix updating of
3272 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
3274 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3276 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
3277 inline clones when edge disappears.
3279 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3282 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
3283 Split 64-bit moves into 2 patterns. Do not allow the use of
3284 direct move for TDmode in little endian, since the decimal value
3285 has little endian bytes within a word, but the 64-bit pieces are
3286 ordered in a big endian fashion, and normal subreg's of TDmode are
3288 (mov<mode>_64bit_dm): Likewise.
3289 (movtd_64bit_nodm): Likewise.
3291 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3293 PR tree-optimization/60174
3294 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
3295 statement of an SSA_NAME that occurs in an abnormal PHI node.
3297 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3300 * final.c (SEEN_BB): Remove.
3301 (SEEN_NOTE, SEEN_EMITTED): Renumber.
3302 (final_scan_insn): Don't force_source_line on second
3303 NOTE_INSN_BASIC_BLOCK.
3305 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
3308 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
3309 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
3310 (type_natural_mode): Warn ABI change when %zmm register is not
3311 available for AVX512F vector value passing.
3313 2014-02-18 Kai Tietz <ktietz@redhat.com>
3316 * config/i386/i386.c (ix86_expand_prologue): Use value in
3317 rax register as displacement when restoring %r10 or %rax.
3318 Fix wrong offset when restoring both registers.
3320 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3322 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
3323 assertion with conditional return.
3325 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3326 Uros Bizjak <ubizjak@gmail.com>
3329 * config/i386/driver-i386.c (host_detect_local_cpu): If
3330 YMM state is not saved by the OS, also clear has_f16c. Move
3331 CPUID 0x80000001 handling before YMM state saving checking.
3333 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3335 PR rtl-optimization/58960
3336 * haifa-sched.c (alloc_global_sched_pressure_data): New,
3337 factored out from ...
3338 (sched_init): ... here.
3339 (free_global_sched_pressure_data): New, factored out from ...
3340 (sched_finish): ... here.
3341 * sched-int.h (free_global_sched_pressure_data): Declare.
3342 * sched-rgn.c (nr_regions_initial): New static global.
3343 (haifa_find_rgns): Initialize it.
3344 (schedule_region): Disable sched-pressure for the newly
3347 2014-02-17 Richard Biener <rguenther@suse.de>
3349 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
3350 release SSA defs of pattern stmts.
3352 2014-02-17 Richard Biener <rguenther@suse.de>
3354 * tree-inline.c (expand_call_inline): Release the virtual
3355 operand defined by the call we are about to inline.
3357 2014-02-17 Richard Biener <rguenther@suse.de>
3359 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
3361 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
3362 Ilya Tocar <ilya.tocar@intel.com>
3364 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
3365 arguments order in builtin.
3366 (_mm512_permutexvar_epi64): Ditto.
3367 (_mm512_mask_permutexvar_epi64): Ditto
3368 (_mm512_maskz_permutexvar_epi32): Ditto
3369 (_mm512_permutexvar_epi32): Ditto
3370 (_mm512_mask_permutexvar_epi32): Ditto
3372 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3374 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
3375 (p8_vmrgow): Likewise.
3377 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3379 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
3382 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3385 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
3386 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
3387 into 64-bit and 32-bit moves. On 64-bit moves, add support for
3388 using direct move instructions on ISA 2.07. Also adjust
3389 instruction length for 64-bit.
3390 (mov<mode>_64bit, TFmode/TDmode): Likewise.
3391 (mov<mode>_32bit, TFmode/TDmode): Likewise.
3393 2014-02-15 Alan Modra <amodra@gmail.com>
3397 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
3398 find_replacement on parts of insn rtl that might be reloaded.
3400 2014-02-15 Richard Biener <rguenther@suse.de>
3402 PR tree-optimization/60183
3403 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
3404 (tree_ssa_phiprop): Calculate and free post-dominators.
3406 2014-02-14 Jeff Law <law@redhat.com>
3408 PR rtl-optimization/60131
3409 * ree.c (get_extended_src_reg): New function.
3410 (combine_reaching_defs): Use it rather than assuming location of REG.
3411 (find_and_remove_re): Verify first operand of extension is
3412 a REG before adding the insns to the copy list.
3414 2014-02-14 Roland McGrath <mcgrathr@google.com>
3416 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
3417 * configure: Regenerated.
3418 * config.in: Regenerated.
3419 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
3420 instead of ASM_SHORT.
3422 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
3423 Richard Earnshaw <rearnsha@arm.com>
3425 PR rtl-optimization/59535
3426 * lra-constraints.c (process_alt_operands): Encourage alternative
3427 when unassigned pseudo class is superset of the alternative class.
3428 (inherit_reload_reg): Don't inherit when optimizing for code size.
3429 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
3430 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
3431 modes not less than 4 for Thumb1.
3433 2014-02-14 Kyle McMartin <kyle@redhat.com>
3436 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
3438 2014-02-14 Richard Biener <rguenther@suse.de>
3440 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
3441 (get_frame_arg): Drop the assert with langhook types_compatible_p.
3442 Do not strip INDIRECT_REFs.
3444 2014-02-14 Richard Biener <rguenther@suse.de>
3447 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
3448 DECL_FUNCTION_SPECIFIC_TARGET.
3449 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
3450 * tree-streamer-out.c (pack_ts_target_option): Remove.
3451 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
3452 (write_ts_function_decl_tree_pointers): Do not stream
3453 DECL_FUNCTION_SPECIFIC_TARGET.
3454 * tree-streamer-in.c (unpack_ts_target_option): Remove.
3455 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
3456 (lto_input_ts_function_decl_tree_pointers): Do not stream
3457 DECL_FUNCTION_SPECIFIC_TARGET.
3459 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3461 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3462 (get_initial_def_for_induction, vectorizable_induction): Ignore
3463 debug stmts when looking for exit_phi.
3464 (vectorizable_live_operation): Fix up condition.
3466 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3468 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3469 nreverse() because it changes the content of original tree list.
3471 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3473 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3474 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3476 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3478 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3479 GNU coding standards.
3481 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3484 * dwarf2out.c (gen_subprogram_die): Don't call
3485 add_calling_convention_attribute if subr_die is old_die.
3487 2014-02-13 Sharad Singhai <singhai@google.com>
3489 * doc/optinfo.texi: Fix order of nodes.
3491 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3493 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3494 operands[2], not operands[3].
3496 2014-02-13 Richard Biener <rguenther@suse.de>
3499 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3500 mention the possibility of an in-tree build.
3501 (CLooG): Update recommended version to 0.18.1, mention the
3502 possibility of an in-tree build and clarify that the ISL
3503 bundled with CLooG does not work.
3505 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3508 * expr.c (compress_float_constant): If x is a hard register,
3509 extend into a pseudo and then move to x.
3511 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3513 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3514 caused by bad second argument to warning_at() with -mhotpatch and
3515 nested functions (e.g. with gfortran).
3517 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3519 * opts.c (option_name): Remove "enabled by default" rider.
3521 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3523 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3525 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3526 Uros Bizjak <ubizjak@gmail.com>
3529 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3530 * configure: Regenerated.
3532 2014-02-12 Richard Biener <rguenther@suse.de>
3534 * vec.c (vec_prefix::calculate_allocation): Move as
3535 inline variant to vec.h.
3536 (vec_prefix::calculate_allocation_1): New out-of-line version.
3537 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3538 (vec_prefix::m_has_auto_buf): Rename to ...
3539 (vec_prefix::m_using_auto_storage): ... this.
3540 (vec_prefix::calculate_allocation): Inline the easy cases
3541 and dispatch to calculate_allocation_1 which doesn't need the
3543 (va_heap::reserve): Use gcc_checking_assert.
3544 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3545 m_using_auto_storage.
3546 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3548 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3549 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3550 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3552 2014-02-12 Richard Biener <rguenther@suse.de>
3554 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3555 when we found a dependence.
3557 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3559 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3561 (maybe_fold_stmt): ... into this new function.
3562 * omp-low.c (lower_omp): Update comment.
3564 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3567 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3570 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3572 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3573 identifiers in comments.
3574 (cortexa53_extra_costs): Likewise.
3575 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3576 (cortexa7_extra_costs): Likewise.
3577 (cortexa12_extra_costs): Likewise.
3578 (cortexa15_extra_costs): Likewise.
3579 (v7m_extra_costs): Likewise.
3581 2014-02-12 Richard Biener <rguenther@suse.de>
3584 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3585 of posix_memalign being successful.
3586 (lower_stmt): Restrict lowering of posix_memalign to when
3587 -ftree-bit-ccp is enabled.
3589 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3591 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3593 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3595 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3597 PR rtl-optimization/60116
3598 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3599 other_insn once the combination has been validated.
3601 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3604 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3606 * ipa-devirt.c: Include demangle.h
3607 (odr_violation_reported): New static variable.
3608 (add_type_duplicate): Update odr_violations.
3609 (maybe_record_node): Add completep parameter; update it.
3610 (record_target_from_binfo): Add COMPLETEP parameter;
3611 update it as needed.
3612 (possible_polymorphic_call_targets_1): Likewise.
3613 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3614 rename FINAL to COMPLETE.
3615 (record_targets_from_bases): Sanity check we found the binfo;
3616 fix COMPLETEP updating.
3617 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3618 parameter, fix computing of COMPLETEP.
3619 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3620 at LTO time do demangling.
3621 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3622 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3624 (gimple_get_virt_method_for_binfo): Likewise.
3625 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3626 gimple_get_virt_method_for_vtable): Update prototypes.
3628 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3631 * genautomata.c (add_presence_absence): Fix typo with
3632 {final_}presence_list.
3634 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3637 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3638 for VSX/Altivec vectors that land in GPR registers.
3640 2014-02-11 Richard Henderson <rth@redhat.com>
3641 Jakub Jelinek <jakub@redhat.com>
3644 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3645 around drhs if type conversion to lacc->type is not useless.
3647 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3649 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3651 (cortex-a57.cortex-a53): Likewise.
3652 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3654 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3656 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3659 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3661 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3662 add_options_for_arm_vfp3.
3664 2014-02-11 Jeff Law <law@redhat.com>
3667 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3668 object with an undesirable mode.
3670 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3673 * config/i386/sol2-9.h: New file.
3674 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3675 *-*-solaris2.9*): Use it.
3677 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3679 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3680 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3682 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3684 * config/microblaze/microblaze.c: Extend mcpu version format
3686 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3688 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3690 2014-02-10 Richard Henderson <rth@redhat.com>
3693 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3694 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3695 ms-abi vs -mno-accumulate-outgoing-args.
3696 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3697 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3700 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3703 * cfgexpand.c (expand_asm_operands): Attach source location to
3704 ASM_INPUT rtx objects.
3705 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3707 2014-02-10 Nick Clifton <nickc@redhat.com>
3709 * config/mn10300/mn10300.c (popcount): New function.
3710 (mn10300_expand_prologue): Include saved registers in stack usage
3713 2014-02-10 Jeff Law <law@redhat.com>
3716 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3717 when changing the SET_DEST of a prior insn to avoid an input reload.
3719 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3721 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3722 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3723 -mcall-openbsd, or -mcall-linux.
3724 (CC1_ENDIAN_BIG_SPEC): Remove.
3725 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3726 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3727 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3728 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3729 and %cc1_endian_default.
3730 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3732 2014-02-10 Richard Biener <rguenther@suse.de>
3734 PR tree-optimization/60115
3735 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3736 MEM_REF handling. Properly verify that the accesses are not
3737 out of the objects bound.
3739 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3741 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3744 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3746 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3747 proper constants and fix formatting.
3748 (possible_polymorphic_call_targets): Fix formatting.
3750 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3751 Ilya Tocar <ilya.tocar@intel.com>
3753 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3754 (_mm512_loadu_epi32): Renamed into...
3755 (_mm512_loadu_si512): This.
3756 (_mm512_storeu_epi32): Renamed into...
3757 (_mm512_storeu_si512): This.
3758 (_mm512_maskz_ceil_ps): Removed.
3759 (_mm512_maskz_ceil_pd): Ditto.
3760 (_mm512_maskz_floor_ps): Ditto.
3761 (_mm512_maskz_floor_pd): Ditto.
3762 (_mm512_floor_round_ps): Ditto.
3763 (_mm512_floor_round_pd): Ditto.
3764 (_mm512_ceil_round_ps): Ditto.
3765 (_mm512_ceil_round_pd): Ditto.
3766 (_mm512_mask_floor_round_ps): Ditto.
3767 (_mm512_mask_floor_round_pd): Ditto.
3768 (_mm512_mask_ceil_round_ps): Ditto.
3769 (_mm512_mask_ceil_round_pd): Ditto.
3770 (_mm512_maskz_floor_round_ps): Ditto.
3771 (_mm512_maskz_floor_round_pd): Ditto.
3772 (_mm512_maskz_ceil_round_ps): Ditto.
3773 (_mm512_maskz_ceil_round_pd): Ditto.
3774 (_mm512_expand_pd): Ditto.
3775 (_mm512_expand_ps): Ditto.
3776 * config/i386/i386.c (ix86_builtins): Remove
3777 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3778 (bdesc_args): Ditto.
3779 * config/i386/predicates.md (const1256_operand): New.
3780 (const_1_to_2_operand): Ditto.
3781 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3782 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3783 (*avx512pf_gatherpf<mode>sf): Ditto.
3784 (avx512pf_gatherpf<mode>df): Ditto.
3785 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3786 (*avx512pf_gatherpf<mode>df): Ditto.
3787 (avx512pf_scatterpf<mode>sf): Ditto.
3788 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3789 (*avx512pf_scatterpf<mode>sf): Ditto.
3790 (avx512pf_scatterpf<mode>df): Ditto.
3791 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3792 (*avx512pf_scatterpf<mode>df): Ditto.
3793 (avx512f_expand<mode>): Removed.
3794 (<shift_insn><mode>3<mask_name>): Change predicate type.
3796 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3798 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3799 not at the end of datarefs vector use ordered_remove to avoid
3800 reordering datarefs vector.
3803 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3804 mark local addressable non-static vars as GOVD_PRIVATE
3805 instead of GOVD_LOCAL.
3806 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3807 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3811 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3812 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3813 assume_aligned or alloc_align attributes.
3814 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3815 arguments. Handle also assume_aligned and alloc_align attributes.
3816 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3817 calls to functions with assume_aligned or alloc_align attributes.
3818 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3820 2014-02-08 Terry Guo <terry.guo@arm.com>
3822 * doc/invoke.texi: Document ARM -march=armv7e-m.
3824 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3826 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3827 flag on __cilkrts_rethrow builtin.
3830 * ipa-cp.c (determine_versionability): Fail at -O0
3831 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3832 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3835 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3838 * tree-inline.c (copy_forbidden): Fail for
3839 __attribute__((optimize (0))) functions.
3841 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3843 * varpool.c: Include pointer-set.h.
3844 (varpool_remove_unreferenced_decls): Variables in other partitions
3845 will not be output; be however careful to not lose information
3848 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3850 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3851 lookup in the vtable constructor.
3853 2014-02-07 Jeff Law <law@redhat.com>
3856 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3857 define_insn_and_split.
3859 * ipa-inline.c (inline_small_functions): Fix typos.
3861 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3863 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3864 (s390_can_use_return_insn): Declare.
3865 * config/s390/s390.h (EPILOGUE_USES): Define.
3866 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3868 (s390_chunkify_start): Handle return JUMP_LABELs.
3869 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3870 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3871 (s390_can_use_return_insn): New functions.
3872 (s390_fix_long_loop_prediction): Handle conditional returns.
3873 (TARGET_SET_UP_BY_PROLOGUE): Define.
3874 * config/s390/s390.md (ANY_RETURN): New code iterator.
3875 (*creturn, *csimple_return, return, simple_return): New patterns.
3877 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3879 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3880 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3881 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3882 REG_CFA_RESTORE list when deciding not to restore a register.
3884 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3886 * config/s390/s390.c: Include tree-pass.h and context.h.
3887 (s390_early_mach): New function, split out from...
3888 (s390_emit_prologue): ...here.
3889 (pass_data_s390_early_mach): New pass structure.
3890 (pass_s390_early_mach): New class.
3891 (s390_option_override): Create and register early_mach pass.
3892 Move to end of file.
3894 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3896 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3897 to match for the exit block.
3899 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3901 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3902 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3903 Reject misaligned operands.
3905 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3907 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3909 2014-02-07 Richard Biener <rguenther@suse.de>
3912 * gimple-low.c (lower_builtin_posix_memalign): New function.
3913 (lower_stmt): Call it to lower posix_memalign in a way
3914 to make alignment info accessible.
3916 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3919 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3920 __builtin_setjmp_receiver.
3922 2014-02-07 Richard Biener <rguenther@suse.de>
3925 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3926 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3927 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3928 Handle BUILT_IN_POSIX_MEMALIGN.
3929 (find_func_clobbers): Likewise.
3930 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3931 (call_may_clobber_ref_p_1): Likewise.
3933 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3936 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3939 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3942 * lto-cgraph.c (lto_output_node): Use
3943 symtab_get_symbol_partitioning_class.
3944 (lto_output_varpool_node): likewise.
3945 (symtab_get_symbol_partitioning_class): Move here from
3947 * cgraph.h (symbol_partitioning_class): Likewise.
3948 (symtab_get_symbol_partitioning_class): Declare.
3950 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3952 * ggc.h (ggc_internal_cleared_alloc): New macro.
3953 * vec.h (vec_safe_copy): Handle memory stats.
3954 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3955 * target-globals.c (save_target_globals): Likewise.
3957 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3960 * expr.c (emit_move_resolve_push): Export; be bit more selective
3961 on when to clear alias set.
3962 * expr.h (emit_move_resolve_push): Declare.
3963 * function.h (struct function): Add tail_call_marked.
3964 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3965 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3966 * config/i386/i386.md (TImode move expander): De not call
3968 (FP push expanders): Preserve memory attributes.
3969 * config/i386/sse.md (push<mode>1): Remove.
3970 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3971 (ix86_expand_push): Remove.
3972 * config/i386/mmx.md (push<mode>1): Remove.
3974 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3976 PR rtl-optimization/60030
3977 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3978 lopart with paradoxical subreg before shifting it up by hprec.
3980 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3982 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3983 Remove extra newline at end of file.
3984 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3985 (arm_issue_rate): Handle cortexa57.
3986 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3987 (cortex-a57.cortex-a53): Likewise.
3989 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3992 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3993 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3995 (arm_expand_prologue): Adjust all callers.
3996 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3997 info, registers also at the lowest numbered registers side. Use
3998 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
4002 * var-tracking.c (adjust_mems): Before adding a SET to
4003 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
4005 2014-02-06 Alan Modra <amodra@gmail.com>
4008 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
4009 change SDmode to DDmode when lra_in_progress.
4011 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4014 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
4015 free_data_ref on the dr first, and before goto again also set dr
4016 to the next dr. For simd_lane_access, free old datarefs[i] before
4017 overwriting it. For get_vectype_for_scalar_type failure, don't
4018 free_data_ref if simd_lane_access.
4020 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
4023 * tree.h (opts_for_fn): New inline function.
4024 (opt_for_fn): Define.
4025 * config/i386/i386.c (ix86_function_regparm): Use
4026 opt_for_fn (decl, optimize) instead of optimize.
4028 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
4030 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4031 for SYMBOL_REF in large memory model.
4033 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4035 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
4037 (cortex-a57): Likewise.
4038 (cortex-a57.cortex-a53): Likewise.
4040 2014-02-06 Yury Gribov <y.gribov@samsung.com>
4041 Kugan Vivekanandarajah <kuganv@linaro.org>
4043 * config/arm/arm.c (arm_vector_alignment_reachable): Check
4045 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
4047 2014-02-06 Richard Biener <rguenther@suse.de>
4049 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
4050 set_loop_copy and initialize_original_copy_tables.
4052 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
4054 * config/aarch64/aarch64-simd.md
4055 (aarch64_ashr_simddi): Change QI to SI.
4057 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4058 Jakub Jelinek <jakub@redhat.com>
4061 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
4064 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4066 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
4067 CODE_FOR_altivec_vpku[hw]um to
4068 CODE_FOR_altivec_vpku[hw]um_direct.
4069 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
4070 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
4071 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
4072 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
4074 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4076 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
4077 generation for -maltivec=be.
4078 (altivec_vsumsws): Simplify redundant test.
4080 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4082 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
4083 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
4084 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
4085 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
4086 gen_altivec_vpkuwum.
4087 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
4089 (altivec_vpks<VI_char>ss): Likewise.
4090 (altivec_vpks<VI_char>us): Likewise.
4091 (altivec_vpku<VI_char>us): Likewise.
4092 (altivec_vpku<VI_char>um): Likewise.
4093 (altivec_vpku<VI_char>um_direct): New (copy of
4094 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
4096 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
4097 target is little endian and -maltivec=be is not specified.
4098 (*altivec_vupkhs<VU_char>_direct): New (copy of
4099 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
4100 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
4101 target is little endian and -maltivec=be is not specified.
4102 (*altivec_vupkls<VU_char>_direct): New (copy of
4103 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
4104 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
4105 little endian and -maltivec=be is not specified.
4106 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
4107 little endian and -maltivec=be is not specified.
4109 2014-02-05 Richard Henderson <rth@redhat.com>
4112 * combine-stack-adj.c: Revert r206943.
4113 * sched-int.h (struct deps_desc): Add last_args_size.
4114 * sched-deps.c (init_deps): Initialize it.
4115 (sched_analyze_insn): Add OUTPUT dependencies between insns that
4116 contain REG_ARGS_SIZE notes.
4118 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4120 * lto-cgraph.c (asm_nodes_output): Make global.
4121 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
4122 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
4123 (driver_handle_option): Handle OPT_fwpa.
4125 2014-02-05 Jakub Jelinek <jakub@redhat.com>
4128 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
4129 a comment typo and formatting issue. If odr_hash hasn't been
4130 created, return vNULL and set *completep to false.
4133 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
4134 bb with no successors.
4136 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
4139 * doc/invoke.texi (-march): Clarify documentation for ARM.
4143 2014-02-05 Richard Biener <rguenther@suse.de>
4145 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
4146 when not vectorizing because of too many alias checks.
4147 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
4148 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
4150 2014-02-05 Nick Clifton <nickc@redhat.com>
4152 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
4153 accept extended registers in any mode when compiling for the MN10300.
4155 2014-02-05 Yury Gribov <y.gribov@samsung.com>
4157 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
4158 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
4159 sanitization attributes.
4160 (can_inline_edge_p): Likewise.
4161 (sanitize_attrs_match_for_inline_p): New function.
4163 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4165 * ipa-prop.c (detect_type_change): Shor circuit testing of
4166 type changes on THIS pointer.
4168 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4171 * config/pa/pa.c (legitimize_tls_address): Return original address
4172 if not passed a SYMBOL_REF rtx.
4173 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
4175 (pa_emit_move_sequence): Simplify TLS source operands.
4176 (pa_legitimate_constant_p): Reject all TLS constants.
4177 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
4178 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
4180 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4182 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
4183 groups when we know they are controlled by LTO.
4184 * varasm.c (default_binds_local_p_1): If object is in other partition,
4185 it will be resolved locally.
4187 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4189 * config/host-linux.c (linux_gt_pch_use_address): Don't
4190 use SSIZE_MAX because it is not always defined.
4192 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
4195 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
4196 threshold for pseudo splitting.
4197 (update_ebb_live_info): Process call argument hard registers and
4198 hard registers from insn definition too.
4199 (max_small_class_regs_num): New constant.
4200 (inherit_in_ebb): Update live hard regs through EBBs. Update
4201 reloads_num only for small register classes. Don't split for
4204 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
4207 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
4210 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4212 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
4215 2014-02-04 Marek Polacek <polacek@redhat.com>
4217 * gdbinit.in (pel): Define.
4219 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4221 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
4224 2014-02-04 Richard Biener <rguenther@suse.de>
4227 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
4228 in function context local.
4229 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
4230 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
4231 similar to LTO_imported_decl_ref.
4233 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4235 PR tree-optimization/60002
4236 * cgraphclones.c (build_function_decl_skip_args): Clear
4239 PR tree-optimization/60023
4240 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
4241 false to gsi_replace.
4242 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
4243 has been in some EH region and vec_stmt could throw, add
4244 vec_stmt into the same EH region.
4245 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
4246 has no lhs, ignore it.
4247 * internal-fn.c (expand_MASK_LOAD): Likewise.
4250 * tree-inline.c (copy_forbidden): Fail for
4251 __attribute__((optimize (0))) functions.
4254 * omp-low.c (simd_clone_struct_copy): If from->inbranch
4255 is set, copy one less argument.
4256 (expand_simd_clones): Don't subtract clone_info->inbranch
4257 from simd_clone_struct_alloc argument.
4259 PR rtl-optimization/57915
4260 * recog.c (simplify_while_replacing): If all unary/binary/relational
4261 operation arguments are constant, attempt to simplify those.
4264 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
4265 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
4267 2014-02-04 Richard Biener <rguenther@suse.de>
4269 PR tree-optimization/60012
4270 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
4271 TBAA disambiguation to all DDRs.
4273 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4276 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
4277 (LINK_SPEC): Use it for -shared, -shared-libgcc.
4279 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4282 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
4284 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4286 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
4287 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
4289 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4292 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
4293 to figure out targets of polymorphic calls with known decl.
4294 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4295 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
4296 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
4297 (get_polymorphic_call_info): ... here.
4298 (get_polymorphic_call_info_from_invariant): New function.
4300 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4302 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
4303 lookup via vtable pointer; check for type consistency
4304 and turn inconsitent facts into UNREACHABLE.
4305 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4306 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
4307 type inconsistent querries; return UNREACHABLE instead.
4309 2014-02-03 Richard Henderson <rth@twiddle.net>
4312 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
4313 already processed this node.
4314 (normalize_one_pred_1): Pass along mark_set.
4315 (normalize_one_pred): Create and destroy a pointer_set_t.
4316 (normalize_one_pred_chain): Likewise.
4318 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
4320 PR gcov-profile/58602
4321 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
4323 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4326 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
4327 -fno-devirtualize; try to devirtualize by the knowledge of
4328 virtual table pointer given by aggregate propagation.
4329 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4330 (ipa_print_node_jump_functions): Dump also offset that
4331 is relevant for polymorphic calls.
4332 (determine_known_aggregate_parts): Add arg_type parameter; use it
4333 instead of determining the type from pointer type.
4334 (ipa_compute_jump_functions_for_edge): Update call of
4335 determine_known_aggregate_parts.
4336 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
4337 (gimple_get_virt_method_for_binfo): ... here; simplify using
4338 vtable_pointer_value_to_vtable.
4339 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
4340 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
4341 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
4342 (vtable_pointer_value_to_vtable): Break out from ...; handle also
4344 (vtable_pointer_value_to_binfo): ... here.
4345 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
4347 2014-02-03 Teresa Johnson <tejohnson@google.com>
4349 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
4350 redef of outer loop index variable.
4352 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
4356 * doc/extend.texi (Function Attributes): Typo.
4358 2014-02-03 Cong Hou <congh@google.com>
4360 PR tree-optimization/60000
4361 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
4362 if the vectorized statement is a store. A store statement can only
4363 appear at the end of pattern statements.
4365 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4367 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
4368 (ix86_option_override_internal): Default long double to 64-bit for
4369 32-bit Bionic and to 128-bit for 64-bit Bionic.
4371 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
4372 TARGET_LONG_DOUBLE_128 is true.
4373 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
4375 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
4376 (mlong-double-64): Negate -mlong-double-128.
4377 (mlong-double-128): New option.
4379 * config/i386/i386-c.c (ix86_target_macros): Define
4380 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
4382 * doc/invoke.texi: Document -mlong-double-128.
4384 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4386 PR rtl-optimization/60024
4387 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
4389 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
4391 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
4393 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4395 PR rtl-optimization/57662
4396 * sel-sched.c (code_motion_path_driver): Do not mark already not
4397 existing blocks in the visiting bitmap.
4399 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4401 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
4402 on the insn being emitted.
4404 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
4405 Will Deacon <will.deacon@arm.com>
4407 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
4409 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4411 * config/arm/arm-tables.opt: Regenerate.
4413 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4415 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
4416 for vector types other than V16QImode.
4417 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
4418 define_expand, and call altivec_expand_vec_perm_le when producing
4419 code with little endian element order.
4420 (*altivec_vperm_<mode>_internal): New insn having previous
4421 behavior of altivec_vperm_<mode>.
4422 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
4423 altivec_expand_vec_perm_le when producing code with little endian
4425 (*altivec_vperm_<mode>_uns_internal): New insn having previous
4426 behavior of altivec_vperm_<mode>_uns.
4428 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4430 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
4431 (altivec_vsumsws): Add handling for -maltivec=be with a little
4433 (altivec_vsumsws_direct): New.
4434 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
4435 gen_altivec_vsumsws.
4437 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4439 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
4440 vtable_pointer_value_to_binfo): New functions.
4441 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
4442 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
4444 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
4446 * config/nios2/nios2.md (load_got_register): Initialize GOT
4447 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
4448 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
4450 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4452 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
4453 preserverd by passthrough, do not propagate the type.
4455 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4457 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
4458 (mips_atomic_assign_expand_fenv): New function.
4459 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4461 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4463 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4464 (__builtin_mips_set_fcsr): Likewise.
4465 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4466 MIPS_USI_FTYPE_VOID.
4467 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4468 (mips16_expand_set_fcsr): Likewise.
4469 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4470 (mips16_set_fcsr_stub): Likewise.
4471 (mips16_get_fcsr_one_only_stub): New class.
4472 (mips16_set_fcsr_one_only_stub): Likewise.
4473 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4474 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4475 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4476 (hard_float): New availability predicate.
4477 (mips_builtins): Add get_fcsr and set_fcsr.
4478 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4479 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4480 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4481 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4482 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4485 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4487 * config/mips/mips.c (mips_one_only_stub): New class.
4488 (mips_need_mips16_rdhwr_p): Replace with...
4489 (mips16_rdhwr_stub): ...this new variable.
4490 (mips16_stub_call_address): New function.
4491 (mips16_rdhwr_one_only_stub): New class.
4492 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4493 (mips_output_mips16_rdhwr): Delete.
4494 (mips_finish_stub): New function.
4495 (mips_code_end): Use it to handle rdhwr stubs.
4497 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4500 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4501 when calculating size of integer atomic types.
4503 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4505 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4507 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4509 PR tree-optimization/60003
4510 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4511 * profile.c (branch_prob): Use gimple_call_builtin_p
4512 to check for BUILT_IN_SETJMP_RECEIVER.
4513 * tree-inline.c (copy_bb): Call notice_special_calls.
4515 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4518 * lra-constraints.c (process_alt_operands): Update reload_sum only
4521 2014-01-31 Richard Henderson <rth@redhat.com>
4524 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4525 until after else_eh is processed.
4527 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4529 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4530 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4531 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4532 in smmintrin.h, remove them.
4533 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4534 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4535 * config/i386/i386.md (ROUND_SAE): Fix value.
4536 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4537 (const48_operand): New.
4538 * config/i386/subst.md (round), (round_expand): Use
4539 const_4_or_8_to_11_operand.
4540 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4542 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4544 * config/i386/constraints.md (Yk): Swap meaning with k.
4545 * config/i386/i386.md (movhi_internal): Change Yk to k.
4546 (movqi_internal): Ditto.
4547 (*k<logic><mode>): Ditto.
4550 (kandn<mode>): Ditto.
4551 (*<code>hi_1): Ditto.
4552 (*<code>qi_1): Ditto.
4553 (kxnor<mode>): Ditto.
4554 (kortestzhi): Ditto.
4555 (kortestchi): Ditto.
4557 (*one_cmplhi2_1): Ditto.
4558 (*one_cmplqi2_1): Ditto.
4559 * config/i386/sse.md (): Change k to Yk.
4560 (avx512f_load<mode>_mask): Ditto.
4561 (avx512f_blendm<mode>): Ditto.
4562 (avx512f_store<mode>_mask): Ditto.
4563 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4564 (avx512f_storedqu<mode>_mask): Ditto.
4565 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4567 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4568 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4569 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4570 (avx512f_maskcmp<mode>3): Ditto.
4571 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4572 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4573 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4574 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4575 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4576 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4577 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4578 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4579 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4580 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4581 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4582 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4583 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4584 (vec_extract_lo_<mode>_maskm): Ditto.
4585 (vec_extract_hi_<mode>_maskm): Ditto.
4586 (avx512f_vternlog<mode>_mask): Ditto.
4587 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4588 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4589 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4590 (avx512f_<code>v8div16qi2_mask): Ditto.
4591 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4592 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4593 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4594 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4595 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4596 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4597 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4598 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4599 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4600 (avx512cd_maskb_vec_dupv8di): Ditto.
4601 (avx512cd_maskw_vec_dupv16si): Ditto.
4602 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4603 (avx512f_vpermi2var<mode>3_mask): Ditto.
4604 (avx512f_vpermi2var<mode>3_mask): Ditto.
4605 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4606 (*avx512f_gathersi<mode>): Ditto.
4607 (*avx512f_gathersi<mode>_2): Ditto.
4608 (*avx512f_gatherdi<mode>): Ditto.
4609 (*avx512f_gatherdi<mode>_2): Ditto.
4610 (*avx512f_scattersi<mode>): Ditto.
4611 (*avx512f_scatterdi<mode>): Ditto.
4612 (avx512f_compress<mode>_mask): Ditto.
4613 (avx512f_compressstore<mode>_mask): Ditto.
4614 (avx512f_expand<mode>_mask): Ditto.
4615 * config/i386/subst.md (mask): Change k to Yk.
4616 (mask_scalar_merge): Ditto.
4619 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4621 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4623 2014-01-31 Richard Biener <rguenther@suse.de>
4626 * builtins.c (fold_builtin_memory_op): Make sure to not
4627 use a floating-point mode or a boolean or enumeral type for
4630 2014-01-30 DJ Delorie <dj@redhat.com>
4632 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4633 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4634 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4635 whenever main() has an epilogue.
4637 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4639 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4640 unused variable "field".
4641 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4642 (vsx_mergeh_<mode>): Likewise.
4643 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4644 (altivec_vmrghh): Likewise.
4645 (altivec_vmrghw): Likewise.
4646 (altivec_vmrglb): Likewise.
4647 (altivec_vmrglh): Likewise.
4648 (altivec_vmrglw): Likewise.
4649 (altivec_vspltb): Add missing uses.
4650 (altivec_vsplth): Likewise.
4651 (altivec_vspltw): Likewise.
4652 (altivec_vspltsf): Likewise.
4654 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4657 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4658 frame related instructions.
4660 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4662 PR rtl-optimization/59959
4663 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4664 any reload of register whose subreg is invalid.
4666 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4668 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4669 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4670 Add missing return type - void.
4672 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4674 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4675 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4676 remove element index adjustment for endian (now handled in vsx.md
4678 (altivec_expand_vec_perm_const): Use
4679 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4680 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4681 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4682 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4683 define_expand and a new define_insn *altivec_vspltb_internal;
4684 adjust for -maltivec=be on a little endian target.
4685 (altivec_vspltb_direct): New.
4686 (altivec_vsplth): Divide into a define_expand and a new
4687 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4688 little endian target.
4689 (altivec_vsplth_direct): New.
4690 (altivec_vspltw): Divide into a define_expand and a new
4691 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4692 little endian target.
4693 (altivec_vspltw_direct): New.
4694 (altivec_vspltsf): Divide into a define_expand and a new
4695 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4696 a little endian target.
4698 2014-01-30 Richard Biener <rguenther@suse.de>
4700 PR tree-optimization/59993
4701 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4702 can propagate form the earlier stmt and avoid the transform
4703 when the intermediate result is needed.
4705 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4707 * README.Portability: Fix typo.
4709 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4711 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4712 comparison_operator with ordered_comparison_operator.
4714 2014-01-30 Nick Clifton <nickc@redhat.com>
4716 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4717 Rename to mn10300_store_multiple_regs.
4718 * config/mn10300/mn10300.c: Likewise.
4719 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4720 store_multiple_regs.
4721 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4722 Call mn10300_store_multiple_regs.
4724 2014-01-30 Nick Clifton <nickc@redhat.com>
4725 DJ Delorie <dj@redhat.com>
4727 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4728 %fp 2 to keep registers after it properly word-aligned.
4729 (rl78_alloc_physical_registers_umul): Handle the case where both
4730 input operands are the same.
4732 2014-01-30 Richard Biener <rguenther@suse.de>
4734 PR tree-optimization/59903
4735 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4738 2014-01-30 Jason Merrill <jason@redhat.com>
4741 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4744 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4746 2014-01-30 Richard Biener <rguenther@suse.de>
4748 PR tree-optimization/59951
4749 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4751 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4754 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4755 SFmode to DFmode case.
4757 2014-01-29 DJ Delorie <dj@redhat.com>
4759 * config/msp430/msp430.opt (-minrt): New.
4760 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4762 (ENDFILE_SPEC): Likewise.
4764 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4766 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4767 (estimate_function_body_sizes): Use it.
4769 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4772 * dwarf2out.c (is_cxx_auto): New.
4773 (is_base_type): Use it.
4774 (gen_type_die_with_usage): Likewise.
4776 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4778 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4779 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4780 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4781 -maltivec=be with LE targets.
4782 (vsx_mergeh_<mode>): Likewise.
4783 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4784 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4785 (altivec_vmrghb): Replace with define_expand and new
4786 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4787 (altivec_vmrghb_direct): New define_insn.
4788 (altivec_vmrghh): Replace with define_expand and new
4789 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4790 (altivec_vmrghh_direct): New define_insn.
4791 (altivec_vmrghw): Replace with define_expand and new
4792 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4793 (altivec_vmrghw_direct): New define_insn.
4794 (*altivec_vmrghsf): Adjust for endianness.
4795 (altivec_vmrglb): Replace with define_expand and new
4796 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4797 (altivec_vmrglb_direct): New define_insn.
4798 (altivec_vmrglh): Replace with define_expand and new
4799 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4800 (altivec_vmrglh_direct): New define_insn.
4801 (altivec_vmrglw): Replace with define_expand and new
4802 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4803 (altivec_vmrglw_direct): New define_insn.
4804 (*altivec_vmrglsf): Adjust for endianness.
4805 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4806 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4807 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4808 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4809 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4810 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4811 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4812 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4814 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4816 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4817 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4820 2014-01-29 Richard Biener <rguenther@suse.de>
4822 PR tree-optimization/58742
4823 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4824 associate_pointerplus_align.
4825 (associate_pointerplus_diff): New function.
4826 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4827 and associate_pointerplus_diff.
4829 2014-01-29 Richard Biener <rguenther@suse.de>
4831 * lto-streamer.h (LTO_major_version): Bump to 3.
4832 (LTO_minor_version): Reset to 0.
4834 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4836 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4837 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4838 (arm_file_start): Generate correct asm header for armv7ve.
4839 * config/arm/bpabi.h: Add multilib support for armv7ve.
4840 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4841 and cortex-a15 to armv7ve.
4842 * config/arm/t-aprofile: Add multilib support for armv7ve.
4843 * doc/invoke.texi: Document -march=armv7ve.
4845 2014-01-29 Richard Biener <rguenther@suse.de>
4847 PR tree-optimization/58742
4848 * tree-ssa-forwprop.c (associate_plusminus): Return true
4849 if we changed sth, defer EH cleanup to ...
4850 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4851 (simplify_mult): New function.
4853 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4856 PR tree-optimization/59920
4857 * tree.c (build_common_builtin_nodes): Remove
4858 __builtin_setjmp_dispatcher initialization.
4859 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4860 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4861 instead of gsi_after_labels + manually skipping debug stmts.
4862 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4863 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4864 * tree-inline.c (copy_edges_for_bb): Remove
4865 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4866 argument. Ignore computed_goto_p stmts. Don't call
4867 make_abnormal_goto_edges. If a call might need abnormal edges
4868 for non-local gotos, see if it already has an edge to
4869 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4870 with true argument, don't do anything then, otherwise add
4871 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4872 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4874 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4875 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4876 (lower_stmt): Don't set data->calls_builtin_setjmp.
4877 (lower_builtin_setjmp): Adjust comment.
4878 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4879 * tree-cfg.c (found_computed_goto): Remove.
4880 (factor_computed_gotos): Remove.
4881 (make_goto_expr_edges): Return bool, true for computed gotos.
4882 Don't call make_abnormal_goto_edges.
4883 (build_gimple_cfg): Don't set found_computed_goto, don't call
4884 factor_computed_gotos.
4885 (computed_goto_p): No longer static.
4886 (make_blocks): Don't set found_computed_goto.
4887 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4888 (make_edges): If make_goto_expr_edges returns true, push bb
4889 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4890 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4891 vector. Record mapping between bbs and OpenMP regions if there
4892 are any, adjust make_gimple_omp_edges caller. Call
4893 handle_abnormal_edges.
4894 (make_abnormal_goto_edges): Remove.
4895 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4896 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4897 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4898 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4899 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4900 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4901 filling *region also set *region_idx to (*region)->entry->index.
4904 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4905 For REGs set ORIGINAL_REGNO.
4907 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4909 * doc/md.texi: Mention that a target shouldn't implement
4910 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4913 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4915 PR tree-optimization/59594
4916 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4917 a copy of the datarefs vector rather than the vector itself.
4919 2014-01-28 Jason Merrill <jason@redhat.com>
4922 * dwarf2out.c (auto_die): New static.
4923 (gen_type_die_with_usage): Handle C++1y 'auto'.
4924 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4927 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4930 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4931 (SPEC_X32): Likewise.
4932 (SPEC_64): Likewise.
4933 * config/i386/i386.c (ix86_option_override_internal): Turn off
4934 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4936 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4937 * config/i386/i386.h (TARGET_16BIT): New macro.
4938 (TARGET_16BIT_P): Likewise.
4939 * config/i386/i386.opt: Add m16.
4940 * doc/invoke.texi: Document -m16.
4942 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4944 PR preprocessor/59935
4945 * input.c (location_get_source_line): Bail out on when line number
4946 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4948 2014-01-28 Richard Biener <rguenther@suse.de>
4950 PR tree-optimization/58742
4951 * tree-ssa-forwprop.c (associate_plusminus): Handle
4952 pointer subtraction of the form (T)(P + A) - (T)P.
4954 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4956 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4959 2014-01-28 Richard Biener <rguenther@suse.de>
4962 2014-01-28 Richard Biener <rguenther@suse.de>
4964 PR rtl-optimization/45364
4965 PR rtl-optimization/59890
4966 * var-tracking.c (local_get_addr_clear_given_value): Handle
4967 already cleared slot.
4968 (val_reset): Handle not allocated local_get_addr_cache.
4969 (vt_find_locations): Use post-order on the inverted CFG.
4971 2014-01-28 Richard Biener <rguenther@suse.de>
4973 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4975 2014-01-28 Richard Biener <rguenther@suse.de>
4977 PR rtl-optimization/45364
4978 PR rtl-optimization/59890
4979 * var-tracking.c (local_get_addr_clear_given_value): Handle
4980 already cleared slot.
4981 (val_reset): Handle not allocated local_get_addr_cache.
4982 (vt_find_locations): Use post-order on the inverted CFG.
4984 2014-01-28 Alan Modra <amodra@gmail.com>
4986 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4987 * configure.ac <recursive call for build != host>: Define
4988 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4989 and LD_FOR_BUILD too.
4990 * configure: Regenerate.
4992 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4994 * config/i386/i386.c (get_builtin_code_for_version): Separate
4995 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4996 Broadwell from Haswell.
4998 2014-01-27 Steve Ellcey <sellcey@mips.com>
5000 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
5001 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
5002 * config/mips/mips.c (mips_option_override): Change setting
5004 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
5005 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
5006 Change from Mask to Var.
5008 2014-01-27 Jeff Law <law@redhat.com>
5010 * ipa-inline.c (inline_small_functions): Fix typo.
5012 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
5014 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
5015 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
5016 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
5017 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
5018 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
5019 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
5020 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
5021 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
5022 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
5023 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
5024 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
5025 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
5026 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
5027 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
5028 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
5029 (_mm512_storeu_epi64): Ditto.
5030 (_mm512_cmpge_epi32_mask): Ditto.
5031 (_mm512_cmpge_epu32_mask): Ditto.
5032 (_mm512_cmpge_epi64_mask): Ditto.
5033 (_mm512_cmpge_epu64_mask): Ditto.
5034 (_mm512_cmple_epi32_mask): Ditto.
5035 (_mm512_cmple_epu32_mask): Ditto.
5036 (_mm512_cmple_epi64_mask): Ditto.
5037 (_mm512_cmple_epu64_mask): Ditto.
5038 (_mm512_cmplt_epi32_mask): Ditto.
5039 (_mm512_cmplt_epu32_mask): Ditto.
5040 (_mm512_cmplt_epi64_mask): Ditto.
5041 (_mm512_cmplt_epu64_mask): Ditto.
5042 (_mm512_cmpneq_epi32_mask): Ditto.
5043 (_mm512_cmpneq_epu32_mask): Ditto.
5044 (_mm512_cmpneq_epi64_mask): Ditto.
5045 (_mm512_cmpneq_epu64_mask): Ditto.
5046 (_mm512_expand_pd): Ditto.
5047 (_mm512_expand_ps): Ditto.
5048 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
5049 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
5050 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
5051 * config/i386/i386.c (ix86_builtins): Add
5052 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
5053 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
5054 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
5055 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
5056 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
5057 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
5058 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
5059 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
5060 IX86_BUILTIN_PMOVUSQW512_MEM.
5061 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
5062 __builtin_ia32_pmovsqd512mem_mask,
5063 __builtin_ia32_pmovqd512mem_mask,
5064 __builtin_ia32_pmovusqw512mem_mask,
5065 __builtin_ia32_pmovsqw512mem_mask,
5066 __builtin_ia32_pmovqw512mem_mask,
5067 __builtin_ia32_pmovusdw512mem_mask,
5068 __builtin_ia32_pmovsdw512mem_mask,
5069 __builtin_ia32_pmovdw512mem_mask,
5070 __builtin_ia32_pmovqb512mem_mask,
5071 __builtin_ia32_pmovusqb512mem_mask,
5072 __builtin_ia32_pmovsqb512mem_mask,
5073 __builtin_ia32_pmovusdb512mem_mask,
5074 __builtin_ia32_pmovsdb512mem_mask,
5075 __builtin_ia32_pmovdb512mem_mask.
5076 (bdesc_args): Add __builtin_ia32_expanddf512,
5077 __builtin_ia32_expandsf512.
5078 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
5079 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
5080 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
5081 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
5082 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
5083 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
5084 (avx512f_<code>v8div16qi2_mask_store): This.
5085 (avx512f_expand<mode>): New.
5087 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
5089 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
5091 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5092 (_mm512_prefetch_i32scatter_pd): Ditto.
5093 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5094 (_mm512_prefetch_i64scatter_pd): Ditto.
5095 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5096 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
5097 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5098 (_mm512_prefetch_i32scatter_ps): Ditto.
5099 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5100 (_mm512_prefetch_i64scatter_ps): Ditto.
5101 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5102 * config/i386/i386-builtin-types.def: Define
5103 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
5104 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
5105 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
5106 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
5107 IX86_BUILTIN_SCATTERPFQPD.
5108 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
5109 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
5110 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
5111 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
5112 __builtin_ia32_scatterpfqps.
5113 (ix86_expand_builtin): Expand new built-ins.
5114 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
5115 fix memory access data type.
5116 (*avx512pf_gatherpf<mode>_mask): Ditto.
5117 (*avx512pf_gatherpf<mode>): Ditto.
5118 (avx512pf_scatterpf<mode>): Ditto.
5119 (*avx512pf_scatterpf<mode>_mask): Ditto.
5120 (*avx512pf_scatterpf<mode>): Ditto.
5121 (GATHER_SCATTER_SF_MEM_MODE): New.
5122 (avx512pf_gatherpf<mode>df): Ditto.
5123 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5124 (*avx512pf_scatterpf<mode>df): Ditto.
5126 2014-01-27 Jakub Jelinek <jakub@redhat.com>
5129 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
5130 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
5133 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5135 * common/config/arm/arm-common.c
5136 (arm_rewrite_mcpu): Handle multiple names.
5138 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5140 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5142 * gimple-builder.h (create_gimple_tmp): Delete.
5144 2014-01-27 Christian Bruel <christian.bruel@st.com>
5146 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
5149 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
5151 * config/pa/pa.md (call): Generate indirect long calls to non-local
5152 functions when outputing 32-bit code.
5153 (call_value): Likewise except for special call to buggy powf function.
5155 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
5156 portable runtime and PIC indirect calls.
5157 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
5158 and PIC call sequences. Use ldo instead of blr to set return register
5159 in PIC call sequence.
5161 2014-01-25 Walter Lee <walt@tilera.com>
5163 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
5164 avoid clobbering a live register.
5166 2014-01-25 Walter Lee <walt@tilera.com>
5168 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
5169 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
5170 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
5171 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
5173 2014-01-25 Walter Lee <walt@tilera.com>
5175 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
5176 arguments on even registers.
5177 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
5179 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
5180 (BIGGEST_ALIGNMENT): Ditto.
5181 (BIGGEST_FIELD_ALIGNMENT): Ditto.
5183 2014-01-25 Walter Lee <walt@tilera.com>
5185 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
5186 insns before bundling.
5187 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
5189 2014-01-25 Walter Lee <walt@tilera.com>
5191 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
5192 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
5193 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
5195 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5197 * config/mips/constraints.md (kl): Delete.
5198 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
5199 define expands, using...
5200 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
5201 instructions for MIPS16.
5202 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
5203 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
5205 2014-01-25 Walter Lee <walt@tilera.com>
5207 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
5211 2014-01-25 Walter Lee <walt@tilera.com>
5213 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
5214 (TARGET_EXPAND_TO_RTL_HOOK): Define.
5216 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5218 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
5221 2014-01-25 Jakub Jelinek <jakub@redhat.com>
5223 * print-rtl.c (in_call_function_usage): New var.
5224 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
5225 EXPR_LIST mode as mode and not as reg note name.
5228 * cfgloopmanip.c (copy_loop_info): If
5229 loop->warned_aggressive_loop_optimizations, make sure
5230 the flag is set in target loop too.
5232 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
5234 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
5236 * builtins.def: Likewise.
5237 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
5238 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
5239 * ira.c (ira_setup_eliminable_regset): Likewise.
5240 * omp-low.c (gate_expand_omp): Likewise.
5241 (execute_lower_omp): Likewise.
5242 (diagnose_sb_0): Likewise.
5243 (gate_diagnose_omp_blocks): Likewise.
5244 (simd_clone_clauses_extract): Likewise.
5247 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5249 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
5250 correction for little endian...
5251 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
5254 2014-01-24 Jeff Law <law@redhat.com>
5256 PR tree-optimization/59919
5257 * tree-vrp.c (find_assert_locations_1): Do not register asserts
5258 for non-returning calls.
5260 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
5262 * common/config/aarch64/aarch64-common.c
5263 (aarch64_rewrite_mcpu): Handle multiple names.
5264 * config/aarch64/aarch64.h
5265 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5267 2014-01-24 Dodji Seketeli <dodji@redhat.com>
5269 * input.c (add_file_to_cache_tab): Handle the case where fopen
5272 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
5275 * config/i386/i386.md (pushsf splitter): Get stack adjustment
5276 from push operand if code of push isn't PRE_DEC.
5278 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
5281 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
5282 -mquad-memory-atomic. Update -mquad-memory documentation to say
5283 it is only used for non-atomic loads/stores.
5285 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
5286 -mquad-memory or -mquad-memory-atomic switches.
5288 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
5289 -mquad-memory-atomic to ISA 2.07 support.
5291 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
5292 to separate support of normal quad word memory operations (ldq, stq)
5293 from the atomic quad word memory operations.
5295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5296 support to separate non-atomic quad word operations from atomic
5297 quad word operations. Disable non-atomic quad word operations in
5298 little endian mode so that we don't have to swap words after the
5299 load and before the store.
5300 (quad_load_store_p): Add comment about atomic quad word support.
5301 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
5302 options printed with -mdebug=reg.
5304 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
5305 -mquad-memory-atomic as the test for whether we have quad word
5306 atomic instructions.
5307 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
5308 or -mp8-vector are used, allow byte/half-word atomic operations.
5310 * config/rs6000/sync.md (load_lockedti): Insure that the address
5311 is a proper indexed or indirect address for the lqarx instruction.
5312 On little endian systems, swap the hi/lo registers after the lqarx
5314 (load_lockedpti): Use indexed_or_indirect_operand predicate to
5315 insure the address is valid for the lqarx instruction.
5316 (store_conditionalti): Insure that the address is a proper indexed
5317 or indirect address for the stqcrx. instruction. On little endian
5318 systems, swap the hi/lo registers before doing the stqcrx.
5320 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
5321 insure the address is valid for the stqcrx. instruction.
5323 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
5324 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
5325 type of quad memory support is available.
5327 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
5330 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
5331 there is a danger of looping.
5333 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
5335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5336 force flag_ira_loop_pressure if set via command line.
5338 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5340 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
5341 (ashr_simd): New builtin handling DI mode.
5342 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
5343 (aarch64_sshr_simddi): New match pattern.
5344 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
5345 (vshrd_n_s64): Likewise.
5346 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
5348 2014-01-23 Nick Clifton <nickc@redhat.com>
5350 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
5351 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
5352 favour of mcu specific scripts.
5353 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
5356 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5357 Alex Velenko <Alex.Velenko@arm.com>
5359 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
5360 (vaddv_s16): Likewise.
5361 (vaddv_s32): Likewise.
5362 (vaddv_u8): Likewise.
5363 (vaddv_u16): Likewise.
5364 (vaddv_u32): Likewise.
5365 (vaddvq_s8): Likewise.
5366 (vaddvq_s16): Likewise.
5367 (vaddvq_s32): Likewise.
5368 (vaddvq_s64): Likewise.
5369 (vaddvq_u8): Likewise.
5370 (vaddvq_u16): Likewise.
5371 (vaddvq_u32): Likewise.
5372 (vaddvq_u64): Likewise.
5373 (vaddv_f32): Likewise.
5374 (vaddvq_f32): Likewise.
5375 (vaddvq_f64): Likewise.
5376 (vmaxv_f32): Likewise.
5377 (vmaxv_s8): Likewise.
5378 (vmaxv_s16): Likewise.
5379 (vmaxv_s32): Likewise.
5380 (vmaxv_u8): Likewise.
5381 (vmaxv_u16): Likewise.
5382 (vmaxv_u32): Likewise.
5383 (vmaxvq_f32): Likewise.
5384 (vmaxvq_f64): Likewise.
5385 (vmaxvq_s8): Likewise.
5386 (vmaxvq_s16): Likewise.
5387 (vmaxvq_s32): Likewise.
5388 (vmaxvq_u8): Likewise.
5389 (vmaxvq_u16): Likewise.
5390 (vmaxvq_u32): Likewise.
5391 (vmaxnmv_f32): Likewise.
5392 (vmaxnmvq_f32): Likewise.
5393 (vmaxnmvq_f64): Likewise.
5394 (vminv_f32): Likewise.
5395 (vminv_s8): Likewise.
5396 (vminv_s16): Likewise.
5397 (vminv_s32): Likewise.
5398 (vminv_u8): Likewise.
5399 (vminv_u16): Likewise.
5400 (vminv_u32): Likewise.
5401 (vminvq_f32): Likewise.
5402 (vminvq_f64): Likewise.
5403 (vminvq_s8): Likewise.
5404 (vminvq_s16): Likewise.
5405 (vminvq_s32): Likewise.
5406 (vminvq_u8): Likewise.
5407 (vminvq_u16): Likewise.
5408 (vminvq_u32): Likewise.
5409 (vminnmv_f32): Likewise.
5410 (vminnmvq_f32): Likewise.
5411 (vminnmvq_f64): Likewise.
5413 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5415 * config/aarch64/aarch64-simd.md
5416 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
5417 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
5418 (*aarch64_mul3_elt<mode>): Likewise.
5419 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
5420 (*aarch64_mul3_elt_to_64v2df): Likewise.
5421 (*aarch64_mla_elt<mode>): Likewise.
5422 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5423 (*aarch64_mls_elt<mode>): Likewise.
5424 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5425 (*aarch64_fma4_elt<mode>): Likewise.
5426 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5427 (*aarch64_fma4_elt_to_64v2df): Likewise.
5428 (*aarch64_fnma4_elt<mode>): Likewise.
5429 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5430 (*aarch64_fnma4_elt_to_64v2df): Likewise.
5431 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
5432 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
5433 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5434 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5435 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5436 (aarch64_sqdmull_lane<mode>_internal): Likewise.
5437 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5439 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
5441 * config/aarch64/aarch64-simd.md
5442 (aarch64_be_checked_get_lane<mode>): New define_expand.
5443 * config/aarch64/aarch64-simd-builtins.def
5444 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
5445 New builtin definition.
5446 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
5447 Use new safe be builtin.
5449 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5451 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
5453 (aarch64_be_st1<mode>): Likewise.
5454 (aarch_ld1<VALL:mode>): Define_expand modified.
5455 (aarch_st1<VALL:mode>): Likewise.
5456 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
5457 (UNSPEC_ST1): Likewise.
5459 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5461 * config/microblaze/microblaze.md: Add trap insn and attribute
5463 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5465 PR preprocessor/58580
5466 * input.h (location_get_source_line): Take an additional line_size
5468 (void diagnostics_file_cache_fini): Declare new function.
5469 * input.c (struct fcache): New type.
5470 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5471 New static constants.
5472 (diagnostic_file_cache_init, total_lines_num)
5473 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5474 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5475 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5476 (get_next_line, read_next_line, goto_next_line, read_line_num):
5477 New static function definitions.
5478 (diagnostic_file_cache_fini): New function.
5479 (location_get_source_line): Take an additional output line_len
5480 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5482 * diagnostic.c (diagnostic_finish): Call
5483 diagnostic_file_cache_fini.
5484 (adjust_line): Take an additional input parameter for the length
5485 of the line, rather than calculating it with strlen.
5486 (diagnostic_show_locus): Adjust the use of
5487 location_get_source_line and adjust_line with respect to their new
5488 signature. While displaying a line now, do not stop at the first
5489 null byte. Rather, display the zero byte as a space and keep
5490 going until we reach the size of the line.
5491 * Makefile.in: Add vec.o to OBJS-libcommon
5493 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5494 Ilya Tocar <ilya.tocar@intel.com>
5496 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5497 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5498 (__builtin_ia32_kmov16): Ditto.
5499 * config/i386/i386.md (UNSPEC_KMOV): New.
5502 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5504 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5505 (_mm512_storeu_si512): Ditto.
5507 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5510 * rtl.h (get_referenced_operands): Declare.
5511 * recog.c (get_referenced_operands): New function.
5512 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5513 operands have been referenced when recording LO_SUM references.
5515 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5517 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5519 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5521 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5522 Enable for generic and recent AMD targets.
5524 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5526 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5527 ARG_SIZE note when adjustment was eliminated.
5529 2014-01-22 Jeff Law <law@redhat.com>
5531 PR tree-optimization/59597
5532 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5533 in file. Accept new argument REGISTERING and use it to modify
5534 dump output appropriately.
5535 (register_jump_thread): Corresponding changes.
5536 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5537 thread paths involving joiner blocks. Add code to dump cancelled
5538 jump threading paths.
5540 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5542 PR rtl-optimization/59477
5543 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5544 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5546 2014-01-22 Tom Tromey <tromey@redhat.com>
5548 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5550 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5552 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5554 PR rtl-optimization/59896
5555 * lra-constraints.c (process_alt_operands): Check unused note for
5556 matched operands of insn with no output reloads.
5558 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5560 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5561 (mips_move_from_gpr_cost): Likewise.
5563 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5565 PR rtl-optimization/59858
5566 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5567 ira_class_hard_regs_num.
5568 (process_alt_operands): Increase reject for dying matched operand.
5570 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5573 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5574 smaller than size, perform several stores or loads and stores
5575 at dst + count - size to store or copy all of size bytes, rather
5576 than just last modesize bytes.
5578 2014-01-20 DJ Delorie <dj@redhat.com>
5580 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5581 that CLOBBERs are REGs before propogating their values.
5583 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5586 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5587 (cgraph_inline_failed_type): New function.
5588 * cgraph.h (DEFCIFCODE): Add type.
5589 (cgraph_inline_failed_type_t): New enum.
5590 (cgraph_inline_failed_type): New prototype.
5591 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5592 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5593 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5594 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5595 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5596 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5597 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5598 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5599 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5600 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5601 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5602 OPTIMIZATION_MISMATCH.
5603 * tree-inline.c (expand_call_inline): Emit errors during
5604 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5606 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5609 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5610 mode attribute in insn output.
5612 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5614 * output.h (output_constant): Delete.
5615 * varasm.c (output_constant): Make private.
5617 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5619 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5621 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5624 * tree.h (fold_builtin_strcat): New prototype.
5625 * builtins.c (fold_builtin_strcat): No longer static. Add len
5626 argument, if non-NULL, don't call c_strlen. Optimize
5627 directly into __builtin_memcpy instead of __builtin_strcpy.
5628 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5629 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5631 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5633 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5634 for SImode_address_operand operands, having only a REG argument.
5636 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5638 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5639 loader name using mbig-endian.
5640 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5642 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5644 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5648 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5650 * config/aarch64/aarch64-protos.h
5651 (aarch64_cannot_change_mode_class_ptr): Declare.
5652 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5653 aarch64_cannot_change_mode_class_ptr): New.
5654 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5655 backend hook aarch64_cannot_change_mode_class.
5657 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5659 * common/config/aarch64/aarch64-common.c
5660 (aarch64_handle_option): Don't handle any option order logic here.
5661 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5662 selected_cpu, warn on architecture version mismatch.
5663 (aarch64_override_options): Fix parsing order for option strings.
5665 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5666 Iain Sandoe <iain@codesourcery.com>
5669 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5670 warning. Amend comment to reflect current functionality.
5672 2014-01-20 Richard Biener <rguenther@suse.de>
5675 * builtins.c (fold_builtin_strcat): Remove case better handled
5676 by tree-ssa-strlen.c.
5678 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5680 * config/aarch64/aarch64.opt
5681 (mcpu, march, mtune): Make case-insensitive.
5683 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5686 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5687 if operands[1] is a REG or ZERO_EXTEND of a REG.
5689 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5691 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5693 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5695 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5696 long non-pic millicode calls.
5698 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5700 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5702 2014-01-19 Kito Cheng <kito@0xlab.org>
5704 * builtins.c (expand_movstr): Check movstr expand done or fail.
5706 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5707 H.J. Lu <hongjiu.lu@intel.com>
5710 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5711 to DImode for zero-extended addresses.
5713 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5715 PR rtl-optimization/57763
5716 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5717 on the new indirect jump_insn and increment LABEL_NUSES (label).
5719 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5723 * config.gcc (x86_archs): New variable.
5724 (x86_64_archs): Likewise.
5725 (x86_cpus): Likewise.
5726 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5727 --with-arch/--with-cpu= options.
5728 Support --with-arch=/--with-cpu={nehalem,westmere,
5729 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5731 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5733 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5734 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5736 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5738 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5740 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5743 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5744 clear cpp_get_options (parse_in)->warn_unused_macros for
5745 ix86_target_macros_internal with cpp_define.
5747 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5749 * jump.c (delete_related_insns): Keep (use (insn))s.
5750 * reorg.c (redundant_insn): Check for barriers too.
5752 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5754 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5756 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5758 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5759 call to $$dyncall when TARGET_LONG_CALLS is true.
5761 2014-01-17 Jeff Law <law@redhat.com>
5763 * ree.c (combine_set_extension): Temporarily disable test for
5764 changing number of hard registers.
5766 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5769 * ipa-inline-analysis.c (inline_free_summary):
5770 Do not free summary of aliases.
5772 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5775 * gimplify.c (gimplify_expr): Use create_tmp_var
5776 instead of create_tmp_var_raw. If cond doesn't have
5777 integral type, don't add the IFN_ANNOTATE builtin at all.
5779 2014-01-17 Martin Jambor <mjambor@suse.cz>
5782 * ipa-cp.c (prev_edge_clone): New variable.
5783 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5784 Also resize prev_edge_clone vector.
5785 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5786 (ipcp_edge_removal_hook): New function.
5787 (ipcp_driver): Register ipcp_edge_removal_hook.
5789 2014-01-17 Andrew Pinski <apinski@cavium.com>
5790 Steve Ellcey <sellcey@mips.com>
5793 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5796 2014-01-17 Jeff Law <law@redhat.com>
5799 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5800 so that pass_ccp runs first.
5802 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5804 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5805 (ix86_adjust_cost): Use !TARGET_XXX.
5806 (do_reorder_for_imul): Likewise.
5807 (swap_top_of_ready_list): Likewise.
5808 (ix86_sched_reorder): Likewise.
5810 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5812 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5813 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5814 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5815 (intel_memset): New. Duplicate slm_memset.
5816 (intel_cost): New. Duplicate slm_cost.
5817 (m_INTEL): New macro.
5818 (processor_target_table): Add "intel".
5819 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5820 with PROCESSOR_INTEL for "intel".
5821 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5822 PROCESSOR_SILVERMONT.
5823 (ix86_issue_rate): Likewise.
5824 (ix86_adjust_cost): Likewise.
5825 (ia32_multipass_dfa_lookahead): Likewise.
5826 (swap_top_of_ready_list): Likewise.
5827 (ix86_sched_reorder): Likewise.
5828 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5829 instead of TARGET_OPT_AGU.
5830 * config/i386/i386.h (TARGET_INTEL): New.
5831 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5832 (processor_type): Add PROCESSOR_INTEL.
5833 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5834 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5836 2014-01-17 Marek Polacek <polacek@redhat.com>
5839 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5842 2014-01-17 Richard Biener <rguenther@suse.de>
5844 PR tree-optimization/46590
5845 * opts.c (default_options_table): Add entries for
5846 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5847 all enabled at -O1 but not for -Og.
5848 * common.opt (fbranch-count-reg): Remove Init(1).
5849 (fmove-loop-invariants): Likewise.
5850 (ftree-pta): Likewise.
5852 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5854 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5855 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5856 decls to at least the GCC 4.8 used alignments.
5859 * tree-nested.c (convert_nonlocal_reference_stmt,
5860 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5861 of GIMPLE_BIND stmts, adjust associated decls.
5863 2014-01-17 Richard Biener <rguenther@suse.de>
5865 PR tree-optimization/46590
5866 * vec.h (vec<>::bseach): New member function implementing
5867 binary search according to C89 bsearch.
5868 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5869 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5870 bitmap pointer again. Make accesses_in_loop a flat array.
5871 (mem_ref_obstack): New global.
5872 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5873 (mark_ref_stored): Likewise.
5874 (ref_indep_loop_p_2): Likewise.
5875 (set_ref_stored_in_loop): New helper function.
5876 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5877 (memref_free): Adjust.
5878 (record_mem_ref_loc): Simplify.
5879 (gather_mem_refs_stmt): Adjust.
5880 (sort_locs_in_loop_postorder_cmp): New function.
5881 (analyze_memory_references): Sort accesses_in_loop after
5882 loop postorder number.
5883 (find_ref_loc_in_loop_cmp): New function.
5884 (for_all_locs_in_loop): Find relevant cluster of locs in
5885 accesses_in_loop and iterate without recursion.
5886 (execute_sm): Avoid uninit warning.
5887 (struct ref_always_accessed): Simplify.
5888 (ref_always_accessed::operator ()): Likewise.
5889 (ref_always_accessed_p): Likewise.
5890 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5891 loop postorder numbers here.
5892 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5895 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5898 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5899 on decls for which assemble_alias has been called.
5901 2014-01-17 Nick Clifton <nickc@redhat.com>
5903 * config/msp430/msp430.opt: (mcpu): New option.
5904 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5905 (msp430_option_override): Parse target_cpu. If the MCU name
5906 matches a generic string, clear target_mcu.
5907 (msp430_attr): Allow numeric interrupt values up to 63.
5908 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5909 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5911 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5913 * config/msp430/msp430.md (popm): Use %J rather than %I.
5914 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5915 (addhi_cy_i): Use immediate_operand for operand 2.
5916 * doc/invoke.texi: Document -mcpu option.
5918 2014-01-17 Richard Biener <rguenther@suse.de>
5920 PR rtl-optimization/38518
5921 * df.h (df_analyze_loop): Declare.
5922 * df-core.c: Include cfgloop.h.
5923 (df_analyze_1): Split out main part of df_analyze.
5924 (df_analyze): Adjust.
5925 (loop_inverted_post_order_compute): New function.
5926 (loop_post_order_compute): Likewise.
5927 (df_analyze_loop): New function avoiding whole-function
5929 * loop-invariant.c (find_defs): Use df_analyze_loop.
5930 (find_invariants): Adjust.
5931 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5933 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5935 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5936 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5938 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5940 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5941 traversal when removing references.
5943 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5946 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5948 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5951 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5952 pushing a reload for an autoinc when we had previously reloaded an
5953 inner part of the address.
5955 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5957 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5959 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5960 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5961 when not giving up or versioning for alias only because of
5963 (vect_analyze_data_ref_dependences): Set to true.
5964 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5966 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5967 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5971 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5974 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5975 operand 0 predicate for gathers, use a new pseudo as subtarget.
5977 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5980 * lra-constraints.c (process_alt_operands): Add printing debug info.
5981 Check absence of input/output reloads for matched operands too.
5983 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5985 PR rtl-optimization/59835
5986 * ira.c (ira_init_register_move_cost): Increase cost for
5989 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5991 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5993 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5996 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5997 non-register objects. Use gen_(high/low)part more consistently.
6000 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
6003 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
6004 endian support, remove tests for WORDS_BIG_ENDIAN.
6005 (p8_mfvsrd_3_<mode>): Likewise.
6006 (reload_gpr_from_vsx<mode>): Likewise.
6007 (reload_gpr_from_vsxsf): Likewise.
6008 (p8_mfvsrd_4_disf): Likewise.
6010 2014-01-16 Richard Biener <rguenther@suse.de>
6012 PR rtl-optimization/46590
6013 * lcm.c (compute_antinout_edge): Use postorder iteration.
6014 (compute_laterin): Use inverted postorder iteration.
6016 2014-01-16 Nick Clifton <nickc@redhat.com>
6019 * varasm.c (output_constant): Return the number of bytes actually
6021 (output_constructor_array_range): Update the field size with the
6022 number of bytes emitted by output_constant.
6023 (output_constructor_regular_field): Likewise. Also do not
6024 complain if the total number of bytes emitted is now greater
6025 than the expected fieldpos.
6026 * output.h (output_constant): Update prototype and descriptive comment.
6028 2014-01-16 Marek Polacek <polacek@redhat.com>
6031 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
6032 it is error_mark_node.
6034 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
6036 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
6037 VALID_AVX256_REG_OR_OI_MODE.
6039 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
6041 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
6042 current procedure should be profiled.
6044 2014-01-15 Andrew Pinski <apinski@cavium.com>
6046 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
6047 of moving from/to the STACK_REG register class.
6049 2014-01-15 Richard Henderson <rth@redhat.com>
6052 * reginfo.c (global_regs_decl): Globalize.
6053 * rtl.h (global_regs_decl): Declare.
6054 * ira.c (do_reload): Diagnose frame_pointer_needed and it
6055 reserved via global_regs.
6057 2014-01-15 Teresa Johnson <tejohnson@google.com>
6059 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
6061 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6063 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
6064 and vmulosh rather than call gen_vec_widen_smult_*.
6065 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
6066 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
6067 (vec_widen_smult_even_v16qi): Likewise.
6068 (vec_widen_umult_even_v8hi): Likewise.
6069 (vec_widen_smult_even_v8hi): Likewise.
6070 (vec_widen_umult_odd_v16qi): Likewise.
6071 (vec_widen_smult_odd_v16qi): Likewise.
6072 (vec_widen_umult_odd_v8hi): Likewise.
6073 (vec_widen_smult_odd_v8hi): Likewise.
6074 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
6075 vmuloub rather than call gen_vec_widen_umult_*.
6076 (vec_widen_umult_lo_v16qi): Likewise.
6077 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
6078 vmulosb rather than call gen_vec_widen_smult_*.
6079 (vec_widen_smult_lo_v16qi): Likewise.
6080 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
6081 rather than call gen_vec_widen_umult_*.
6082 (vec_widen_umult_lo_v8hi): Likewise.
6083 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
6084 rather than call gen_vec_widen_smult_*.
6085 (vec_widen_smult_lo_v8hi): Likewise.
6087 2014-01-15 Jeff Law <law@redhat.com>
6089 PR tree-optimization/59747
6090 * ree.c (find_and_remove_re): Properly handle case where a second
6091 eliminated extension requires widening a copy created for elimination
6092 of a prior extension.
6093 (combine_set_extension): Ensure that the number of hard regs needed
6094 for a destination register does not change when we widen it.
6096 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
6098 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
6099 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
6100 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
6101 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
6102 (avr-*-rtems*): Likewise.
6103 (bfin*-rtems*): Likewise.
6104 (moxie-*-rtems*): Likewise.
6105 (h8300-*-rtems*): Likewise.
6106 (i[34567]86-*-rtems*): Likewise.
6107 (lm32-*-rtems*): Likewise.
6108 (m32r-*-rtems*): Likewise.
6109 (m68k-*-rtems*): Likewise.
6110 (microblaze*-*-rtems*): Likewise.
6111 (mips*-*-rtems*): Likewise.
6112 (powerpc-*-rtems*): Likewise.
6113 (sh-*-rtems*): Likewise.
6114 (sparc-*-rtems*): Likewise.
6115 (sparc64-*-rtems*): Likewise.
6116 (v850-*-rtems*): Likewise.
6117 (m32c-*-rtems*): Likewise.
6119 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
6121 PR rtl-optimization/59511
6122 * ira.c (ira_init_register_move_cost): Use memory costs for some
6123 cases of register move cost calculations.
6124 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
6125 instead of BB frequency.
6126 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
6127 * lra-assigns.c (find_hard_regno_for): Ditto.
6129 2014-01-15 Richard Biener <rguenther@suse.de>
6131 PR tree-optimization/59822
6132 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
6133 (vectorizable_load): Use it to hoist defs of uses of invariant
6134 loads out of the loop.
6136 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
6137 Kugan Vivekanandarajah <kuganv@linaro.org>
6140 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
6143 2014-01-15 Richard Biener <rguenther@suse.de>
6145 PR rtl-optimization/59802
6146 * lcm.c (compute_available): Use inverted postorder to seed
6147 the initial worklist.
6149 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6152 * config/s390/s390.c (s390_preferred_reload_class): Don't return
6153 ADDR_REGS for invalid symrefs in non-PIC code.
6155 2014-01-15 Jakub Jelinek <jakub@redhat.com>
6158 * builtins.c (determine_block_size): Initialize *probable_max_size
6159 even if len_rtx is CONST_INT.
6161 2014-01-14 Andrew Pinski <apinski@cavium.com>
6163 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
6164 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
6165 (cortexa53_tunings): Likewise.
6166 (aarch64_sched_issue_rate): New function.
6167 (TARGET_SCHED_ISSUE_RATE): Define.
6169 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6171 * ira-costs.c (find_costs_and_classes): Add missed
6172 ira_init_register_move_cost_if_necessary.
6174 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6177 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
6179 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
6182 * config/i386/i386.c (type_natural_mode): Add a bool parameter
6183 to indicate if type is used for function return value. Warn ABI
6184 change if the vector mode isn't available for function return value.
6185 (ix86_function_arg_advance): Pass false to type_natural_mode.
6186 (ix86_function_arg): Likewise.
6187 (ix86_gimplify_va_arg): Likewise.
6188 (function_arg_32): Don't warn ABI change.
6189 (ix86_function_value): Pass true to type_natural_mode.
6190 (ix86_return_in_memory): Likewise.
6191 (ix86_struct_value_rtx): Removed.
6192 (TARGET_STRUCT_VALUE_RTX): Likewise.
6194 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6196 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
6197 converting a conditional jump into a conditional return.
6199 2014-01-14 Richard Biener <rguenther@suse.de>
6201 PR tree-optimization/58921
6202 PR tree-optimization/59006
6203 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
6204 hoisting invariant stmts.
6205 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
6206 invariant loads on the preheader edge if possible.
6208 2014-01-14 Joey Ye <joey.ye@arm.com>
6210 * doc/plugin.texi (Building GCC plugins): Update to C++.
6212 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
6214 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
6215 (_mm_rcp28_round_ss): Ditto.
6216 (_mm_rsqrt28_round_sd): Ditto.
6217 (_mm_rsqrt28_round_ss): Ditto.
6218 (_mm_rcp28_sd): Ditto.
6219 (_mm_rcp28_ss): Ditto.
6220 (_mm_rsqrt28_sd): Ditto.
6221 (_mm_rsqrt28_ss): Ditto.
6222 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
6223 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
6224 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
6225 (IX86_BUILTIN_RCP28SD): Ditto.
6226 (IX86_BUILTIN_RCP28SS): Ditto.
6227 (IX86_BUILTIN_RSQRT28SD): Ditto.
6228 (IX86_BUILTIN_RSQRT28SS): Ditto.
6229 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
6230 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
6231 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
6232 (ix86_expand_special_args_builtin): Expand new FTYPE.
6233 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
6234 (srcp14<mode>): Make insn unary.
6235 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
6236 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
6237 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
6238 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
6239 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
6240 Fix rounding: make it SAE only.
6241 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6243 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6245 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
6246 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
6247 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
6248 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
6249 (round_saeonly_mask_scalar_operand4): Ditto.
6250 (round_saeonly_mask_scalar_op3): Ditto.
6251 (round_saeonly_mask_scalar_op4): Ditto.
6253 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6255 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6256 Implement -maltivec=be for vec_insert and vec_extract.
6258 2014-01-10 DJ Delorie <dj@redhat.com>
6260 * config/msp430/msp430.md (call_internal): Don't allow memory
6261 references with SP as the base register.
6262 (call_value_internal): Likewise.
6263 * config/msp430/constraints.md (Yc): New. For memory references
6264 that don't use SP as a base register.
6266 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
6267 "an integer without a # prefix"
6268 * config/msp430/msp430.md (epilogue_helper): Use it.
6270 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6273 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
6274 AVX512F gather builtins.
6275 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
6276 on gather decls with INTEGER_TYPE masktype.
6277 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
6278 directly into the builtin rather than hoisting it before loop.
6280 PR tree-optimization/59387
6281 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
6282 (scev_const_prop): If folded_casts and type has undefined overflow,
6283 use force_gimple_operand instead of force_gimple_operand_gsi and
6284 for each added stmt if it is assign with
6285 arith_code_with_undefined_signed_overflow, call
6286 rewrite_to_defined_overflow.
6287 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
6288 gimple-fold.h instead.
6289 (arith_code_with_undefined_signed_overflow,
6290 rewrite_to_defined_overflow): Moved to ...
6291 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
6292 rewrite_to_defined_overflow): ... here. No longer static.
6293 Include gimplify-me.h.
6294 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
6295 rewrite_to_defined_overflow): New prototypes.
6297 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6299 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
6301 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
6303 * builtins.c (get_object_alignment_2): Minor tweak.
6304 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
6306 2014-01-13 Christian Bruel <christian.bruel@st.com>
6308 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
6309 optimized non constant lengths.
6311 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6314 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
6315 load as __atomic_load_N if possible.
6317 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6319 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
6321 (rs6000_expand_builtin): Adjust call.
6323 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6326 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
6327 * config/rs6000/rs6000.c: Include target-globals.h.
6328 (rs6000_set_current_function): Instead of doing target_reinit
6329 unconditionally, use save_target_globals_default_opts and
6330 restore_target_globals.
6332 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
6334 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
6335 (rs6000_expand_builtin): Handle mffs and mtfsf.
6336 (rs6000_init_builtins): Define mffs and mtfsf.
6337 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
6338 (rs6000_mffs): New pattern.
6339 (rs6000_mtfsf): New pattern.
6341 2014-01-11 Bin Cheng <bin.cheng@arm.com>
6343 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
6344 Start narrowing with START. Apply candidate-use pair
6345 and check overall cost in narrowing.
6346 (iv_ca_prune): Pass new argument.
6348 2014-01-10 Jeff Law <law@redhat.com>
6351 * ree.c (combine_reaching_defs): Ensure the defining statement
6352 occurs before the extension when optimizing extensions with
6353 different source and destination hard registers.
6355 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6358 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
6359 vtables into the type inheritance graph.
6361 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6363 PR rtl-optimization/59754
6364 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
6365 modes in the REGNO != REGNO case.
6367 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6369 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
6371 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6373 PR tree-optimization/59745
6374 * tree-predcom.c (tree_predictive_commoning_loop): Call
6375 free_affine_expand_cache if giving up because components is NULL.
6377 * target-globals.c (save_target_globals): Allocate < 4KB structs using
6378 GC in payload of target_globals struct instead of allocating them on
6379 the heap and the larger structs separately using GC.
6380 * target-globals.h (struct target_globals): Make regs, hard_regs,
6381 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
6382 of GTY((skip)) and change type to void *.
6383 (reset_target_globals): Cast loads from those fields to corresponding
6386 2014-01-10 Steve Ellcey <sellcey@mips.com>
6389 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
6390 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
6391 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
6393 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
6396 * aarch64-modes.def (CC_Zmode): New flags mode.
6397 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
6398 represents an equality.
6399 (aarch64_get_condition_code): Handle CC_Zmode.
6400 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
6402 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6404 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
6405 extraction in good case.
6407 2014-01-10 Richard Biener <rguenther@suse.de>
6409 PR tree-optimization/59374
6410 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
6411 checking after SLP discovery. Mark stmts not participating
6412 in any SLP instance properly.
6414 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6416 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
6417 when handling a SET rtx.
6419 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6421 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
6422 (cortex-a57): Likewise.
6423 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
6425 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6427 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
6428 non-iwmmxt builtins.
6430 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6434 * ipa-devirt.c record_target_from_binfo): Take as argument
6435 stack of binfos and lookup matching one for virtual inheritance.
6436 (possible_polymorphic_call_targets_1): Update.
6438 2014-01-10 Huacai Chen <chenhc@lemote.com>
6440 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
6441 kernel strings for Loongson-2E/2F/3A.
6443 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6446 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
6447 is_gimple_call before calling gimple_call_internal_p.
6449 2014-01-09 Steve Ellcey <sellcey@mips.com>
6451 * Makefile.in (TREE_FLOW_H): Remove.
6452 (TREE_SSA_H): Add file names from tree-flow.h.
6453 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
6454 * tree.h: Remove tree-flow.h reference.
6455 * hash-table.h: Remove tree-flow.h reference.
6456 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
6457 reference with tree-ssa-loop.h.
6459 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6461 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6462 default element-order behavior for -maltivec.
6463 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6464 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6465 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6466 when targeting big endian, at least for now.
6467 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6469 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6472 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6473 var satisfies use_register_for_decl, just take into account type
6474 alignment, rather than decl alignment.
6476 PR tree-optimization/59622
6477 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6478 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6479 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6480 Don't devirtualize for inplace at all. For targets.length () == 1,
6481 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6483 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6485 * config/i386/i386.md (cpu): Remove the unused btver1.
6487 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6489 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6491 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6494 * tree-core.h (struct target_globals): New forward declaration.
6495 (struct tree_target_option): Add globals field.
6496 * tree.h (TREE_TARGET_GLOBALS): Define.
6497 (prepare_target_option_nodes_for_pch): New prototype.
6498 * target-globals.h (struct target_globals): Define even if
6500 * tree.c (prepare_target_option_node_for_pch,
6501 prepare_target_option_nodes_for_pch): New functions.
6502 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6503 * config/i386/i386.c: Include target-globals.h.
6504 (ix86_set_current_function): Instead of doing target_reinit
6505 unconditionally, use save_target_globals_default_opts and
6506 restore_target_globals.
6508 2014-01-09 Richard Biener <rguenther@suse.de>
6510 PR tree-optimization/59715
6511 * tree-cfg.h (split_critical_edges): Declare.
6512 * tree-cfg.c (split_critical_edges): Export.
6513 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6515 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6517 * cfgexpand.c (expand_stack_vars): Optionally disable
6518 asan stack protection.
6519 (expand_used_vars): Likewise.
6520 (partition_stack_vars): Likewise.
6521 * asan.c (asan_emit_stack_protection): Optionally disable
6522 after return stack usage.
6523 (instrument_derefs): Optionally disable memory access instrumentation.
6524 (instrument_builtin_call): Likewise.
6525 (instrument_strlen_call): Likewise.
6526 (asan_protect_global): Optionally disable global variables protection.
6527 * doc/invoke.texi: Added doc for new options.
6528 * params.def: Added new options.
6529 * params.h: Likewise.
6531 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6533 PR rtl-optimization/59724
6534 * ifcvt.c (cond_exec_process_if_block): Don't call
6535 flow_find_head_matching_sequence with 0 longest_match.
6536 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6537 non-active insns if !stop_after.
6538 (try_head_merge_bb): Revert 2014-01-07 changes.
6540 2014-01-08 Jeff Law <law@redhat.com>
6542 * ree.c (get_sub_rtx): New function, extracted from...
6543 (merge_def_and_ext): Here.
6544 (combine_reaching_defs): Use get_sub_rtx.
6546 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6548 * cgraph.h (varpool_variable_node): Do not choke on null node.
6550 2014-01-08 Catherine Moore <clm@codesourcery.com>
6552 * config/mips/mips.md (simple_return): Attempt to use JRC
6554 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6556 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6558 PR rtl-optimization/59137
6559 * reorg.c (steal_delay_list_from_target): Call update_block for
6561 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6563 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6565 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6566 two duplicate entries.
6568 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6571 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6573 * config/mips/mips.c (mips_truncated_op_cost): New function.
6574 (mips_rtx_costs): Adjust test for BADDU.
6575 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6577 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6579 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6580 (*baddu_si): ...this new pattern.
6582 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6585 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6587 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6590 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6592 (expand_expr, expand_normal): Adjust.
6593 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6594 inner_reference_p. Use inner_reference_p to expand inner references.
6595 (store_expr): Adjust.
6596 * cfgexpand.c (expand_call_stmt): Adjust.
6598 2014-01-08 Rong Xu <xur@google.com>
6600 * gcov-io.c (gcov_var): Move from gcov-io.h.
6601 (gcov_position): Ditto.
6602 (gcov_is_error): Ditto.
6603 (gcov_rewrite): Ditto.
6604 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6605 only part to libgcc/libgcov.h.
6607 2014-01-08 Marek Polacek <polacek@redhat.com>
6610 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6612 2014-01-08 Marek Polacek <polacek@redhat.com>
6615 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6617 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6619 PR rtl-optimization/59649
6620 * stor-layout.c (get_mode_bounds): For BImode return
6621 0 and STORE_FLAG_VALUE.
6623 2014-01-08 Richard Biener <rguenther@suse.de>
6626 * gimple.h (is_gimple_builtin_call): Remove.
6627 (gimple_builtin_call_types_compatible_p): New.
6628 (gimple_call_builtin_p): New overload.
6629 * gimple.c (is_gimple_builtin_call): Remove.
6630 (validate_call): Rename to ...
6631 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6633 (validate_type): New static function.
6634 (gimple_call_builtin_p): New overload and adjust.
6635 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6636 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6637 (gimple_fold_stmt_to_constant_1): Likewise.
6638 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6640 2014-01-08 Richard Biener <rguenther@suse.de>
6643 * gimplify.c (gimplify_expr): Gimplify register-register type
6644 VIEW_CONVERT_EXPRs to separate stmts.
6646 2014-01-07 Jeff Law <law@redhat.com>
6649 * ree.c (combine_set_extension): Handle case where source
6650 and destination registers in an extension insn are different.
6651 (combine_reaching_defs): Allow source and destination registers
6652 in extension to be different under limited circumstances.
6653 (add_removable_extension): Remove restriction that the
6654 source and destination registers in the extension are the same.
6655 (find_and_remove_re): Emit a copy from the extension's
6656 destination to its source after the defining insn if
6657 the source and destination registers are different.
6660 * ifcvt.c (merge_if_block): If we are merging a block with more than
6661 one successor with a block with no successors, remove any BARRIER
6662 after the second block.
6664 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6666 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6668 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6671 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6672 for 14-bit register offsets when INT14_OK_STRICT is false.
6674 2014-01-07 Roland Stigge <stigge@antcom.de>
6675 Michael Meissner <meissner@linux.vnet.ibm.com>
6678 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6679 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6681 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6683 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6686 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6688 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6689 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6690 rtx is const0_rtx or not.
6692 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6695 * target-globals.c (save_target_globals): Remove this_fn_optab
6697 * toplev.c: Include optabs.h.
6698 (target_reinit): Temporarily restore the global options if another
6699 set of options are in force.
6701 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6703 PR rtl-optimization/58668
6704 * cfgcleanup.c (flow_find_cross_jump): Don't count
6705 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6706 to determine what is counted.
6707 (flow_find_head_matching_sequence): Use active_insn_p to determine
6709 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6711 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6712 determine what is counted.
6714 PR tree-optimization/59643
6715 * tree-predcom.c (split_data_refs_to_components): If one dr is
6716 read and one write, determine_offset fails and the write isn't
6717 in the bad component, just put the read into the bad component.
6719 2014-01-07 Mike Stump <mikestump@comcast.net>
6720 Jakub Jelinek <jakub@redhat.com>
6723 * tree-core.h (struct tree_optimization_option): Change optabs
6724 type from unsigned char * to void *.
6725 * optabs.c (init_tree_optimization_optabs): Adjust
6726 TREE_OPTIMIZATION_OPTABS initialization.
6728 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6731 * config/i386/i386.h (struct machine_function): Add
6732 no_drap_save_restore field.
6733 * config/i386/i386.c (ix86_save_reg): Use
6734 !cfun->machine->no_drap_save_restore instead of
6735 crtl->stack_realign_needed.
6736 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6737 this function clears frame_pointer_needed. Set
6738 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6739 and DRAP reg is needed.
6741 2014-01-06 Marek Polacek <polacek@redhat.com>
6744 * doc/implement-c.texi: Mention that other integer types are
6745 permitted as bit-field types in strictly conforming mode.
6747 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6749 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6752 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6754 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6756 2014-01-06 Martin Jambor <mjambor@suse.cz>
6759 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6761 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6763 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6767 * var-tracking.c (add_stores): Preserve the value of the source even if
6768 we don't record the store.
6770 2014-01-06 Terry Guo <terry.guo@arm.com>
6772 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6774 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6777 * config/darwin.c (darwin_function_section): Adjust return values to
6778 correspond to optimisation changes made in r206070.
6780 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6782 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6783 from prefetch_block tune setting.
6784 (nocona_cost): Correct size of prefetch block to 64.
6786 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6788 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6789 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6790 used to save the static chain register in the computation of the offset
6791 from which the FP registers need to be restored.
6793 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6795 PR tree-optimization/59519
6796 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6797 ICE if get_current_def (current_new_name) is already non-NULL, as long
6798 as it is a phi result of some other phi in *new_exit_bb that has
6801 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6802 or vmovdqu* for misaligned_operand.
6803 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6804 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6805 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6806 aligned_mem for AVX512F masked aligned load and store builtins and for
6809 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6811 PR tree-optimization/59651
6812 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6813 Address range for negative step should be added by TYPE_SIZE_UNIT.
6815 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6817 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6818 overlapping registers also for registers other than the stack pointer.
6820 2014-01-03 Marek Polacek <polacek@redhat.com>
6823 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6826 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6829 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6832 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6833 (push splitter): Use <P:MODE_SIZE> instead of
6834 GET_MODE_SIZE (<P:MODE>mode).
6835 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6836 (mov -1, reg peephole2): Likewise.
6837 * config/i386/sse.md (*mov<mode>_internal,
6838 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6839 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6840 *<code><mode>3, *andnot<mode>3<mask_name>,
6841 <mask_codefor><code><mode>3<mask_name>): Likewise.
6842 * config/i386/subst.md (mask_mode512bit_condition,
6843 sd_mask_mode512bit_condition): Likewise.
6845 2014-01-02 Xinliang David Li <davidxl@google.com>
6847 PR tree-optimization/59303
6848 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6849 (dump_predicates): Better output format.
6850 (pred_equal_p): New function.
6851 (is_neq_relop_p): Ditto.
6852 (is_neq_zero_form_p): Ditto.
6853 (pred_expr_equal_p): Ditto.
6854 (pred_neg_p): Ditto.
6855 (simplify_pred): Ditto.
6856 (simplify_preds_2): Ditto.
6857 (simplify_preds_3): Ditto.
6858 (simplify_preds_4): Ditto.
6859 (simplify_preds): Ditto.
6861 (push_to_worklist): Ditto.
6862 (get_pred_info_from_cmp): Ditto.
6863 (is_degenerated_phi): Ditto.
6864 (normalize_one_pred_1): Ditto.
6865 (normalize_one_pred): Ditto.
6866 (normalize_one_pred_chain): Ditto.
6867 (normalize_preds): Ditto.
6868 (normalize_cond_1): Remove function.
6869 (normalize_cond): Ditto.
6870 (is_gcond_subset_of): Ditto.
6871 (is_subset_of_any): Ditto.
6872 (is_or_set_subset_of): Ditto.
6873 (is_and_set_subset_of): Ditto.
6874 (is_norm_cond_subset_of): Ditto.
6875 (pred_chain_length_cmp): Ditto.
6876 (convert_control_dep_chain_into_preds): Type change.
6877 (find_predicates): Ditto.
6878 (find_def_preds): Ditto.
6879 (destroy_predicates_vecs): Ditto.
6880 (find_matching_predicates_in_rest_chains): Ditto.
6881 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6882 (is_pred_expr_subset): Ditto.
6883 (is_pred_chain_subset_of): Ditto.
6884 (is_included_in): Ditto.
6885 (is_superset_of): Ditto.
6887 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6889 Update copyright years.
6891 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6893 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6894 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6895 config/arc/arc.md, config/arc/arc.opt,
6896 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6897 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6898 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6899 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6900 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6901 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6902 vtable-verify.c, vtable-verify.h: Use the standard form for the
6905 2014-01-02 Tobias Burnus <burnus@net-b.de>
6907 * gcc.c (process_command): Update copyright notice dates.
6908 * gcov-dump.c: Ditto.
6910 * doc/cpp.texi: Bump @copying's copyright year.
6911 * doc/cppinternals.texi: Ditto.
6912 * doc/gcc.texi: Ditto.
6913 * doc/gccint.texi: Ditto.
6914 * doc/gcov.texi: Ditto.
6915 * doc/install.texi: Ditto.
6916 * doc/invoke.texi: Ditto.
6918 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6920 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6922 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6924 * config/i386/sse.md (*mov<mode>_internal): Guard
6925 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6927 PR rtl-optimization/59647
6928 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6929 new_rtx into UNSIGNED_FLOAT rtxes.
6931 Copyright (C) 2014 Free Software Foundation, Inc.
6933 Copying and distribution of this file, with or without modification,
6934 are permitted in any medium without royalty provided the copyright
6935 notice and this notice are preserved.