1 2014-04-22 Marcus Shawcroft <marcus.shawcroft@arm.com>
3 * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
6 2014-04-22 Richard Sandiford <rdsandiford@googlemail.com>
8 * machmode.h (bitwise_mode_for_mode): Declare.
9 * stor-layout.h (bitwise_type_for_mode): Likewise.
10 * stor-layout.c (bitwise_mode_for_mode): New function.
11 (bitwise_type_for_mode): Likewise.
12 * builtins.c (fold_builtin_memory_op): Use it instead of
13 int_mode_for_mode and build_nonstandard_integer_type.
15 2014-04-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17 * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
18 (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
19 (*-*-solaris2*): Simplify.
20 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
21 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
22 *-*-solaris2.9* handling.
24 * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
26 (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
27 (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
29 (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
30 * configure: Regenerate.
32 * config/i386/sol2-9.h: Remove.
34 * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
35 (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
36 Remove Solaris 9 references.
38 2014-04-22 Vidya Praveen <vidyapraveen@arm.com>
40 * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
41 (floatuns<GPI:mode><GPF:mode>2): Remove.
42 (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
43 and floatuns conversions.
44 (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
45 and floatuns conversions.
46 * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
47 (w1,w2): New mode attributes for inequal width conversions.
49 2014-04-22 Renlin Li <Renlin.Li@arm.com>
51 * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
52 the output asm format.
54 2014-04-22 James Greenhalgh <james.greenhalgh@arm.com>
56 * config/aarch64/aarch64-simd.md
57 (aarch64_cm<optab>di): Always split.
58 (*aarch64_cm<optab>di): New.
59 (aarch64_cmtstdi): Always split.
60 (*aarch64_cmtstdi): New.
62 2014-04-22 Jakub Jelinek <jakub@redhat.com>
64 PR tree-optimization/60823
65 * omp-low.c (ipa_simd_modify_function_body): Go through
66 all SSA_NAMEs and for those refering to vector arguments
67 which are going to be replaced adjust SSA_NAME_VAR and,
68 if it is a default definition, change it into a non-default
69 definition assigned at the beginning of function from new_decl.
70 (ipa_simd_modify_stmt_ops): Rewritten.
71 * tree-dfa.c (set_ssa_default_def): When removing default def,
72 check for NULL loc instead of NULL *loc.
74 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
76 * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
77 restrictions on core registers for DImode values in Thumb2.
79 2014-04-22 Ian Bolton <ian.bolton@arm.com>
81 * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
82 * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
84 2014-04-22 Ian Bolton <ian.bolton@arm.com>
86 * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
87 (*iordi_notzesidi_di): Likewise.
88 (*iordi_notsesidi_di): Likewise.
90 2014-04-22 Ian Bolton <ian.bolton@arm.com>
92 * config/arm/arm-protos.h (tune_params): New struct members.
93 * config/arm/arm.c: Initialise tune_params per processor.
94 (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
95 for speed, based on new tune_params.
97 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
99 * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
101 * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
103 * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
105 * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
106 * config/aarch64/arm_neon.h (vrnd_f64): Added.
107 (vrnda_f64): Likewise.
108 (vrndi_f64): Likewise.
109 (vrndm_f64): Likewise.
110 (vrndn_f64): Likewise.
111 (vrndp_f64): Likewise.
112 (vrndx_f64): Likewise.
114 2014-04-22 Zhenqiang Chen <zhenqiang.chen@linaro.org>
116 * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
117 GET_MODE_SIZE argument is enum machine_mode.
119 2014-04-22 Jakub Jelinek <jakub@redhat.com>
122 * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
123 value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
125 2014-04-22 Lin Zuojian <manjian2006@gmail.com>
128 * asan.c (asan_emit_stack_protection): Force the base to align to
129 appropriate bits if STRICT_ALIGNMENT. Set shadow_mem align to
130 appropriate bits if STRICT_ALIGNMENT.
131 * cfgexpand.c (expand_stack_vars): Set base_align appropriately
133 (expand_used_vars): Leave a space in the stack frame for alignment
136 2014-04-21 David Malcolm <dmalcolm@redhat.com>
138 * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
140 (gimple_store_p): Likewise.
141 (gimple_assign_load_p): Likewise.
142 (gimple_assign_cast_p): Likewise.
143 (gimple_clobber_p): Likewise.
145 * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
146 rather than a gimple.
147 (gimple_assign_cast_p): Likewise.
149 2014-04-21 Michael Meissner <meissner@linux.vnet.ibm.com>
152 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
153 If mode is DDmode and TARGET_E500_DOUBLE allow move.
155 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
156 more debug information for E500 if -mdebug=reg.
158 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
161 * config/i386/i386.c (ix86_expand_builtin)
162 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
163 register for target RTX.
164 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
166 2014-04-18 Cong Hou <congh@google.com>
168 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
169 the widen-mult pattern by handling two operands with different sizes,
170 and operands whose size is smaller than half of the result type.
172 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
174 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
175 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
176 (do_estimate_edge_time): Compute it.
177 * ipa-inline.c (want_inline_small_function_p): Bypass
178 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
180 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
182 * ipa-inline.c (spec_rem): New static variable.
183 (dump_overall_stats): New function.
184 (dump_inline_stats): New function.
186 2014-04-18 Richard Henderson <rth@redhat.com>
188 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
189 to GET_MODE_SIZE, not a reg_class_t.
191 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
193 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
194 (vsx_xxmrglw_<mode>): Likewise.
196 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
199 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
200 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
201 (rs6000_init_hard_regno_mode_ok): Likewise.
203 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
205 * ipa-inline.c (inline_small_functions): Account only non-cold
207 * doc/invoke.texi (inline-unit-growth): Update documentation.
209 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
211 * config/rs6000/rs6000.md (addti3, subti3): New.
213 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
216 * config/i386/i386.c (ix86_expand_clear): Remove outdated
217 comment. Check optimize_insn_for_size_p instead of
218 optimize_insn_for_speed_p.
220 2014-04-17 Martin Jambor <mjambor@suse.cz>
222 * gimple-iterator.c (gsi_start_edge): New function.
223 * gimple-iterator.h (gsi_start_edge): Declare.
224 * tree-sra.c (single_non_eh_succ): New function.
225 (disqualify_ops_if_throwing_stmt): Renamed to
226 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
227 having one non-EH successor BB.
228 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
229 generate loads into replacements.
230 (sra_modify_assign): Likewise and and also use the simple path for
232 (sra_modify_function_body): Commit statements on edges.
234 2014-04-17 Richard Biener <rguenther@suse.de>
237 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
238 comparison results and add clarifying comment.
240 2014-04-17 Jakub Jelinek <jakub@redhat.com>
242 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
243 (blank_mode): Initialize it.
244 (emit_mode_size_inline, emit_mode_nunits_inline,
245 emit_mode_inner_inline): New functions.
246 (emit_insn_modes_h): Call them and surround their output with
247 #if GCC_VERSION >= 4001 ... #endif.
248 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
249 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
250 mode_* arrays if the argument is __builtin_constant_p.
251 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
252 is enum machine_mode.
254 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
256 * passes.c (opt_pass::execute): Adjust.
257 (pass_manager::execute_pass_mode_switching): Likewise.
258 (early_local_passes::execute): Likewise.
259 (execute_one_pass): Pass cfun to the pass's execute method.
260 * tree-pass.h (opt_pass::execute): Add function * argument.
261 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
262 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
263 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
264 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
265 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
266 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
267 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
268 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
269 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
270 ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
271 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
272 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
273 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
274 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
275 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
276 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
277 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
278 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
279 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
280 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
281 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
282 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
283 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
284 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
285 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
286 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
287 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
288 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
289 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
292 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
294 * passes.c (opt_pass::gate): Take function * argument.
295 (gate_all_early_local_passes): Merge into
296 (early_local_passes::gate): this.
297 (gate_all_early_optimizations): Merge into
298 (all_early_optimizations::gate): this.
299 (gate_all_optimizations): Mege into
300 (all_optimizations::gate): this.
301 (gate_all_optimizations_g): Merge into
302 (all_optimizations_g::gate): this.
303 (gate_rest_of_compilation): Mege into
304 (rest_of_compilation::gate): this.
305 (gate_postreload): Merge into
306 (postreload::gate): this.
307 (dump_one_pass): Pass cfun to the pass's gate method.
308 (execute_ipa_summary_passes): Likewise.
309 (execute_one_pass): Likewise.
310 (ipa_write_summaries_2): Likewise.
311 (ipa_write_optimization_summaries_1): Likewise.
312 (ipa_read_summaries_1): Likewise.
313 (ipa_read_optimization_summaries_1): Likewise.
314 (execute_ipa_stmt_fixups): Likewise.
315 * tree-pass.h (opt_pass::gate): Add function * argument.
316 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
317 combine-stack-adj.c, combine.c, compare-elim.c,
318 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
319 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
320 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
321 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
322 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
323 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
324 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
325 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
326 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
327 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
328 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
329 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
330 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
331 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
332 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
333 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
334 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
335 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
336 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
337 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
338 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
339 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
340 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
341 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
342 var-tracking.c, vtable-verify.c, web.c: Adjust.
344 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
346 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
347 * configure: Regenerate.
349 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
351 * passes.c (dump_one_pass): don't check pass->has_gate.
352 (execute_ipa_summary_passes): Likewise.
353 (execute_one_pass): Likewise.
354 (ipa_write_summaries_2): Likewise.
355 (ipa_write_optimization_summaries_1): Likewise.
356 (ipa_read_optimization_summaries_1): Likewise.
357 (execute_ipa_stmt_fixups): Likewise.
358 * tree-pass.h (pass_data::has_gate): Remove.
359 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
360 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
361 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
362 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
363 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
364 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
365 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
366 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
367 gimple-low.c, gimple-ssa-isolate-paths.c,
368 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
369 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
370 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
371 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
372 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
373 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
374 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
375 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
376 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
377 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
378 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
379 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
380 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
381 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
382 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
383 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
384 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
385 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
386 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
387 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
388 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
389 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
392 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
394 * pass_manager.h (pass_manager::register_dump_files_1): Remove
396 * passes.c (pass_manager::register_dump_files_1): Merge into
397 (pass_manager::register_dump_files): this, and remove its handling of
398 properties since the pass always has the properties anyway.
399 (pass_manager::pass_manager): Adjust.
401 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
403 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
404 * passes.c (pass_manager::register_dump_files_1): Remove dead code
405 dealing with properties.
406 (pass_manager::register_dump_files): Adjust.
408 2014-03-20 Mark Wielaard <mjw@redhat.com>
410 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
411 then represent the bound as normal constant value.
413 2014-04-17 Jakub Jelinek <jakub@redhat.com>
416 Forward port from 4.8 branch
417 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
419 * config/i386/bmiintrin.h (_blsi_u32): New.
423 (_blsmsk_u32): Ditto.
424 (_blsmsk_u64): Ditto.
428 2014-04-17 Kito Cheng <kito@0xlab.org>
430 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
432 2014-04-17 Richard Biener <rguenther@suse.de>
435 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
436 boolean results for comparisons.
438 2014-04-17 Richard Biener <rguenther@suse.de>
440 PR tree-optimization/60836
441 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
442 initial PHI args to be gimple values.
444 2014-04-17 Richard Biener <rguenther@suse.de>
446 PR tree-optimization/60841
447 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
448 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
449 of stmts to SLP build.
450 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
451 (vect_analyze_slp): Likewise.
452 (vect_analyze_slp_instance): Likewise.
453 (vect_build_slp_tree): Limit overall SLP tree growth.
454 * tree-vectorizer.h (vect_analyze_data_refs,
455 vect_analyze_slp): Adjust prototypes.
457 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
459 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
462 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
464 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
465 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
466 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
467 for TARGET_SLOW_PSHUFB
469 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
471 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
472 * config/i386/i386.c (intel_cost): Ditto.
474 2014-04-17 Joey Ye <joey.ye@arm.com>
476 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
478 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
480 * opts.c (common_handle_option): Disable -fipa-reference coorectly
483 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
485 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
486 (type_all_derivations_known_p): New predicate.
487 (type_all_ctors_visible_p): New predicate.
488 (type_possibly_instantiated_p): New predicate.
489 (get_odr_type): Compute all_derivations_known.
490 (dump_odr_type): Dump the flag.
491 (maybe_record_type): Cleanup.
492 (record_target_from_binfo): Add bases_to_consider array;
493 record bases for types w/o instances and skip CXX destructor.
494 (possible_polymorphic_call_targets_1): Add bases_to_consider
495 and consider_construction parameters; check if type may have instance.
496 (get_polymorphic_call_info): Set maybe_in_construction to true
497 when we know nothing.
498 (record_targets_from_bases): Skip CXX destructors; they are
499 never called for types in construction.
500 (possible_polymorphic_call_targets): Do not record target when
501 type may not have instance.
503 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
506 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
507 external aliases alive, too.
509 2014-04-16 Andrew Pinski <apinski@cavium.com>
511 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
514 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
516 * final.c (compute_alignments): Do not apply loop alignment to a block
517 falling through to the exit.
519 2014-04-16 Catherine Moore <clm@codesourcery.com>
521 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
522 Adjust constraints for microMIPS store patterns.
524 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
526 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
528 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
530 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
531 (append_use): Run at -O0.
532 (append_vdef): Likewise.
533 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
534 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
536 2014-04-16 Jakub Jelinek <jakub@redhat.com>
538 PR tree-optimization/60844
539 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
540 (propagate_op_to_single_use, remove_visited_stmt_chain,
541 linearize_expr, repropagate_negates, reassociate_bb): Use it
542 instead of gsi_remove.
544 2014-04-16 Martin Jambor <mjambor@suse.cz>
546 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
547 ipa_transforms_to_apply.
548 (cgraph_function_versioning): Assert that old_node has empty
549 ipa_transforms_to_apply.
550 * trans-mem.c (ipa_tm_create_version): Likewise.
551 * tree-inline.c (tree_function_versioning): Do not duplicate
552 ipa_transforms_to_apply.
554 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
557 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
559 Pass necessary as flags on 64-bit Solaris/x86.
560 Use lowercase relocs for x86_64-*-*.
561 * configure: Regenerate.
563 2014-04-15 Jan Hubicka <jh@suse.cz>
565 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
566 (maybe_record_node, likely_target_p): Use it.
568 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
571 Revert following patch
573 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
576 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
577 software floating point or no floating point registers, do not
578 allow any type in the FPRs. Eliminate a test for SPE SIMD types
579 in GPRs that occurs after we tested for GPRs that would never be
582 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
583 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
584 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
585 specifically allow DDmode, since that does not use the SPE SIMD
588 2014-03-21 Mark Wielaard <mjw@redhat.com>
590 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
591 as unsigned or int depending on type and value used.
593 2014-04-15 Richard Biener <rguenther@suse.de>
595 PR rtl-optimization/56965
596 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
597 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
599 * alias.c (true_dependence_1): Do not call
600 nonoverlapping_component_refs_p.
601 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
602 nonoverlapping_component_refs_p.
603 (indirect_refs_may_alias_p): Likewise.
605 2014-04-15 Teresa Johnson <tejohnson@google.com>
607 * cfg.c (dump_bb_info): Fix flags check.
608 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
610 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
612 PR rtl-optimization/60663
613 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
616 2014-04-15 Richard Biener <rguenther@suse.de>
618 * lto-streamer.h (LTO_major_version): Bump to 4.
620 2014-04-15 Richard Biener <rguenther@suse.de>
622 * common.opt (lto_partition_model): New enum.
623 (flto-partition=): Merge separate options with a single with argument,
624 add -flto-partition=one support.
625 * flag-types.h (enum lto_partition_model): Declare.
626 * opts.c (finish_options): Remove duplicate -flto-partition=
628 * lto-wrapper.c (run_gcc): Adjust.
630 2014-04-15 Richard Biener <rguenther@suse.de>
632 * alias.c (ncr_compar): New function.
633 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
635 2014-04-15 Richard Biener <rguenther@suse.de>
637 * alias.c (record_component_aliases): Do not walk BINFOs.
639 2014-04-15 Richard Biener <rguenther@suse.de>
641 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
642 Add struct function argument and adjust.
643 (find_func_aliases_for_call): Likewise.
644 (find_func_aliases): Likewise.
645 (find_func_clobbers): Likewise.
646 (intra_create_variable_infos): Likewise.
647 (compute_points_to_sets): Likewise.
648 (ipa_pta_execute): Adjust. Do not push/pop cfun.
650 2014-04-15 Richard Biener <rguenther@suse.de>
652 * tree.c (iterative_hash_expr): Use enum tree_code_class
653 to store TREE_CODE_CLASS.
654 (tree_block): Likewise.
655 (tree_set_block): Likewise.
656 * tree.h (fold_build_pointer_plus_loc): Use
657 convert_to_ptrofftype_loc.
659 2014-04-15 Jakub Jelinek <jakub@redhat.com>
662 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
665 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
667 * cfgloop.h (struct loop): Move force_vectorize down.
668 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
669 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
670 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
671 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
672 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
673 * tree-core.h (enum annot_expr_kind): Add new kind values.
674 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
675 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
677 * tree.def (ANNOTATE_EXPR): Tweak comment.
679 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
681 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
684 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
686 * tree.h (TYPE_IDENTIFIER): Declare.
687 * tree.c (subrange_type_for_debug_p): Use it.
688 * godump.c (go_format_type): Likewise.
689 * dwarf2out.c (is_cxx_auto, modified_type_die,
690 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
691 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
693 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
696 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
698 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
700 * config/i386/i386.c (examine_argument): Return bool. Return true if
701 parameter should be passed in memory.
702 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
703 (construct_container): Update calls to examine_argument.
704 (function_arg_advance_64): Ditto.
705 (return_in_memory_32): Merge with ix86_return_in_memory.
706 (return_in_memory_64): Ditto.
707 (return_in_memory_ms_64): Ditto.
709 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
711 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
712 * coverage.c (coverage_compute_profile_id): Handle externally visible
715 2014-04-14 Martin Jambor <mjambor@suse.cz>
717 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
718 DECL_DISREGARD_INLINE_LIMITS functions.
720 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
723 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
725 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
728 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
729 optimize_insn_for_speed_p instead of
730 optimize_function_for_speed_p.
732 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
734 * doc/invoke.texi (free): Document AArch64.
736 2014-04-14 Richard Biener <rguenther@suse.de>
738 PR tree-optimization/60042
739 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
740 (insert_into_preds_of_block): Do not prevent PHI insertion
741 for REFERENCE exprs here ...
742 (eliminate_dom_walker::before_dom_children): ... but prevent
743 their use here under similar conditions when applied to the
744 IL after PRE optimizations.
746 2014-04-14 Richard Biener <rguenther@suse.de>
748 * passes.def: Move early points-to after early SRA.
750 2014-04-14 Richard Biener <rguenther@suse.de>
752 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
753 check for which sign-changes we allow when forwarding
754 a converted value into a switch.
756 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
758 * stor-layout.c (place_field): Finalize non-constant offset for the
761 2014-04-14 Richard Biener <rguenther@suse.de>
763 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
765 (expand_switch_using_bit_tests_p): Likewise.
766 (process_switch): Compute and pass on speed_p based on the
768 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
769 optimize_bb_for_speed_p.
771 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
773 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
774 * function.h (struct function): Rename has_force_vect_loops into
775 has_force_vectorize_loops.
776 * lto-streamer-in.c (input_cfg): Adjust for renaming.
777 (input_struct_function_base): Likewise.
778 * lto-streamer-out.c (output_cfg): Likewise.
779 (output_struct_function_base): Likewise.
780 * omp-low.c (expand_omp_simd): Likewise.
781 * tree-cfg.c (move_sese_region_to_fn): Likewise.
782 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
783 (version_loop_for_if_conversion): Likewise.
784 (tree_if_conversion): Likewise.
785 (main_tree_if_conversion): Likewise.
786 (gate_tree_if_conversion): Likewise.
787 * tree-inline.c (copy_loops): Likewise.
788 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
789 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
790 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
791 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
792 * tree-vectorizer.c (vectorize_loops): Likewise.
793 * tree-vectorizer.h (unlimited_cost_model): Likewise.
795 2014-04-14 Richard Biener <rguenther@suse.de>
798 * lto-streamer-out.c (wrap_refs): New function.
799 (lto_output): Wrap symbol references in global initializes in
800 type-preserving MEM_REFs.
802 2014-04-14 Christian Bruel <christian.bruel@st.com>
804 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
806 2014-04-14 Christian Bruel <christian.bruel@st.com>
808 * config/sh/sh.md (setmemqi): New expand pattern.
809 * config/sh/sh.h (CLEAR_RATIO): Define.
810 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
811 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
813 2014-04-14 Richard Biener <rguenther@suse.de>
816 * fold-const.c (negate_expr_p): Don't negate directional rounding
818 (fold_negate_expr): Likewise.
820 2014-04-14 Richard Biener <rguenther@suse.de>
822 PR tree-optimization/59817
823 PR tree-optimization/60453
824 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
825 recursion to catch all CHRECs in the scalar evolution and restrict
826 the predicate for the remains appropriately.
828 2014-04-12 Catherine Moore <clm@codesourcery.com>
830 * config/mips/constraints.md: Add new register constraint "kb".
831 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
832 (*movhi_internal): Likewise.
833 (*movqi_internal): Likewise.
834 * config/mips/mips.h (M16_STORE_REGS): New register class.
835 (REG_CLASS_NAMES): Add M16_STORE_REGS.
836 (REG_CLASS_CONTENTS): Likewise.
837 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
839 2014-04-11 Tobias Burnus <burnus@net-b.de>
842 * doc/invoke.texi (-Wformat-signedness): Document it.
843 (Wformat=2): Mention that this enables -Wformat-signedness.
845 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
847 * common/config/epiphany/epiphany-common.c
848 (epiphany_option_optimization_table): Enable section anchors by
849 default at -O1 or higher.
850 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
851 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
852 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
853 carries no extra cost.
854 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
855 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
856 * config/epiphany/predicates.md (memclob_operand): New predicate.
857 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
858 Use memclob_operand predicate and X constraint for operand 3.
860 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
862 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
863 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
866 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
868 PR rtl-optimization/60651
869 * mode-switching.c (optimize_mode_switching): Make sure to emit
870 sets of a lower numbered entity before sets of a higher numbered
871 entity to a mode of the same or lower priority.
872 When creating a seginfo for a basic block that starts with a code
873 label, move the insertion point past the code label.
874 (new_seginfo): Document and enforce requirement that
875 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
876 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
877 * doc/tm.texi: Regenerate.
879 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
882 * config/arc/arc.c (arc_save_restore): Fix assert typo.
884 2013-04-11 Jakub Jelinek <jakub@redhat.com>
886 * BASE-VER: Set to 4.10.0.
888 2014-04-11 Tobias Burnus <burnus@net-b.de>
891 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
892 * doc/gcc.texi (Service): Update description in the @menu
893 * doc/invoke.texi (Option Summary): Remove misplaced and
896 2014-04-11 Steve Ellcey <sellcey@mips.com>
897 Jakub Jelinek <jakub@redhat.com>
900 * expr.c (convert_move): Use emit_store_flag_force instead of
901 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
904 2014-04-11 Richard Biener <rguenther@suse.de>
907 * varasm.c (assemble_alias): Avoid endless error reporting
908 recursion by setting TREE_ASM_WRITTEN.
910 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
912 * config/s390/s390.md: Add a splitter for NOT rtx.
914 2014-04-11 Jakub Jelinek <jakub@redhat.com>
916 PR rtl-optimization/60663
917 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
919 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
920 Jakub Jelinek <jakub@redhat.com>
923 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
924 flag from decl_node to node.
926 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
929 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
930 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
931 ameliorating the cases where it can be.
933 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
936 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
938 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
939 (loadsync_<mode>): Change mode.
940 (load_quadpti, store_quadpti): New.
941 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
942 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
943 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
945 2014-04-09 Cong Hou <congh@google.com>
948 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
951 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
953 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
954 instead of vnor to exploit possible fusion opportunity in the
956 (altivec_expand_vec_perm_const_le): Likewise.
958 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
960 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
961 (loadsync_<mode>): Change mode.
962 (load_quadpti, store_quadpti): New.
963 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
964 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
966 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
969 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
970 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
971 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
973 2014-04-08 Richard Biener <rguenther@suse.de>
976 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
977 a 64bit widest int print double-int similar to on HWI64 hosts.
979 2014-04-08 Richard Biener <rguenther@suse.de>
981 PR tree-optimization/60785
982 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
983 default defs properly.
985 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
987 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
990 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
992 * ipa-devirt.c (maybe_record_node): When node is not recorded,
993 set completep to false rather than true.
995 2014-04-07 Douglas B Rupp <rupp@adacore.com>
998 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
999 ARM_TARGET2_DWARF_FORMAT.
1001 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
1004 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
1005 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
1008 2014-04-07 Richard Biener <rguenther@suse.de>
1010 PR tree-optimization/60766
1011 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
1012 (may_eliminate_iv): Convert cand_value_at result to desired type.
1014 2014-04-07 Jason Merrill <jason@redhat.com>
1017 * common.opt (-fno-gnu-unique): Add.
1018 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
1020 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1022 * haifa-sched.c: Fix outdated function reference and minor
1023 grammar errors in introductory comment.
1025 2014-04-07 Richard Biener <rguenther@suse.de>
1028 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
1030 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
1032 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
1035 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
1036 size accounting for thunks.
1037 (pa_asm_output_mi_thunk): Use final_start_function() and
1038 final_end_function() to output function start and end directives.
1040 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
1042 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
1043 device specific ISA/ feature information. Remove short_sp and
1044 errata_skip ds. Add avr_device_specific_features enum to have device
1046 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
1047 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
1048 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
1049 updated device specific info.
1050 * config/avr/avr-mcus.def: Merge device specific details to
1051 dev_attribute field.
1052 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
1054 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
1055 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
1056 assembler if RMW isa supported by current device.
1057 * config/avr/genmultilib.awk: Update as device info structure changed.
1058 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
1060 2014-04-04 Cong Hou <congh@google.com>
1062 PR tree-optimization/60656
1063 * tree-vect-stmts.c (supportable_widening_operation):
1064 Fix a bug that elements in a vector with vect_used_by_reduction
1065 property are incorrectly reordered when the operation on it is not
1066 consistant with the one in reduction operation.
1068 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
1070 PR rtl-optimization/60155
1071 * gcse.c (record_set_data): New function.
1072 (single_set_gcse): New function.
1073 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
1074 (hoist_code): Likewise.
1075 (get_pressure_class_and_nregs): Likewise.
1077 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
1079 * explow.c (probe_stack_range): Emit a final optimization blockage.
1081 2014-04-04 Anthony Green <green@moxielogic.com>
1083 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
1086 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
1089 * lto-cgraph.c (input_overwrite_node): Check that partitioning
1090 flags are set only during streaming.
1091 * ipa.c (process_references, walk_polymorphic_call_targets,
1092 symtab_remove_unreachable_nodes): Drop bodies of always inline
1093 after early inlining.
1094 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
1096 2014-04-04 Jakub Jelinek <jakub@redhat.com>
1097 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1100 * dwarf2out.c (const_ok_for_output_1): Reject expressions
1103 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1106 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
1108 (cortex_a53_fdivd): Likewise.
1110 2014-04-04 Martin Jambor <mjambor@suse.cz>
1113 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
1115 * cgraph.c (clone_of_p): Also return true if thunks match.
1116 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
1117 cgraph_function_or_thunk_node and an obsolete comment.
1118 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
1120 (build_function_decl_skip_args): Likewise.
1121 (set_new_clone_decl_and_node_flags): New function.
1122 (duplicate_thunk_for_node): Likewise.
1123 (redirect_edge_duplicating_thunks): Likewise.
1124 (cgraph_clone_node): New parameter args_to_skip, pass it to
1125 redirect_edge_duplicating_thunks which is called instead of
1126 cgraph_redirect_edge_callee.
1127 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
1128 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
1130 2014-04-04 Jeff Law <law@redhat.com>
1133 * config/arm/predicates.md (const_int_I_operand): New predicate.
1134 (const_int_M_operand): Similarly.
1135 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
1137 (insv_t2, extv_reg, extzv_t2): Likewise.
1138 (load_multiple_with_writeback): Similarly for const_int_I_operand.
1139 (pop_multiple_with_writeback_and_return): Likewise.
1140 (vfp_pop_multiple_with_writeback): Likewise
1142 2014-04-04 Richard Biener <rguenther@suse.de>
1145 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
1146 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
1148 * gimplify.h (gimple_add_tmp_var_fn): Declare.
1149 * gimplify.c (gimple_add_tmp_var_fn): New function.
1150 * gimple-expr.h (create_tmp_reg_fn): Declare.
1151 * gimple-expr.c (create_tmp_reg_fn): New function.
1152 * gimple-low.c (record_vars_into): Don't change cfun.
1153 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
1154 code generation without cfun.
1156 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
1159 * Makefile.in (install-driver): Fix shell scripting.
1161 2014-04-03 Cong Hou <congh@google.com>
1163 PR tree-optimization/60505
1164 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
1165 threshold of number of iterations below which no vectorization
1167 * tree-vect-loop.c (new_loop_vec_info):
1168 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
1169 * tree-vect-loop.c (vect_analyze_loop_operations):
1170 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
1171 * tree-vect-loop.c (vect_transform_loop):
1172 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
1173 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
1174 of iterations of the loop and see if we should build the epilogue.
1176 2014-04-03 Richard Biener <rguenther@suse.de>
1178 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
1179 (streamer_tree_cache_create): Adjust.
1180 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
1181 to allow optional nodes array.
1182 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
1183 (streamer_tree_cache_append): Likewise.
1184 (streamer_tree_cache_create): Create nodes array optionally
1185 as specified by parameter.
1186 * lto-streamer-out.c (create_output_block): Avoid maintaining
1187 the node array in the writer cache.
1188 (DFS_write_tree): Remove assertion.
1189 (produce_asm_for_decls): Free the out decl state hash table early.
1190 * lto-streamer-in.c (lto_data_in_create): Adjust for
1191 streamer_tree_cache_create prototype change.
1193 2014-04-03 Richard Biener <rguenther@suse.de>
1195 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
1196 set TREE_CHAIN to NULL_TREE.
1198 2014-04-03 Richard Biener <rguenther@suse.de>
1200 PR tree-optimization/60740
1201 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
1202 over all GIMPLE_COND operands.
1204 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
1206 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
1207 (Weffc++): Remove Scott's numbering, merge lists and reference
1210 2014-04-03 Nick Clifton <nickc@redhat.com>
1212 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
1215 2014-04-03 Martin Jambor <mjambor@suse.cz>
1217 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
1218 mention gcc_unreachable before failing.
1219 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
1222 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
1225 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
1226 inconsistent code and instead mark the context inconsistent.
1227 (possible_polymorphic_call_targets): For inconsistent contexts
1228 return empty complete list.
1230 2014-04-02 Anthony Green <green@moxielogic.com>
1232 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
1233 (extendqisi2, extendhisi2): Define.
1234 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
1235 (WCHAR_TYPE): Change to unsigned int.
1237 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1239 PR tree-optimization/60733
1240 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
1241 insertion point for PHI candidates to be the end of the feeding
1242 block for the PHI argument.
1244 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
1246 PR rtl-optimization/60650
1247 * lra-constraints.c (process_alt_operands): Decrease reject for
1248 earlyclobber matching.
1250 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1252 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
1254 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1256 * config/spu/spu.c (pad_bb): Do not crash when the last
1257 insn is CODE_FOR_blockage.
1259 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1261 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
1262 lies outside the target mode.
1264 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
1267 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1268 software floating point or no floating point registers, do not
1269 allow any type in the FPRs. Eliminate a test for SPE SIMD types
1270 in GPRs that occurs after we tested for GPRs that would never be
1273 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1274 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1275 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
1276 specifically allow DDmode, since that does not use the SPE SIMD
1279 2014-04-02 Richard Biener <rguenther@suse.de>
1282 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
1283 MODE_INTs. Properly use negv_optab.
1284 (expand_abs): Likewise.
1286 2014-04-02 Richard Biener <rguenther@suse.de>
1289 * Makefile.in (install-driver): Guard extra installs with special
1292 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1294 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1297 2014-04-01 Richard Henderson <rth@redhat.com>
1300 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
1301 alternative enabled before register allocation.
1303 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
1305 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
1306 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
1308 (nios2_large_got_address): Remove unneeded 'sym' parameter.
1309 (nios2_got_address): Update nios2_large_got_address call site.
1310 (nios2_delegitimize_address): New function.
1311 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
1312 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
1313 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
1315 2014-04-01 Martin Husemann <martin@duskware.de>
1317 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
1320 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
1322 PR rtl-optimization/60604
1323 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
1324 check from register_operand.
1325 (register_operand): Redefine in terms of general_operand.
1326 (nonmemory_operand): Use register_operand for the non-constant cases.
1328 2014-04-01 Richard Biener <rguenther@suse.de>
1330 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
1332 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
1334 * doc/invoke.texi (mapp-regs): Clarify.
1336 2014-03-31 Ulrich Drepper <drepper@gmail.com>
1338 * config/i386/avx512fintrin.h (__v32hi): Define type.
1339 (__v64qi): Likewise.
1340 (_mm512_set1_epi8): Define.
1341 (_mm512_set1_epi16): Define.
1342 (_mm512_set4_epi32): Define.
1343 (_mm512_set4_epi64): Define.
1344 (_mm512_set4_pd): Define.
1345 (_mm512_set4_ps): Define.
1346 (_mm512_setr4_epi64): Define.
1347 (_mm512_setr4_epi32): Define.
1348 (_mm512_setr4_pd): Define.
1349 (_mm512_setr4_ps): Define.
1350 (_mm512_setzero_epi32): Define.
1352 2014-03-31 Martin Jambor <mjambor@suse.cz>
1355 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
1356 callsite_arguments_match_p. Updated all callers. Also check types of
1357 corresponding formal parameters and actual arguments.
1358 (not_all_callers_have_enough_arguments_p) Renamed to
1359 some_callers_have_mismatched_arguments_p.
1361 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
1363 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
1365 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
1368 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
1371 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
1373 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
1374 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
1376 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
1377 Use FMAMODE_NOVF512 mode iterator.
1378 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
1379 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
1380 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
1381 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
1383 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
1384 Use VF_128_256 mode iterator.
1385 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
1388 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1390 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
1391 static chain if needed.
1393 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1396 * lra-constraints.c (index_part_to_reg): New.
1397 (process_address): Use it.
1399 2014-03-27 Jeff Law <law@redhat.com>
1400 Jakub Jelinek <jakub@redhat.com>
1403 * expr.c (do_tablejump): Use simplify_gen_binary rather than
1404 gen_rtx_{PLUS,MULT} to build up the address expression.
1406 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
1407 creating non-canonical RTL.
1409 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1412 * ipa-inline.c (want_inline_small_function_p): Short circuit large
1413 functions; reorganize to make cheap checks first.
1414 (inline_small_functions): Do not estimate growth when dumping;
1416 * ipa-inline.h (inline_summary): Add min_size.
1417 (growth_likely_positive): New function.
1418 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
1419 (set_cond_stmt_execution_predicate): Cleanup.
1420 (estimate_edge_size_and_time): Compute min_size.
1421 (estimate_calls_size_and_time): Likewise.
1422 (estimate_node_size_and_time): Likewise.
1423 (inline_update_overall_summary): Update min_size.
1424 (do_estimate_edge_time): Likewise.
1425 (do_estimate_edge_size): Update.
1426 (do_estimate_edge_hints): Update.
1427 (growth_likely_positive): New function.
1429 2014-03-28 Jakub Jelinek <jakub@redhat.com>
1432 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
1433 also if addr has VOIDmode.
1435 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1437 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
1438 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
1440 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
1441 instructions as well as AdvancedSIMD loads.
1443 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1445 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1446 Use crypto_aese type.
1447 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
1448 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
1449 crypto_aese, crypto_aesmc. Move to types.md.
1450 * config/arm/types.md (crypto_aes): Split into crypto_aese,
1452 * config/arm/iterators.md (crypto_type): Likewise.
1454 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
1456 * cgraph.c: Include expr.h and tree-dfa.h.
1457 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
1460 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
1463 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
1464 regs from checking multi-reg pseudos.
1466 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1468 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
1470 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1472 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
1473 if it would clobber the stack pointer, even temporarily.
1475 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
1477 * mode-switching.c: Make small adjustments to the top comment.
1479 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1481 * config/rs6000/constraints.md (wD constraint): New constraint to
1482 match the constant integer to get the top DImode/DFmode out of a
1483 vector in a VSX register.
1485 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
1486 match the constant integer to get the top DImode/DFmode out of a
1487 vector in a VSX register.
1489 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
1492 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1495 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
1496 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
1498 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
1499 Optimize vec_extract of 64-bit values, where the value being
1500 extracted is in the top word, where we can use scalar
1501 instructions. Add direct move and store support. Combine the big
1502 endian/little endian vector select load support into a single insn.
1503 (vsx_extract_<mode>_internal1): Likewise.
1504 (vsx_extract_<mode>_internal2): Likewise.
1505 (vsx_extract_<mode>_load): Likewise.
1506 (vsx_extract_<mode>_store): Likewise.
1507 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
1508 combined into vsx_extract_<mode>_load.
1509 (vsx_extract_<mode>_one_le): Likewise.
1511 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
1512 define the top 64-bit vector element.
1514 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
1517 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1518 Document vec_vbpermq builtin.
1521 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
1522 enable use of xxsldwi and xxpermdi builtin functions.
1523 (vec_xxpermdi): Likewise.
1525 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1526 Document use of vec_xxsldwi and vec_xxpermdi builtins.
1528 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
1530 PR rtl-optimization/60650
1531 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
1533 (find_spills_for): New.
1534 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
1535 Spill all pseudos on the second iteration.
1537 2014-03-27 Marek Polacek <polacek@redhat.com>
1540 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
1543 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1545 * config/s390/s390.c (s390_can_use_return_insn): Check for
1546 call-saved FPRs on 31 bit.
1548 2014-03-27 Jakub Jelinek <jakub@redhat.com>
1551 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
1552 if they need regimplification, just drop them instead of
1553 calling gimple_regimplify_operands on them.
1555 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
1558 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
1559 (aarch64_frame_pointer_required): Adjust logic.
1560 (aarch64_can_eliminate): Adjust logic.
1561 (aarch64_override_options_after_change): Adjust logic.
1563 2014-03-27 Dehao Chen <dehao@google.com>
1565 * ipa-inline.c (early_inliner): Update node's inline info.
1567 2014-03-26 Dehao Chen <dehao@google.com>
1569 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
1570 compiler inserted conditional jumps for NAN float check.
1572 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1574 * ubsan.h (ubsan_create_data): Change second argument's type
1575 to const location_t *.
1576 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
1578 (ubsan_create_data): Change second argument to const location_t *PLOC.
1579 Create Loc field whenever PLOC is non-NULL.
1580 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
1581 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
1585 * real.c (real_to_integer2): Change type of low to UHWI.
1587 2014-03-26 Tobias Burnus <burnus@net-b.de>
1589 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
1590 (CILK_SELF_SPECS): New define.
1591 (driver_self_specs): Use it.
1593 2014-03-26 Richard Biener <rguenther@suse.de>
1595 * tree-pretty-print.c (percent_K_format): Implement special
1596 case for LTO and its stripped down BLOCK tree.
1598 2014-03-26 Jakub Jelinek <jakub@redhat.com>
1601 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
1603 * tree-vrp.c (simplify_internal_call_using_ranges): If only
1604 one range is range_int_cst_p, but not both, at least optimize
1605 addition/subtraction of 0 and multiplication by 0 or 1.
1606 * gimple-fold.c (gimple_fold_call): Fold
1607 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
1608 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
1609 INTEGER_CSTs, try to fold at least x * 0 and y - y.
1611 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
1613 PR rtl-optimization/60452
1614 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
1615 <case REG>: Return 1 for invalid offsets from the frame pointer.
1617 2014-03-26 Marek Polacek <polacek@redhat.com>
1620 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1623 2014-03-26 Marek Polacek <polacek@redhat.com>
1626 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1629 2014-03-26 Marek Polacek <polacek@redhat.com>
1632 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1633 multiplication in unsigned type.
1635 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1637 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1639 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1641 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1643 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1646 * cif-code.def (UNREACHABLE) New code.
1647 * ipa-inline.c (inline_small_functions): Skip edges to
1648 __builtlin_unreachable.
1649 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1650 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1651 predicate to __bulitin_unreachable.
1652 (set_cond_stmt_execution_predicate): Fix issue when
1653 invert_tree_comparison returns ERROR_MARK.
1654 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1655 propagate to inline clones.
1656 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1658 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1659 * cgraphclones.c (cgraph_clone_node): If call destination is already
1660 ureachable, do not redirect it back.
1661 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1664 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1666 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1667 Do not modify inline clones.
1669 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1671 * config/i386/i386.md (general_sext_operand): New mode attr.
1672 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1673 don't generate (sign_extend (const_int)).
1674 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1675 operands[2]. Use We constraint instead of <i> and
1676 <general_sext_operand> predicate instead of <general_operand>.
1677 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1678 * config/i386/constraints.md (We): New constraint.
1679 * config/i386/predicates.md (x86_64_sext_operand,
1680 sext_operand): New predicates.
1682 2014-03-25 Martin Jambor <mjambor@suse.cz>
1685 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1686 inconsistent devirtualizations to __builtin_unreachable.
1688 2014-03-25 Marek Polacek <polacek@redhat.com>
1691 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1693 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1695 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1696 order of elements for big-endian.
1698 2014-03-25 Richard Biener <rguenther@suse.de>
1701 * gimplify-me.c (gimple_regimplify_operands): Update the
1704 2014-03-25 Martin Jambor <mjambor@suse.cz>
1707 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1708 (lto_output_varpool_node): Likewise.
1709 (input_overwrite_node): Likewise.
1710 (input_varpool_node): Likewise.
1712 2014-03-25 Richard Biener <rguenther@suse.de>
1714 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1715 (run_gcc): Likewise.
1717 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1719 * combine.c (simplify_compare_const): Add MODE argument.
1720 Handle mode_width 0 as very large mode_width.
1721 (try_combine, simplify_comparison): Adjust callers.
1723 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1724 type to avoid signed integer overflow.
1725 * explow.c (plus_constant): Likewise.
1727 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1729 * doc/generic.texi: Correct typos.
1731 2014-03-24 Tobias Burnus <burnus@net-b.de>
1733 * doc/invoke.texi (-flto): Expand section about
1734 using static libraries with LTO.
1736 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1738 PR rtl-optimization/60501
1739 * optabs.def (addptr3_optab): New optab.
1740 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1741 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1742 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1744 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1746 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1748 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1750 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1753 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1754 (_mm256_undefined_ps): Define.
1755 (_mm256_undefined_pd): Define.
1756 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1757 (_mm_undefined_pd): Define.
1758 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1759 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1760 (_mm512_undefined_ps): Define.
1761 (_mm512_undefined_pd): Define.
1762 Use _mm*_undefined_*.
1763 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1765 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1767 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1768 (lshr_simd): DI mode added.
1769 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1770 (aarch64_ushr_simddi): Likewise.
1771 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1772 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1773 (vshrd_n_u64): Likewise.
1775 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1777 * Makefile.in (s-macro_list): Depend on cc1.
1779 2014-03-23 Teresa Johnson <tejohnson@google.com>
1781 * ipa-utils.c (ipa_print_order): Use specified dump file.
1783 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1785 PR rtl-optimization/60601
1786 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1788 * gcc.c (eval_spec_function): Initialize save_growing_value.
1790 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1793 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1794 code == MINUS_EXPR, never swap op0 with op1.
1796 * toplev.c (init_local_tick): Avoid signed integer multiplication
1798 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1799 shift by first operand's bitsize.
1801 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1804 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1806 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1807 TARGET_ISA_64BIT_P(x).
1809 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1811 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1812 pattern for vector nor instead of subtract from splat(-1).
1813 (altivec_expand_vec_perm_const_le): Likewise.
1815 2014-03-21 Richard Henderson <rth@twiddle.net>
1818 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1819 related insns after epilogue_completed.
1821 2014-03-21 Martin Jambor <mjambor@suse.cz>
1824 * cgraph.h (symtab_node): New flag body_removed.
1825 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1826 when removing bodies.
1827 * symtab.c (dump_symtab_base): Dump body_removed flag.
1828 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1829 had their bodies removed.
1831 2014-03-21 Martin Jambor <mjambor@suse.cz>
1834 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1837 2014-03-21 Richard Biener <rguenther@suse.de>
1839 PR tree-optimization/60577
1840 * tree-core.h (struct tree_base): Document nothrow_flag use
1841 in VAR_DECL_NONALIASED.
1842 * tree.h (VAR_DECL_NONALIASED): New.
1843 (may_be_aliased): Adjust.
1844 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1846 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1848 * expr.c (expand_expr_real_1): Remove outdated comment.
1850 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1853 * ira.c (adjust_cleared_regs): Call copy_rtx on
1854 *reg_equiv[REGNO (loc)].src_p before passing it to
1855 simplify_replace_fn_rtx.
1858 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1859 into CONST, put pic register as first operand of PLUS. Use
1860 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1862 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1864 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1866 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1868 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1869 around for store forwarding issue in the FPU on the UT699.
1870 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1871 loads and operations if -mfix-ut699 is specified.
1872 (divtf3_hq): Tweak attribute.
1873 (sqrttf2_hq): Likewise.
1875 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1877 * calls.c (store_one_arg): Remove incorrect const qualification on the
1878 type of the temporary.
1879 * cfgexpand.c (expand_return): Likewise.
1880 * expr.c (expand_constructor): Likewise.
1881 (expand_expr_real_1): Likewise.
1883 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1885 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1888 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1891 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1893 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1895 * config/arm/aarch-common-protos.h
1896 (alu_cost_table): Fix spelling of "extend".
1897 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1899 2014-03-19 Richard Biener <rguenther@suse.de>
1902 * tree-core.h (tree_type_common): Re-order pointer members
1903 to reduce recursion depth during GC walks.
1905 2014-03-19 Marek Polacek <polacek@redhat.com>
1908 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1909 before accessing it.
1911 2014-03-19 Richard Biener <rguenther@suse.de>
1914 * lto-streamer-in.c (input_function): In WPA stage do not drop
1917 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1919 PR tree-optimization/60559
1920 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1921 with build_zero_cst assignment.
1923 2014-03-18 Kai Tietz <ktietz@redhat.com>
1925 PR rtl-optimization/56356
1926 * sdbout.c (sdbout_parms): Verify that parms'
1927 incoming argument is valid.
1928 (sdbout_reg_parms): Likewise.
1930 2014-03-18 Richard Henderson <rth@redhat.com>
1933 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1934 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1935 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1937 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1939 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1940 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1941 Italicize plugin event names in description. Explain that
1942 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1943 Remind that no GCC functions should be called after PLUGIN_FINISH.
1944 Explain what pragmas with expansion are.
1946 2014-03-18 Martin Liska <mliska@suse.cz>
1948 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1949 gimple call statement is update.
1950 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1951 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1953 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1956 * ubsan.c (ubsan_instrument_unreachable): Call
1957 initialize_sanitizer_builtins.
1958 (ubsan_pass): Likewise.
1961 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1962 varpool_finalize_decl instead of rest_of_decl_compilation.
1964 2014-03-18 Richard Biener <rguenther@suse.de>
1966 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1967 by using bitmap_and_compl instead of bitmap_and_compl_into.
1968 (df_rd_transfer_function): Likewise.
1970 2014-03-18 Richard Biener <rguenther@suse.de>
1972 * doc/lto.texi (fresolution): Fix typo.
1974 2014-03-18 Richard Biener <rguenther@suse.de>
1976 * doc/invoke.texi (flto): Update for changes in 4.9.
1978 2014-03-18 Richard Biener <rguenther@suse.de>
1980 * doc/loop.texi: Remove section on the removed lambda framework.
1981 Update loop docs with recent changes in preserving loop structure.
1983 2014-03-18 Richard Biener <rguenther@suse.de>
1985 * doc/lto.texi (-fresolution): Document.
1987 2014-03-18 Richard Biener <rguenther@suse.de>
1989 * doc/contrib.texi: Adjust my name.
1991 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1994 * internal-fn.c: Include diagnostic-core.h.
1995 (expand_BUILTIN_EXPECT): New function.
1996 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1997 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1998 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1999 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
2001 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
2002 Revert 3 argument __builtin_expect code.
2003 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
2004 * gimple-fold.c (gimple_fold_call): Likewise.
2005 * tree.h (fold_builtin_expect): New prototype.
2006 * builtins.c (build_builtin_expect_predicate): Add predictor
2007 argument, if non-NULL, create 3 argument __builtin_expect.
2008 (fold_builtin_expect): No longer static. Add ARG2 argument,
2009 pass it through to build_builtin_expect_predicate.
2010 (fold_builtin_2): Adjust caller.
2011 (fold_builtin_3): Handle BUILT_IN_EXPECT.
2012 * internal-fn.def (BUILTIN_EXPECT): New.
2014 2014-03-18 Tobias Burnus <burnus@net-b.de>
2017 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
2018 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
2019 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
2021 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
2024 * predict.c (combine_predictions_for_bb): Fix up formatting.
2025 (expr_expected_value_1, expr_expected_value): Add predictor argument,
2026 fill what it points to if non-NULL.
2027 (tree_predict_by_opcode): Adjust caller, use the predictor.
2028 * predict.def (PRED_COMPARE_AND_SWAP): Add.
2030 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
2032 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
2033 proper constant for the store mode.
2035 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
2037 * symtab.c (change_decl_assembler_name): Fix transparent alias
2040 2014-03-16 Renlin Li <Renlin.Li@arm.com>
2042 * config/aarch64/aarch64.c: Correct the comments about the
2043 aarch64 stack layout.
2045 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
2047 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
2048 check for GF_OMP_FOR_KIND_FOR.
2050 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
2052 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
2053 ymm and zmm register names.
2055 2014-03-17 Jakub Jelinek <jakub@redhat.com>
2058 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
2059 note creation for the 2010-08-31 changes.
2061 2014-03-17 Marek Polacek <polacek@redhat.com>
2064 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
2065 as -fno-tree-loop-vectorize.
2066 (expand_omp_simd): Likewise.
2068 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
2070 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
2071 (eligible_for_call_delay): New prototype.
2072 * config/sparc/sparc.c (tls_call_delay): Rename into...
2073 (eligible_for_call_delay): ...this. Return false if the instruction
2074 cannot be put in the delay slot of a branch.
2075 (eligible_for_restore_insn): Simplify.
2076 (eligible_for_return_delay): Return false if the instruction cannot be
2077 put in the delay slot of a branch and simplify.
2078 (eligible_for_sibcall_delay): Return false if the instruction cannot be
2079 put in the delay slot of a branch.
2080 * config/sparc/sparc.md (fix_ut699): New attribute.
2081 (tls_call_delay): Delete.
2082 (in_call_delay): Reimplement.
2083 (eligible_for_sibcall_delay): Rename into...
2084 (in_sibcall_delay): ...this.
2085 (eligible_for_return_delay): Rename into...
2086 (in_return_delay): ...this.
2087 (in_branch_delay): Reimplement.
2088 (in_uncond_branch_delay): Delete.
2089 (in_annul_branch_delay): Delete.
2091 2014-03-14 Richard Henderson <rth@redhat.com>
2094 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
2095 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
2096 (*floathi<X87MODEF>2_i387_with_temp): Remove.
2097 (floathi splitters): Remove.
2098 (float<SWI48x>xf2): New pattern.
2099 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
2100 code that tried to handle DImode for 32-bit, but which was excluded
2101 by the pattern's condition. Drop allocation of stack temporary.
2102 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
2103 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
2104 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
2105 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
2106 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
2107 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
2108 (*float<SWI48><MODEF>2_sse_interunit): Remove.
2109 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
2110 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
2111 (*float<SWI48x><X87MODEF>2_i387): Remove.
2112 (all float _with_temp splitters): Remove.
2113 (*float<SWI48x><MODEF>2_i387): New pattern.
2114 (*float<SWI48><MODEF>2_sse): New pattern.
2115 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
2116 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
2118 2014-03-14 Jakub Jelinek <jakub@redhat.com>
2119 Marek Polacek <polacek@redhat.com>
2122 * common.opt (dump_base_name_prefixed): New Variable.
2123 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
2124 if x_dump_base_name_prefixed is already set, set it at the end.
2126 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
2128 PR rtl-optimization/60508
2129 * lra-constraints.c (get_reload_reg): Add new parameter
2131 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
2132 Pass the new parameter values.
2134 2014-03-14 Richard Biener <rguenther@suse.de>
2136 * common.opt: Revert unintented changes from r205065.
2139 2014-03-14 Richard Biener <rguenther@suse.de>
2142 * cfghooks.c (split_block): Properly adjust all loops the
2143 block was a latch of.
2145 2014-03-14 Martin Jambor <mjambor@suse.cz>
2148 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
2151 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
2154 * config/avr/avr.c (avr_set_current_function): Pass function name
2155 through default_strip_name_encoding before sanity checking instead
2156 of skipping the first char of the assembler name.
2158 2014-03-13 Richard Henderson <rth@redhat.com>
2161 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
2162 (ix86_force_to_memory, ix86_free_from_memory): Remove.
2163 * config/i386/i386-protos.h: Likewise.
2164 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
2165 in the expander instead of a splitter.
2166 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
2167 any possibility of requiring a memory.
2168 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
2169 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
2170 (fp branch splitters): Update for ix86_split_fp_branch.
2171 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
2172 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
2173 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
2174 (*fop_<MODEF>_2_i387): Remove f/r alternative.
2175 (*fop_<MODEF>_3_i387): Likewise.
2176 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
2177 (splitters for the fop_* register patterns): Remove.
2178 (fscalexf4_i387): Rename from *fscalexf4_i387.
2179 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
2181 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2183 PR tree-optimization/59779
2184 * tree-dfa.c (get_ref_base_and_extent): Use double_int
2185 type for bitsize and maxsize instead of HOST_WIDE_INT.
2187 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
2189 PR rtl-optimization/57320
2190 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
2191 the CFG after thread_prologue_and_epilogue_insns.
2193 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
2195 PR rtl-optimization/57189
2196 * lra-constraints.c (process_alt_operands): Disfavor spilling
2199 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
2201 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
2203 2014-03-13 Jakub Jelinek <jakub@redhat.com>
2205 PR tree-optimization/59025
2207 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
2208 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
2210 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
2213 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
2214 calls of avr_out_plus_1.
2216 2014-03-13 Bin Cheng <bin.cheng@arm.com>
2218 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
2219 BB's single pred and update the father loop's latch info later.
2221 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2223 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
2227 (VEC_base): Likewise.
2228 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
2229 registers, we need to swap double words in little endian mode.
2231 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
2232 to be a container mode for 128-bit integer operations added in ISA
2233 2.07. Unlike TImode and PTImode, the preferred register set is
2234 the Altivec/VMX registers for the 128-bit operations.
2236 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
2238 (rs6000_split_128bit_ok_p): Likewise.
2240 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
2241 macros for creating ISA 2.07 normal and overloaded builtin
2242 functions with 3 arguments.
2243 (BU_P8V_OVERLOAD_3): Likewise.
2244 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
2245 for use as overloaded functions.
2246 (VPERM_1TI_UNS): Likewise.
2247 (VSEL_1TI): Likewise.
2248 (VSEL_1TI_UNS): Likewise.
2249 (ST_INTERNAL_1ti): Likewise.
2250 (LD_INTERNAL_1ti): Likewise.
2251 (XXSEL_1TI): Likewise.
2252 (XXSEL_1TI_UNS): Likewise.
2253 (VPERM_1TI): Likewise.
2254 (VPERM_1TI_UNS): Likewise.
2255 (XXPERMDI_1TI): Likewise.
2256 (SET_1TI): Likewise.
2257 (LXVD2X_V1TI): Likewise.
2258 (STXVD2X_V1TI): Likewise.
2259 (VEC_INIT_V1TI): Likewise.
2260 (VEC_SET_V1TI): Likewise.
2261 (VEC_EXT_V1TI): Likewise.
2262 (EQV_V1TI): Likewise.
2263 (NAND_V1TI): Likewise.
2264 (ORC_V1TI): Likewise.
2265 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
2266 added in ISA 2.07. Add both normal 'altivec' builtins, and the
2268 (VADDUQM): Likewise.
2269 (VSUBCUQ): Likewise.
2270 (VADDEUQM): Likewise.
2271 (VADDECUQ): Likewise.
2272 (VSUBEUQM): Likewise.
2273 (VSUBECUQ): Likewise.
2275 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
2276 __int128_t and __uint128_t types.
2277 (__uint128_type): Likewise.
2278 (altivec_categorize_keyword): Add support for vector __int128_t,
2279 vector __uint128_t, vector __int128, and vector unsigned __int128
2280 as a container type for TImode operations that need to be done in
2281 VSX/Altivec registers.
2282 (rs6000_macro_to_expand): Likewise.
2283 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
2284 to support 128-bit integer instructions vaddcuq, vadduqm,
2285 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
2286 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
2288 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
2289 for V1TImode, and set up preferences to use VSX/Altivec registers.
2290 Setup VSX reload handlers.
2291 (rs6000_debug_reg_global): Likewise.
2292 (rs6000_init_hard_regno_mode_ok): Likewise.
2293 (rs6000_preferred_simd_mode): Likewise.
2294 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
2295 (easy_altivec_constant): Likewise.
2296 (output_vec_const_move): Likewise.
2297 (rs6000_expand_vector_set): Convert V1TImode set and extract to
2299 (rs6000_expand_vector_extract): Likewise.
2300 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
2302 (rs6000_const_vec): Add support for V1TImode.
2303 (rs6000_emit_le_vsx_load): Swap double words when loading or
2304 storing TImode/V1TImode.
2305 (rs6000_emit_le_vsx_store): Likewise.
2306 (rs6000_emit_le_vsx_move): Likewise.
2307 (rs6000_emit_move): Add support for V1TImode.
2308 (altivec_expand_ld_builtin): Likewise.
2309 (altivec_expand_st_builtin): Likewise.
2310 (altivec_expand_vec_init_builtin): Likewise.
2311 (altivec_expand_builtin): Likewise.
2312 (rs6000_init_builtins): Add support for V1TImode type. Add
2313 support for ISA 2.07 128-bit integer builtins. Define type names
2314 for the VSX/Altivec vector types.
2315 (altivec_init_builtins): Add support for overloaded vector
2316 functions with V1TImode type.
2317 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
2318 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
2320 (rs6000_split_128bit_ok_p): Likewise.
2321 (rs6000_handle_altivec_attribute): Create V1TImode from vector
2322 __int128_t and vector __uint128_t.
2324 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
2325 and mode attributes.
2332 (VS_scalar): Likewise.
2333 (VS_double): Likewise.
2334 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
2336 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
2337 we support the ISA 2.07 128-bit integer arithmetic instructions.
2338 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
2339 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
2340 and TImode types for use with the builtin functions.
2341 (V1TI_type_node): Likewise.
2342 (unsigned_V1TI_type_node): Likewise.
2343 (intTI_type_internal_node): Likewise.
2344 (uintTI_type_internal_node): Likewise.
2346 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
2347 128-bit builtin functions.
2348 (UNSPEC_VADDEUQM): Likewise.
2349 (UNSPEC_VADDECUQ): Likewise.
2350 (UNSPEC_VSUBCUQ): Likewise.
2351 (UNSPEC_VSUBEUQM): Likewise.
2352 (UNSPEC_VSUBECUQ): Likewise.
2353 (VM): Add V1TImode to vector mode iterators.
2355 (VI_unit): Likewise.
2356 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
2357 (altivec_vaddcuq): Likewise.
2358 (altivec_vsubuqm): Likewise.
2359 (altivec_vsubcuq): Likewise.
2360 (altivec_vaddeuqm): Likewise.
2361 (altivec_vaddecuq): Likewise.
2362 (altivec_vsubeuqm): Likewise.
2363 (altivec_vsubecuq): Likewise.
2365 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
2367 (BOOL_128): Likewise.
2368 (BOOL_REGS_OUTPUT): Likewise.
2369 (BOOL_REGS_OP1): Likewise.
2370 (BOOL_REGS_OP2): Likewise.
2371 (BOOL_REGS_UNARY): Likewise.
2372 (BOOL_REGS_AND_CR0): Likewise.
2374 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
2375 128-bit integer builtin support.
2376 (vec_vadduqm): Likewise.
2377 (vec_vaddecuq): Likewise.
2378 (vec_vaddeuqm): Likewise.
2379 (vec_vsubecuq): Likewise.
2380 (vec_vsubeuqm): Likewise.
2381 (vec_vsubcuq): Likewise.
2382 (vec_vsubuqm): Likewise.
2384 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2385 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
2386 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
2387 128-bit integer add/subtract to ISA 2.07.
2389 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
2391 * config/arc/arc.c (arc_predicate_delay_insns):
2392 Fix third argument passed to conditionalize_nonjump.
2394 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
2396 * config/aarch64/aarch64-builtins.c
2397 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
2398 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
2399 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
2400 instead of __builtin_lfloor.
2401 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
2403 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2405 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
2406 (tree_ssa_ifcombine_bb_1): New function.
2407 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
2408 is an empty forwarder block to then_bb or vice versa and then_bb
2409 and else_bb are effectively swapped.
2411 2014-03-12 Christian Bruel <christian.bruel@st.com>
2414 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
2415 REG_CFA_DEF_CFA note.
2416 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
2417 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
2419 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
2421 PR tree-optimization/60454
2422 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
2424 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2426 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
2427 Do not define target_cpu_default2 to generic.
2428 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
2429 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
2430 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
2432 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2433 Marc Glisse <marc.glisse@inria.fr>
2435 PR tree-optimization/60502
2436 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
2437 instead of build_low_bits_mask.
2439 2014-03-12 Jakub Jelinek <jakub@redhat.com>
2442 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
2443 if there are multiple uses, but op doesn't live on E edge.
2444 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
2445 clobber stmts before __builtin_unreachable.
2447 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
2449 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
2450 hard_frame_pointer_rtx.
2451 * cse.c (cse_insn): Remove volatile check.
2452 * cselib.c (cselib_process_insn): Likewise.
2453 * dse.c (scan_insn): Likewise.
2455 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2457 * config/arc/arc.c (conditionalize_nonjump): New function,
2459 (arc_ifcvt): ... this.
2460 (arc_predicate_delay_insns): Use it.
2462 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
2464 * config/arc/predicates.md (extend_operand): During/after reload,
2465 allow const_int_operand.
2466 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
2467 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
2468 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
2470 (umulsi3_highpart_i): Likewise.
2472 2014-03-11 Richard Biener <rguenther@suse.de>
2474 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
2475 Add asserts to guard possible wrong-code bugs.
2477 2014-03-11 Richard Biener <rguenther@suse.de>
2479 PR tree-optimization/60429
2480 PR tree-optimization/60485
2481 * tree-ssa-structalias.c (set_union_with_increment): Properly
2482 take into account all fields that overlap the shifted vars.
2483 (do_sd_constraint): Likewise.
2484 (do_ds_constraint): Likewise.
2485 (get_constraint_for_ptr_offset): Likewise.
2487 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
2489 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
2490 (nios2_compute_frame_layout):
2491 Add calculation of cfun->machine->fp_save_offset.
2492 (nios2_expand_prologue): Correct setting of frame pointer register
2494 (nios2_expand_epilogue): Update recovery of stack pointer from
2495 frame pointer accordingly.
2496 (nios2_initial_elimination_offset): Update calculation of offset
2497 for eliminating to HARD_FRAME_POINTER_REGNUM.
2499 2014-03-10 Jakub Jelinek <jakub@redhat.com>
2502 * ipa.c (symtab_remove_unreachable_nodes): Don't call
2503 cgraph_get_create_node on VAR_DECLs.
2505 2014-03-10 Richard Biener <rguenther@suse.de>
2508 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
2510 2014-03-08 Douglas B Rupp <rupp@gnat.com>
2512 * config/vms/vms.opt (vms_float_format): New variable.
2514 2014-03-08 Tobias Burnus <burnus@net-b.de>
2516 * doc/invoke.texi (-fcilkplus): Update implementation status.
2518 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
2519 Richard Biener <rguenther@suse.de>
2521 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
2522 consistently accross all TUs.
2523 (run_gcc): Enable -fshort-double automatically at link at link-time
2524 and disallow override.
2526 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
2529 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
2530 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
2531 if they can't be used.
2533 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2535 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
2536 for Solaris 11/x86 ld.
2537 * configure: Regenerate.
2539 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2541 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
2542 (LIB_TLS_SPEC): Save as ld_tls_libs.
2543 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
2544 (HAVE_AS_IX86_TLSLDM): New test.
2545 * configure, config.in: Regenerate.
2546 * config/i386/i386.c (legitimize_tls_address): Fall back to
2547 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
2548 cannot support TLS_MODEL_LOCAL_DYNAMIC.
2549 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
2550 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
2552 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
2554 * common.opt (fira-loop-pressure): Mark as optimization.
2556 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
2558 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
2559 an OpenMP mappable type.
2561 2014-03-06 Matthias Klose <doko@ubuntu.com>
2563 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
2564 MULTILIB_OSDIRNAMES is not defined.
2566 2014-03-06 Jakub Jelinek <jakub@redhat.com>
2567 Meador Inge <meadori@codesourcery.com>
2570 * config/arm/arm.c (arm_tls_symbol_p): Remove.
2571 (arm_legitimize_address): Call legitimize_tls_address for any
2572 arm_tls_referenced_p expression, handle constant addend. Call it
2573 before testing for !TARGET_ARM.
2574 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
2576 2014-03-06 Richard Biener <rguenther@suse.de>
2582 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2584 * tree-streamer.c (record_common_node): Assert we don't record
2585 nodes with type double.
2586 (preload_common_node): Skip type double, complex double and double
2587 pointer since it is now frontend dependent due to fshort-double option.
2589 2014-03-06 Richard Biener <rguenther@suse.de>
2591 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
2592 or -fno-lto is specified and the linker has full plugin support.
2593 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
2594 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
2595 * lto-wrapper.c (merge_and_complain): Merge compile-time
2596 optimization levels.
2597 (run_gcc): And pass it through to the link options.
2599 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
2603 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2605 * cselib.c (remove_useless_values): Skip to avoid quadratic
2606 behavior if the condition moved from...
2607 (cselib_process_insn): ... here holds.
2609 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2612 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
2613 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2616 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2617 (TM_H): Add x86-tune.def.
2619 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2621 * config/aarch64/aarch64.c (generic_tunings):
2622 Use cortexa57_extra_costs.
2624 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2627 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2628 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2629 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2632 2014-03-04 Heiher <r@hev.cc>
2634 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2635 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2637 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2639 * config/i386/predicates.md (const2356_operand): Change to ...
2640 (const2367_operand): ... this.
2641 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2643 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2644 (*avx512pf_scatterpf<mode>sf): Ditto.
2645 (avx512pf_scatterpf<mode>df): Ditto.
2646 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2647 (*avx512pf_scatterpf<mode>df): Ditto.
2648 * config/i386/i386.c (ix86_expand_builtin): Update
2649 incorrect hint operand error message.
2651 2014-03-04 Richard Biener <rguenther@suse.de>
2653 * lto-section-in.c (lto_get_section_data): Fix const cast.
2655 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2657 * tree-streamer.c (record_common_node): Assert we don't record
2658 nodes with type double.
2659 (preload_common_node): Skip type double, complex double and double
2660 pointer since it is now frontend dependent due to fshort-double option.
2662 2014-03-04 Richard Biener <rguenther@suse.de>
2665 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2666 (lto_input_toplevel_asms): Likewise.
2667 * lto-section-in.c (lto_get_section_data): Instead do it here
2670 2014-03-04 Richard Biener <rguenther@suse.de>
2672 PR tree-optimization/60382
2673 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2674 dead PHIs a reduction.
2676 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2678 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2680 (_mm_prefetch): Move out of GCC target("sse") pragma.
2681 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2682 GCC target("prfchw") pragma.
2683 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2685 * config/i386/i386.c (ix86_option_override_internal): Enable
2686 -mprfchw with -mprefetchwt1.
2688 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2690 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2693 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2695 * opts.h (CL_PCH_IGNORE): Define.
2696 * targhooks.c (option_affects_pch_p):
2697 Return false for options that have CL_PCH_IGNORE set.
2698 * opt-functions.awk: Process PchIgnore.
2699 * doc/options.texi: Document PchIgnore.
2701 * config/arc/arc.opt (misize): Add PchIgnore property.
2703 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2705 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2706 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2707 constraint on constants to permit them being loaded into
2708 GENERAL_REGS or BASE_REGS.
2710 2014-03-03 Nick Clifton <nickc@redhat.com>
2712 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2713 anti-cacnonical alternatives.
2714 (negandhi3_real): New pattern.
2715 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2717 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2719 * config/avr/avr-mcus.def: Remove atxmega16x1.
2720 * config/avr/avr-tables.opt: Regenerate.
2721 * config/avr/t-multilib: Regenerate.
2722 * doc/avr-mmcu.texi: Regenerate.
2724 2014-03-03 Tobias Grosser <tobias@grosser.es>
2725 Mircea Namolaru <mircea.namolaru@inria.fr>
2727 PR tree-optimization/58028
2728 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2731 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2733 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2734 not handled by recognizers.
2736 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2739 * function.c (expand_function_end): Don't emit
2740 clobber_return_register sequence if clobber_after is a BARRIER.
2741 * cfgexpand.c (construct_exit_block): Append instructions before
2742 return_label to prev_bb.
2744 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2746 * config/rs6000/constraints.md: Document reserved use of "wc".
2748 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2751 * ipa.c (function_and_variable_visibility): When dissolving comdat
2752 group, also set all symbols to local.
2754 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2759 2013-12-14 Jan Hubicka <jh@suse.cz>
2761 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2763 2014-03-02 Jon Beniston <jon@beniston.com>
2769 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2770 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2771 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2772 (simple_return, *simple_return): New patterns
2773 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2774 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2776 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2778 * dwarf2out.c (gen_subprogram_die): Tidy.
2780 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2783 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2784 (*mov_t_msb_neg_negc): ... this new insn.
2786 2014-02-28 Jason Merrill <jason@redhat.com>
2789 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2792 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2795 * dwarf2out.c (decltype_auto_die): New static.
2796 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2797 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2798 (is_cxx_auto): Likewise.
2800 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2802 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2803 we are not using general regs only.
2805 2014-02-28 Richard Biener <rguenther@suse.de>
2808 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2809 previous fix and only allow to remove trivial pre-headers
2810 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2811 (remove_forwarder_block): Properly update the latch of a loop.
2813 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2816 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2817 (cselib_preserved_hash_table): New.
2818 (preserve_constants_and_equivs): Move preserved vals to it.
2819 (cselib_find_slot): Look it up first.
2820 (cselib_init): Initialize it.
2821 (cselib_finish): Release it.
2822 (dump_cselib_table): Dump it.
2824 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2827 * cselib.c (remove_useless_values): Skip to avoid quadratic
2828 behavior if the condition moved from...
2829 (cselib_process_insn): ... here holds.
2831 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2834 * var-tracking.c (vt_initialize): Apply the same condition to
2835 preserve the CFA base value.
2837 2014-02-28 Joey Ye <joey.ye@arm.com>
2840 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2841 if reload in progress or completed.
2843 2014-02-28 Tobias Burnus <burnus@net-b.de>
2846 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2849 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2851 * doc/tm.texi.in (Condition Code Status): Update documention for
2852 relative locations of cc0-setter and cc0-user.
2854 2014-02-27 Jeff Law <law@redhat.com>
2856 PR rtl-optimization/52714
2857 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2858 into two independent simple sets, if I3 is a jump, ensure the
2859 pattern we place into I3 is a (set (pc) ...).
2861 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2862 Jeff Law <law@redhat.com>
2864 PR rtl-optimization/49847
2865 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2866 are in different blocks.
2867 * doc/tm.texi (Condition Code Status): Update documention for
2868 relative locations of cc0-setter and cc0-user.
2870 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2873 * lra.c (lra_emit_add): Check SUBREG too.
2875 2014-02-27 Andreas Schwab <schwab@suse.de>
2877 * config/m68k/m68k.c (m68k_option_override): Disable
2878 -flive-range-shrinkage for classic m68k.
2879 (m68k_override_options_after_change): Likewise.
2881 2014-02-27 Marek Polacek <polacek@redhat.com>
2884 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2885 -Wmaybe-uninitialized.
2887 2014-02-27 Alan Modra <amodra@gmail.com>
2890 * reload1.c (emit_input_reload_insns): When reload_override_in,
2891 set old to rl->in_reg when rl->in_reg is a subreg.
2893 2014-02-26 Richard Biener <rguenther@suse.de>
2896 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2898 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2900 * common/config/i386/predicates.md (const1256_operand): Remove.
2901 (const2356_operand): New.
2902 (const_1_to_2_operand): Remove.
2903 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2904 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2905 (*avx512pf_gatherpf<mode>sf): Ditto.
2906 (avx512pf_gatherpf<mode>df): Ditto.
2907 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2908 (*avx512pf_gatherpf<mode>df): Ditto.
2909 (avx512pf_scatterpf<mode>sf): Ditto.
2910 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2911 (*avx512pf_scatterpf<mode>sf): Ditto.
2912 (avx512pf_scatterpf<mode>df): Ditto.
2913 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2914 (*avx512pf_scatterpf<mode>df): Ditto.
2915 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2917 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2919 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2920 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2921 (_mm512_mask_testn_epi64_mask): Move to ...
2922 * config/i386/avx512cdintrin.h: Here.
2923 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2924 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2925 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2926 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2927 TARGET_AVX512F from TARGET_AVX512CD.
2929 2014-02-26 Richard Biener <rguenther@suse.de>
2932 * ipa.c (walk_polymorphic_call_targets): Properly guard
2933 call to inline_update_overall_summary.
2935 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2938 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2939 and latches only if requested. Fix latch if it is removed.
2940 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2941 LOOPS_HAVE_PREHEADERS.
2943 2014-02-25 Andrew Pinski <apinski@cavium.com>
2945 * builtins.c (expand_builtin_thread_pointer): Create a new target
2946 when the target is NULL.
2948 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2950 PR rtl-optimization/60317
2951 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2952 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2953 * lra-assigns.c: Include params.h.
2954 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2955 other reload pseudos considerations.
2957 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2959 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2960 to use canonical form for nor<mode>3.
2962 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2965 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2968 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2970 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2971 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2972 (ix86_handle_option): Handle OPT_mprefetchwt1.
2973 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2974 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2976 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2977 OPTION_MASK_ISA_PREFETCHWT1.
2978 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2979 (PTA_PREFETCHWT1): New.
2980 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2981 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2982 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2983 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2984 (*prefetch_avx512pf_<mode>_: Change into ...
2985 (*prefetch_prefetchwt1_<mode>: This.
2986 * config/i386/i386.opt (mprefetchwt1): New.
2987 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2988 (_mm_prefetch): Handle intent to write.
2989 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2991 2014-02-25 Richard Biener <rguenther@suse.de>
2994 * emit-rtl.c (mem_attrs_htab): Remove.
2995 (mem_attrs_htab_hash): Likewise.
2996 (mem_attrs_htab_eq): Likewise.
2997 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2998 (init_emit_once): Do not allocate mem_attrs_htab.
3000 2014-02-25 Richard Biener <rguenther@suse.de>
3003 * lto-opts.c (lto_write_options): Output non-explicit conservative
3004 -fwrapv, -fno-trapv and -fno-strict-overflow.
3005 * lto-wrapper.c (merge_and_complain): Handle merging those options.
3006 (run_gcc): And pass them through.
3008 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3010 * sel-sched.c (calculate_new_fences): New parameter ptime.
3011 Calculate it as a maximum over all fence cycles.
3012 (sel_sched_region_2): Adjust the call to calculate_new_fences.
3013 Print the final schedule timing when sched_verbose.
3015 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
3017 PR rtl-optimization/60292
3018 * sel-sched.c (fill_vec_av_set): Do not reset target availability
3019 bit fot the fence instruction.
3021 2014-02-24 Alangi Derick <alangiderick@gmail.com>
3023 * calls.h: Fix typo in comment.
3025 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
3027 * config/pa/pa.c (pa_output_move_double): Don't valididate when
3028 adjusting offsetable addresses.
3030 2014-02-24 Guozhi Wei <carrot@google.com>
3032 * sparseset.h (sparseset_pop): Fix the wrong index.
3034 2014-02-24 Walter Lee <walt@tilera.com>
3036 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
3037 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
3039 * common/config/tilegx/tilegx-common.c
3040 (TARGET_DEFAULT_TARGET_FLAGS): Define.
3041 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
3043 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
3044 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
3045 (tilegx_gimplify_va_arg_expr): Handle big endian.
3046 (tilegx_expand_unaligned_load): Ditto.
3047 (tilegx_expand_unaligned_store): Ditto.
3048 (TARGET_RETURN_IN_MSB): New.
3049 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
3050 (TARGET_ENDIAN_DEFAULT): New.
3051 (TARGET_BIG_ENDIAN): Handle big endian.
3052 (BYTES_BIG_ENDIAN): Ditto.
3053 (WORDS_BIG_ENDIAN): Ditto.
3054 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
3057 * config/tilegx/tilegx.md (extv): Handle big endian.
3059 (insn_st<n>): Ditto.
3060 (insn_st<n>_add<bitsuffix>): Ditto.
3061 (insn_stnt<n>): Ditto.
3062 (insn_stnt<n>_add<bitsuffix>):Ditto.
3063 (vec_interleave_highv8qi): Handle big endian.
3064 (vec_interleave_highv8qi_be): New.
3065 (vec_interleave_highv8qi_le): New.
3066 (insn_v1int_h): Handle big endian.
3067 (vec_interleave_lowv8qi): Handle big endian.
3068 (vec_interleave_lowv8qi_be): New.
3069 (vec_interleave_lowv8qi_le): New.
3070 (insn_v1int_l): Handle big endian.
3071 (vec_interleave_highv4hi): Handle big endian.
3072 (vec_interleave_highv4hi_be): New.
3073 (vec_interleave_highv4hi_le): New.
3074 (insn_v2int_h): Handle big endian.
3075 (vec_interleave_lowv4hi): Handle big endian.
3076 (vec_interleave_lowv4hi_be): New.
3077 (vec_interleave_lowv4hi_le): New.
3078 (insn_v2int_l): Handle big endian.
3079 (vec_interleave_highv2si): Handle big endian.
3080 (vec_interleave_highv2si_be): New.
3081 (vec_interleave_highv2si_le): New.
3082 (insn_v4int_h): Handle big endian.
3083 (vec_interleave_lowv2si): Handle big endian.
3084 (vec_interleave_lowv2si_be): New.
3085 (vec_interleave_lowv2si_le): New.
3086 (insn_v4int_l): Handle big endian.
3087 * config/tilegx/tilegx.opt (mbig-endian): New option.
3088 (mlittle-endian): New option.
3089 * doc/install.texi: Document tilegxbe-linux.
3090 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
3092 2014-02-24 Martin Jambor <mjambor@suse.cz>
3095 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
3096 there are no parameter descriptors.
3098 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
3100 PR rtl-optimization/60268
3101 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
3102 initialization to ...
3103 (sched_rgn_init): ... here.
3104 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
3106 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3108 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
3111 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
3113 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
3116 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3118 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
3119 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
3121 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
3123 * config/microblaze/predicates.md: Add cmp_op predicate.
3124 * config/microblaze/microblaze.md: Add branch_compare instruction
3125 which uses cmp_op predicate and emits cmp insn before branch.
3126 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
3127 to microblaze_expand_conditional_branch and consolidate logic.
3128 (microblaze_expand_conditional_branch): emit branch_compare
3129 insn instead of handling cmp op separate from branch insn.
3131 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3133 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
3136 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3138 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
3139 define_insn with define_expand and new define_insn
3140 *altivec_lve<VI_char>x_internal.
3141 (altivec_stve<VI_char>x): Replace define_insn with define_expand
3142 and new define_insn *altivec_stve<VI_char>x_internal.
3143 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
3145 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
3147 (altivec_expand_stvex_be): New function.
3149 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
3151 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
3152 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
3153 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
3154 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
3156 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
3159 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
3160 instead of emit_move_insn.
3162 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3164 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
3166 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
3167 gen_altivec_vsumsws.
3169 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3171 * config/rs6000/altivec.md (altivec_lvxl): Rename as
3172 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
3173 (altivec_lvxl_<mode>): New define_expand incorporating
3174 -maltivec=be semantics where needed.
3175 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
3176 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
3177 semantics where needed.
3178 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
3179 (altivec_stvx_<mode>): New define_expand incorporating
3180 -maltivec=be semantics where needed.
3181 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
3182 VM2 iterator instead of V4SI.
3183 (altivec_stvxl_<mode>): New define_expand incorporating
3184 -maltivec=be semantics where needed.
3185 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
3186 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
3187 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
3188 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
3189 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
3190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
3191 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
3192 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
3193 ALTIVEC_BUILTIN_STVXL.
3194 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
3195 (altivec_expand_stvx_be): Likewise.
3196 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
3197 (altivec_expand_lvx_be): Likewise.
3198 (altivec_expand_stvx_be): Likewise.
3199 (altivec_expand_builtin): Add cases for
3200 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
3201 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
3202 (altivec_init_builtins): Add definitions for
3203 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
3204 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
3206 2014-02-21 Catherine Moore <clm@codesourcery.com>
3208 * doc/invoke.texi (mvirt, mno-virt): Document.
3209 * config/mips/mips.opt (mvirt): New option.
3210 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
3212 2014-02-21 Richard Biener <rguenther@suse.de>
3214 PR tree-optimization/60276
3215 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
3216 (STMT_VINFO_MIN_NEG_DIST): New macro.
3217 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
3218 STMT_VINFO_MIN_NEG_DIST.
3219 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
3220 made for negative dependence distances still hold.
3222 2014-02-21 Richard Biener <rguenther@suse.de>
3225 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
3226 DECL_INITIAL for globals not in the current function context.
3228 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3230 PR tree-optimization/56490
3231 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
3232 * tree-ssa-uninit.c: Include params.h.
3233 (compute_control_dep_chain): Add num_calls argument, return false
3234 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
3235 num_calls to recursive call.
3236 (find_predicates): Change dep_chain into normal array,
3237 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
3238 variable and adjust compute_control_dep_chain caller.
3239 (find_def_preds): Likewise.
3241 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
3243 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
3244 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
3246 2014-02-21 Nick Clifton <nickc@redhat.com>
3248 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
3249 (pushhi1): Likewise.
3250 (popqi1): Add mode to pre_dec.
3253 2014-02-21 Jakub Jelinek <jakub@redhat.com>
3255 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
3256 mode for mask of V8SFmode permutation.
3258 2014-02-20 Richard Henderson <rth@redhat.com>
3261 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
3262 a new pseudo for OLDVAL.
3264 2014-02-20 Jakub Jelinek <jakub@redhat.com>
3267 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
3268 gen_reg_rtx if d->testing_p.
3269 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
3270 if d->testing_p and we will certainly return true.
3271 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
3274 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
3276 * emit-rtl.c (gen_reg_rtx): Assert that
3277 crtl->emit.regno_pointer_align_length is non-zero.
3279 2014-02-20 Richard Henderson <rth@redhat.com>
3282 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
3283 on failure the store back into EXPECT.
3285 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
3286 Sandra Loosemore <sandra@codesourcery.com>
3288 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
3289 * config/nios2/nios2.c (nios2_function_profiler): Add
3290 -fPIC (flag_pic == 2) support.
3291 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
3292 (nios2_large_offset_p): New function.
3293 (nios2_unspec_reloc_p): Move up position, update to use
3294 nios2_large_offset_p.
3295 (nios2_unspec_address): Remove function.
3296 (nios2_unspec_offset): New function.
3297 (nios2_large_got_address): New function.
3298 (nios2_got_address): Add large offset support.
3299 (nios2_legitimize_tls_address): Update usage of removed and new
3301 (nios2_symbol_binds_local_p): New function.
3302 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
3303 (nios2_legitimize_address): Update to use nios2_large_offset_p.
3304 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
3305 (nios2_print_operand): Merge H/L processing, add hiadj/lo
3306 processing for (const (unspec ...)).
3307 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
3309 2014-02-20 Richard Biener <rguenther@suse.de>
3311 * tree-cfg.c (replace_uses_by): Mark altered BBs before
3312 doing the substitution.
3313 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
3315 2014-02-20 Martin Jambor <mjambor@suse.cz>
3318 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
3319 info when checking whether lattices are bottom.
3321 2014-02-20 Richard Biener <rguenther@suse.de>
3324 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
3327 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
3330 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
3331 parameter specifying the scaling.
3332 (inline_call): Update.
3333 (want_inline_recursively): Guard division by zero.
3334 (recursive_inlining): Update.
3335 * ipa-inline.h (clone_inlined_nodes): Update.
3337 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3340 * config/i386/i386.c (classify_argument): Pass structures of size
3341 64 bytes or less in register.
3343 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
3344 Kirill Yukhin <kirill.yukhin@intel.com>
3346 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
3347 (_mm_rcp28_round_ss): Ditto.
3348 (_mm_rsqrt28_round_sd): Ditto.
3349 (_mm_rsqrt28_round_ss): Ditto.
3350 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
3351 (_mm_rcp14_round_ss): Ditto.
3352 (_mm_rsqrt14_round_sd): Ditto.
3353 (_mm_rsqrt14_round_ss): Ditto.
3354 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
3355 the first input operand, get rid of match_dup.
3356 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
3358 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3360 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
3361 operand as the first input operand, set type attribute.
3362 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3364 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
3365 operand as the first input operand, set type attribute.
3367 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3369 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
3372 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
3375 * config/i386/i386.c (construct_container): Remove TFmode check
3376 for X86_64_INTEGER_CLASS.
3378 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
3381 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
3382 only when -Wpsabi is enabled.
3384 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
3387 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
3388 passing arrays in registers are the same as for structs, so remove the
3389 special case for them.
3391 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
3393 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
3394 destination type, extract only the valid bits if the source type is not
3395 integral and has a different mode.
3397 2014-02-19 Richard Biener <rguenther@suse.de>
3400 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
3403 2014-02-19 Richard Biener <rguenther@suse.de>
3406 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
3407 (ipa_modify_call_arguments): Emit an argument load explicitely and
3408 preserve virtual SSA form there and for the replacement call.
3409 Do not update SSA form nor free dominance info.
3411 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3413 * ipa.c (function_and_variable_visibility): Also clear WEAK
3414 flag when disolving COMDAT_GROUP.
3416 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3418 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
3419 * ipa-prop.c (ipa_set_jf_known_type): Return early when
3421 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
3422 do more sanity checks.
3423 (detect_type_change): Return true when giving up early.
3424 (compute_complex_assign_jump_func): Fix type parameter of
3425 ipa_set_ancestor_jf.
3426 (compute_complex_ancestor_jump_func): Likewise.
3427 (update_jump_functions_after_inlining): Fix updating of
3429 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
3431 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3433 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
3434 inline clones when edge disappears.
3436 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3439 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
3440 Split 64-bit moves into 2 patterns. Do not allow the use of
3441 direct move for TDmode in little endian, since the decimal value
3442 has little endian bytes within a word, but the 64-bit pieces are
3443 ordered in a big endian fashion, and normal subreg's of TDmode are
3445 (mov<mode>_64bit_dm): Likewise.
3446 (movtd_64bit_nodm): Likewise.
3448 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3450 PR tree-optimization/60174
3451 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
3452 statement of an SSA_NAME that occurs in an abnormal PHI node.
3454 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3457 * final.c (SEEN_BB): Remove.
3458 (SEEN_NOTE, SEEN_EMITTED): Renumber.
3459 (final_scan_insn): Don't force_source_line on second
3460 NOTE_INSN_BASIC_BLOCK.
3462 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
3465 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
3466 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
3467 (type_natural_mode): Warn ABI change when %zmm register is not
3468 available for AVX512F vector value passing.
3470 2014-02-18 Kai Tietz <ktietz@redhat.com>
3473 * config/i386/i386.c (ix86_expand_prologue): Use value in
3474 rax register as displacement when restoring %r10 or %rax.
3475 Fix wrong offset when restoring both registers.
3477 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
3479 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
3480 assertion with conditional return.
3482 2014-02-18 Jakub Jelinek <jakub@redhat.com>
3483 Uros Bizjak <ubizjak@gmail.com>
3486 * config/i386/driver-i386.c (host_detect_local_cpu): If
3487 YMM state is not saved by the OS, also clear has_f16c. Move
3488 CPUID 0x80000001 handling before YMM state saving checking.
3490 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
3492 PR rtl-optimization/58960
3493 * haifa-sched.c (alloc_global_sched_pressure_data): New,
3494 factored out from ...
3495 (sched_init): ... here.
3496 (free_global_sched_pressure_data): New, factored out from ...
3497 (sched_finish): ... here.
3498 * sched-int.h (free_global_sched_pressure_data): Declare.
3499 * sched-rgn.c (nr_regions_initial): New static global.
3500 (haifa_find_rgns): Initialize it.
3501 (schedule_region): Disable sched-pressure for the newly
3504 2014-02-17 Richard Biener <rguenther@suse.de>
3506 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
3507 release SSA defs of pattern stmts.
3509 2014-02-17 Richard Biener <rguenther@suse.de>
3511 * tree-inline.c (expand_call_inline): Release the virtual
3512 operand defined by the call we are about to inline.
3514 2014-02-17 Richard Biener <rguenther@suse.de>
3516 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
3518 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
3519 Ilya Tocar <ilya.tocar@intel.com>
3521 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
3522 arguments order in builtin.
3523 (_mm512_permutexvar_epi64): Ditto.
3524 (_mm512_mask_permutexvar_epi64): Ditto
3525 (_mm512_maskz_permutexvar_epi32): Ditto
3526 (_mm512_permutexvar_epi32): Ditto
3527 (_mm512_mask_permutexvar_epi32): Ditto
3529 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3531 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
3532 (p8_vmrgow): Likewise.
3534 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3536 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
3539 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3542 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
3543 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
3544 into 64-bit and 32-bit moves. On 64-bit moves, add support for
3545 using direct move instructions on ISA 2.07. Also adjust
3546 instruction length for 64-bit.
3547 (mov<mode>_64bit, TFmode/TDmode): Likewise.
3548 (mov<mode>_32bit, TFmode/TDmode): Likewise.
3550 2014-02-15 Alan Modra <amodra@gmail.com>
3554 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
3555 find_replacement on parts of insn rtl that might be reloaded.
3557 2014-02-15 Richard Biener <rguenther@suse.de>
3559 PR tree-optimization/60183
3560 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
3561 (tree_ssa_phiprop): Calculate and free post-dominators.
3563 2014-02-14 Jeff Law <law@redhat.com>
3565 PR rtl-optimization/60131
3566 * ree.c (get_extended_src_reg): New function.
3567 (combine_reaching_defs): Use it rather than assuming location of REG.
3568 (find_and_remove_re): Verify first operand of extension is
3569 a REG before adding the insns to the copy list.
3571 2014-02-14 Roland McGrath <mcgrathr@google.com>
3573 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
3574 * configure: Regenerated.
3575 * config.in: Regenerated.
3576 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
3577 instead of ASM_SHORT.
3579 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
3580 Richard Earnshaw <rearnsha@arm.com>
3582 PR rtl-optimization/59535
3583 * lra-constraints.c (process_alt_operands): Encourage alternative
3584 when unassigned pseudo class is superset of the alternative class.
3585 (inherit_reload_reg): Don't inherit when optimizing for code size.
3586 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
3587 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
3588 modes not less than 4 for Thumb1.
3590 2014-02-14 Kyle McMartin <kyle@redhat.com>
3593 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
3595 2014-02-14 Richard Biener <rguenther@suse.de>
3597 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
3598 (get_frame_arg): Drop the assert with langhook types_compatible_p.
3599 Do not strip INDIRECT_REFs.
3601 2014-02-14 Richard Biener <rguenther@suse.de>
3604 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
3605 DECL_FUNCTION_SPECIFIC_TARGET.
3606 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
3607 * tree-streamer-out.c (pack_ts_target_option): Remove.
3608 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
3609 (write_ts_function_decl_tree_pointers): Do not stream
3610 DECL_FUNCTION_SPECIFIC_TARGET.
3611 * tree-streamer-in.c (unpack_ts_target_option): Remove.
3612 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
3613 (lto_input_ts_function_decl_tree_pointers): Do not stream
3614 DECL_FUNCTION_SPECIFIC_TARGET.
3616 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3618 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3619 (get_initial_def_for_induction, vectorizable_induction): Ignore
3620 debug stmts when looking for exit_phi.
3621 (vectorizable_live_operation): Fix up condition.
3623 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3625 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3626 nreverse() because it changes the content of original tree list.
3628 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3630 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3631 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3633 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3635 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3636 GNU coding standards.
3638 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3641 * dwarf2out.c (gen_subprogram_die): Don't call
3642 add_calling_convention_attribute if subr_die is old_die.
3644 2014-02-13 Sharad Singhai <singhai@google.com>
3646 * doc/optinfo.texi: Fix order of nodes.
3648 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3650 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3651 operands[2], not operands[3].
3653 2014-02-13 Richard Biener <rguenther@suse.de>
3656 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3657 mention the possibility of an in-tree build.
3658 (CLooG): Update recommended version to 0.18.1, mention the
3659 possibility of an in-tree build and clarify that the ISL
3660 bundled with CLooG does not work.
3662 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3665 * expr.c (compress_float_constant): If x is a hard register,
3666 extend into a pseudo and then move to x.
3668 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3670 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3671 caused by bad second argument to warning_at() with -mhotpatch and
3672 nested functions (e.g. with gfortran).
3674 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3676 * opts.c (option_name): Remove "enabled by default" rider.
3678 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3680 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3682 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3683 Uros Bizjak <ubizjak@gmail.com>
3686 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3687 * configure: Regenerated.
3689 2014-02-12 Richard Biener <rguenther@suse.de>
3691 * vec.c (vec_prefix::calculate_allocation): Move as
3692 inline variant to vec.h.
3693 (vec_prefix::calculate_allocation_1): New out-of-line version.
3694 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3695 (vec_prefix::m_has_auto_buf): Rename to ...
3696 (vec_prefix::m_using_auto_storage): ... this.
3697 (vec_prefix::calculate_allocation): Inline the easy cases
3698 and dispatch to calculate_allocation_1 which doesn't need the
3700 (va_heap::reserve): Use gcc_checking_assert.
3701 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3702 m_using_auto_storage.
3703 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3705 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3706 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3707 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3709 2014-02-12 Richard Biener <rguenther@suse.de>
3711 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3712 when we found a dependence.
3714 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3716 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3718 (maybe_fold_stmt): ... into this new function.
3719 * omp-low.c (lower_omp): Update comment.
3721 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3724 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3727 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3729 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3730 identifiers in comments.
3731 (cortexa53_extra_costs): Likewise.
3732 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3733 (cortexa7_extra_costs): Likewise.
3734 (cortexa12_extra_costs): Likewise.
3735 (cortexa15_extra_costs): Likewise.
3736 (v7m_extra_costs): Likewise.
3738 2014-02-12 Richard Biener <rguenther@suse.de>
3741 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3742 of posix_memalign being successful.
3743 (lower_stmt): Restrict lowering of posix_memalign to when
3744 -ftree-bit-ccp is enabled.
3746 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3748 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3750 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3752 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3754 PR rtl-optimization/60116
3755 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3756 other_insn once the combination has been validated.
3758 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3761 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3763 * ipa-devirt.c: Include demangle.h
3764 (odr_violation_reported): New static variable.
3765 (add_type_duplicate): Update odr_violations.
3766 (maybe_record_node): Add completep parameter; update it.
3767 (record_target_from_binfo): Add COMPLETEP parameter;
3768 update it as needed.
3769 (possible_polymorphic_call_targets_1): Likewise.
3770 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3771 rename FINAL to COMPLETE.
3772 (record_targets_from_bases): Sanity check we found the binfo;
3773 fix COMPLETEP updating.
3774 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3775 parameter, fix computing of COMPLETEP.
3776 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3777 at LTO time do demangling.
3778 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3779 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3781 (gimple_get_virt_method_for_binfo): Likewise.
3782 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3783 gimple_get_virt_method_for_vtable): Update prototypes.
3785 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3788 * genautomata.c (add_presence_absence): Fix typo with
3789 {final_}presence_list.
3791 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3794 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3795 for VSX/Altivec vectors that land in GPR registers.
3797 2014-02-11 Richard Henderson <rth@redhat.com>
3798 Jakub Jelinek <jakub@redhat.com>
3801 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3802 around drhs if type conversion to lacc->type is not useless.
3804 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3806 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3808 (cortex-a57.cortex-a53): Likewise.
3809 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3811 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3813 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3816 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3818 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3819 add_options_for_arm_vfp3.
3821 2014-02-11 Jeff Law <law@redhat.com>
3824 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3825 object with an undesirable mode.
3827 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3830 * config/i386/sol2-9.h: New file.
3831 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3832 *-*-solaris2.9*): Use it.
3834 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3836 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3837 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3839 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3841 * config/microblaze/microblaze.c: Extend mcpu version format
3843 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3845 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3847 2014-02-10 Richard Henderson <rth@redhat.com>
3850 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3851 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3852 ms-abi vs -mno-accumulate-outgoing-args.
3853 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3854 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3857 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3860 * cfgexpand.c (expand_asm_operands): Attach source location to
3861 ASM_INPUT rtx objects.
3862 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3864 2014-02-10 Nick Clifton <nickc@redhat.com>
3866 * config/mn10300/mn10300.c (popcount): New function.
3867 (mn10300_expand_prologue): Include saved registers in stack usage
3870 2014-02-10 Jeff Law <law@redhat.com>
3873 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3874 when changing the SET_DEST of a prior insn to avoid an input reload.
3876 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3878 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3879 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3880 -mcall-openbsd, or -mcall-linux.
3881 (CC1_ENDIAN_BIG_SPEC): Remove.
3882 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3883 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3884 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3885 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3886 and %cc1_endian_default.
3887 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3889 2014-02-10 Richard Biener <rguenther@suse.de>
3891 PR tree-optimization/60115
3892 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3893 MEM_REF handling. Properly verify that the accesses are not
3894 out of the objects bound.
3896 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3898 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3901 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3903 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3904 proper constants and fix formatting.
3905 (possible_polymorphic_call_targets): Fix formatting.
3907 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3908 Ilya Tocar <ilya.tocar@intel.com>
3910 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3911 (_mm512_loadu_epi32): Renamed into...
3912 (_mm512_loadu_si512): This.
3913 (_mm512_storeu_epi32): Renamed into...
3914 (_mm512_storeu_si512): This.
3915 (_mm512_maskz_ceil_ps): Removed.
3916 (_mm512_maskz_ceil_pd): Ditto.
3917 (_mm512_maskz_floor_ps): Ditto.
3918 (_mm512_maskz_floor_pd): Ditto.
3919 (_mm512_floor_round_ps): Ditto.
3920 (_mm512_floor_round_pd): Ditto.
3921 (_mm512_ceil_round_ps): Ditto.
3922 (_mm512_ceil_round_pd): Ditto.
3923 (_mm512_mask_floor_round_ps): Ditto.
3924 (_mm512_mask_floor_round_pd): Ditto.
3925 (_mm512_mask_ceil_round_ps): Ditto.
3926 (_mm512_mask_ceil_round_pd): Ditto.
3927 (_mm512_maskz_floor_round_ps): Ditto.
3928 (_mm512_maskz_floor_round_pd): Ditto.
3929 (_mm512_maskz_ceil_round_ps): Ditto.
3930 (_mm512_maskz_ceil_round_pd): Ditto.
3931 (_mm512_expand_pd): Ditto.
3932 (_mm512_expand_ps): Ditto.
3933 * config/i386/i386.c (ix86_builtins): Remove
3934 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3935 (bdesc_args): Ditto.
3936 * config/i386/predicates.md (const1256_operand): New.
3937 (const_1_to_2_operand): Ditto.
3938 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3939 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3940 (*avx512pf_gatherpf<mode>sf): Ditto.
3941 (avx512pf_gatherpf<mode>df): Ditto.
3942 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3943 (*avx512pf_gatherpf<mode>df): Ditto.
3944 (avx512pf_scatterpf<mode>sf): Ditto.
3945 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3946 (*avx512pf_scatterpf<mode>sf): Ditto.
3947 (avx512pf_scatterpf<mode>df): Ditto.
3948 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3949 (*avx512pf_scatterpf<mode>df): Ditto.
3950 (avx512f_expand<mode>): Removed.
3951 (<shift_insn><mode>3<mask_name>): Change predicate type.
3953 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3955 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3956 not at the end of datarefs vector use ordered_remove to avoid
3957 reordering datarefs vector.
3960 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3961 mark local addressable non-static vars as GOVD_PRIVATE
3962 instead of GOVD_LOCAL.
3963 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3964 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3968 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3969 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3970 assume_aligned or alloc_align attributes.
3971 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3972 arguments. Handle also assume_aligned and alloc_align attributes.
3973 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3974 calls to functions with assume_aligned or alloc_align attributes.
3975 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3977 2014-02-08 Terry Guo <terry.guo@arm.com>
3979 * doc/invoke.texi: Document ARM -march=armv7e-m.
3981 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3983 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3984 flag on __cilkrts_rethrow builtin.
3987 * ipa-cp.c (determine_versionability): Fail at -O0
3988 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3989 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3992 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3995 * tree-inline.c (copy_forbidden): Fail for
3996 __attribute__((optimize (0))) functions.
3998 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4000 * varpool.c: Include pointer-set.h.
4001 (varpool_remove_unreferenced_decls): Variables in other partitions
4002 will not be output; be however careful to not lose information
4005 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
4007 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
4008 lookup in the vtable constructor.
4010 2014-02-07 Jeff Law <law@redhat.com>
4013 * config/m68k/m68k.md (ashldi_extsi): Turn into a
4014 define_insn_and_split.
4016 * ipa-inline.c (inline_small_functions): Fix typos.
4018 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4020 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
4021 (s390_can_use_return_insn): Declare.
4022 * config/s390/s390.h (EPILOGUE_USES): Define.
4023 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
4025 (s390_chunkify_start): Handle return JUMP_LABELs.
4026 (s390_early_mach): Emit a main_pool instruction on the entry edge.
4027 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
4028 (s390_can_use_return_insn): New functions.
4029 (s390_fix_long_loop_prediction): Handle conditional returns.
4030 (TARGET_SET_UP_BY_PROLOGUE): Define.
4031 * config/s390/s390.md (ANY_RETURN): New code iterator.
4032 (*creturn, *csimple_return, return, simple_return): New patterns.
4034 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4036 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
4037 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
4038 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
4039 REG_CFA_RESTORE list when deciding not to restore a register.
4041 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4043 * config/s390/s390.c: Include tree-pass.h and context.h.
4044 (s390_early_mach): New function, split out from...
4045 (s390_emit_prologue): ...here.
4046 (pass_data_s390_early_mach): New pass structure.
4047 (pass_s390_early_mach): New class.
4048 (s390_option_override): Create and register early_mach pass.
4049 Move to end of file.
4051 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4053 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
4054 to match for the exit block.
4056 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4058 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
4059 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
4060 Reject misaligned operands.
4062 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4064 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
4066 2014-02-07 Richard Biener <rguenther@suse.de>
4069 * gimple-low.c (lower_builtin_posix_memalign): New function.
4070 (lower_stmt): Call it to lower posix_memalign in a way
4071 to make alignment info accessible.
4073 2014-02-07 Jakub Jelinek <jakub@redhat.com>
4076 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
4077 __builtin_setjmp_receiver.
4079 2014-02-07 Richard Biener <rguenther@suse.de>
4082 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
4083 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
4084 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4085 Handle BUILT_IN_POSIX_MEMALIGN.
4086 (find_func_clobbers): Likewise.
4087 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4088 (call_may_clobber_ref_p_1): Likewise.
4090 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4093 * ipa-devirt.c (record_target_from_binfo): Remove overactive
4096 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4099 * lto-cgraph.c (lto_output_node): Use
4100 symtab_get_symbol_partitioning_class.
4101 (lto_output_varpool_node): likewise.
4102 (symtab_get_symbol_partitioning_class): Move here from
4104 * cgraph.h (symbol_partitioning_class): Likewise.
4105 (symtab_get_symbol_partitioning_class): Declare.
4107 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4109 * ggc.h (ggc_internal_cleared_alloc): New macro.
4110 * vec.h (vec_safe_copy): Handle memory stats.
4111 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
4112 * target-globals.c (save_target_globals): Likewise.
4114 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4117 * expr.c (emit_move_resolve_push): Export; be bit more selective
4118 on when to clear alias set.
4119 * expr.h (emit_move_resolve_push): Declare.
4120 * function.h (struct function): Add tail_call_marked.
4121 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
4122 * config/i386/i386-protos.h (ix86_expand_push): Remove.
4123 * config/i386/i386.md (TImode move expander): De not call
4125 (FP push expanders): Preserve memory attributes.
4126 * config/i386/sse.md (push<mode>1): Remove.
4127 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
4128 (ix86_expand_push): Remove.
4129 * config/i386/mmx.md (push<mode>1): Remove.
4131 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4133 PR rtl-optimization/60030
4134 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
4135 lopart with paradoxical subreg before shifting it up by hprec.
4137 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4139 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
4140 Remove extra newline at end of file.
4141 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
4142 (arm_issue_rate): Handle cortexa57.
4143 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
4144 (cortex-a57.cortex-a53): Likewise.
4146 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4149 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
4150 don't record in REG_FRAME_RELATED_EXPR registers not set in that
4152 (arm_expand_prologue): Adjust all callers.
4153 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
4154 info, registers also at the lowest numbered registers side. Use
4155 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
4159 * var-tracking.c (adjust_mems): Before adding a SET to
4160 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
4162 2014-02-06 Alan Modra <amodra@gmail.com>
4165 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
4166 change SDmode to DDmode when lra_in_progress.
4168 2014-02-06 Jakub Jelinek <jakub@redhat.com>
4171 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
4172 free_data_ref on the dr first, and before goto again also set dr
4173 to the next dr. For simd_lane_access, free old datarefs[i] before
4174 overwriting it. For get_vectype_for_scalar_type failure, don't
4175 free_data_ref if simd_lane_access.
4177 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
4180 * tree.h (opts_for_fn): New inline function.
4181 (opt_for_fn): Define.
4182 * config/i386/i386.c (ix86_function_regparm): Use
4183 opt_for_fn (decl, optimize) instead of optimize.
4185 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
4187 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4188 for SYMBOL_REF in large memory model.
4190 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4192 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
4194 (cortex-a57): Likewise.
4195 (cortex-a57.cortex-a53): Likewise.
4197 2014-02-06 Yury Gribov <y.gribov@samsung.com>
4198 Kugan Vivekanandarajah <kuganv@linaro.org>
4200 * config/arm/arm.c (arm_vector_alignment_reachable): Check
4202 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
4204 2014-02-06 Richard Biener <rguenther@suse.de>
4206 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
4207 set_loop_copy and initialize_original_copy_tables.
4209 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
4211 * config/aarch64/aarch64-simd.md
4212 (aarch64_ashr_simddi): Change QI to SI.
4214 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4215 Jakub Jelinek <jakub@redhat.com>
4218 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
4221 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4223 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
4224 CODE_FOR_altivec_vpku[hw]um to
4225 CODE_FOR_altivec_vpku[hw]um_direct.
4226 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
4227 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
4228 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
4229 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
4231 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4233 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
4234 generation for -maltivec=be.
4235 (altivec_vsumsws): Simplify redundant test.
4237 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4239 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
4240 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
4241 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
4242 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
4243 gen_altivec_vpkuwum.
4244 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
4246 (altivec_vpks<VI_char>ss): Likewise.
4247 (altivec_vpks<VI_char>us): Likewise.
4248 (altivec_vpku<VI_char>us): Likewise.
4249 (altivec_vpku<VI_char>um): Likewise.
4250 (altivec_vpku<VI_char>um_direct): New (copy of
4251 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
4253 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
4254 target is little endian and -maltivec=be is not specified.
4255 (*altivec_vupkhs<VU_char>_direct): New (copy of
4256 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
4257 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
4258 target is little endian and -maltivec=be is not specified.
4259 (*altivec_vupkls<VU_char>_direct): New (copy of
4260 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
4261 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
4262 little endian and -maltivec=be is not specified.
4263 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
4264 little endian and -maltivec=be is not specified.
4266 2014-02-05 Richard Henderson <rth@redhat.com>
4269 * combine-stack-adj.c: Revert r206943.
4270 * sched-int.h (struct deps_desc): Add last_args_size.
4271 * sched-deps.c (init_deps): Initialize it.
4272 (sched_analyze_insn): Add OUTPUT dependencies between insns that
4273 contain REG_ARGS_SIZE notes.
4275 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
4277 * lto-cgraph.c (asm_nodes_output): Make global.
4278 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
4279 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
4280 (driver_handle_option): Handle OPT_fwpa.
4282 2014-02-05 Jakub Jelinek <jakub@redhat.com>
4285 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
4286 a comment typo and formatting issue. If odr_hash hasn't been
4287 created, return vNULL and set *completep to false.
4290 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
4291 bb with no successors.
4293 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
4296 * doc/invoke.texi (-march): Clarify documentation for ARM.
4300 2014-02-05 Richard Biener <rguenther@suse.de>
4302 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
4303 when not vectorizing because of too many alias checks.
4304 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
4305 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
4307 2014-02-05 Nick Clifton <nickc@redhat.com>
4309 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
4310 accept extended registers in any mode when compiling for the MN10300.
4312 2014-02-05 Yury Gribov <y.gribov@samsung.com>
4314 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
4315 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
4316 sanitization attributes.
4317 (can_inline_edge_p): Likewise.
4318 (sanitize_attrs_match_for_inline_p): New function.
4320 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4322 * ipa-prop.c (detect_type_change): Shor circuit testing of
4323 type changes on THIS pointer.
4325 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
4328 * config/pa/pa.c (legitimize_tls_address): Return original address
4329 if not passed a SYMBOL_REF rtx.
4330 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
4332 (pa_emit_move_sequence): Simplify TLS source operands.
4333 (pa_legitimate_constant_p): Reject all TLS constants.
4334 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
4335 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
4337 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4339 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
4340 groups when we know they are controlled by LTO.
4341 * varasm.c (default_binds_local_p_1): If object is in other partition,
4342 it will be resolved locally.
4344 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4346 * config/host-linux.c (linux_gt_pch_use_address): Don't
4347 use SSIZE_MAX because it is not always defined.
4349 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
4352 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
4353 threshold for pseudo splitting.
4354 (update_ebb_live_info): Process call argument hard registers and
4355 hard registers from insn definition too.
4356 (max_small_class_regs_num): New constant.
4357 (inherit_in_ebb): Update live hard regs through EBBs. Update
4358 reloads_num only for small register classes. Don't split for
4361 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
4364 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
4367 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
4369 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
4372 2014-02-04 Marek Polacek <polacek@redhat.com>
4374 * gdbinit.in (pel): Define.
4376 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4378 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
4381 2014-02-04 Richard Biener <rguenther@suse.de>
4384 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
4385 in function context local.
4386 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
4387 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
4388 similar to LTO_imported_decl_ref.
4390 2014-02-04 Jakub Jelinek <jakub@redhat.com>
4392 PR tree-optimization/60002
4393 * cgraphclones.c (build_function_decl_skip_args): Clear
4396 PR tree-optimization/60023
4397 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
4398 false to gsi_replace.
4399 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
4400 has been in some EH region and vec_stmt could throw, add
4401 vec_stmt into the same EH region.
4402 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
4403 has no lhs, ignore it.
4404 * internal-fn.c (expand_MASK_LOAD): Likewise.
4407 * tree-inline.c (copy_forbidden): Fail for
4408 __attribute__((optimize (0))) functions.
4411 * omp-low.c (simd_clone_struct_copy): If from->inbranch
4412 is set, copy one less argument.
4413 (expand_simd_clones): Don't subtract clone_info->inbranch
4414 from simd_clone_struct_alloc argument.
4416 PR rtl-optimization/57915
4417 * recog.c (simplify_while_replacing): If all unary/binary/relational
4418 operation arguments are constant, attempt to simplify those.
4421 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
4422 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
4424 2014-02-04 Richard Biener <rguenther@suse.de>
4426 PR tree-optimization/60012
4427 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
4428 TBAA disambiguation to all DDRs.
4430 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4433 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
4434 (LINK_SPEC): Use it for -shared, -shared-libgcc.
4436 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4439 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
4441 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4443 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
4444 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
4446 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4449 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
4450 to figure out targets of polymorphic calls with known decl.
4451 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4452 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
4453 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
4454 (get_polymorphic_call_info): ... here.
4455 (get_polymorphic_call_info_from_invariant): New function.
4457 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4459 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
4460 lookup via vtable pointer; check for type consistency
4461 and turn inconsitent facts into UNREACHABLE.
4462 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4463 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
4464 type inconsistent querries; return UNREACHABLE instead.
4466 2014-02-03 Richard Henderson <rth@twiddle.net>
4469 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
4470 already processed this node.
4471 (normalize_one_pred_1): Pass along mark_set.
4472 (normalize_one_pred): Create and destroy a pointer_set_t.
4473 (normalize_one_pred_chain): Likewise.
4475 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
4477 PR gcov-profile/58602
4478 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
4480 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
4483 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
4484 -fno-devirtualize; try to devirtualize by the knowledge of
4485 virtual table pointer given by aggregate propagation.
4486 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
4487 (ipa_print_node_jump_functions): Dump also offset that
4488 is relevant for polymorphic calls.
4489 (determine_known_aggregate_parts): Add arg_type parameter; use it
4490 instead of determining the type from pointer type.
4491 (ipa_compute_jump_functions_for_edge): Update call of
4492 determine_known_aggregate_parts.
4493 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
4494 (gimple_get_virt_method_for_binfo): ... here; simplify using
4495 vtable_pointer_value_to_vtable.
4496 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
4497 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
4498 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
4499 (vtable_pointer_value_to_vtable): Break out from ...; handle also
4501 (vtable_pointer_value_to_binfo): ... here.
4502 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
4504 2014-02-03 Teresa Johnson <tejohnson@google.com>
4506 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
4507 redef of outer loop index variable.
4509 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
4513 * doc/extend.texi (Function Attributes): Typo.
4515 2014-02-03 Cong Hou <congh@google.com>
4517 PR tree-optimization/60000
4518 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
4519 if the vectorized statement is a store. A store statement can only
4520 appear at the end of pattern statements.
4522 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4524 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
4525 (ix86_option_override_internal): Default long double to 64-bit for
4526 32-bit Bionic and to 128-bit for 64-bit Bionic.
4528 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
4529 TARGET_LONG_DOUBLE_128 is true.
4530 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
4532 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
4533 (mlong-double-64): Negate -mlong-double-128.
4534 (mlong-double-128): New option.
4536 * config/i386/i386-c.c (ix86_target_macros): Define
4537 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
4539 * doc/invoke.texi: Document -mlong-double-128.
4541 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
4543 PR rtl-optimization/60024
4544 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
4546 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
4548 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
4550 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4552 PR rtl-optimization/57662
4553 * sel-sched.c (code_motion_path_driver): Do not mark already not
4554 existing blocks in the visiting bitmap.
4556 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
4558 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
4559 on the insn being emitted.
4561 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
4562 Will Deacon <will.deacon@arm.com>
4564 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
4566 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4568 * config/arm/arm-tables.opt: Regenerate.
4570 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4572 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
4573 for vector types other than V16QImode.
4574 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
4575 define_expand, and call altivec_expand_vec_perm_le when producing
4576 code with little endian element order.
4577 (*altivec_vperm_<mode>_internal): New insn having previous
4578 behavior of altivec_vperm_<mode>.
4579 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
4580 altivec_expand_vec_perm_le when producing code with little endian
4582 (*altivec_vperm_<mode>_uns_internal): New insn having previous
4583 behavior of altivec_vperm_<mode>_uns.
4585 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4587 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
4588 (altivec_vsumsws): Add handling for -maltivec=be with a little
4590 (altivec_vsumsws_direct): New.
4591 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
4592 gen_altivec_vsumsws.
4594 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4596 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
4597 vtable_pointer_value_to_binfo): New functions.
4598 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
4599 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
4601 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
4603 * config/nios2/nios2.md (load_got_register): Initialize GOT
4604 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
4605 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
4607 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
4609 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
4610 preserverd by passthrough, do not propagate the type.
4612 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4614 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
4615 (mips_atomic_assign_expand_fenv): New function.
4616 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4618 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4620 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4621 (__builtin_mips_set_fcsr): Likewise.
4622 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4623 MIPS_USI_FTYPE_VOID.
4624 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4625 (mips16_expand_set_fcsr): Likewise.
4626 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4627 (mips16_set_fcsr_stub): Likewise.
4628 (mips16_get_fcsr_one_only_stub): New class.
4629 (mips16_set_fcsr_one_only_stub): Likewise.
4630 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4631 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4632 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4633 (hard_float): New availability predicate.
4634 (mips_builtins): Add get_fcsr and set_fcsr.
4635 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4636 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4637 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4638 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4639 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4642 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4644 * config/mips/mips.c (mips_one_only_stub): New class.
4645 (mips_need_mips16_rdhwr_p): Replace with...
4646 (mips16_rdhwr_stub): ...this new variable.
4647 (mips16_stub_call_address): New function.
4648 (mips16_rdhwr_one_only_stub): New class.
4649 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4650 (mips_output_mips16_rdhwr): Delete.
4651 (mips_finish_stub): New function.
4652 (mips_code_end): Use it to handle rdhwr stubs.
4654 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4657 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4658 when calculating size of integer atomic types.
4660 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4662 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4664 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4666 PR tree-optimization/60003
4667 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4668 * profile.c (branch_prob): Use gimple_call_builtin_p
4669 to check for BUILT_IN_SETJMP_RECEIVER.
4670 * tree-inline.c (copy_bb): Call notice_special_calls.
4672 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4675 * lra-constraints.c (process_alt_operands): Update reload_sum only
4678 2014-01-31 Richard Henderson <rth@redhat.com>
4681 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4682 until after else_eh is processed.
4684 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4686 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4687 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4688 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4689 in smmintrin.h, remove them.
4690 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4691 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4692 * config/i386/i386.md (ROUND_SAE): Fix value.
4693 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4694 (const48_operand): New.
4695 * config/i386/subst.md (round), (round_expand): Use
4696 const_4_or_8_to_11_operand.
4697 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4699 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4701 * config/i386/constraints.md (Yk): Swap meaning with k.
4702 * config/i386/i386.md (movhi_internal): Change Yk to k.
4703 (movqi_internal): Ditto.
4704 (*k<logic><mode>): Ditto.
4707 (kandn<mode>): Ditto.
4708 (*<code>hi_1): Ditto.
4709 (*<code>qi_1): Ditto.
4710 (kxnor<mode>): Ditto.
4711 (kortestzhi): Ditto.
4712 (kortestchi): Ditto.
4714 (*one_cmplhi2_1): Ditto.
4715 (*one_cmplqi2_1): Ditto.
4716 * config/i386/sse.md (): Change k to Yk.
4717 (avx512f_load<mode>_mask): Ditto.
4718 (avx512f_blendm<mode>): Ditto.
4719 (avx512f_store<mode>_mask): Ditto.
4720 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4721 (avx512f_storedqu<mode>_mask): Ditto.
4722 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4724 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4725 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4726 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4727 (avx512f_maskcmp<mode>3): Ditto.
4728 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4729 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4730 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4731 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4732 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4733 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4734 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4735 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4736 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4737 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4738 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4739 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4740 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4741 (vec_extract_lo_<mode>_maskm): Ditto.
4742 (vec_extract_hi_<mode>_maskm): Ditto.
4743 (avx512f_vternlog<mode>_mask): Ditto.
4744 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4745 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4746 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4747 (avx512f_<code>v8div16qi2_mask): Ditto.
4748 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4749 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4750 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4751 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4752 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4753 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4754 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4755 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4756 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4757 (avx512cd_maskb_vec_dupv8di): Ditto.
4758 (avx512cd_maskw_vec_dupv16si): Ditto.
4759 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4760 (avx512f_vpermi2var<mode>3_mask): Ditto.
4761 (avx512f_vpermi2var<mode>3_mask): Ditto.
4762 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4763 (*avx512f_gathersi<mode>): Ditto.
4764 (*avx512f_gathersi<mode>_2): Ditto.
4765 (*avx512f_gatherdi<mode>): Ditto.
4766 (*avx512f_gatherdi<mode>_2): Ditto.
4767 (*avx512f_scattersi<mode>): Ditto.
4768 (*avx512f_scatterdi<mode>): Ditto.
4769 (avx512f_compress<mode>_mask): Ditto.
4770 (avx512f_compressstore<mode>_mask): Ditto.
4771 (avx512f_expand<mode>_mask): Ditto.
4772 * config/i386/subst.md (mask): Change k to Yk.
4773 (mask_scalar_merge): Ditto.
4776 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4778 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4780 2014-01-31 Richard Biener <rguenther@suse.de>
4783 * builtins.c (fold_builtin_memory_op): Make sure to not
4784 use a floating-point mode or a boolean or enumeral type for
4787 2014-01-30 DJ Delorie <dj@redhat.com>
4789 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4790 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4791 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4792 whenever main() has an epilogue.
4794 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4796 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4797 unused variable "field".
4798 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4799 (vsx_mergeh_<mode>): Likewise.
4800 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4801 (altivec_vmrghh): Likewise.
4802 (altivec_vmrghw): Likewise.
4803 (altivec_vmrglb): Likewise.
4804 (altivec_vmrglh): Likewise.
4805 (altivec_vmrglw): Likewise.
4806 (altivec_vspltb): Add missing uses.
4807 (altivec_vsplth): Likewise.
4808 (altivec_vspltw): Likewise.
4809 (altivec_vspltsf): Likewise.
4811 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4814 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4815 frame related instructions.
4817 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4819 PR rtl-optimization/59959
4820 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4821 any reload of register whose subreg is invalid.
4823 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4825 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4826 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4827 Add missing return type - void.
4829 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4831 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4832 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4833 remove element index adjustment for endian (now handled in vsx.md
4835 (altivec_expand_vec_perm_const): Use
4836 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4837 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4838 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4839 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4840 define_expand and a new define_insn *altivec_vspltb_internal;
4841 adjust for -maltivec=be on a little endian target.
4842 (altivec_vspltb_direct): New.
4843 (altivec_vsplth): Divide into a define_expand and a new
4844 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4845 little endian target.
4846 (altivec_vsplth_direct): New.
4847 (altivec_vspltw): Divide into a define_expand and a new
4848 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4849 little endian target.
4850 (altivec_vspltw_direct): New.
4851 (altivec_vspltsf): Divide into a define_expand and a new
4852 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4853 a little endian target.
4855 2014-01-30 Richard Biener <rguenther@suse.de>
4857 PR tree-optimization/59993
4858 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4859 can propagate form the earlier stmt and avoid the transform
4860 when the intermediate result is needed.
4862 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4864 * README.Portability: Fix typo.
4866 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4868 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4869 comparison_operator with ordered_comparison_operator.
4871 2014-01-30 Nick Clifton <nickc@redhat.com>
4873 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4874 Rename to mn10300_store_multiple_regs.
4875 * config/mn10300/mn10300.c: Likewise.
4876 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4877 store_multiple_regs.
4878 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4879 Call mn10300_store_multiple_regs.
4881 2014-01-30 Nick Clifton <nickc@redhat.com>
4882 DJ Delorie <dj@redhat.com>
4884 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4885 %fp 2 to keep registers after it properly word-aligned.
4886 (rl78_alloc_physical_registers_umul): Handle the case where both
4887 input operands are the same.
4889 2014-01-30 Richard Biener <rguenther@suse.de>
4891 PR tree-optimization/59903
4892 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4895 2014-01-30 Jason Merrill <jason@redhat.com>
4898 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4901 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4903 2014-01-30 Richard Biener <rguenther@suse.de>
4905 PR tree-optimization/59951
4906 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4908 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4911 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4912 SFmode to DFmode case.
4914 2014-01-29 DJ Delorie <dj@redhat.com>
4916 * config/msp430/msp430.opt (-minrt): New.
4917 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4919 (ENDFILE_SPEC): Likewise.
4921 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4923 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4924 (estimate_function_body_sizes): Use it.
4926 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4929 * dwarf2out.c (is_cxx_auto): New.
4930 (is_base_type): Use it.
4931 (gen_type_die_with_usage): Likewise.
4933 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4935 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4936 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4937 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4938 -maltivec=be with LE targets.
4939 (vsx_mergeh_<mode>): Likewise.
4940 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4941 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4942 (altivec_vmrghb): Replace with define_expand and new
4943 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4944 (altivec_vmrghb_direct): New define_insn.
4945 (altivec_vmrghh): Replace with define_expand and new
4946 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4947 (altivec_vmrghh_direct): New define_insn.
4948 (altivec_vmrghw): Replace with define_expand and new
4949 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4950 (altivec_vmrghw_direct): New define_insn.
4951 (*altivec_vmrghsf): Adjust for endianness.
4952 (altivec_vmrglb): Replace with define_expand and new
4953 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4954 (altivec_vmrglb_direct): New define_insn.
4955 (altivec_vmrglh): Replace with define_expand and new
4956 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4957 (altivec_vmrglh_direct): New define_insn.
4958 (altivec_vmrglw): Replace with define_expand and new
4959 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4960 (altivec_vmrglw_direct): New define_insn.
4961 (*altivec_vmrglsf): Adjust for endianness.
4962 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4963 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4964 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4965 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4966 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4967 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4968 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4969 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4971 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4973 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4974 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4977 2014-01-29 Richard Biener <rguenther@suse.de>
4979 PR tree-optimization/58742
4980 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4981 associate_pointerplus_align.
4982 (associate_pointerplus_diff): New function.
4983 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4984 and associate_pointerplus_diff.
4986 2014-01-29 Richard Biener <rguenther@suse.de>
4988 * lto-streamer.h (LTO_major_version): Bump to 3.
4989 (LTO_minor_version): Reset to 0.
4991 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4993 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4994 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4995 (arm_file_start): Generate correct asm header for armv7ve.
4996 * config/arm/bpabi.h: Add multilib support for armv7ve.
4997 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4998 and cortex-a15 to armv7ve.
4999 * config/arm/t-aprofile: Add multilib support for armv7ve.
5000 * doc/invoke.texi: Document -march=armv7ve.
5002 2014-01-29 Richard Biener <rguenther@suse.de>
5004 PR tree-optimization/58742
5005 * tree-ssa-forwprop.c (associate_plusminus): Return true
5006 if we changed sth, defer EH cleanup to ...
5007 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
5008 (simplify_mult): New function.
5010 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5013 PR tree-optimization/59920
5014 * tree.c (build_common_builtin_nodes): Remove
5015 __builtin_setjmp_dispatcher initialization.
5016 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
5017 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
5018 instead of gsi_after_labels + manually skipping debug stmts.
5019 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
5020 ignore bbs with IFN_ABNORMAL_DISPATCHER.
5021 * tree-inline.c (copy_edges_for_bb): Remove
5022 can_make_abnormal_goto argument, instead add abnormal_goto_dest
5023 argument. Ignore computed_goto_p stmts. Don't call
5024 make_abnormal_goto_edges. If a call might need abnormal edges
5025 for non-local gotos, see if it already has an edge to
5026 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
5027 with true argument, don't do anything then, otherwise add
5028 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
5029 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
5031 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
5032 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
5033 (lower_stmt): Don't set data->calls_builtin_setjmp.
5034 (lower_builtin_setjmp): Adjust comment.
5035 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
5036 * tree-cfg.c (found_computed_goto): Remove.
5037 (factor_computed_gotos): Remove.
5038 (make_goto_expr_edges): Return bool, true for computed gotos.
5039 Don't call make_abnormal_goto_edges.
5040 (build_gimple_cfg): Don't set found_computed_goto, don't call
5041 factor_computed_gotos.
5042 (computed_goto_p): No longer static.
5043 (make_blocks): Don't set found_computed_goto.
5044 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
5045 (make_edges): If make_goto_expr_edges returns true, push bb
5046 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
5047 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
5048 vector. Record mapping between bbs and OpenMP regions if there
5049 are any, adjust make_gimple_omp_edges caller. Call
5050 handle_abnormal_edges.
5051 (make_abnormal_goto_edges): Remove.
5052 * tree-cfg.h (make_abnormal_goto_edges): Remove.
5053 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
5054 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
5055 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
5056 * internal-fn.def (ABNORMAL_DISPATCHER): New.
5057 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
5058 filling *region also set *region_idx to (*region)->entry->index.
5061 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
5062 For REGs set ORIGINAL_REGNO.
5064 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
5066 * doc/md.texi: Mention that a target shouldn't implement
5067 vec_widen_(s|u)mul_even/odd pair if it is less efficient
5070 2014-01-29 Jakub Jelinek <jakub@redhat.com>
5072 PR tree-optimization/59594
5073 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
5074 a copy of the datarefs vector rather than the vector itself.
5076 2014-01-28 Jason Merrill <jason@redhat.com>
5079 * dwarf2out.c (auto_die): New static.
5080 (gen_type_die_with_usage): Handle C++1y 'auto'.
5081 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
5084 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
5087 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
5088 (SPEC_X32): Likewise.
5089 (SPEC_64): Likewise.
5090 * config/i386/i386.c (ix86_option_override_internal): Turn off
5091 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
5093 (x86_file_start): Output .code16gcc for TARGET_16BIT.
5094 * config/i386/i386.h (TARGET_16BIT): New macro.
5095 (TARGET_16BIT_P): Likewise.
5096 * config/i386/i386.opt: Add m16.
5097 * doc/invoke.texi: Document -m16.
5099 2014-01-28 Jakub Jelinek <jakub@redhat.com>
5101 PR preprocessor/59935
5102 * input.c (location_get_source_line): Bail out on when line number
5103 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
5105 2014-01-28 Richard Biener <rguenther@suse.de>
5107 PR tree-optimization/58742
5108 * tree-ssa-forwprop.c (associate_plusminus): Handle
5109 pointer subtraction of the form (T)(P + A) - (T)P.
5111 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5113 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
5116 2014-01-28 Richard Biener <rguenther@suse.de>
5119 2014-01-28 Richard Biener <rguenther@suse.de>
5121 PR rtl-optimization/45364
5122 PR rtl-optimization/59890
5123 * var-tracking.c (local_get_addr_clear_given_value): Handle
5124 already cleared slot.
5125 (val_reset): Handle not allocated local_get_addr_cache.
5126 (vt_find_locations): Use post-order on the inverted CFG.
5128 2014-01-28 Richard Biener <rguenther@suse.de>
5130 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
5132 2014-01-28 Richard Biener <rguenther@suse.de>
5134 PR rtl-optimization/45364
5135 PR rtl-optimization/59890
5136 * var-tracking.c (local_get_addr_clear_given_value): Handle
5137 already cleared slot.
5138 (val_reset): Handle not allocated local_get_addr_cache.
5139 (vt_find_locations): Use post-order on the inverted CFG.
5141 2014-01-28 Alan Modra <amodra@gmail.com>
5143 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
5144 * configure.ac <recursive call for build != host>: Define
5145 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
5146 and LD_FOR_BUILD too.
5147 * configure: Regenerate.
5149 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
5151 * config/i386/i386.c (get_builtin_code_for_version): Separate
5152 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
5153 Broadwell from Haswell.
5155 2014-01-27 Steve Ellcey <sellcey@mips.com>
5157 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
5158 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
5159 * config/mips/mips.c (mips_option_override): Change setting
5161 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
5162 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
5163 Change from Mask to Var.
5165 2014-01-27 Jeff Law <law@redhat.com>
5167 * ipa-inline.c (inline_small_functions): Fix typo.
5169 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
5171 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
5172 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
5173 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
5174 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
5175 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
5176 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
5177 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
5178 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
5179 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
5180 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
5181 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
5182 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
5183 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
5184 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
5185 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
5186 (_mm512_storeu_epi64): Ditto.
5187 (_mm512_cmpge_epi32_mask): Ditto.
5188 (_mm512_cmpge_epu32_mask): Ditto.
5189 (_mm512_cmpge_epi64_mask): Ditto.
5190 (_mm512_cmpge_epu64_mask): Ditto.
5191 (_mm512_cmple_epi32_mask): Ditto.
5192 (_mm512_cmple_epu32_mask): Ditto.
5193 (_mm512_cmple_epi64_mask): Ditto.
5194 (_mm512_cmple_epu64_mask): Ditto.
5195 (_mm512_cmplt_epi32_mask): Ditto.
5196 (_mm512_cmplt_epu32_mask): Ditto.
5197 (_mm512_cmplt_epi64_mask): Ditto.
5198 (_mm512_cmplt_epu64_mask): Ditto.
5199 (_mm512_cmpneq_epi32_mask): Ditto.
5200 (_mm512_cmpneq_epu32_mask): Ditto.
5201 (_mm512_cmpneq_epi64_mask): Ditto.
5202 (_mm512_cmpneq_epu64_mask): Ditto.
5203 (_mm512_expand_pd): Ditto.
5204 (_mm512_expand_ps): Ditto.
5205 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
5206 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
5207 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
5208 * config/i386/i386.c (ix86_builtins): Add
5209 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
5210 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
5211 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
5212 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
5213 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
5214 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
5215 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
5216 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
5217 IX86_BUILTIN_PMOVUSQW512_MEM.
5218 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
5219 __builtin_ia32_pmovsqd512mem_mask,
5220 __builtin_ia32_pmovqd512mem_mask,
5221 __builtin_ia32_pmovusqw512mem_mask,
5222 __builtin_ia32_pmovsqw512mem_mask,
5223 __builtin_ia32_pmovqw512mem_mask,
5224 __builtin_ia32_pmovusdw512mem_mask,
5225 __builtin_ia32_pmovsdw512mem_mask,
5226 __builtin_ia32_pmovdw512mem_mask,
5227 __builtin_ia32_pmovqb512mem_mask,
5228 __builtin_ia32_pmovusqb512mem_mask,
5229 __builtin_ia32_pmovsqb512mem_mask,
5230 __builtin_ia32_pmovusdb512mem_mask,
5231 __builtin_ia32_pmovsdb512mem_mask,
5232 __builtin_ia32_pmovdb512mem_mask.
5233 (bdesc_args): Add __builtin_ia32_expanddf512,
5234 __builtin_ia32_expandsf512.
5235 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
5236 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
5237 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
5238 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
5239 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
5240 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
5241 (avx512f_<code>v8div16qi2_mask_store): This.
5242 (avx512f_expand<mode>): New.
5244 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
5246 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
5248 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5249 (_mm512_prefetch_i32scatter_pd): Ditto.
5250 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5251 (_mm512_prefetch_i64scatter_pd): Ditto.
5252 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5253 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
5254 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5255 (_mm512_prefetch_i32scatter_ps): Ditto.
5256 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5257 (_mm512_prefetch_i64scatter_ps): Ditto.
5258 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5259 * config/i386/i386-builtin-types.def: Define
5260 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
5261 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
5262 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
5263 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
5264 IX86_BUILTIN_SCATTERPFQPD.
5265 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
5266 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
5267 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
5268 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
5269 __builtin_ia32_scatterpfqps.
5270 (ix86_expand_builtin): Expand new built-ins.
5271 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
5272 fix memory access data type.
5273 (*avx512pf_gatherpf<mode>_mask): Ditto.
5274 (*avx512pf_gatherpf<mode>): Ditto.
5275 (avx512pf_scatterpf<mode>): Ditto.
5276 (*avx512pf_scatterpf<mode>_mask): Ditto.
5277 (*avx512pf_scatterpf<mode>): Ditto.
5278 (GATHER_SCATTER_SF_MEM_MODE): New.
5279 (avx512pf_gatherpf<mode>df): Ditto.
5280 (*avx512pf_gatherpf<mode>df_mask): Ditto.
5281 (*avx512pf_scatterpf<mode>df): Ditto.
5283 2014-01-27 Jakub Jelinek <jakub@redhat.com>
5286 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
5287 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
5290 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5292 * common/config/arm/arm-common.c
5293 (arm_rewrite_mcpu): Handle multiple names.
5295 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5297 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
5299 * gimple-builder.h (create_gimple_tmp): Delete.
5301 2014-01-27 Christian Bruel <christian.bruel@st.com>
5303 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
5306 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
5308 * config/pa/pa.md (call): Generate indirect long calls to non-local
5309 functions when outputing 32-bit code.
5310 (call_value): Likewise except for special call to buggy powf function.
5312 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
5313 portable runtime and PIC indirect calls.
5314 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
5315 and PIC call sequences. Use ldo instead of blr to set return register
5316 in PIC call sequence.
5318 2014-01-25 Walter Lee <walt@tilera.com>
5320 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
5321 avoid clobbering a live register.
5323 2014-01-25 Walter Lee <walt@tilera.com>
5325 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
5326 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
5327 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
5328 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
5330 2014-01-25 Walter Lee <walt@tilera.com>
5332 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
5333 arguments on even registers.
5334 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
5336 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
5337 (BIGGEST_ALIGNMENT): Ditto.
5338 (BIGGEST_FIELD_ALIGNMENT): Ditto.
5340 2014-01-25 Walter Lee <walt@tilera.com>
5342 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
5343 insns before bundling.
5344 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
5346 2014-01-25 Walter Lee <walt@tilera.com>
5348 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
5349 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
5350 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
5352 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5354 * config/mips/constraints.md (kl): Delete.
5355 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
5356 define expands, using...
5357 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
5358 instructions for MIPS16.
5359 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
5360 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
5362 2014-01-25 Walter Lee <walt@tilera.com>
5364 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
5368 2014-01-25 Walter Lee <walt@tilera.com>
5370 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
5371 (TARGET_EXPAND_TO_RTL_HOOK): Define.
5373 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
5375 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
5378 2014-01-25 Jakub Jelinek <jakub@redhat.com>
5380 * print-rtl.c (in_call_function_usage): New var.
5381 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
5382 EXPR_LIST mode as mode and not as reg note name.
5385 * cfgloopmanip.c (copy_loop_info): If
5386 loop->warned_aggressive_loop_optimizations, make sure
5387 the flag is set in target loop too.
5389 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
5391 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
5393 * builtins.def: Likewise.
5394 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
5395 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
5396 * ira.c (ira_setup_eliminable_regset): Likewise.
5397 * omp-low.c (gate_expand_omp): Likewise.
5398 (execute_lower_omp): Likewise.
5399 (diagnose_sb_0): Likewise.
5400 (gate_diagnose_omp_blocks): Likewise.
5401 (simd_clone_clauses_extract): Likewise.
5404 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5406 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
5407 correction for little endian...
5408 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
5411 2014-01-24 Jeff Law <law@redhat.com>
5413 PR tree-optimization/59919
5414 * tree-vrp.c (find_assert_locations_1): Do not register asserts
5415 for non-returning calls.
5417 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
5419 * common/config/aarch64/aarch64-common.c
5420 (aarch64_rewrite_mcpu): Handle multiple names.
5421 * config/aarch64/aarch64.h
5422 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
5424 2014-01-24 Dodji Seketeli <dodji@redhat.com>
5426 * input.c (add_file_to_cache_tab): Handle the case where fopen
5429 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
5432 * config/i386/i386.md (pushsf splitter): Get stack adjustment
5433 from push operand if code of push isn't PRE_DEC.
5435 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
5438 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
5439 -mquad-memory-atomic. Update -mquad-memory documentation to say
5440 it is only used for non-atomic loads/stores.
5442 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
5443 -mquad-memory or -mquad-memory-atomic switches.
5445 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
5446 -mquad-memory-atomic to ISA 2.07 support.
5448 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
5449 to separate support of normal quad word memory operations (ldq, stq)
5450 from the atomic quad word memory operations.
5452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5453 support to separate non-atomic quad word operations from atomic
5454 quad word operations. Disable non-atomic quad word operations in
5455 little endian mode so that we don't have to swap words after the
5456 load and before the store.
5457 (quad_load_store_p): Add comment about atomic quad word support.
5458 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
5459 options printed with -mdebug=reg.
5461 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
5462 -mquad-memory-atomic as the test for whether we have quad word
5463 atomic instructions.
5464 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
5465 or -mp8-vector are used, allow byte/half-word atomic operations.
5467 * config/rs6000/sync.md (load_lockedti): Insure that the address
5468 is a proper indexed or indirect address for the lqarx instruction.
5469 On little endian systems, swap the hi/lo registers after the lqarx
5471 (load_lockedpti): Use indexed_or_indirect_operand predicate to
5472 insure the address is valid for the lqarx instruction.
5473 (store_conditionalti): Insure that the address is a proper indexed
5474 or indirect address for the stqcrx. instruction. On little endian
5475 systems, swap the hi/lo registers before doing the stqcrx.
5477 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
5478 insure the address is valid for the stqcrx. instruction.
5480 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
5481 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
5482 type of quad memory support is available.
5484 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
5487 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
5488 there is a danger of looping.
5490 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
5492 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5493 force flag_ira_loop_pressure if set via command line.
5495 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5497 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
5498 (ashr_simd): New builtin handling DI mode.
5499 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
5500 (aarch64_sshr_simddi): New match pattern.
5501 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
5502 (vshrd_n_s64): Likewise.
5503 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
5505 2014-01-23 Nick Clifton <nickc@redhat.com>
5507 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
5508 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
5509 favour of mcu specific scripts.
5510 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
5513 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5514 Alex Velenko <Alex.Velenko@arm.com>
5516 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
5517 (vaddv_s16): Likewise.
5518 (vaddv_s32): Likewise.
5519 (vaddv_u8): Likewise.
5520 (vaddv_u16): Likewise.
5521 (vaddv_u32): Likewise.
5522 (vaddvq_s8): Likewise.
5523 (vaddvq_s16): Likewise.
5524 (vaddvq_s32): Likewise.
5525 (vaddvq_s64): Likewise.
5526 (vaddvq_u8): Likewise.
5527 (vaddvq_u16): Likewise.
5528 (vaddvq_u32): Likewise.
5529 (vaddvq_u64): Likewise.
5530 (vaddv_f32): Likewise.
5531 (vaddvq_f32): Likewise.
5532 (vaddvq_f64): Likewise.
5533 (vmaxv_f32): Likewise.
5534 (vmaxv_s8): Likewise.
5535 (vmaxv_s16): Likewise.
5536 (vmaxv_s32): Likewise.
5537 (vmaxv_u8): Likewise.
5538 (vmaxv_u16): Likewise.
5539 (vmaxv_u32): Likewise.
5540 (vmaxvq_f32): Likewise.
5541 (vmaxvq_f64): Likewise.
5542 (vmaxvq_s8): Likewise.
5543 (vmaxvq_s16): Likewise.
5544 (vmaxvq_s32): Likewise.
5545 (vmaxvq_u8): Likewise.
5546 (vmaxvq_u16): Likewise.
5547 (vmaxvq_u32): Likewise.
5548 (vmaxnmv_f32): Likewise.
5549 (vmaxnmvq_f32): Likewise.
5550 (vmaxnmvq_f64): Likewise.
5551 (vminv_f32): Likewise.
5552 (vminv_s8): Likewise.
5553 (vminv_s16): Likewise.
5554 (vminv_s32): Likewise.
5555 (vminv_u8): Likewise.
5556 (vminv_u16): Likewise.
5557 (vminv_u32): Likewise.
5558 (vminvq_f32): Likewise.
5559 (vminvq_f64): Likewise.
5560 (vminvq_s8): Likewise.
5561 (vminvq_s16): Likewise.
5562 (vminvq_s32): Likewise.
5563 (vminvq_u8): Likewise.
5564 (vminvq_u16): Likewise.
5565 (vminvq_u32): Likewise.
5566 (vminnmv_f32): Likewise.
5567 (vminnmvq_f32): Likewise.
5568 (vminnmvq_f64): Likewise.
5570 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
5572 * config/aarch64/aarch64-simd.md
5573 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
5574 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
5575 (*aarch64_mul3_elt<mode>): Likewise.
5576 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
5577 (*aarch64_mul3_elt_to_64v2df): Likewise.
5578 (*aarch64_mla_elt<mode>): Likewise.
5579 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5580 (*aarch64_mls_elt<mode>): Likewise.
5581 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5582 (*aarch64_fma4_elt<mode>): Likewise.
5583 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5584 (*aarch64_fma4_elt_to_64v2df): Likewise.
5585 (*aarch64_fnma4_elt<mode>): Likewise.
5586 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5587 (*aarch64_fnma4_elt_to_64v2df): Likewise.
5588 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
5589 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
5590 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5591 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
5592 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5593 (aarch64_sqdmull_lane<mode>_internal): Likewise.
5594 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5596 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
5598 * config/aarch64/aarch64-simd.md
5599 (aarch64_be_checked_get_lane<mode>): New define_expand.
5600 * config/aarch64/aarch64-simd-builtins.def
5601 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
5602 New builtin definition.
5603 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
5604 Use new safe be builtin.
5606 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
5608 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
5610 (aarch64_be_st1<mode>): Likewise.
5611 (aarch_ld1<VALL:mode>): Define_expand modified.
5612 (aarch_st1<VALL:mode>): Likewise.
5613 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
5614 (UNSPEC_ST1): Likewise.
5616 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5618 * config/microblaze/microblaze.md: Add trap insn and attribute
5620 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5622 PR preprocessor/58580
5623 * input.h (location_get_source_line): Take an additional line_size
5625 (void diagnostics_file_cache_fini): Declare new function.
5626 * input.c (struct fcache): New type.
5627 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5628 New static constants.
5629 (diagnostic_file_cache_init, total_lines_num)
5630 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5631 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5632 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5633 (get_next_line, read_next_line, goto_next_line, read_line_num):
5634 New static function definitions.
5635 (diagnostic_file_cache_fini): New function.
5636 (location_get_source_line): Take an additional output line_len
5637 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5639 * diagnostic.c (diagnostic_finish): Call
5640 diagnostic_file_cache_fini.
5641 (adjust_line): Take an additional input parameter for the length
5642 of the line, rather than calculating it with strlen.
5643 (diagnostic_show_locus): Adjust the use of
5644 location_get_source_line and adjust_line with respect to their new
5645 signature. While displaying a line now, do not stop at the first
5646 null byte. Rather, display the zero byte as a space and keep
5647 going until we reach the size of the line.
5648 * Makefile.in: Add vec.o to OBJS-libcommon
5650 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5651 Ilya Tocar <ilya.tocar@intel.com>
5653 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5654 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5655 (__builtin_ia32_kmov16): Ditto.
5656 * config/i386/i386.md (UNSPEC_KMOV): New.
5659 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5661 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5662 (_mm512_storeu_si512): Ditto.
5664 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5667 * rtl.h (get_referenced_operands): Declare.
5668 * recog.c (get_referenced_operands): New function.
5669 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5670 operands have been referenced when recording LO_SUM references.
5672 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5674 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5676 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5678 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5679 Enable for generic and recent AMD targets.
5681 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5683 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5684 ARG_SIZE note when adjustment was eliminated.
5686 2014-01-22 Jeff Law <law@redhat.com>
5688 PR tree-optimization/59597
5689 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5690 in file. Accept new argument REGISTERING and use it to modify
5691 dump output appropriately.
5692 (register_jump_thread): Corresponding changes.
5693 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5694 thread paths involving joiner blocks. Add code to dump cancelled
5695 jump threading paths.
5697 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5699 PR rtl-optimization/59477
5700 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5701 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5703 2014-01-22 Tom Tromey <tromey@redhat.com>
5705 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5707 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5709 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5711 PR rtl-optimization/59896
5712 * lra-constraints.c (process_alt_operands): Check unused note for
5713 matched operands of insn with no output reloads.
5715 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5717 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5718 (mips_move_from_gpr_cost): Likewise.
5720 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5722 PR rtl-optimization/59858
5723 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5724 ira_class_hard_regs_num.
5725 (process_alt_operands): Increase reject for dying matched operand.
5727 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5730 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5731 smaller than size, perform several stores or loads and stores
5732 at dst + count - size to store or copy all of size bytes, rather
5733 than just last modesize bytes.
5735 2014-01-20 DJ Delorie <dj@redhat.com>
5737 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5738 that CLOBBERs are REGs before propogating their values.
5740 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5743 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5744 (cgraph_inline_failed_type): New function.
5745 * cgraph.h (DEFCIFCODE): Add type.
5746 (cgraph_inline_failed_type_t): New enum.
5747 (cgraph_inline_failed_type): New prototype.
5748 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5749 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5750 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5751 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5752 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5753 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5754 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5755 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5756 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5757 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5758 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5759 OPTIMIZATION_MISMATCH.
5760 * tree-inline.c (expand_call_inline): Emit errors during
5761 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5763 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5766 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5767 mode attribute in insn output.
5769 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5771 * output.h (output_constant): Delete.
5772 * varasm.c (output_constant): Make private.
5774 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5776 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5778 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5781 * tree.h (fold_builtin_strcat): New prototype.
5782 * builtins.c (fold_builtin_strcat): No longer static. Add len
5783 argument, if non-NULL, don't call c_strlen. Optimize
5784 directly into __builtin_memcpy instead of __builtin_strcpy.
5785 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5786 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5788 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5790 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5791 for SImode_address_operand operands, having only a REG argument.
5793 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5795 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5796 loader name using mbig-endian.
5797 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5799 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5801 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5805 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5807 * config/aarch64/aarch64-protos.h
5808 (aarch64_cannot_change_mode_class_ptr): Declare.
5809 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5810 aarch64_cannot_change_mode_class_ptr): New.
5811 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5812 backend hook aarch64_cannot_change_mode_class.
5814 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5816 * common/config/aarch64/aarch64-common.c
5817 (aarch64_handle_option): Don't handle any option order logic here.
5818 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5819 selected_cpu, warn on architecture version mismatch.
5820 (aarch64_override_options): Fix parsing order for option strings.
5822 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5823 Iain Sandoe <iain@codesourcery.com>
5826 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5827 warning. Amend comment to reflect current functionality.
5829 2014-01-20 Richard Biener <rguenther@suse.de>
5832 * builtins.c (fold_builtin_strcat): Remove case better handled
5833 by tree-ssa-strlen.c.
5835 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5837 * config/aarch64/aarch64.opt
5838 (mcpu, march, mtune): Make case-insensitive.
5840 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5843 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5844 if operands[1] is a REG or ZERO_EXTEND of a REG.
5846 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5848 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5850 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5852 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5853 long non-pic millicode calls.
5855 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5857 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5859 2014-01-19 Kito Cheng <kito@0xlab.org>
5861 * builtins.c (expand_movstr): Check movstr expand done or fail.
5863 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5864 H.J. Lu <hongjiu.lu@intel.com>
5867 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5868 to DImode for zero-extended addresses.
5870 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5872 PR rtl-optimization/57763
5873 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5874 on the new indirect jump_insn and increment LABEL_NUSES (label).
5876 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5880 * config.gcc (x86_archs): New variable.
5881 (x86_64_archs): Likewise.
5882 (x86_cpus): Likewise.
5883 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5884 --with-arch/--with-cpu= options.
5885 Support --with-arch=/--with-cpu={nehalem,westmere,
5886 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5888 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5890 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5891 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5893 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5895 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5897 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5900 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5901 clear cpp_get_options (parse_in)->warn_unused_macros for
5902 ix86_target_macros_internal with cpp_define.
5904 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5906 * jump.c (delete_related_insns): Keep (use (insn))s.
5907 * reorg.c (redundant_insn): Check for barriers too.
5909 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5911 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5913 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5915 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5916 call to $$dyncall when TARGET_LONG_CALLS is true.
5918 2014-01-17 Jeff Law <law@redhat.com>
5920 * ree.c (combine_set_extension): Temporarily disable test for
5921 changing number of hard registers.
5923 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5926 * ipa-inline-analysis.c (inline_free_summary):
5927 Do not free summary of aliases.
5929 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5932 * gimplify.c (gimplify_expr): Use create_tmp_var
5933 instead of create_tmp_var_raw. If cond doesn't have
5934 integral type, don't add the IFN_ANNOTATE builtin at all.
5936 2014-01-17 Martin Jambor <mjambor@suse.cz>
5939 * ipa-cp.c (prev_edge_clone): New variable.
5940 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5941 Also resize prev_edge_clone vector.
5942 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5943 (ipcp_edge_removal_hook): New function.
5944 (ipcp_driver): Register ipcp_edge_removal_hook.
5946 2014-01-17 Andrew Pinski <apinski@cavium.com>
5947 Steve Ellcey <sellcey@mips.com>
5950 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5953 2014-01-17 Jeff Law <law@redhat.com>
5956 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5957 so that pass_ccp runs first.
5959 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5961 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5962 (ix86_adjust_cost): Use !TARGET_XXX.
5963 (do_reorder_for_imul): Likewise.
5964 (swap_top_of_ready_list): Likewise.
5965 (ix86_sched_reorder): Likewise.
5967 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5969 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5970 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5971 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5972 (intel_memset): New. Duplicate slm_memset.
5973 (intel_cost): New. Duplicate slm_cost.
5974 (m_INTEL): New macro.
5975 (processor_target_table): Add "intel".
5976 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5977 with PROCESSOR_INTEL for "intel".
5978 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5979 PROCESSOR_SILVERMONT.
5980 (ix86_issue_rate): Likewise.
5981 (ix86_adjust_cost): Likewise.
5982 (ia32_multipass_dfa_lookahead): Likewise.
5983 (swap_top_of_ready_list): Likewise.
5984 (ix86_sched_reorder): Likewise.
5985 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5986 instead of TARGET_OPT_AGU.
5987 * config/i386/i386.h (TARGET_INTEL): New.
5988 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5989 (processor_type): Add PROCESSOR_INTEL.
5990 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5991 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5993 2014-01-17 Marek Polacek <polacek@redhat.com>
5996 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5999 2014-01-17 Richard Biener <rguenther@suse.de>
6001 PR tree-optimization/46590
6002 * opts.c (default_options_table): Add entries for
6003 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
6004 all enabled at -O1 but not for -Og.
6005 * common.opt (fbranch-count-reg): Remove Init(1).
6006 (fmove-loop-invariants): Likewise.
6007 (ftree-pta): Likewise.
6009 2014-01-17 Jakub Jelinek <jakub@redhat.com>
6011 * config/i386/i386.c (ix86_data_alignment): For compatibility with
6012 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
6013 decls to at least the GCC 4.8 used alignments.
6016 * tree-nested.c (convert_nonlocal_reference_stmt,
6017 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
6018 of GIMPLE_BIND stmts, adjust associated decls.
6020 2014-01-17 Richard Biener <rguenther@suse.de>
6022 PR tree-optimization/46590
6023 * vec.h (vec<>::bseach): New member function implementing
6024 binary search according to C89 bsearch.
6025 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
6026 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
6027 bitmap pointer again. Make accesses_in_loop a flat array.
6028 (mem_ref_obstack): New global.
6029 (outermost_indep_loop): Adjust for mem_ref->stored changes.
6030 (mark_ref_stored): Likewise.
6031 (ref_indep_loop_p_2): Likewise.
6032 (set_ref_stored_in_loop): New helper function.
6033 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
6034 (memref_free): Adjust.
6035 (record_mem_ref_loc): Simplify.
6036 (gather_mem_refs_stmt): Adjust.
6037 (sort_locs_in_loop_postorder_cmp): New function.
6038 (analyze_memory_references): Sort accesses_in_loop after
6039 loop postorder number.
6040 (find_ref_loc_in_loop_cmp): New function.
6041 (for_all_locs_in_loop): Find relevant cluster of locs in
6042 accesses_in_loop and iterate without recursion.
6043 (execute_sm): Avoid uninit warning.
6044 (struct ref_always_accessed): Simplify.
6045 (ref_always_accessed::operator ()): Likewise.
6046 (ref_always_accessed_p): Likewise.
6047 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
6048 loop postorder numbers here.
6049 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
6052 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
6055 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
6056 on decls for which assemble_alias has been called.
6058 2014-01-17 Nick Clifton <nickc@redhat.com>
6060 * config/msp430/msp430.opt: (mcpu): New option.
6061 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
6062 (msp430_option_override): Parse target_cpu. If the MCU name
6063 matches a generic string, clear target_mcu.
6064 (msp430_attr): Allow numeric interrupt values up to 63.
6065 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
6066 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
6068 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
6070 * config/msp430/msp430.md (popm): Use %J rather than %I.
6071 (addsi3): Use msp430_nonimmediate_operand for operand 2.
6072 (addhi_cy_i): Use immediate_operand for operand 2.
6073 * doc/invoke.texi: Document -mcpu option.
6075 2014-01-17 Richard Biener <rguenther@suse.de>
6077 PR rtl-optimization/38518
6078 * df.h (df_analyze_loop): Declare.
6079 * df-core.c: Include cfgloop.h.
6080 (df_analyze_1): Split out main part of df_analyze.
6081 (df_analyze): Adjust.
6082 (loop_inverted_post_order_compute): New function.
6083 (loop_post_order_compute): Likewise.
6084 (df_analyze_loop): New function avoiding whole-function
6086 * loop-invariant.c (find_defs): Use df_analyze_loop.
6087 (find_invariants): Adjust.
6088 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
6090 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
6092 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
6093 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
6095 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
6097 * ipa-ref.c (ipa_remove_stmt_references): Fix references
6098 traversal when removing references.
6100 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
6103 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
6105 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
6108 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
6109 pushing a reload for an autoinc when we had previously reloaded an
6110 inner part of the address.
6112 2014-01-16 Jakub Jelinek <jakub@redhat.com>
6114 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
6116 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
6117 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
6118 when not giving up or versioning for alias only because of
6120 (vect_analyze_data_ref_dependences): Set to true.
6121 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
6123 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
6124 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
6128 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
6131 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
6132 operand 0 predicate for gathers, use a new pseudo as subtarget.
6134 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6137 * lra-constraints.c (process_alt_operands): Add printing debug info.
6138 Check absence of input/output reloads for matched operands too.
6140 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
6142 PR rtl-optimization/59835
6143 * ira.c (ira_init_register_move_cost): Increase cost for
6146 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
6148 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
6150 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
6153 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
6154 non-register objects. Use gen_(high/low)part more consistently.
6157 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
6160 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
6161 endian support, remove tests for WORDS_BIG_ENDIAN.
6162 (p8_mfvsrd_3_<mode>): Likewise.
6163 (reload_gpr_from_vsx<mode>): Likewise.
6164 (reload_gpr_from_vsxsf): Likewise.
6165 (p8_mfvsrd_4_disf): Likewise.
6167 2014-01-16 Richard Biener <rguenther@suse.de>
6169 PR rtl-optimization/46590
6170 * lcm.c (compute_antinout_edge): Use postorder iteration.
6171 (compute_laterin): Use inverted postorder iteration.
6173 2014-01-16 Nick Clifton <nickc@redhat.com>
6176 * varasm.c (output_constant): Return the number of bytes actually
6178 (output_constructor_array_range): Update the field size with the
6179 number of bytes emitted by output_constant.
6180 (output_constructor_regular_field): Likewise. Also do not
6181 complain if the total number of bytes emitted is now greater
6182 than the expected fieldpos.
6183 * output.h (output_constant): Update prototype and descriptive comment.
6185 2014-01-16 Marek Polacek <polacek@redhat.com>
6188 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
6189 it is error_mark_node.
6191 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
6193 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
6194 VALID_AVX256_REG_OR_OI_MODE.
6196 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
6198 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
6199 current procedure should be profiled.
6201 2014-01-15 Andrew Pinski <apinski@cavium.com>
6203 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
6204 of moving from/to the STACK_REG register class.
6206 2014-01-15 Richard Henderson <rth@redhat.com>
6209 * reginfo.c (global_regs_decl): Globalize.
6210 * rtl.h (global_regs_decl): Declare.
6211 * ira.c (do_reload): Diagnose frame_pointer_needed and it
6212 reserved via global_regs.
6214 2014-01-15 Teresa Johnson <tejohnson@google.com>
6216 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
6218 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6220 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
6221 and vmulosh rather than call gen_vec_widen_smult_*.
6222 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
6223 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
6224 (vec_widen_smult_even_v16qi): Likewise.
6225 (vec_widen_umult_even_v8hi): Likewise.
6226 (vec_widen_smult_even_v8hi): Likewise.
6227 (vec_widen_umult_odd_v16qi): Likewise.
6228 (vec_widen_smult_odd_v16qi): Likewise.
6229 (vec_widen_umult_odd_v8hi): Likewise.
6230 (vec_widen_smult_odd_v8hi): Likewise.
6231 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
6232 vmuloub rather than call gen_vec_widen_umult_*.
6233 (vec_widen_umult_lo_v16qi): Likewise.
6234 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
6235 vmulosb rather than call gen_vec_widen_smult_*.
6236 (vec_widen_smult_lo_v16qi): Likewise.
6237 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
6238 rather than call gen_vec_widen_umult_*.
6239 (vec_widen_umult_lo_v8hi): Likewise.
6240 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
6241 rather than call gen_vec_widen_smult_*.
6242 (vec_widen_smult_lo_v8hi): Likewise.
6244 2014-01-15 Jeff Law <law@redhat.com>
6246 PR tree-optimization/59747
6247 * ree.c (find_and_remove_re): Properly handle case where a second
6248 eliminated extension requires widening a copy created for elimination
6249 of a prior extension.
6250 (combine_set_extension): Ensure that the number of hard regs needed
6251 for a destination register does not change when we widen it.
6253 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
6255 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
6256 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
6257 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
6258 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
6259 (avr-*-rtems*): Likewise.
6260 (bfin*-rtems*): Likewise.
6261 (moxie-*-rtems*): Likewise.
6262 (h8300-*-rtems*): Likewise.
6263 (i[34567]86-*-rtems*): Likewise.
6264 (lm32-*-rtems*): Likewise.
6265 (m32r-*-rtems*): Likewise.
6266 (m68k-*-rtems*): Likewise.
6267 (microblaze*-*-rtems*): Likewise.
6268 (mips*-*-rtems*): Likewise.
6269 (powerpc-*-rtems*): Likewise.
6270 (sh-*-rtems*): Likewise.
6271 (sparc-*-rtems*): Likewise.
6272 (sparc64-*-rtems*): Likewise.
6273 (v850-*-rtems*): Likewise.
6274 (m32c-*-rtems*): Likewise.
6276 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
6278 PR rtl-optimization/59511
6279 * ira.c (ira_init_register_move_cost): Use memory costs for some
6280 cases of register move cost calculations.
6281 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
6282 instead of BB frequency.
6283 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
6284 * lra-assigns.c (find_hard_regno_for): Ditto.
6286 2014-01-15 Richard Biener <rguenther@suse.de>
6288 PR tree-optimization/59822
6289 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
6290 (vectorizable_load): Use it to hoist defs of uses of invariant
6291 loads out of the loop.
6293 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
6294 Kugan Vivekanandarajah <kuganv@linaro.org>
6297 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
6300 2014-01-15 Richard Biener <rguenther@suse.de>
6302 PR rtl-optimization/59802
6303 * lcm.c (compute_available): Use inverted postorder to seed
6304 the initial worklist.
6306 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6309 * config/s390/s390.c (s390_preferred_reload_class): Don't return
6310 ADDR_REGS for invalid symrefs in non-PIC code.
6312 2014-01-15 Jakub Jelinek <jakub@redhat.com>
6315 * builtins.c (determine_block_size): Initialize *probable_max_size
6316 even if len_rtx is CONST_INT.
6318 2014-01-14 Andrew Pinski <apinski@cavium.com>
6320 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
6321 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
6322 (cortexa53_tunings): Likewise.
6323 (aarch64_sched_issue_rate): New function.
6324 (TARGET_SCHED_ISSUE_RATE): Define.
6326 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6328 * ira-costs.c (find_costs_and_classes): Add missed
6329 ira_init_register_move_cost_if_necessary.
6331 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
6334 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
6336 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
6339 * config/i386/i386.c (type_natural_mode): Add a bool parameter
6340 to indicate if type is used for function return value. Warn ABI
6341 change if the vector mode isn't available for function return value.
6342 (ix86_function_arg_advance): Pass false to type_natural_mode.
6343 (ix86_function_arg): Likewise.
6344 (ix86_gimplify_va_arg): Likewise.
6345 (function_arg_32): Don't warn ABI change.
6346 (ix86_function_value): Pass true to type_natural_mode.
6347 (ix86_return_in_memory): Likewise.
6348 (ix86_struct_value_rtx): Removed.
6349 (TARGET_STRUCT_VALUE_RTX): Likewise.
6351 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6353 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
6354 converting a conditional jump into a conditional return.
6356 2014-01-14 Richard Biener <rguenther@suse.de>
6358 PR tree-optimization/58921
6359 PR tree-optimization/59006
6360 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
6361 hoisting invariant stmts.
6362 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
6363 invariant loads on the preheader edge if possible.
6365 2014-01-14 Joey Ye <joey.ye@arm.com>
6367 * doc/plugin.texi (Building GCC plugins): Update to C++.
6369 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
6371 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
6372 (_mm_rcp28_round_ss): Ditto.
6373 (_mm_rsqrt28_round_sd): Ditto.
6374 (_mm_rsqrt28_round_ss): Ditto.
6375 (_mm_rcp28_sd): Ditto.
6376 (_mm_rcp28_ss): Ditto.
6377 (_mm_rsqrt28_sd): Ditto.
6378 (_mm_rsqrt28_ss): Ditto.
6379 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
6380 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
6381 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
6382 (IX86_BUILTIN_RCP28SD): Ditto.
6383 (IX86_BUILTIN_RCP28SS): Ditto.
6384 (IX86_BUILTIN_RSQRT28SD): Ditto.
6385 (IX86_BUILTIN_RSQRT28SS): Ditto.
6386 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
6387 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
6388 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
6389 (ix86_expand_special_args_builtin): Expand new FTYPE.
6390 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
6391 (srcp14<mode>): Make insn unary.
6392 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
6393 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
6394 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
6395 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
6396 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
6397 Fix rounding: make it SAE only.
6398 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
6400 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
6402 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
6403 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
6404 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
6405 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
6406 (round_saeonly_mask_scalar_operand4): Ditto.
6407 (round_saeonly_mask_scalar_op3): Ditto.
6408 (round_saeonly_mask_scalar_op4): Ditto.
6410 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6412 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6413 Implement -maltivec=be for vec_insert and vec_extract.
6415 2014-01-10 DJ Delorie <dj@redhat.com>
6417 * config/msp430/msp430.md (call_internal): Don't allow memory
6418 references with SP as the base register.
6419 (call_value_internal): Likewise.
6420 * config/msp430/constraints.md (Yc): New. For memory references
6421 that don't use SP as a base register.
6423 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
6424 "an integer without a # prefix"
6425 * config/msp430/msp430.md (epilogue_helper): Use it.
6427 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6430 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
6431 AVX512F gather builtins.
6432 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
6433 on gather decls with INTEGER_TYPE masktype.
6434 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
6435 directly into the builtin rather than hoisting it before loop.
6437 PR tree-optimization/59387
6438 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
6439 (scev_const_prop): If folded_casts and type has undefined overflow,
6440 use force_gimple_operand instead of force_gimple_operand_gsi and
6441 for each added stmt if it is assign with
6442 arith_code_with_undefined_signed_overflow, call
6443 rewrite_to_defined_overflow.
6444 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
6445 gimple-fold.h instead.
6446 (arith_code_with_undefined_signed_overflow,
6447 rewrite_to_defined_overflow): Moved to ...
6448 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
6449 rewrite_to_defined_overflow): ... here. No longer static.
6450 Include gimplify-me.h.
6451 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
6452 rewrite_to_defined_overflow): New prototypes.
6454 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6456 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
6458 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
6460 * builtins.c (get_object_alignment_2): Minor tweak.
6461 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
6463 2014-01-13 Christian Bruel <christian.bruel@st.com>
6465 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
6466 optimized non constant lengths.
6468 2014-01-13 Jakub Jelinek <jakub@redhat.com>
6471 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
6472 load as __atomic_load_N if possible.
6474 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6476 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
6478 (rs6000_expand_builtin): Adjust call.
6480 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
6483 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
6484 * config/rs6000/rs6000.c: Include target-globals.h.
6485 (rs6000_set_current_function): Instead of doing target_reinit
6486 unconditionally, use save_target_globals_default_opts and
6487 restore_target_globals.
6489 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
6491 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
6492 (rs6000_expand_builtin): Handle mffs and mtfsf.
6493 (rs6000_init_builtins): Define mffs and mtfsf.
6494 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
6495 (rs6000_mffs): New pattern.
6496 (rs6000_mtfsf): New pattern.
6498 2014-01-11 Bin Cheng <bin.cheng@arm.com>
6500 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
6501 Start narrowing with START. Apply candidate-use pair
6502 and check overall cost in narrowing.
6503 (iv_ca_prune): Pass new argument.
6505 2014-01-10 Jeff Law <law@redhat.com>
6508 * ree.c (combine_reaching_defs): Ensure the defining statement
6509 occurs before the extension when optimizing extensions with
6510 different source and destination hard registers.
6512 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6515 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
6516 vtables into the type inheritance graph.
6518 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6520 PR rtl-optimization/59754
6521 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
6522 modes in the REGNO != REGNO case.
6524 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6526 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
6528 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6530 PR tree-optimization/59745
6531 * tree-predcom.c (tree_predictive_commoning_loop): Call
6532 free_affine_expand_cache if giving up because components is NULL.
6534 * target-globals.c (save_target_globals): Allocate < 4KB structs using
6535 GC in payload of target_globals struct instead of allocating them on
6536 the heap and the larger structs separately using GC.
6537 * target-globals.h (struct target_globals): Make regs, hard_regs,
6538 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
6539 of GTY((skip)) and change type to void *.
6540 (reset_target_globals): Cast loads from those fields to corresponding
6543 2014-01-10 Steve Ellcey <sellcey@mips.com>
6546 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
6547 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
6548 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
6550 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
6553 * aarch64-modes.def (CC_Zmode): New flags mode.
6554 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
6555 represents an equality.
6556 (aarch64_get_condition_code): Handle CC_Zmode.
6557 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
6559 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6561 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
6562 extraction in good case.
6564 2014-01-10 Richard Biener <rguenther@suse.de>
6566 PR tree-optimization/59374
6567 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
6568 checking after SLP discovery. Mark stmts not participating
6569 in any SLP instance properly.
6571 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6573 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
6574 when handling a SET rtx.
6576 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6578 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
6579 (cortex-a57): Likewise.
6580 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
6582 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6584 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
6585 non-iwmmxt builtins.
6587 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
6591 * ipa-devirt.c record_target_from_binfo): Take as argument
6592 stack of binfos and lookup matching one for virtual inheritance.
6593 (possible_polymorphic_call_targets_1): Update.
6595 2014-01-10 Huacai Chen <chenhc@lemote.com>
6597 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
6598 kernel strings for Loongson-2E/2F/3A.
6600 2014-01-10 Jakub Jelinek <jakub@redhat.com>
6603 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
6604 is_gimple_call before calling gimple_call_internal_p.
6606 2014-01-09 Steve Ellcey <sellcey@mips.com>
6608 * Makefile.in (TREE_FLOW_H): Remove.
6609 (TREE_SSA_H): Add file names from tree-flow.h.
6610 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
6611 * tree.h: Remove tree-flow.h reference.
6612 * hash-table.h: Remove tree-flow.h reference.
6613 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
6614 reference with tree-ssa-loop.h.
6616 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6618 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6619 default element-order behavior for -maltivec.
6620 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6622 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6623 when targeting big endian, at least for now.
6624 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6626 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6629 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6630 var satisfies use_register_for_decl, just take into account type
6631 alignment, rather than decl alignment.
6633 PR tree-optimization/59622
6634 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6635 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6636 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6637 Don't devirtualize for inplace at all. For targets.length () == 1,
6638 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6640 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6642 * config/i386/i386.md (cpu): Remove the unused btver1.
6644 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6646 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6648 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6651 * tree-core.h (struct target_globals): New forward declaration.
6652 (struct tree_target_option): Add globals field.
6653 * tree.h (TREE_TARGET_GLOBALS): Define.
6654 (prepare_target_option_nodes_for_pch): New prototype.
6655 * target-globals.h (struct target_globals): Define even if
6657 * tree.c (prepare_target_option_node_for_pch,
6658 prepare_target_option_nodes_for_pch): New functions.
6659 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6660 * config/i386/i386.c: Include target-globals.h.
6661 (ix86_set_current_function): Instead of doing target_reinit
6662 unconditionally, use save_target_globals_default_opts and
6663 restore_target_globals.
6665 2014-01-09 Richard Biener <rguenther@suse.de>
6667 PR tree-optimization/59715
6668 * tree-cfg.h (split_critical_edges): Declare.
6669 * tree-cfg.c (split_critical_edges): Export.
6670 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6672 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6674 * cfgexpand.c (expand_stack_vars): Optionally disable
6675 asan stack protection.
6676 (expand_used_vars): Likewise.
6677 (partition_stack_vars): Likewise.
6678 * asan.c (asan_emit_stack_protection): Optionally disable
6679 after return stack usage.
6680 (instrument_derefs): Optionally disable memory access instrumentation.
6681 (instrument_builtin_call): Likewise.
6682 (instrument_strlen_call): Likewise.
6683 (asan_protect_global): Optionally disable global variables protection.
6684 * doc/invoke.texi: Added doc for new options.
6685 * params.def: Added new options.
6686 * params.h: Likewise.
6688 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6690 PR rtl-optimization/59724
6691 * ifcvt.c (cond_exec_process_if_block): Don't call
6692 flow_find_head_matching_sequence with 0 longest_match.
6693 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6694 non-active insns if !stop_after.
6695 (try_head_merge_bb): Revert 2014-01-07 changes.
6697 2014-01-08 Jeff Law <law@redhat.com>
6699 * ree.c (get_sub_rtx): New function, extracted from...
6700 (merge_def_and_ext): Here.
6701 (combine_reaching_defs): Use get_sub_rtx.
6703 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6705 * cgraph.h (varpool_variable_node): Do not choke on null node.
6707 2014-01-08 Catherine Moore <clm@codesourcery.com>
6709 * config/mips/mips.md (simple_return): Attempt to use JRC
6711 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6713 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6715 PR rtl-optimization/59137
6716 * reorg.c (steal_delay_list_from_target): Call update_block for
6718 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6720 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6722 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6723 two duplicate entries.
6725 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6728 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6730 * config/mips/mips.c (mips_truncated_op_cost): New function.
6731 (mips_rtx_costs): Adjust test for BADDU.
6732 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6734 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6736 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6737 (*baddu_si): ...this new pattern.
6739 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6742 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6744 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6747 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6749 (expand_expr, expand_normal): Adjust.
6750 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6751 inner_reference_p. Use inner_reference_p to expand inner references.
6752 (store_expr): Adjust.
6753 * cfgexpand.c (expand_call_stmt): Adjust.
6755 2014-01-08 Rong Xu <xur@google.com>
6757 * gcov-io.c (gcov_var): Move from gcov-io.h.
6758 (gcov_position): Ditto.
6759 (gcov_is_error): Ditto.
6760 (gcov_rewrite): Ditto.
6761 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6762 only part to libgcc/libgcov.h.
6764 2014-01-08 Marek Polacek <polacek@redhat.com>
6767 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6769 2014-01-08 Marek Polacek <polacek@redhat.com>
6772 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6774 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6776 PR rtl-optimization/59649
6777 * stor-layout.c (get_mode_bounds): For BImode return
6778 0 and STORE_FLAG_VALUE.
6780 2014-01-08 Richard Biener <rguenther@suse.de>
6783 * gimple.h (is_gimple_builtin_call): Remove.
6784 (gimple_builtin_call_types_compatible_p): New.
6785 (gimple_call_builtin_p): New overload.
6786 * gimple.c (is_gimple_builtin_call): Remove.
6787 (validate_call): Rename to ...
6788 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6790 (validate_type): New static function.
6791 (gimple_call_builtin_p): New overload and adjust.
6792 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6793 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6794 (gimple_fold_stmt_to_constant_1): Likewise.
6795 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6797 2014-01-08 Richard Biener <rguenther@suse.de>
6800 * gimplify.c (gimplify_expr): Gimplify register-register type
6801 VIEW_CONVERT_EXPRs to separate stmts.
6803 2014-01-07 Jeff Law <law@redhat.com>
6806 * ree.c (combine_set_extension): Handle case where source
6807 and destination registers in an extension insn are different.
6808 (combine_reaching_defs): Allow source and destination registers
6809 in extension to be different under limited circumstances.
6810 (add_removable_extension): Remove restriction that the
6811 source and destination registers in the extension are the same.
6812 (find_and_remove_re): Emit a copy from the extension's
6813 destination to its source after the defining insn if
6814 the source and destination registers are different.
6817 * ifcvt.c (merge_if_block): If we are merging a block with more than
6818 one successor with a block with no successors, remove any BARRIER
6819 after the second block.
6821 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6823 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6825 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6828 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6829 for 14-bit register offsets when INT14_OK_STRICT is false.
6831 2014-01-07 Roland Stigge <stigge@antcom.de>
6832 Michael Meissner <meissner@linux.vnet.ibm.com>
6835 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6836 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6838 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6840 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6843 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6845 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6846 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6847 rtx is const0_rtx or not.
6849 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6852 * target-globals.c (save_target_globals): Remove this_fn_optab
6854 * toplev.c: Include optabs.h.
6855 (target_reinit): Temporarily restore the global options if another
6856 set of options are in force.
6858 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6860 PR rtl-optimization/58668
6861 * cfgcleanup.c (flow_find_cross_jump): Don't count
6862 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6863 to determine what is counted.
6864 (flow_find_head_matching_sequence): Use active_insn_p to determine
6866 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6868 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6869 determine what is counted.
6871 PR tree-optimization/59643
6872 * tree-predcom.c (split_data_refs_to_components): If one dr is
6873 read and one write, determine_offset fails and the write isn't
6874 in the bad component, just put the read into the bad component.
6876 2014-01-07 Mike Stump <mikestump@comcast.net>
6877 Jakub Jelinek <jakub@redhat.com>
6880 * tree-core.h (struct tree_optimization_option): Change optabs
6881 type from unsigned char * to void *.
6882 * optabs.c (init_tree_optimization_optabs): Adjust
6883 TREE_OPTIMIZATION_OPTABS initialization.
6885 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6888 * config/i386/i386.h (struct machine_function): Add
6889 no_drap_save_restore field.
6890 * config/i386/i386.c (ix86_save_reg): Use
6891 !cfun->machine->no_drap_save_restore instead of
6892 crtl->stack_realign_needed.
6893 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6894 this function clears frame_pointer_needed. Set
6895 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6896 and DRAP reg is needed.
6898 2014-01-06 Marek Polacek <polacek@redhat.com>
6901 * doc/implement-c.texi: Mention that other integer types are
6902 permitted as bit-field types in strictly conforming mode.
6904 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6906 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6909 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6911 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6913 2014-01-06 Martin Jambor <mjambor@suse.cz>
6916 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6918 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6920 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6924 * var-tracking.c (add_stores): Preserve the value of the source even if
6925 we don't record the store.
6927 2014-01-06 Terry Guo <terry.guo@arm.com>
6929 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6931 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6934 * config/darwin.c (darwin_function_section): Adjust return values to
6935 correspond to optimisation changes made in r206070.
6937 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6939 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6940 from prefetch_block tune setting.
6941 (nocona_cost): Correct size of prefetch block to 64.
6943 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6945 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6946 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6947 used to save the static chain register in the computation of the offset
6948 from which the FP registers need to be restored.
6950 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6952 PR tree-optimization/59519
6953 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6954 ICE if get_current_def (current_new_name) is already non-NULL, as long
6955 as it is a phi result of some other phi in *new_exit_bb that has
6958 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6959 or vmovdqu* for misaligned_operand.
6960 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6961 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6962 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6963 aligned_mem for AVX512F masked aligned load and store builtins and for
6966 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6968 PR tree-optimization/59651
6969 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6970 Address range for negative step should be added by TYPE_SIZE_UNIT.
6972 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6974 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6975 overlapping registers also for registers other than the stack pointer.
6977 2014-01-03 Marek Polacek <polacek@redhat.com>
6980 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6983 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6986 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6989 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6990 (push splitter): Use <P:MODE_SIZE> instead of
6991 GET_MODE_SIZE (<P:MODE>mode).
6992 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6993 (mov -1, reg peephole2): Likewise.
6994 * config/i386/sse.md (*mov<mode>_internal,
6995 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6996 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6997 *<code><mode>3, *andnot<mode>3<mask_name>,
6998 <mask_codefor><code><mode>3<mask_name>): Likewise.
6999 * config/i386/subst.md (mask_mode512bit_condition,
7000 sd_mask_mode512bit_condition): Likewise.
7002 2014-01-02 Xinliang David Li <davidxl@google.com>
7004 PR tree-optimization/59303
7005 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
7006 (dump_predicates): Better output format.
7007 (pred_equal_p): New function.
7008 (is_neq_relop_p): Ditto.
7009 (is_neq_zero_form_p): Ditto.
7010 (pred_expr_equal_p): Ditto.
7011 (pred_neg_p): Ditto.
7012 (simplify_pred): Ditto.
7013 (simplify_preds_2): Ditto.
7014 (simplify_preds_3): Ditto.
7015 (simplify_preds_4): Ditto.
7016 (simplify_preds): Ditto.
7018 (push_to_worklist): Ditto.
7019 (get_pred_info_from_cmp): Ditto.
7020 (is_degenerated_phi): Ditto.
7021 (normalize_one_pred_1): Ditto.
7022 (normalize_one_pred): Ditto.
7023 (normalize_one_pred_chain): Ditto.
7024 (normalize_preds): Ditto.
7025 (normalize_cond_1): Remove function.
7026 (normalize_cond): Ditto.
7027 (is_gcond_subset_of): Ditto.
7028 (is_subset_of_any): Ditto.
7029 (is_or_set_subset_of): Ditto.
7030 (is_and_set_subset_of): Ditto.
7031 (is_norm_cond_subset_of): Ditto.
7032 (pred_chain_length_cmp): Ditto.
7033 (convert_control_dep_chain_into_preds): Type change.
7034 (find_predicates): Ditto.
7035 (find_def_preds): Ditto.
7036 (destroy_predicates_vecs): Ditto.
7037 (find_matching_predicates_in_rest_chains): Ditto.
7038 (use_pred_not_overlap_with_undef_path_pred): Ditto.
7039 (is_pred_expr_subset): Ditto.
7040 (is_pred_chain_subset_of): Ditto.
7041 (is_included_in): Ditto.
7042 (is_superset_of): Ditto.
7044 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7046 Update copyright years.
7048 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
7050 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
7051 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
7052 config/arc/arc.md, config/arc/arc.opt,
7053 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
7054 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
7055 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
7056 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
7057 config/linux-protos.h, config/linux.c, config/winnt-c.c,
7058 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
7059 vtable-verify.c, vtable-verify.h: Use the standard form for the
7062 2014-01-02 Tobias Burnus <burnus@net-b.de>
7064 * gcc.c (process_command): Update copyright notice dates.
7065 * gcov-dump.c: Ditto.
7067 * doc/cpp.texi: Bump @copying's copyright year.
7068 * doc/cppinternals.texi: Ditto.
7069 * doc/gcc.texi: Ditto.
7070 * doc/gccint.texi: Ditto.
7071 * doc/gcov.texi: Ditto.
7072 * doc/install.texi: Ditto.
7073 * doc/invoke.texi: Ditto.
7075 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7077 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
7079 2014-01-01 Jakub Jelinek <jakub@redhat.com>
7081 * config/i386/sse.md (*mov<mode>_internal): Guard
7082 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
7084 PR rtl-optimization/59647
7085 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
7086 new_rtx into UNSIGNED_FLOAT rtxes.
7088 Copyright (C) 2014 Free Software Foundation, Inc.
7090 Copying and distribution of this file, with or without modification,
7091 are permitted in any medium without royalty provided the copyright
7092 notice and this notice are preserved.