1 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
3 * symtab.c (change_decl_assembler_name): Fix transparent alias
6 2014-03-16 Renlin Li <Renlin.Li@arm.com>
8 * config/aarch64/aarch64.c: Correct the comments about the
11 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
13 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
14 check for GF_OMP_FOR_KIND_FOR.
16 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
18 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
19 ymm and zmm register names.
21 2014-03-17 Jakub Jelinek <jakub@redhat.com>
24 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
25 note creation for the 2010-08-31 changes.
27 2014-03-17 Marek Polacek <polacek@redhat.com>
30 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
31 as -fno-tree-loop-vectorize.
32 (expand_omp_simd): Likewise.
34 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
36 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
37 (eligible_for_call_delay): New prototype.
38 * config/sparc/sparc.c (tls_call_delay): Rename into...
39 (eligible_for_call_delay): ...this. Return false if the instruction
40 cannot be put in the delay slot of a branch.
41 (eligible_for_restore_insn): Simplify.
42 (eligible_for_return_delay): Return false if the instruction cannot be
43 put in the delay slot of a branch and simplify.
44 (eligible_for_sibcall_delay): Return false if the instruction cannot be
45 put in the delay slot of a branch.
46 * config/sparc/sparc.md (fix_ut699): New attribute.
47 (tls_call_delay): Delete.
48 (in_call_delay): Reimplement.
49 (eligible_for_sibcall_delay): Rename into...
50 (in_sibcall_delay): ...this.
51 (eligible_for_return_delay): Rename into...
52 (in_return_delay): ...this.
53 (in_branch_delay): Reimplement.
54 (in_uncond_branch_delay): Delete.
55 (in_annul_branch_delay): Delete.
57 2014-03-14 Richard Henderson <rth@redhat.com>
60 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
61 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
62 (*floathi<X87MODEF>2_i387_with_temp): Remove.
63 (floathi splitters): Remove.
64 (float<SWI48x>xf2): New pattern.
65 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
66 code that tried to handle DImode for 32-bit, but which was excluded
67 by the pattern's condition. Drop allocation of stack temporary.
68 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
69 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
70 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
71 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
72 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
73 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
74 (*float<SWI48><MODEF>2_sse_interunit): Remove.
75 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
76 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
77 (*float<SWI48x><X87MODEF>2_i387): Remove.
78 (all float _with_temp splitters): Remove.
79 (*float<SWI48x><MODEF>2_i387): New pattern.
80 (*float<SWI48><MODEF>2_sse): New pattern.
81 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
82 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
84 2014-03-14 Jakub Jelinek <jakub@redhat.com>
85 Marek Polacek <polacek@redhat.com>
88 * common.opt (dump_base_name_prefixed): New Variable.
89 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
90 if x_dump_base_name_prefixed is already set, set it at the end.
92 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
94 PR rtl-optimization/60508
95 * lra-constraints.c (get_reload_reg): Add new parameter
97 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
98 Pass the new parameter values.
100 2014-03-14 Richard Biener <rguenther@suse.de>
102 * common.opt: Revert unintented changes from r205065.
105 2014-03-14 Richard Biener <rguenther@suse.de>
108 * cfghooks.c (split_block): Properly adjust all loops the
109 block was a latch of.
111 2014-03-14 Martin Jambor <mjambor@suse.cz>
114 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
117 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
120 * config/avr/avr.c (avr_set_current_function): Pass function name
121 through default_strip_name_encoding before sanity checking instead
122 of skipping the first char of the assembler name.
124 2014-03-13 Richard Henderson <rth@redhat.com>
127 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
128 (ix86_force_to_memory, ix86_free_from_memory): Remove.
129 * config/i386/i386-protos.h: Likewise.
130 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
131 in the expander instead of a splitter.
132 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
133 any possibility of requiring a memory.
134 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
135 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
136 (fp branch splitters): Update for ix86_split_fp_branch.
137 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
138 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
139 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
140 (*fop_<MODEF>_2_i387): Remove f/r alternative.
141 (*fop_<MODEF>_3_i387): Likewise.
142 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
143 (splitters for the fop_* register patterns): Remove.
144 (fscalexf4_i387): Rename from *fscalexf4_i387.
145 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
147 2014-03-13 Jakub Jelinek <jakub@redhat.com>
149 PR tree-optimization/59779
150 * tree-dfa.c (get_ref_base_and_extent): Use double_int
151 type for bitsize and maxsize instead of HOST_WIDE_INT.
153 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
155 PR rtl-optimization/57320
156 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
157 the CFG after thread_prologue_and_epilogue_insns.
159 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
161 PR rtl-optimization/57189
162 * lra-constraints.c (process_alt_operands): Disfavor spilling
165 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
167 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
170 2014-03-13 Jakub Jelinek <jakub@redhat.com>
172 PR tree-optimization/59025
174 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
175 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
177 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
180 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
181 calls of avr_out_plus_1.
183 2014-03-13 Bin Cheng <bin.cheng@arm.com>
185 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
186 BB's single pred and update the father loop's latch info later.
188 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
190 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
194 (VEC_base): Likewise.
195 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
196 registers, we need to swap double words in little endian mode.
198 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
199 to be a container mode for 128-bit integer operations added in ISA
200 2.07. Unlike TImode and PTImode, the preferred register set is
201 the Altivec/VMX registers for the 128-bit operations.
203 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
205 (rs6000_split_128bit_ok_p): Likewise.
207 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
208 macros for creating ISA 2.07 normal and overloaded builtin
209 functions with 3 arguments.
210 (BU_P8V_OVERLOAD_3): Likewise.
211 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
212 for use as overloaded functions.
213 (VPERM_1TI_UNS): Likewise.
214 (VSEL_1TI): Likewise.
215 (VSEL_1TI_UNS): Likewise.
216 (ST_INTERNAL_1ti): Likewise.
217 (LD_INTERNAL_1ti): Likewise.
218 (XXSEL_1TI): Likewise.
219 (XXSEL_1TI_UNS): Likewise.
220 (VPERM_1TI): Likewise.
221 (VPERM_1TI_UNS): Likewise.
222 (XXPERMDI_1TI): Likewise.
224 (LXVD2X_V1TI): Likewise.
225 (STXVD2X_V1TI): Likewise.
226 (VEC_INIT_V1TI): Likewise.
227 (VEC_SET_V1TI): Likewise.
228 (VEC_EXT_V1TI): Likewise.
229 (EQV_V1TI): Likewise.
230 (NAND_V1TI): Likewise.
231 (ORC_V1TI): Likewise.
232 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
233 added in ISA 2.07. Add both normal 'altivec' builtins, and the
237 (VADDEUQM): Likewise.
238 (VADDECUQ): Likewise.
239 (VSUBEUQM): Likewise.
240 (VSUBECUQ): Likewise.
242 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
243 __int128_t and __uint128_t types.
244 (__uint128_type): Likewise.
245 (altivec_categorize_keyword): Add support for vector __int128_t,
246 vector __uint128_t, vector __int128, and vector unsigned __int128
247 as a container type for TImode operations that need to be done in
248 VSX/Altivec registers.
249 (rs6000_macro_to_expand): Likewise.
250 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
251 to support 128-bit integer instructions vaddcuq, vadduqm,
252 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
253 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
255 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
256 for V1TImode, and set up preferences to use VSX/Altivec registers.
257 Setup VSX reload handlers.
258 (rs6000_debug_reg_global): Likewise.
259 (rs6000_init_hard_regno_mode_ok): Likewise.
260 (rs6000_preferred_simd_mode): Likewise.
261 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
262 (easy_altivec_constant): Likewise.
263 (output_vec_const_move): Likewise.
264 (rs6000_expand_vector_set): Convert V1TImode set and extract to
266 (rs6000_expand_vector_extract): Likewise.
267 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
269 (rs6000_const_vec): Add support for V1TImode.
270 (rs6000_emit_le_vsx_load): Swap double words when loading or
271 storing TImode/V1TImode.
272 (rs6000_emit_le_vsx_store): Likewise.
273 (rs6000_emit_le_vsx_move): Likewise.
274 (rs6000_emit_move): Add support for V1TImode.
275 (altivec_expand_ld_builtin): Likewise.
276 (altivec_expand_st_builtin): Likewise.
277 (altivec_expand_vec_init_builtin): Likewise.
278 (altivec_expand_builtin): Likewise.
279 (rs6000_init_builtins): Add support for V1TImode type. Add
280 support for ISA 2.07 128-bit integer builtins. Define type names
281 for the VSX/Altivec vector types.
282 (altivec_init_builtins): Add support for overloaded vector
283 functions with V1TImode type.
284 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
285 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
287 (rs6000_split_128bit_ok_p): Likewise.
288 (rs6000_handle_altivec_attribute): Create V1TImode from vector
289 __int128_t and vector __uint128_t.
291 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
299 (VS_scalar): Likewise.
300 (VS_double): Likewise.
301 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
303 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
304 we support the ISA 2.07 128-bit integer arithmetic instructions.
305 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
306 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
307 and TImode types for use with the builtin functions.
308 (V1TI_type_node): Likewise.
309 (unsigned_V1TI_type_node): Likewise.
310 (intTI_type_internal_node): Likewise.
311 (uintTI_type_internal_node): Likewise.
313 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
314 128-bit builtin functions.
315 (UNSPEC_VADDEUQM): Likewise.
316 (UNSPEC_VADDECUQ): Likewise.
317 (UNSPEC_VSUBCUQ): Likewise.
318 (UNSPEC_VSUBEUQM): Likewise.
319 (UNSPEC_VSUBECUQ): Likewise.
320 (VM): Add V1TImode to vector mode iterators.
323 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
324 (altivec_vaddcuq): Likewise.
325 (altivec_vsubuqm): Likewise.
326 (altivec_vsubcuq): Likewise.
327 (altivec_vaddeuqm): Likewise.
328 (altivec_vaddecuq): Likewise.
329 (altivec_vsubeuqm): Likewise.
330 (altivec_vsubecuq): Likewise.
332 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
334 (BOOL_128): Likewise.
335 (BOOL_REGS_OUTPUT): Likewise.
336 (BOOL_REGS_OP1): Likewise.
337 (BOOL_REGS_OP2): Likewise.
338 (BOOL_REGS_UNARY): Likewise.
339 (BOOL_REGS_AND_CR0): Likewise.
341 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
342 128-bit integer builtin support.
343 (vec_vadduqm): Likewise.
344 (vec_vaddecuq): Likewise.
345 (vec_vaddeuqm): Likewise.
346 (vec_vsubecuq): Likewise.
347 (vec_vsubeuqm): Likewise.
348 (vec_vsubcuq): Likewise.
349 (vec_vsubuqm): Likewise.
351 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
352 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
353 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
354 128-bit integer add/subtract to ISA 2.07.
356 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
358 * config/arc/arc.c (arc_predicate_delay_insns):
359 Fix third argument passed to conditionalize_nonjump.
361 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
363 * config/aarch64/aarch64-builtins.c
364 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
365 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
366 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
367 instead of __builtin_lfloor.
368 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
370 2014-03-12 Jakub Jelinek <jakub@redhat.com>
372 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
373 (tree_ssa_ifcombine_bb_1): New function.
374 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
375 is an empty forwarder block to then_bb or vice versa and then_bb
376 and else_bb are effectively swapped.
378 2014-03-12 Christian Bruel <christian.bruel@st.com>
381 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
382 REG_CFA_DEF_CFA note.
383 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
384 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
386 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
388 PR tree-optimization/60454
389 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
391 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
393 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
394 Do not define target_cpu_default2 to generic.
395 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
396 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
397 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
399 2014-03-12 Jakub Jelinek <jakub@redhat.com>
400 Marc Glisse <marc.glisse@inria.fr>
402 PR tree-optimization/60502
403 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
404 instead of build_low_bits_mask.
406 2014-03-12 Jakub Jelinek <jakub@redhat.com>
409 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
410 if there are multiple uses, but op doesn't live on E edge.
411 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
412 clobber stmts before __builtin_unreachable.
414 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
416 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
417 hard_frame_pointer_rtx.
418 * cse.c (cse_insn): Remove volatile check.
419 * cselib.c (cselib_process_insn): Likewise.
420 * dse.c (scan_insn): Likewise.
422 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
424 * config/arc/arc.c (conditionalize_nonjump): New function,
426 (arc_ifcvt): ... this.
427 (arc_predicate_delay_insns): Use it.
429 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
431 * config/arc/predicates.md (extend_operand): During/after reload,
432 allow const_int_operand.
433 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
434 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
435 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
437 (umulsi3_highpart_i): Likewise.
439 2014-03-11 Richard Biener <rguenther@suse.de>
441 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
442 Add asserts to guard possible wrong-code bugs.
444 2014-03-11 Richard Biener <rguenther@suse.de>
446 PR tree-optimization/60429
447 PR tree-optimization/60485
448 * tree-ssa-structalias.c (set_union_with_increment): Properly
449 take into account all fields that overlap the shifted vars.
450 (do_sd_constraint): Likewise.
451 (do_ds_constraint): Likewise.
452 (get_constraint_for_ptr_offset): Likewise.
454 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
456 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
457 (nios2_compute_frame_layout):
458 Add calculation of cfun->machine->fp_save_offset.
459 (nios2_expand_prologue): Correct setting of frame pointer register
461 (nios2_expand_epilogue): Update recovery of stack pointer from
462 frame pointer accordingly.
463 (nios2_initial_elimination_offset): Update calculation of offset
464 for eliminating to HARD_FRAME_POINTER_REGNUM.
466 2014-03-10 Jakub Jelinek <jakub@redhat.com>
469 * ipa.c (symtab_remove_unreachable_nodes): Don't call
470 cgraph_get_create_node on VAR_DECLs.
472 2014-03-10 Richard Biener <rguenther@suse.de>
475 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
477 2014-03-08 Douglas B Rupp <rupp@gnat.com>
479 * config/vms/vms.opt (vms_float_format): New variable.
481 2014-03-08 Tobias Burnus <burnus@net-b.de>
483 * doc/invoke.texi (-fcilkplus): Update implementation status.
485 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
486 Richard Biener <rguenther@suse.de>
488 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
489 consistently accross all TUs.
490 (run_gcc): Enable -fshort-double automatically at link at link-time
491 and disallow override.
493 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
496 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
497 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
498 if they can't be used.
500 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
502 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
503 for Solaris 11/x86 ld.
504 * configure: Regenerate.
506 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
508 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
509 (LIB_TLS_SPEC): Save as ld_tls_libs.
510 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
511 (HAVE_AS_IX86_TLSLDM): New test.
512 * configure, config.in: Regenerate.
513 * config/i386/i386.c (legitimize_tls_address): Fall back to
514 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
515 cannot support TLS_MODEL_LOCAL_DYNAMIC.
516 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
517 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
519 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
521 * common.opt (fira-loop-pressure): Mark as optimization.
523 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
525 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
526 an OpenMP mappable type.
528 2014-03-06 Matthias Klose <doko@ubuntu.com>
530 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
531 MULTILIB_OSDIRNAMES is not defined.
533 2014-03-06 Jakub Jelinek <jakub@redhat.com>
534 Meador Inge <meadori@codesourcery.com>
537 * config/arm/arm.c (arm_tls_symbol_p): Remove.
538 (arm_legitimize_address): Call legitimize_tls_address for any
539 arm_tls_referenced_p expression, handle constant addend. Call it
540 before testing for !TARGET_ARM.
541 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
543 2014-03-06 Richard Biener <rguenther@suse.de>
549 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
551 * tree-streamer.c (record_common_node): Assert we don't record
552 nodes with type double.
553 (preload_common_node): Skip type double, complex double and double
554 pointer since it is now frontend dependent due to fshort-double option.
556 2014-03-06 Richard Biener <rguenther@suse.de>
558 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
559 or -fno-lto is specified and the linker has full plugin support.
560 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
561 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
562 * lto-wrapper.c (merge_and_complain): Merge compile-time
564 (run_gcc): And pass it through to the link options.
566 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
570 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
572 * cselib.c (remove_useless_values): Skip to avoid quadratic
573 behavior if the condition moved from...
574 (cselib_process_insn): ... here holds.
576 2014-03-05 Jakub Jelinek <jakub@redhat.com>
579 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
580 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
583 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
584 (TM_H): Add x86-tune.def.
586 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
588 * config/aarch64/aarch64.c (generic_tunings):
589 Use cortexa57_extra_costs.
591 2014-03-05 Jakub Jelinek <jakub@redhat.com>
594 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
595 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
596 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
599 2014-03-04 Heiher <r@hev.cc>
601 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
602 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
604 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
606 * config/i386/predicates.md (const2356_operand): Change to ...
607 (const2367_operand): ... this.
608 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
610 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
611 (*avx512pf_scatterpf<mode>sf): Ditto.
612 (avx512pf_scatterpf<mode>df): Ditto.
613 (*avx512pf_scatterpf<mode>df_mask): Ditto.
614 (*avx512pf_scatterpf<mode>df): Ditto.
615 * config/i386/i386.c (ix86_expand_builtin): Update
616 incorrect hint operand error message.
618 2014-03-04 Richard Biener <rguenther@suse.de>
620 * lto-section-in.c (lto_get_section_data): Fix const cast.
622 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
624 * tree-streamer.c (record_common_node): Assert we don't record
625 nodes with type double.
626 (preload_common_node): Skip type double, complex double and double
627 pointer since it is now frontend dependent due to fshort-double option.
629 2014-03-04 Richard Biener <rguenther@suse.de>
632 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
633 (lto_input_toplevel_asms): Likewise.
634 * lto-section-in.c (lto_get_section_data): Instead do it here
637 2014-03-04 Richard Biener <rguenther@suse.de>
639 PR tree-optimization/60382
640 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
641 dead PHIs a reduction.
643 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
645 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
647 (_mm_prefetch): Move out of GCC target("sse") pragma.
648 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
649 GCC target("prfchw") pragma.
650 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
652 * config/i386/i386.c (ix86_option_override_internal): Enable
653 -mprfchw with -mprefetchwt1.
655 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
657 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
660 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
662 * opts.h (CL_PCH_IGNORE): Define.
663 * targhooks.c (option_affects_pch_p):
664 Return false for options that have CL_PCH_IGNORE set.
665 * opt-functions.awk: Process PchIgnore.
666 * doc/options.texi: Document PchIgnore.
668 * config/arc/arc.opt (misize): Add PchIgnore property.
670 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
672 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
673 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
674 constraint on constants to permit them being loaded into
675 GENERAL_REGS or BASE_REGS.
677 2014-03-03 Nick Clifton <nickc@redhat.com>
679 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
680 anti-cacnonical alternatives.
681 (negandhi3_real): New pattern.
682 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
684 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
686 * config/avr/avr-mcus.def: Remove atxmega16x1.
687 * config/avr/avr-tables.opt: Regenerate.
688 * config/avr/t-multilib: Regenerate.
689 * doc/avr-mmcu.texi: Regenerate.
691 2014-03-03 Tobias Grosser <tobias@grosser.es>
692 Mircea Namolaru <mircea.namolaru@inria.fr>
694 PR tree-optimization/58028
695 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
698 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
700 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
701 not handled by recognizers.
703 2014-03-03 Jakub Jelinek <jakub@redhat.com>
706 * function.c (expand_function_end): Don't emit
707 clobber_return_register sequence if clobber_after is a BARRIER.
708 * cfgexpand.c (construct_exit_block): Append instructions before
709 return_label to prev_bb.
711 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
713 * config/rs6000/constraints.md: Document reserved use of "wc".
715 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
718 * ipa.c (function_and_variable_visibility): When dissolving comdat
719 group, also set all symbols to local.
721 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
726 2013-12-14 Jan Hubicka <jh@suse.cz>
728 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
730 2014-03-02 Jon Beniston <jon@beniston.com>
736 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
737 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
738 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
739 (simple_return, *simple_return): New patterns
740 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
741 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
743 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
745 * dwarf2out.c (gen_subprogram_die): Tidy.
747 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
750 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
751 (*mov_t_msb_neg_negc): ... this new insn.
753 2014-02-28 Jason Merrill <jason@redhat.com>
756 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
759 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
762 * dwarf2out.c (decltype_auto_die): New static.
763 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
764 (gen_type_die_with_usage): Handle 'decltype(auto)'.
765 (is_cxx_auto): Likewise.
767 2014-02-28 Ian Bolton <ian.bolton@arm.com>
769 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
770 we are not using general regs only.
772 2014-02-28 Richard Biener <rguenther@suse.de>
775 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
776 previous fix and only allow to remove trivial pre-headers
777 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
778 (remove_forwarder_block): Properly update the latch of a loop.
780 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
783 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
784 (cselib_preserved_hash_table): New.
785 (preserve_constants_and_equivs): Move preserved vals to it.
786 (cselib_find_slot): Look it up first.
787 (cselib_init): Initialize it.
788 (cselib_finish): Release it.
789 (dump_cselib_table): Dump it.
791 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
794 * cselib.c (remove_useless_values): Skip to avoid quadratic
795 behavior if the condition moved from...
796 (cselib_process_insn): ... here holds.
798 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
801 * var-tracking.c (vt_initialize): Apply the same condition to
802 preserve the CFA base value.
804 2014-02-28 Joey Ye <joey.ye@arm.com>
807 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
808 if reload in progress or completed.
810 2014-02-28 Tobias Burnus <burnus@net-b.de>
813 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
816 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
818 * doc/tm.texi.in (Condition Code Status): Update documention for
819 relative locations of cc0-setter and cc0-user.
821 2014-02-27 Jeff Law <law@redhat.com>
823 PR rtl-optimization/52714
824 * combine.c (try_combine): When splitting an unrecognized PARALLEL
825 into two independent simple sets, if I3 is a jump, ensure the
826 pattern we place into I3 is a (set (pc) ...).
828 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
829 Jeff Law <law@redhat.com>
831 PR rtl-optimization/49847
832 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
833 are in different blocks.
834 * doc/tm.texi (Condition Code Status): Update documention for
835 relative locations of cc0-setter and cc0-user.
837 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
840 * lra.c (lra_emit_add): Check SUBREG too.
842 2014-02-27 Andreas Schwab <schwab@suse.de>
844 * config/m68k/m68k.c (m68k_option_override): Disable
845 -flive-range-shrinkage for classic m68k.
846 (m68k_override_options_after_change): Likewise.
848 2014-02-27 Marek Polacek <polacek@redhat.com>
851 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
852 -Wmaybe-uninitialized.
854 2014-02-27 Alan Modra <amodra@gmail.com>
857 * reload1.c (emit_input_reload_insns): When reload_override_in,
858 set old to rl->in_reg when rl->in_reg is a subreg.
860 2014-02-26 Richard Biener <rguenther@suse.de>
863 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
865 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
867 * common/config/i386/predicates.md (const1256_operand): Remove.
868 (const2356_operand): New.
869 (const_1_to_2_operand): Remove.
870 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
871 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
872 (*avx512pf_gatherpf<mode>sf): Ditto.
873 (avx512pf_gatherpf<mode>df): Ditto.
874 (*avx512pf_gatherpf<mode>df_mask): Ditto.
875 (*avx512pf_gatherpf<mode>df): Ditto.
876 (avx512pf_scatterpf<mode>sf): Ditto.
877 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
878 (*avx512pf_scatterpf<mode>sf): Ditto.
879 (avx512pf_scatterpf<mode>df): Ditto.
880 (*avx512pf_scatterpf<mode>df_mask): Ditto.
881 (*avx512pf_scatterpf<mode>df): Ditto.
882 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
884 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
886 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
887 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
888 (_mm512_mask_testn_epi64_mask): Move to ...
889 * config/i386/avx512cdintrin.h: Here.
890 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
891 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
892 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
893 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
894 TARGET_AVX512F from TARGET_AVX512CD.
896 2014-02-26 Richard Biener <rguenther@suse.de>
899 * ipa.c (walk_polymorphic_call_targets): Properly guard
900 call to inline_update_overall_summary.
902 2014-02-26 Bin Cheng <bin.cheng@arm.com>
905 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
906 and latches only if requested. Fix latch if it is removed.
907 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
908 LOOPS_HAVE_PREHEADERS.
910 2014-02-25 Andrew Pinski <apinski@cavium.com>
912 * builtins.c (expand_builtin_thread_pointer): Create a new target
913 when the target is NULL.
915 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
917 PR rtl-optimization/60317
918 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
919 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
920 * lra-assigns.c: Include params.h.
921 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
922 other reload pseudos considerations.
924 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
926 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
927 to use canonical form for nor<mode>3.
929 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
932 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
935 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
937 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
938 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
939 (ix86_handle_option): Handle OPT_mprefetchwt1.
940 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
941 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
943 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
944 OPTION_MASK_ISA_PREFETCHWT1.
945 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
946 (PTA_PREFETCHWT1): New.
947 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
948 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
949 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
950 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
951 (*prefetch_avx512pf_<mode>_: Change into ...
952 (*prefetch_prefetchwt1_<mode>: This.
953 * config/i386/i386.opt (mprefetchwt1): New.
954 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
955 (_mm_prefetch): Handle intent to write.
956 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
958 2014-02-25 Richard Biener <rguenther@suse.de>
961 * emit-rtl.c (mem_attrs_htab): Remove.
962 (mem_attrs_htab_hash): Likewise.
963 (mem_attrs_htab_eq): Likewise.
964 (set_mem_attrs): Always allocate new mem-attrs when something changed.
965 (init_emit_once): Do not allocate mem_attrs_htab.
967 2014-02-25 Richard Biener <rguenther@suse.de>
970 * lto-opts.c (lto_write_options): Output non-explicit conservative
971 -fwrapv, -fno-trapv and -fno-strict-overflow.
972 * lto-wrapper.c (merge_and_complain): Handle merging those options.
973 (run_gcc): And pass them through.
975 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
977 * sel-sched.c (calculate_new_fences): New parameter ptime.
978 Calculate it as a maximum over all fence cycles.
979 (sel_sched_region_2): Adjust the call to calculate_new_fences.
980 Print the final schedule timing when sched_verbose.
982 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
984 PR rtl-optimization/60292
985 * sel-sched.c (fill_vec_av_set): Do not reset target availability
986 bit fot the fence instruction.
988 2014-02-24 Alangi Derick <alangiderick@gmail.com>
990 * calls.h: Fix typo in comment.
992 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
994 * config/pa/pa.c (pa_output_move_double): Don't valididate when
995 adjusting offsetable addresses.
997 2014-02-24 Guozhi Wei <carrot@google.com>
999 * sparseset.h (sparseset_pop): Fix the wrong index.
1001 2014-02-24 Walter Lee <walt@tilera.com>
1003 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1004 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1006 * common/config/tilegx/tilegx-common.c
1007 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1008 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1010 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1011 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1012 (tilegx_gimplify_va_arg_expr): Handle big endian.
1013 (tilegx_expand_unaligned_load): Ditto.
1014 (tilegx_expand_unaligned_store): Ditto.
1015 (TARGET_RETURN_IN_MSB): New.
1016 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1017 (TARGET_ENDIAN_DEFAULT): New.
1018 (TARGET_BIG_ENDIAN): Handle big endian.
1019 (BYTES_BIG_ENDIAN): Ditto.
1020 (WORDS_BIG_ENDIAN): Ditto.
1021 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1024 * config/tilegx/tilegx.md (extv): Handle big endian.
1026 (insn_st<n>): Ditto.
1027 (insn_st<n>_add<bitsuffix>): Ditto.
1028 (insn_stnt<n>): Ditto.
1029 (insn_stnt<n>_add<bitsuffix>):Ditto.
1030 (vec_interleave_highv8qi): Handle big endian.
1031 (vec_interleave_highv8qi_be): New.
1032 (vec_interleave_highv8qi_le): New.
1033 (insn_v1int_h): Handle big endian.
1034 (vec_interleave_lowv8qi): Handle big endian.
1035 (vec_interleave_lowv8qi_be): New.
1036 (vec_interleave_lowv8qi_le): New.
1037 (insn_v1int_l): Handle big endian.
1038 (vec_interleave_highv4hi): Handle big endian.
1039 (vec_interleave_highv4hi_be): New.
1040 (vec_interleave_highv4hi_le): New.
1041 (insn_v2int_h): Handle big endian.
1042 (vec_interleave_lowv4hi): Handle big endian.
1043 (vec_interleave_lowv4hi_be): New.
1044 (vec_interleave_lowv4hi_le): New.
1045 (insn_v2int_l): Handle big endian.
1046 (vec_interleave_highv2si): Handle big endian.
1047 (vec_interleave_highv2si_be): New.
1048 (vec_interleave_highv2si_le): New.
1049 (insn_v4int_h): Handle big endian.
1050 (vec_interleave_lowv2si): Handle big endian.
1051 (vec_interleave_lowv2si_be): New.
1052 (vec_interleave_lowv2si_le): New.
1053 (insn_v4int_l): Handle big endian.
1054 * config/tilegx/tilegx.opt (mbig-endian): New option.
1055 (mlittle-endian): New option.
1056 * doc/install.texi: Document tilegxbe-linux.
1057 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1059 2014-02-24 Martin Jambor <mjambor@suse.cz>
1062 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1063 there are no parameter descriptors.
1065 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1067 PR rtl-optimization/60268
1068 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1069 initialization to ...
1070 (sched_rgn_init): ... here.
1071 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1073 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1075 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1078 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1080 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1083 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1085 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1086 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1088 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1090 * config/microblaze/predicates.md: Add cmp_op predicate.
1091 * config/microblaze/microblaze.md: Add branch_compare instruction
1092 which uses cmp_op predicate and emits cmp insn before branch.
1093 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1094 to microblaze_expand_conditional_branch and consolidate logic.
1095 (microblaze_expand_conditional_branch): emit branch_compare
1096 insn instead of handling cmp op separate from branch insn.
1098 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1100 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1103 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1105 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1106 define_insn with define_expand and new define_insn
1107 *altivec_lve<VI_char>x_internal.
1108 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1109 and new define_insn *altivec_stve<VI_char>x_internal.
1110 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1112 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1114 (altivec_expand_stvex_be): New function.
1116 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1118 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1119 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1120 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1121 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1123 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1126 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1127 instead of emit_move_insn.
1129 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1131 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1133 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1134 gen_altivec_vsumsws.
1136 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1138 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1139 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1140 (altivec_lvxl_<mode>): New define_expand incorporating
1141 -maltivec=be semantics where needed.
1142 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1143 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1144 semantics where needed.
1145 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1146 (altivec_stvx_<mode>): New define_expand incorporating
1147 -maltivec=be semantics where needed.
1148 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1149 VM2 iterator instead of V4SI.
1150 (altivec_stvxl_<mode>): New define_expand incorporating
1151 -maltivec=be semantics where needed.
1152 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1153 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1154 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1155 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1156 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1157 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1158 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1159 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1160 ALTIVEC_BUILTIN_STVXL.
1161 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1162 (altivec_expand_stvx_be): Likewise.
1163 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1164 (altivec_expand_lvx_be): Likewise.
1165 (altivec_expand_stvx_be): Likewise.
1166 (altivec_expand_builtin): Add cases for
1167 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1168 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1169 (altivec_init_builtins): Add definitions for
1170 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1171 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1173 2014-02-21 Catherine Moore <clm@codesourcery.com>
1175 * doc/invoke.texi (mvirt, mno-virt): Document.
1176 * config/mips/mips.opt (mvirt): New option.
1177 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1179 2014-02-21 Richard Biener <rguenther@suse.de>
1181 PR tree-optimization/60276
1182 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1183 (STMT_VINFO_MIN_NEG_DIST): New macro.
1184 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1185 STMT_VINFO_MIN_NEG_DIST.
1186 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1187 made for negative dependence distances still hold.
1189 2014-02-21 Richard Biener <rguenther@suse.de>
1192 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1193 DECL_INITIAL for globals not in the current function context.
1195 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1197 PR tree-optimization/56490
1198 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1199 * tree-ssa-uninit.c: Include params.h.
1200 (compute_control_dep_chain): Add num_calls argument, return false
1201 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1202 num_calls to recursive call.
1203 (find_predicates): Change dep_chain into normal array,
1204 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1205 variable and adjust compute_control_dep_chain caller.
1206 (find_def_preds): Likewise.
1208 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1210 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1211 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1213 2014-02-21 Nick Clifton <nickc@redhat.com>
1215 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1216 (pushhi1): Likewise.
1217 (popqi1): Add mode to pre_dec.
1220 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1222 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1223 mode for mask of V8SFmode permutation.
1225 2014-02-20 Richard Henderson <rth@redhat.com>
1228 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1229 a new pseudo for OLDVAL.
1231 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1234 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1235 gen_reg_rtx if d->testing_p.
1236 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1237 if d->testing_p and we will certainly return true.
1238 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1241 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1243 * emit-rtl.c (gen_reg_rtx): Assert that
1244 crtl->emit.regno_pointer_align_length is non-zero.
1246 2014-02-20 Richard Henderson <rth@redhat.com>
1249 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1250 on failure the store back into EXPECT.
1252 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1253 Sandra Loosemore <sandra@codesourcery.com>
1255 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1256 * config/nios2/nios2.c (nios2_function_profiler): Add
1257 -fPIC (flag_pic == 2) support.
1258 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1259 (nios2_large_offset_p): New function.
1260 (nios2_unspec_reloc_p): Move up position, update to use
1261 nios2_large_offset_p.
1262 (nios2_unspec_address): Remove function.
1263 (nios2_unspec_offset): New function.
1264 (nios2_large_got_address): New function.
1265 (nios2_got_address): Add large offset support.
1266 (nios2_legitimize_tls_address): Update usage of removed and new
1268 (nios2_symbol_binds_local_p): New function.
1269 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1270 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1271 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1272 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1273 processing for (const (unspec ...)).
1274 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1276 2014-02-20 Richard Biener <rguenther@suse.de>
1278 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1279 doing the substitution.
1280 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1282 2014-02-20 Martin Jambor <mjambor@suse.cz>
1285 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1286 info when checking whether lattices are bottom.
1288 2014-02-20 Richard Biener <rguenther@suse.de>
1291 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1294 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1297 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1298 parameter specifying the scaling.
1299 (inline_call): Update.
1300 (want_inline_recursively): Guard division by zero.
1301 (recursive_inlining): Update.
1302 * ipa-inline.h (clone_inlined_nodes): Update.
1304 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1307 * config/i386/i386.c (classify_argument): Pass structures of size
1308 64 bytes or less in register.
1310 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1311 Kirill Yukhin <kirill.yukhin@intel.com>
1313 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1314 (_mm_rcp28_round_ss): Ditto.
1315 (_mm_rsqrt28_round_sd): Ditto.
1316 (_mm_rsqrt28_round_ss): Ditto.
1317 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1318 (_mm_rcp14_round_ss): Ditto.
1319 (_mm_rsqrt14_round_sd): Ditto.
1320 (_mm_rsqrt14_round_ss): Ditto.
1321 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1322 the first input operand, get rid of match_dup.
1323 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1325 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1327 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1328 operand as the first input operand, set type attribute.
1329 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1331 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1332 operand as the first input operand, set type attribute.
1334 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1336 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1339 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1342 * config/i386/i386.c (construct_container): Remove TFmode check
1343 for X86_64_INTEGER_CLASS.
1345 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1348 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1349 only when -Wpsabi is enabled.
1351 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1354 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1355 passing arrays in registers are the same as for structs, so remove the
1356 special case for them.
1358 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1360 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1361 destination type, extract only the valid bits if the source type is not
1362 integral and has a different mode.
1364 2014-02-19 Richard Biener <rguenther@suse.de>
1367 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1370 2014-02-19 Richard Biener <rguenther@suse.de>
1373 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1374 (ipa_modify_call_arguments): Emit an argument load explicitely and
1375 preserve virtual SSA form there and for the replacement call.
1376 Do not update SSA form nor free dominance info.
1378 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1380 * ipa.c (function_and_variable_visibility): Also clear WEAK
1381 flag when disolving COMDAT_GROUP.
1383 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1385 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1386 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1388 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1389 do more sanity checks.
1390 (detect_type_change): Return true when giving up early.
1391 (compute_complex_assign_jump_func): Fix type parameter of
1392 ipa_set_ancestor_jf.
1393 (compute_complex_ancestor_jump_func): Likewise.
1394 (update_jump_functions_after_inlining): Fix updating of
1396 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1398 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1400 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1401 inline clones when edge disappears.
1403 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1406 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1407 Split 64-bit moves into 2 patterns. Do not allow the use of
1408 direct move for TDmode in little endian, since the decimal value
1409 has little endian bytes within a word, but the 64-bit pieces are
1410 ordered in a big endian fashion, and normal subreg's of TDmode are
1412 (mov<mode>_64bit_dm): Likewise.
1413 (movtd_64bit_nodm): Likewise.
1415 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1417 PR tree-optimization/60174
1418 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1419 statement of an SSA_NAME that occurs in an abnormal PHI node.
1421 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1424 * final.c (SEEN_BB): Remove.
1425 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1426 (final_scan_insn): Don't force_source_line on second
1427 NOTE_INSN_BASIC_BLOCK.
1429 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1432 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1433 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1434 (type_natural_mode): Warn ABI change when %zmm register is not
1435 available for AVX512F vector value passing.
1437 2014-02-18 Kai Tietz <ktietz@redhat.com>
1440 * config/i386/i386.c (ix86_expand_prologue): Use value in
1441 rax register as displacement when restoring %r10 or %rax.
1442 Fix wrong offset when restoring both registers.
1444 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1446 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1447 assertion with conditional return.
1449 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1450 Uros Bizjak <ubizjak@gmail.com>
1453 * config/i386/driver-i386.c (host_detect_local_cpu): If
1454 YMM state is not saved by the OS, also clear has_f16c. Move
1455 CPUID 0x80000001 handling before YMM state saving checking.
1457 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1459 PR rtl-optimization/58960
1460 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1461 factored out from ...
1462 (sched_init): ... here.
1463 (free_global_sched_pressure_data): New, factored out from ...
1464 (sched_finish): ... here.
1465 * sched-int.h (free_global_sched_pressure_data): Declare.
1466 * sched-rgn.c (nr_regions_initial): New static global.
1467 (haifa_find_rgns): Initialize it.
1468 (schedule_region): Disable sched-pressure for the newly
1471 2014-02-17 Richard Biener <rguenther@suse.de>
1473 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1474 release SSA defs of pattern stmts.
1476 2014-02-17 Richard Biener <rguenther@suse.de>
1478 * tree-inline.c (expand_call_inline): Release the virtual
1479 operand defined by the call we are about to inline.
1481 2014-02-17 Richard Biener <rguenther@suse.de>
1483 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1485 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1486 Ilya Tocar <ilya.tocar@intel.com>
1488 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1489 arguments order in builtin.
1490 (_mm512_permutexvar_epi64): Ditto.
1491 (_mm512_mask_permutexvar_epi64): Ditto
1492 (_mm512_maskz_permutexvar_epi32): Ditto
1493 (_mm512_permutexvar_epi32): Ditto
1494 (_mm512_mask_permutexvar_epi32): Ditto
1496 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1498 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1499 (p8_vmrgow): Likewise.
1501 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1503 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1506 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1509 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1510 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1511 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1512 using direct move instructions on ISA 2.07. Also adjust
1513 instruction length for 64-bit.
1514 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1515 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1517 2014-02-15 Alan Modra <amodra@gmail.com>
1521 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1522 find_replacement on parts of insn rtl that might be reloaded.
1524 2014-02-15 Richard Biener <rguenther@suse.de>
1526 PR tree-optimization/60183
1527 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1528 (tree_ssa_phiprop): Calculate and free post-dominators.
1530 2014-02-14 Jeff Law <law@redhat.com>
1532 PR rtl-optimization/60131
1533 * ree.c (get_extended_src_reg): New function.
1534 (combine_reaching_defs): Use it rather than assuming location of REG.
1535 (find_and_remove_re): Verify first operand of extension is
1536 a REG before adding the insns to the copy list.
1538 2014-02-14 Roland McGrath <mcgrathr@google.com>
1540 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1541 * configure: Regenerated.
1542 * config.in: Regenerated.
1543 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1544 instead of ASM_SHORT.
1546 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1547 Richard Earnshaw <rearnsha@arm.com>
1549 PR rtl-optimization/59535
1550 * lra-constraints.c (process_alt_operands): Encourage alternative
1551 when unassigned pseudo class is superset of the alternative class.
1552 (inherit_reload_reg): Don't inherit when optimizing for code size.
1553 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1554 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1555 modes not less than 4 for Thumb1.
1557 2014-02-14 Kyle McMartin <kyle@redhat.com>
1560 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1562 2014-02-14 Richard Biener <rguenther@suse.de>
1564 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1565 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1566 Do not strip INDIRECT_REFs.
1568 2014-02-14 Richard Biener <rguenther@suse.de>
1571 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1572 DECL_FUNCTION_SPECIFIC_TARGET.
1573 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1574 * tree-streamer-out.c (pack_ts_target_option): Remove.
1575 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1576 (write_ts_function_decl_tree_pointers): Do not stream
1577 DECL_FUNCTION_SPECIFIC_TARGET.
1578 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1579 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1580 (lto_input_ts_function_decl_tree_pointers): Do not stream
1581 DECL_FUNCTION_SPECIFIC_TARGET.
1583 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1585 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1586 (get_initial_def_for_induction, vectorizable_induction): Ignore
1587 debug stmts when looking for exit_phi.
1588 (vectorizable_live_operation): Fix up condition.
1590 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1592 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1593 nreverse() because it changes the content of original tree list.
1595 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1597 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1598 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1600 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1602 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1603 GNU coding standards.
1605 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1608 * dwarf2out.c (gen_subprogram_die): Don't call
1609 add_calling_convention_attribute if subr_die is old_die.
1611 2014-02-13 Sharad Singhai <singhai@google.com>
1613 * doc/optinfo.texi: Fix order of nodes.
1615 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1617 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1618 operands[2], not operands[3].
1620 2014-02-13 Richard Biener <rguenther@suse.de>
1623 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1624 mention the possibility of an in-tree build.
1625 (CLooG): Update recommended version to 0.18.1, mention the
1626 possibility of an in-tree build and clarify that the ISL
1627 bundled with CLooG does not work.
1629 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1632 * expr.c (compress_float_constant): If x is a hard register,
1633 extend into a pseudo and then move to x.
1635 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1637 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1638 caused by bad second argument to warning_at() with -mhotpatch and
1639 nested functions (e.g. with gfortran).
1641 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1643 * opts.c (option_name): Remove "enabled by default" rider.
1645 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1647 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1649 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1650 Uros Bizjak <ubizjak@gmail.com>
1653 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1654 * configure: Regenerated.
1656 2014-02-12 Richard Biener <rguenther@suse.de>
1658 * vec.c (vec_prefix::calculate_allocation): Move as
1659 inline variant to vec.h.
1660 (vec_prefix::calculate_allocation_1): New out-of-line version.
1661 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1662 (vec_prefix::m_has_auto_buf): Rename to ...
1663 (vec_prefix::m_using_auto_storage): ... this.
1664 (vec_prefix::calculate_allocation): Inline the easy cases
1665 and dispatch to calculate_allocation_1 which doesn't need the
1667 (va_heap::reserve): Use gcc_checking_assert.
1668 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1669 m_using_auto_storage.
1670 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1672 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1673 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1674 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1676 2014-02-12 Richard Biener <rguenther@suse.de>
1678 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1679 when we found a dependence.
1681 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1683 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1685 (maybe_fold_stmt): ... into this new function.
1686 * omp-low.c (lower_omp): Update comment.
1688 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1691 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1694 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1696 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1697 identifiers in comments.
1698 (cortexa53_extra_costs): Likewise.
1699 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1700 (cortexa7_extra_costs): Likewise.
1701 (cortexa12_extra_costs): Likewise.
1702 (cortexa15_extra_costs): Likewise.
1703 (v7m_extra_costs): Likewise.
1705 2014-02-12 Richard Biener <rguenther@suse.de>
1708 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1709 of posix_memalign being successful.
1710 (lower_stmt): Restrict lowering of posix_memalign to when
1711 -ftree-bit-ccp is enabled.
1713 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1715 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1717 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1719 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1721 PR rtl-optimization/60116
1722 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1723 other_insn once the combination has been validated.
1725 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1728 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1730 * ipa-devirt.c: Include demangle.h
1731 (odr_violation_reported): New static variable.
1732 (add_type_duplicate): Update odr_violations.
1733 (maybe_record_node): Add completep parameter; update it.
1734 (record_target_from_binfo): Add COMPLETEP parameter;
1735 update it as needed.
1736 (possible_polymorphic_call_targets_1): Likewise.
1737 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1738 rename FINAL to COMPLETE.
1739 (record_targets_from_bases): Sanity check we found the binfo;
1740 fix COMPLETEP updating.
1741 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1742 parameter, fix computing of COMPLETEP.
1743 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1744 at LTO time do demangling.
1745 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1746 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1748 (gimple_get_virt_method_for_binfo): Likewise.
1749 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1750 gimple_get_virt_method_for_vtable): Update prototypes.
1752 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1755 * genautomata.c (add_presence_absence): Fix typo with
1756 {final_}presence_list.
1758 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1761 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1762 for VSX/Altivec vectors that land in GPR registers.
1764 2014-02-11 Richard Henderson <rth@redhat.com>
1765 Jakub Jelinek <jakub@redhat.com>
1768 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1769 around drhs if type conversion to lacc->type is not useless.
1771 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1773 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1775 (cortex-a57.cortex-a53): Likewise.
1776 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1778 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1780 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1783 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1785 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1786 add_options_for_arm_vfp3.
1788 2014-02-11 Jeff Law <law@redhat.com>
1791 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1792 object with an undesirable mode.
1794 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1797 * config/i386/sol2-9.h: New file.
1798 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1799 *-*-solaris2.9*): Use it.
1801 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1803 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1804 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1806 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1808 * config/microblaze/microblaze.c: Extend mcpu version format
1810 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1812 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1814 2014-02-10 Richard Henderson <rth@redhat.com>
1817 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1818 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1819 ms-abi vs -mno-accumulate-outgoing-args.
1820 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1821 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1824 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1827 * cfgexpand.c (expand_asm_operands): Attach source location to
1828 ASM_INPUT rtx objects.
1829 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1831 2014-02-10 Nick Clifton <nickc@redhat.com>
1833 * config/mn10300/mn10300.c (popcount): New function.
1834 (mn10300_expand_prologue): Include saved registers in stack usage
1837 2014-02-10 Jeff Law <law@redhat.com>
1840 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1841 when changing the SET_DEST of a prior insn to avoid an input reload.
1843 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1845 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1846 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1847 -mcall-openbsd, or -mcall-linux.
1848 (CC1_ENDIAN_BIG_SPEC): Remove.
1849 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1850 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1851 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1852 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1853 and %cc1_endian_default.
1854 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1856 2014-02-10 Richard Biener <rguenther@suse.de>
1858 PR tree-optimization/60115
1859 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1860 MEM_REF handling. Properly verify that the accesses are not
1861 out of the objects bound.
1863 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1868 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1870 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1871 proper constants and fix formatting.
1872 (possible_polymorphic_call_targets): Fix formatting.
1874 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1875 Ilya Tocar <ilya.tocar@intel.com>
1877 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1878 (_mm512_loadu_epi32): Renamed into...
1879 (_mm512_loadu_si512): This.
1880 (_mm512_storeu_epi32): Renamed into...
1881 (_mm512_storeu_si512): This.
1882 (_mm512_maskz_ceil_ps): Removed.
1883 (_mm512_maskz_ceil_pd): Ditto.
1884 (_mm512_maskz_floor_ps): Ditto.
1885 (_mm512_maskz_floor_pd): Ditto.
1886 (_mm512_floor_round_ps): Ditto.
1887 (_mm512_floor_round_pd): Ditto.
1888 (_mm512_ceil_round_ps): Ditto.
1889 (_mm512_ceil_round_pd): Ditto.
1890 (_mm512_mask_floor_round_ps): Ditto.
1891 (_mm512_mask_floor_round_pd): Ditto.
1892 (_mm512_mask_ceil_round_ps): Ditto.
1893 (_mm512_mask_ceil_round_pd): Ditto.
1894 (_mm512_maskz_floor_round_ps): Ditto.
1895 (_mm512_maskz_floor_round_pd): Ditto.
1896 (_mm512_maskz_ceil_round_ps): Ditto.
1897 (_mm512_maskz_ceil_round_pd): Ditto.
1898 (_mm512_expand_pd): Ditto.
1899 (_mm512_expand_ps): Ditto.
1900 * config/i386/i386.c (ix86_builtins): Remove
1901 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1902 (bdesc_args): Ditto.
1903 * config/i386/predicates.md (const1256_operand): New.
1904 (const_1_to_2_operand): Ditto.
1905 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1906 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1907 (*avx512pf_gatherpf<mode>sf): Ditto.
1908 (avx512pf_gatherpf<mode>df): Ditto.
1909 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1910 (*avx512pf_gatherpf<mode>df): Ditto.
1911 (avx512pf_scatterpf<mode>sf): Ditto.
1912 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1913 (*avx512pf_scatterpf<mode>sf): Ditto.
1914 (avx512pf_scatterpf<mode>df): Ditto.
1915 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1916 (*avx512pf_scatterpf<mode>df): Ditto.
1917 (avx512f_expand<mode>): Removed.
1918 (<shift_insn><mode>3<mask_name>): Change predicate type.
1920 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1922 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1923 not at the end of datarefs vector use ordered_remove to avoid
1924 reordering datarefs vector.
1927 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1928 mark local addressable non-static vars as GOVD_PRIVATE
1929 instead of GOVD_LOCAL.
1930 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1931 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1935 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1936 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1937 assume_aligned or alloc_align attributes.
1938 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1939 arguments. Handle also assume_aligned and alloc_align attributes.
1940 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1941 calls to functions with assume_aligned or alloc_align attributes.
1942 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1944 2014-02-08 Terry Guo <terry.guo@arm.com>
1946 * doc/invoke.texi: Document ARM -march=armv7e-m.
1948 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1950 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1951 flag on __cilkrts_rethrow builtin.
1954 * ipa-cp.c (determine_versionability): Fail at -O0
1955 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1956 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1959 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1962 * tree-inline.c (copy_forbidden): Fail for
1963 __attribute__((optimize (0))) functions.
1965 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1967 * varpool.c: Include pointer-set.h.
1968 (varpool_remove_unreferenced_decls): Variables in other partitions
1969 will not be output; be however careful to not lose information
1972 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1974 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1975 lookup in the vtable constructor.
1977 2014-02-07 Jeff Law <law@redhat.com>
1980 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1981 define_insn_and_split.
1983 * ipa-inline.c (inline_small_functions): Fix typos.
1985 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1987 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1988 (s390_can_use_return_insn): Declare.
1989 * config/s390/s390.h (EPILOGUE_USES): Define.
1990 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1992 (s390_chunkify_start): Handle return JUMP_LABELs.
1993 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1994 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1995 (s390_can_use_return_insn): New functions.
1996 (s390_fix_long_loop_prediction): Handle conditional returns.
1997 (TARGET_SET_UP_BY_PROLOGUE): Define.
1998 * config/s390/s390.md (ANY_RETURN): New code iterator.
1999 (*creturn, *csimple_return, return, simple_return): New patterns.
2001 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2003 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2004 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2005 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2006 REG_CFA_RESTORE list when deciding not to restore a register.
2008 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2010 * config/s390/s390.c: Include tree-pass.h and context.h.
2011 (s390_early_mach): New function, split out from...
2012 (s390_emit_prologue): ...here.
2013 (pass_data_s390_early_mach): New pass structure.
2014 (pass_s390_early_mach): New class.
2015 (s390_option_override): Create and register early_mach pass.
2016 Move to end of file.
2018 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2020 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2021 to match for the exit block.
2023 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2025 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2026 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2027 Reject misaligned operands.
2029 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2031 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2033 2014-02-07 Richard Biener <rguenther@suse.de>
2036 * gimple-low.c (lower_builtin_posix_memalign): New function.
2037 (lower_stmt): Call it to lower posix_memalign in a way
2038 to make alignment info accessible.
2040 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2043 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2044 __builtin_setjmp_receiver.
2046 2014-02-07 Richard Biener <rguenther@suse.de>
2049 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2050 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2051 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2052 Handle BUILT_IN_POSIX_MEMALIGN.
2053 (find_func_clobbers): Likewise.
2054 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2055 (call_may_clobber_ref_p_1): Likewise.
2057 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2060 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2063 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2066 * lto-cgraph.c (lto_output_node): Use
2067 symtab_get_symbol_partitioning_class.
2068 (lto_output_varpool_node): likewise.
2069 (symtab_get_symbol_partitioning_class): Move here from
2071 * cgraph.h (symbol_partitioning_class): Likewise.
2072 (symtab_get_symbol_partitioning_class): Declare.
2074 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2076 * ggc.h (ggc_internal_cleared_alloc): New macro.
2077 * vec.h (vec_safe_copy): Handle memory stats.
2078 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2079 * target-globals.c (save_target_globals): Likewise.
2081 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2084 * expr.c (emit_move_resolve_push): Export; be bit more selective
2085 on when to clear alias set.
2086 * expr.h (emit_move_resolve_push): Declare.
2087 * function.h (struct function): Add tail_call_marked.
2088 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2089 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2090 * config/i386/i386.md (TImode move expander): De not call
2092 (FP push expanders): Preserve memory attributes.
2093 * config/i386/sse.md (push<mode>1): Remove.
2094 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2095 (ix86_expand_push): Remove.
2096 * config/i386/mmx.md (push<mode>1): Remove.
2098 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2100 PR rtl-optimization/60030
2101 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2102 lopart with paradoxical subreg before shifting it up by hprec.
2104 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2106 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2107 Remove extra newline at end of file.
2108 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2109 (arm_issue_rate): Handle cortexa57.
2110 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2111 (cortex-a57.cortex-a53): Likewise.
2113 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2116 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2117 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2119 (arm_expand_prologue): Adjust all callers.
2120 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2121 info, registers also at the lowest numbered registers side. Use
2122 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2126 * var-tracking.c (adjust_mems): Before adding a SET to
2127 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2129 2014-02-06 Alan Modra <amodra@gmail.com>
2132 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2133 change SDmode to DDmode when lra_in_progress.
2135 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2138 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2139 free_data_ref on the dr first, and before goto again also set dr
2140 to the next dr. For simd_lane_access, free old datarefs[i] before
2141 overwriting it. For get_vectype_for_scalar_type failure, don't
2142 free_data_ref if simd_lane_access.
2144 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2147 * tree.h (opts_for_fn): New inline function.
2148 (opt_for_fn): Define.
2149 * config/i386/i386.c (ix86_function_regparm): Use
2150 opt_for_fn (decl, optimize) instead of optimize.
2152 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2154 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2155 for SYMBOL_REF in large memory model.
2157 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2159 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2161 (cortex-a57): Likewise.
2162 (cortex-a57.cortex-a53): Likewise.
2164 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2165 Kugan Vivekanandarajah <kuganv@linaro.org>
2167 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2169 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2171 2014-02-06 Richard Biener <rguenther@suse.de>
2173 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2174 set_loop_copy and initialize_original_copy_tables.
2176 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2178 * config/aarch64/aarch64-simd.md
2179 (aarch64_ashr_simddi): Change QI to SI.
2181 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2182 Jakub Jelinek <jakub@redhat.com>
2185 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2188 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2190 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2191 CODE_FOR_altivec_vpku[hw]um to
2192 CODE_FOR_altivec_vpku[hw]um_direct.
2193 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2194 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2195 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2196 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2198 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2200 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2201 generation for -maltivec=be.
2202 (altivec_vsumsws): Simplify redundant test.
2204 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2206 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2207 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2208 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2209 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2210 gen_altivec_vpkuwum.
2211 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2213 (altivec_vpks<VI_char>ss): Likewise.
2214 (altivec_vpks<VI_char>us): Likewise.
2215 (altivec_vpku<VI_char>us): Likewise.
2216 (altivec_vpku<VI_char>um): Likewise.
2217 (altivec_vpku<VI_char>um_direct): New (copy of
2218 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2220 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2221 target is little endian and -maltivec=be is not specified.
2222 (*altivec_vupkhs<VU_char>_direct): New (copy of
2223 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2224 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2225 target is little endian and -maltivec=be is not specified.
2226 (*altivec_vupkls<VU_char>_direct): New (copy of
2227 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2228 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2229 little endian and -maltivec=be is not specified.
2230 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2231 little endian and -maltivec=be is not specified.
2233 2014-02-05 Richard Henderson <rth@redhat.com>
2236 * combine-stack-adj.c: Revert r206943.
2237 * sched-int.h (struct deps_desc): Add last_args_size.
2238 * sched-deps.c (init_deps): Initialize it.
2239 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2240 contain REG_ARGS_SIZE notes.
2242 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2244 * lto-cgraph.c (asm_nodes_output): Make global.
2245 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2246 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2247 (driver_handle_option): Handle OPT_fwpa.
2249 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2252 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2253 a comment typo and formatting issue. If odr_hash hasn't been
2254 created, return vNULL and set *completep to false.
2257 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2258 bb with no successors.
2260 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2263 * doc/invoke.texi (-march): Clarify documentation for ARM.
2267 2014-02-05 Richard Biener <rguenther@suse.de>
2269 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2270 when not vectorizing because of too many alias checks.
2271 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2272 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2274 2014-02-05 Nick Clifton <nickc@redhat.com>
2276 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2277 accept extended registers in any mode when compiling for the MN10300.
2279 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2281 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2282 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2283 sanitization attributes.
2284 (can_inline_edge_p): Likewise.
2285 (sanitize_attrs_match_for_inline_p): New function.
2287 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2289 * ipa-prop.c (detect_type_change): Shor circuit testing of
2290 type changes on THIS pointer.
2292 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2295 * config/pa/pa.c (legitimize_tls_address): Return original address
2296 if not passed a SYMBOL_REF rtx.
2297 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2299 (pa_emit_move_sequence): Simplify TLS source operands.
2300 (pa_legitimate_constant_p): Reject all TLS constants.
2301 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2302 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2304 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2306 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2307 groups when we know they are controlled by LTO.
2308 * varasm.c (default_binds_local_p_1): If object is in other partition,
2309 it will be resolved locally.
2311 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2313 * config/host-linux.c (linux_gt_pch_use_address): Don't
2314 use SSIZE_MAX because it is not always defined.
2316 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2319 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2320 threshold for pseudo splitting.
2321 (update_ebb_live_info): Process call argument hard registers and
2322 hard registers from insn definition too.
2323 (max_small_class_regs_num): New constant.
2324 (inherit_in_ebb): Update live hard regs through EBBs. Update
2325 reloads_num only for small register classes. Don't split for
2328 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2331 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2334 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2336 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2339 2014-02-04 Marek Polacek <polacek@redhat.com>
2341 * gdbinit.in (pel): Define.
2343 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2345 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2348 2014-02-04 Richard Biener <rguenther@suse.de>
2351 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2352 in function context local.
2353 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2354 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2355 similar to LTO_imported_decl_ref.
2357 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2359 PR tree-optimization/60002
2360 * cgraphclones.c (build_function_decl_skip_args): Clear
2363 PR tree-optimization/60023
2364 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2365 false to gsi_replace.
2366 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2367 has been in some EH region and vec_stmt could throw, add
2368 vec_stmt into the same EH region.
2369 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2370 has no lhs, ignore it.
2371 * internal-fn.c (expand_MASK_LOAD): Likewise.
2374 * tree-inline.c (copy_forbidden): Fail for
2375 __attribute__((optimize (0))) functions.
2378 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2379 is set, copy one less argument.
2380 (expand_simd_clones): Don't subtract clone_info->inbranch
2381 from simd_clone_struct_alloc argument.
2383 PR rtl-optimization/57915
2384 * recog.c (simplify_while_replacing): If all unary/binary/relational
2385 operation arguments are constant, attempt to simplify those.
2388 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2389 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2391 2014-02-04 Richard Biener <rguenther@suse.de>
2393 PR tree-optimization/60012
2394 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2395 TBAA disambiguation to all DDRs.
2397 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2400 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2401 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2403 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2406 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2408 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2410 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2411 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2413 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2416 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2417 to figure out targets of polymorphic calls with known decl.
2418 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2419 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2420 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2421 (get_polymorphic_call_info): ... here.
2422 (get_polymorphic_call_info_from_invariant): New function.
2424 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2426 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2427 lookup via vtable pointer; check for type consistency
2428 and turn inconsitent facts into UNREACHABLE.
2429 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2430 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2431 type inconsistent querries; return UNREACHABLE instead.
2433 2014-02-03 Richard Henderson <rth@twiddle.net>
2436 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2437 already processed this node.
2438 (normalize_one_pred_1): Pass along mark_set.
2439 (normalize_one_pred): Create and destroy a pointer_set_t.
2440 (normalize_one_pred_chain): Likewise.
2442 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2444 PR gcov-profile/58602
2445 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2447 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2450 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2451 -fno-devirtualize; try to devirtualize by the knowledge of
2452 virtual table pointer given by aggregate propagation.
2453 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2454 (ipa_print_node_jump_functions): Dump also offset that
2455 is relevant for polymorphic calls.
2456 (determine_known_aggregate_parts): Add arg_type parameter; use it
2457 instead of determining the type from pointer type.
2458 (ipa_compute_jump_functions_for_edge): Update call of
2459 determine_known_aggregate_parts.
2460 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2461 (gimple_get_virt_method_for_binfo): ... here; simplify using
2462 vtable_pointer_value_to_vtable.
2463 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2464 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2465 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2466 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2468 (vtable_pointer_value_to_binfo): ... here.
2469 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2471 2014-02-03 Teresa Johnson <tejohnson@google.com>
2473 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2474 redef of outer loop index variable.
2476 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2480 * doc/extend.texi (Function Attributes): Typo.
2482 2014-02-03 Cong Hou <congh@google.com>
2484 PR tree-optimization/60000
2485 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2486 if the vectorized statement is a store. A store statement can only
2487 appear at the end of pattern statements.
2489 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2491 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2492 (ix86_option_override_internal): Default long double to 64-bit for
2493 32-bit Bionic and to 128-bit for 64-bit Bionic.
2495 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2496 TARGET_LONG_DOUBLE_128 is true.
2497 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2499 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2500 (mlong-double-64): Negate -mlong-double-128.
2501 (mlong-double-128): New option.
2503 * config/i386/i386-c.c (ix86_target_macros): Define
2504 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2506 * doc/invoke.texi: Document -mlong-double-128.
2508 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2510 PR rtl-optimization/60024
2511 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2513 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2515 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2517 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2519 PR rtl-optimization/57662
2520 * sel-sched.c (code_motion_path_driver): Do not mark already not
2521 existing blocks in the visiting bitmap.
2523 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2525 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2526 on the insn being emitted.
2528 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2529 Will Deacon <will.deacon@arm.com>
2531 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2533 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2535 * config/arm/arm-tables.opt: Regenerate.
2537 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2539 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2540 for vector types other than V16QImode.
2541 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2542 define_expand, and call altivec_expand_vec_perm_le when producing
2543 code with little endian element order.
2544 (*altivec_vperm_<mode>_internal): New insn having previous
2545 behavior of altivec_vperm_<mode>.
2546 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2547 altivec_expand_vec_perm_le when producing code with little endian
2549 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2550 behavior of altivec_vperm_<mode>_uns.
2552 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2554 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2555 (altivec_vsumsws): Add handling for -maltivec=be with a little
2557 (altivec_vsumsws_direct): New.
2558 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2559 gen_altivec_vsumsws.
2561 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2563 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2564 vtable_pointer_value_to_binfo): New functions.
2565 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2566 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2568 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2570 * config/nios2/nios2.md (load_got_register): Initialize GOT
2571 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2572 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2574 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2576 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2577 preserverd by passthrough, do not propagate the type.
2579 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2581 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2582 (mips_atomic_assign_expand_fenv): New function.
2583 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2585 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2587 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2588 (__builtin_mips_set_fcsr): Likewise.
2589 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2590 MIPS_USI_FTYPE_VOID.
2591 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2592 (mips16_expand_set_fcsr): Likewise.
2593 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2594 (mips16_set_fcsr_stub): Likewise.
2595 (mips16_get_fcsr_one_only_stub): New class.
2596 (mips16_set_fcsr_one_only_stub): Likewise.
2597 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2598 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2599 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2600 (hard_float): New availability predicate.
2601 (mips_builtins): Add get_fcsr and set_fcsr.
2602 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2603 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2604 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2605 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2606 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2609 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2611 * config/mips/mips.c (mips_one_only_stub): New class.
2612 (mips_need_mips16_rdhwr_p): Replace with...
2613 (mips16_rdhwr_stub): ...this new variable.
2614 (mips16_stub_call_address): New function.
2615 (mips16_rdhwr_one_only_stub): New class.
2616 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2617 (mips_output_mips16_rdhwr): Delete.
2618 (mips_finish_stub): New function.
2619 (mips_code_end): Use it to handle rdhwr stubs.
2621 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2624 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2625 when calculating size of integer atomic types.
2627 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2629 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2631 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2633 PR tree-optimization/60003
2634 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2635 * profile.c (branch_prob): Use gimple_call_builtin_p
2636 to check for BUILT_IN_SETJMP_RECEIVER.
2637 * tree-inline.c (copy_bb): Call notice_special_calls.
2639 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2642 * lra-constraints.c (process_alt_operands): Update reload_sum only
2645 2014-01-31 Richard Henderson <rth@redhat.com>
2648 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2649 until after else_eh is processed.
2651 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2653 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2654 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2655 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2656 in smmintrin.h, remove them.
2657 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2658 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2659 * config/i386/i386.md (ROUND_SAE): Fix value.
2660 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2661 (const48_operand): New.
2662 * config/i386/subst.md (round), (round_expand): Use
2663 const_4_or_8_to_11_operand.
2664 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2666 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2668 * config/i386/constraints.md (Yk): Swap meaning with k.
2669 * config/i386/i386.md (movhi_internal): Change Yk to k.
2670 (movqi_internal): Ditto.
2671 (*k<logic><mode>): Ditto.
2674 (kandn<mode>): Ditto.
2675 (*<code>hi_1): Ditto.
2676 (*<code>qi_1): Ditto.
2677 (kxnor<mode>): Ditto.
2678 (kortestzhi): Ditto.
2679 (kortestchi): Ditto.
2681 (*one_cmplhi2_1): Ditto.
2682 (*one_cmplqi2_1): Ditto.
2683 * config/i386/sse.md (): Change k to Yk.
2684 (avx512f_load<mode>_mask): Ditto.
2685 (avx512f_blendm<mode>): Ditto.
2686 (avx512f_store<mode>_mask): Ditto.
2687 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2688 (avx512f_storedqu<mode>_mask): Ditto.
2689 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2691 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2692 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2693 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2694 (avx512f_maskcmp<mode>3): Ditto.
2695 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2696 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2697 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2698 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2699 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2700 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2701 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2702 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2703 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2704 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2705 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2706 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2707 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2708 (vec_extract_lo_<mode>_maskm): Ditto.
2709 (vec_extract_hi_<mode>_maskm): Ditto.
2710 (avx512f_vternlog<mode>_mask): Ditto.
2711 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2712 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2713 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2714 (avx512f_<code>v8div16qi2_mask): Ditto.
2715 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2716 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2717 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2718 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2719 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2720 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2721 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2722 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2723 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2724 (avx512cd_maskb_vec_dupv8di): Ditto.
2725 (avx512cd_maskw_vec_dupv16si): Ditto.
2726 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2727 (avx512f_vpermi2var<mode>3_mask): Ditto.
2728 (avx512f_vpermi2var<mode>3_mask): Ditto.
2729 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2730 (*avx512f_gathersi<mode>): Ditto.
2731 (*avx512f_gathersi<mode>_2): Ditto.
2732 (*avx512f_gatherdi<mode>): Ditto.
2733 (*avx512f_gatherdi<mode>_2): Ditto.
2734 (*avx512f_scattersi<mode>): Ditto.
2735 (*avx512f_scatterdi<mode>): Ditto.
2736 (avx512f_compress<mode>_mask): Ditto.
2737 (avx512f_compressstore<mode>_mask): Ditto.
2738 (avx512f_expand<mode>_mask): Ditto.
2739 * config/i386/subst.md (mask): Change k to Yk.
2740 (mask_scalar_merge): Ditto.
2743 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2745 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2747 2014-01-31 Richard Biener <rguenther@suse.de>
2750 * builtins.c (fold_builtin_memory_op): Make sure to not
2751 use a floating-point mode or a boolean or enumeral type for
2754 2014-01-30 DJ Delorie <dj@redhat.com>
2756 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2757 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2758 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2759 whenever main() has an epilogue.
2761 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2763 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2764 unused variable "field".
2765 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2766 (vsx_mergeh_<mode>): Likewise.
2767 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2768 (altivec_vmrghh): Likewise.
2769 (altivec_vmrghw): Likewise.
2770 (altivec_vmrglb): Likewise.
2771 (altivec_vmrglh): Likewise.
2772 (altivec_vmrglw): Likewise.
2773 (altivec_vspltb): Add missing uses.
2774 (altivec_vsplth): Likewise.
2775 (altivec_vspltw): Likewise.
2776 (altivec_vspltsf): Likewise.
2778 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2781 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2782 frame related instructions.
2784 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2786 PR rtl-optimization/59959
2787 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2788 any reload of register whose subreg is invalid.
2790 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2792 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2793 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2794 Add missing return type - void.
2796 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2798 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2799 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2800 remove element index adjustment for endian (now handled in vsx.md
2802 (altivec_expand_vec_perm_const): Use
2803 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2804 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2805 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2806 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2807 define_expand and a new define_insn *altivec_vspltb_internal;
2808 adjust for -maltivec=be on a little endian target.
2809 (altivec_vspltb_direct): New.
2810 (altivec_vsplth): Divide into a define_expand and a new
2811 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2812 little endian target.
2813 (altivec_vsplth_direct): New.
2814 (altivec_vspltw): Divide into a define_expand and a new
2815 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2816 little endian target.
2817 (altivec_vspltw_direct): New.
2818 (altivec_vspltsf): Divide into a define_expand and a new
2819 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2820 a little endian target.
2822 2014-01-30 Richard Biener <rguenther@suse.de>
2824 PR tree-optimization/59993
2825 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2826 can propagate form the earlier stmt and avoid the transform
2827 when the intermediate result is needed.
2829 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2831 * README.Portability: Fix typo.
2833 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2835 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2836 comparison_operator with ordered_comparison_operator.
2838 2014-01-30 Nick Clifton <nickc@redhat.com>
2840 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2841 Rename to mn10300_store_multiple_regs.
2842 * config/mn10300/mn10300.c: Likewise.
2843 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2844 store_multiple_regs.
2845 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2846 Call mn10300_store_multiple_regs.
2848 2014-01-30 Nick Clifton <nickc@redhat.com>
2849 DJ Delorie <dj@redhat.com>
2851 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2852 %fp 2 to keep registers after it properly word-aligned.
2853 (rl78_alloc_physical_registers_umul): Handle the case where both
2854 input operands are the same.
2856 2014-01-30 Richard Biener <rguenther@suse.de>
2858 PR tree-optimization/59903
2859 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2862 2014-01-30 Jason Merrill <jason@redhat.com>
2865 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2868 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2870 2014-01-30 Richard Biener <rguenther@suse.de>
2872 PR tree-optimization/59951
2873 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2875 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2878 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2879 SFmode to DFmode case.
2881 2014-01-29 DJ Delorie <dj@redhat.com>
2883 * config/msp430/msp430.opt (-minrt): New.
2884 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2886 (ENDFILE_SPEC): Likewise.
2888 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2890 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2891 (estimate_function_body_sizes): Use it.
2893 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2896 * dwarf2out.c (is_cxx_auto): New.
2897 (is_base_type): Use it.
2898 (gen_type_die_with_usage): Likewise.
2900 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2902 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2903 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2904 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2905 -maltivec=be with LE targets.
2906 (vsx_mergeh_<mode>): Likewise.
2907 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2908 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2909 (altivec_vmrghb): Replace with define_expand and new
2910 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2911 (altivec_vmrghb_direct): New define_insn.
2912 (altivec_vmrghh): Replace with define_expand and new
2913 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2914 (altivec_vmrghh_direct): New define_insn.
2915 (altivec_vmrghw): Replace with define_expand and new
2916 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2917 (altivec_vmrghw_direct): New define_insn.
2918 (*altivec_vmrghsf): Adjust for endianness.
2919 (altivec_vmrglb): Replace with define_expand and new
2920 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2921 (altivec_vmrglb_direct): New define_insn.
2922 (altivec_vmrglh): Replace with define_expand and new
2923 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2924 (altivec_vmrglh_direct): New define_insn.
2925 (altivec_vmrglw): Replace with define_expand and new
2926 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2927 (altivec_vmrglw_direct): New define_insn.
2928 (*altivec_vmrglsf): Adjust for endianness.
2929 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2930 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2931 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2932 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2933 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2934 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2935 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2936 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2938 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2940 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2941 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2944 2014-01-29 Richard Biener <rguenther@suse.de>
2946 PR tree-optimization/58742
2947 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2948 associate_pointerplus_align.
2949 (associate_pointerplus_diff): New function.
2950 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2951 and associate_pointerplus_diff.
2953 2014-01-29 Richard Biener <rguenther@suse.de>
2955 * lto-streamer.h (LTO_major_version): Bump to 3.
2956 (LTO_minor_version): Reset to 0.
2958 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2960 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2961 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2962 (arm_file_start): Generate correct asm header for armv7ve.
2963 * config/arm/bpabi.h: Add multilib support for armv7ve.
2964 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2965 and cortex-a15 to armv7ve.
2966 * config/arm/t-aprofile: Add multilib support for armv7ve.
2967 * doc/invoke.texi: Document -march=armv7ve.
2969 2014-01-29 Richard Biener <rguenther@suse.de>
2971 PR tree-optimization/58742
2972 * tree-ssa-forwprop.c (associate_plusminus): Return true
2973 if we changed sth, defer EH cleanup to ...
2974 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2975 (simplify_mult): New function.
2977 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2980 PR tree-optimization/59920
2981 * tree.c (build_common_builtin_nodes): Remove
2982 __builtin_setjmp_dispatcher initialization.
2983 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2984 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2985 instead of gsi_after_labels + manually skipping debug stmts.
2986 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2987 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2988 * tree-inline.c (copy_edges_for_bb): Remove
2989 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2990 argument. Ignore computed_goto_p stmts. Don't call
2991 make_abnormal_goto_edges. If a call might need abnormal edges
2992 for non-local gotos, see if it already has an edge to
2993 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2994 with true argument, don't do anything then, otherwise add
2995 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2996 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2998 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2999 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3000 (lower_stmt): Don't set data->calls_builtin_setjmp.
3001 (lower_builtin_setjmp): Adjust comment.
3002 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3003 * tree-cfg.c (found_computed_goto): Remove.
3004 (factor_computed_gotos): Remove.
3005 (make_goto_expr_edges): Return bool, true for computed gotos.
3006 Don't call make_abnormal_goto_edges.
3007 (build_gimple_cfg): Don't set found_computed_goto, don't call
3008 factor_computed_gotos.
3009 (computed_goto_p): No longer static.
3010 (make_blocks): Don't set found_computed_goto.
3011 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3012 (make_edges): If make_goto_expr_edges returns true, push bb
3013 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3014 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3015 vector. Record mapping between bbs and OpenMP regions if there
3016 are any, adjust make_gimple_omp_edges caller. Call
3017 handle_abnormal_edges.
3018 (make_abnormal_goto_edges): Remove.
3019 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3020 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3021 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3022 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3023 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3024 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3025 filling *region also set *region_idx to (*region)->entry->index.
3028 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3029 For REGs set ORIGINAL_REGNO.
3031 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3033 * doc/md.texi: Mention that a target shouldn't implement
3034 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3037 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3039 PR tree-optimization/59594
3040 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3041 a copy of the datarefs vector rather than the vector itself.
3043 2014-01-28 Jason Merrill <jason@redhat.com>
3046 * dwarf2out.c (auto_die): New static.
3047 (gen_type_die_with_usage): Handle C++1y 'auto'.
3048 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3051 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3054 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3055 (SPEC_X32): Likewise.
3056 (SPEC_64): Likewise.
3057 * config/i386/i386.c (ix86_option_override_internal): Turn off
3058 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3060 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3061 * config/i386/i386.h (TARGET_16BIT): New macro.
3062 (TARGET_16BIT_P): Likewise.
3063 * config/i386/i386.opt: Add m16.
3064 * doc/invoke.texi: Document -m16.
3066 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3068 PR preprocessor/59935
3069 * input.c (location_get_source_line): Bail out on when line number
3070 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3072 2014-01-28 Richard Biener <rguenther@suse.de>
3074 PR tree-optimization/58742
3075 * tree-ssa-forwprop.c (associate_plusminus): Handle
3076 pointer subtraction of the form (T)(P + A) - (T)P.
3078 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3080 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3083 2014-01-28 Richard Biener <rguenther@suse.de>
3086 2014-01-28 Richard Biener <rguenther@suse.de>
3088 PR rtl-optimization/45364
3089 PR rtl-optimization/59890
3090 * var-tracking.c (local_get_addr_clear_given_value): Handle
3091 already cleared slot.
3092 (val_reset): Handle not allocated local_get_addr_cache.
3093 (vt_find_locations): Use post-order on the inverted CFG.
3095 2014-01-28 Richard Biener <rguenther@suse.de>
3097 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3099 2014-01-28 Richard Biener <rguenther@suse.de>
3101 PR rtl-optimization/45364
3102 PR rtl-optimization/59890
3103 * var-tracking.c (local_get_addr_clear_given_value): Handle
3104 already cleared slot.
3105 (val_reset): Handle not allocated local_get_addr_cache.
3106 (vt_find_locations): Use post-order on the inverted CFG.
3108 2014-01-28 Alan Modra <amodra@gmail.com>
3110 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3111 * configure.ac <recursive call for build != host>: Define
3112 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3113 and LD_FOR_BUILD too.
3114 * configure: Regenerate.
3116 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3118 * config/i386/i386.c (get_builtin_code_for_version): Separate
3119 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3120 Broadwell from Haswell.
3122 2014-01-27 Steve Ellcey <sellcey@mips.com>
3124 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3125 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3126 * config/mips/mips.c (mips_option_override): Change setting
3128 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3129 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3130 Change from Mask to Var.
3132 2014-01-27 Jeff Law <law@redhat.com>
3134 * ipa-inline.c (inline_small_functions): Fix typo.
3136 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3138 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3139 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3140 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3141 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3142 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3143 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3144 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3145 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3146 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3147 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3148 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3149 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3150 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3151 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3152 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3153 (_mm512_storeu_epi64): Ditto.
3154 (_mm512_cmpge_epi32_mask): Ditto.
3155 (_mm512_cmpge_epu32_mask): Ditto.
3156 (_mm512_cmpge_epi64_mask): Ditto.
3157 (_mm512_cmpge_epu64_mask): Ditto.
3158 (_mm512_cmple_epi32_mask): Ditto.
3159 (_mm512_cmple_epu32_mask): Ditto.
3160 (_mm512_cmple_epi64_mask): Ditto.
3161 (_mm512_cmple_epu64_mask): Ditto.
3162 (_mm512_cmplt_epi32_mask): Ditto.
3163 (_mm512_cmplt_epu32_mask): Ditto.
3164 (_mm512_cmplt_epi64_mask): Ditto.
3165 (_mm512_cmplt_epu64_mask): Ditto.
3166 (_mm512_cmpneq_epi32_mask): Ditto.
3167 (_mm512_cmpneq_epu32_mask): Ditto.
3168 (_mm512_cmpneq_epi64_mask): Ditto.
3169 (_mm512_cmpneq_epu64_mask): Ditto.
3170 (_mm512_expand_pd): Ditto.
3171 (_mm512_expand_ps): Ditto.
3172 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3173 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3174 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3175 * config/i386/i386.c (ix86_builtins): Add
3176 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3177 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3178 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3179 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3180 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3181 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3182 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3183 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3184 IX86_BUILTIN_PMOVUSQW512_MEM.
3185 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3186 __builtin_ia32_pmovsqd512mem_mask,
3187 __builtin_ia32_pmovqd512mem_mask,
3188 __builtin_ia32_pmovusqw512mem_mask,
3189 __builtin_ia32_pmovsqw512mem_mask,
3190 __builtin_ia32_pmovqw512mem_mask,
3191 __builtin_ia32_pmovusdw512mem_mask,
3192 __builtin_ia32_pmovsdw512mem_mask,
3193 __builtin_ia32_pmovdw512mem_mask,
3194 __builtin_ia32_pmovqb512mem_mask,
3195 __builtin_ia32_pmovusqb512mem_mask,
3196 __builtin_ia32_pmovsqb512mem_mask,
3197 __builtin_ia32_pmovusdb512mem_mask,
3198 __builtin_ia32_pmovsdb512mem_mask,
3199 __builtin_ia32_pmovdb512mem_mask.
3200 (bdesc_args): Add __builtin_ia32_expanddf512,
3201 __builtin_ia32_expandsf512.
3202 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3203 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3204 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3205 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3206 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3207 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3208 (avx512f_<code>v8div16qi2_mask_store): This.
3209 (avx512f_expand<mode>): New.
3211 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3213 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3215 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3216 (_mm512_prefetch_i32scatter_pd): Ditto.
3217 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3218 (_mm512_prefetch_i64scatter_pd): Ditto.
3219 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3220 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3221 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3222 (_mm512_prefetch_i32scatter_ps): Ditto.
3223 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3224 (_mm512_prefetch_i64scatter_ps): Ditto.
3225 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3226 * config/i386/i386-builtin-types.def: Define
3227 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3228 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3229 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3230 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3231 IX86_BUILTIN_SCATTERPFQPD.
3232 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3233 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3234 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3235 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3236 __builtin_ia32_scatterpfqps.
3237 (ix86_expand_builtin): Expand new built-ins.
3238 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3239 fix memory access data type.
3240 (*avx512pf_gatherpf<mode>_mask): Ditto.
3241 (*avx512pf_gatherpf<mode>): Ditto.
3242 (avx512pf_scatterpf<mode>): Ditto.
3243 (*avx512pf_scatterpf<mode>_mask): Ditto.
3244 (*avx512pf_scatterpf<mode>): Ditto.
3245 (GATHER_SCATTER_SF_MEM_MODE): New.
3246 (avx512pf_gatherpf<mode>df): Ditto.
3247 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3248 (*avx512pf_scatterpf<mode>df): Ditto.
3250 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3253 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3254 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3257 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3259 * common/config/arm/arm-common.c
3260 (arm_rewrite_mcpu): Handle multiple names.
3262 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3264 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3266 * gimple-builder.h (create_gimple_tmp): Delete.
3268 2014-01-27 Christian Bruel <christian.bruel@st.com>
3270 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3273 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3275 * config/pa/pa.md (call): Generate indirect long calls to non-local
3276 functions when outputing 32-bit code.
3277 (call_value): Likewise except for special call to buggy powf function.
3279 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3280 portable runtime and PIC indirect calls.
3281 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3282 and PIC call sequences. Use ldo instead of blr to set return register
3283 in PIC call sequence.
3285 2014-01-25 Walter Lee <walt@tilera.com>
3287 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3288 avoid clobbering a live register.
3290 2014-01-25 Walter Lee <walt@tilera.com>
3292 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3293 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3294 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3295 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3297 2014-01-25 Walter Lee <walt@tilera.com>
3299 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3300 arguments on even registers.
3301 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3303 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3304 (BIGGEST_ALIGNMENT): Ditto.
3305 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3307 2014-01-25 Walter Lee <walt@tilera.com>
3309 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3310 insns before bundling.
3311 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3313 2014-01-25 Walter Lee <walt@tilera.com>
3315 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3316 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3317 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3319 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3321 * config/mips/constraints.md (kl): Delete.
3322 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3323 define expands, using...
3324 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3325 instructions for MIPS16.
3326 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3327 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3329 2014-01-25 Walter Lee <walt@tilera.com>
3331 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3335 2014-01-25 Walter Lee <walt@tilera.com>
3337 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3338 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3340 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3342 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3345 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3347 * print-rtl.c (in_call_function_usage): New var.
3348 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3349 EXPR_LIST mode as mode and not as reg note name.
3352 * cfgloopmanip.c (copy_loop_info): If
3353 loop->warned_aggressive_loop_optimizations, make sure
3354 the flag is set in target loop too.
3356 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3358 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3360 * builtins.def: Likewise.
3361 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3362 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3363 * ira.c (ira_setup_eliminable_regset): Likewise.
3364 * omp-low.c (gate_expand_omp): Likewise.
3365 (execute_lower_omp): Likewise.
3366 (diagnose_sb_0): Likewise.
3367 (gate_diagnose_omp_blocks): Likewise.
3368 (simd_clone_clauses_extract): Likewise.
3371 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3373 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3374 correction for little endian...
3375 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3378 2014-01-24 Jeff Law <law@redhat.com>
3380 PR tree-optimization/59919
3381 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3382 for non-returning calls.
3384 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3386 * common/config/aarch64/aarch64-common.c
3387 (aarch64_rewrite_mcpu): Handle multiple names.
3388 * config/aarch64/aarch64.h
3389 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3391 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3393 * input.c (add_file_to_cache_tab): Handle the case where fopen
3396 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3399 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3400 from push operand if code of push isn't PRE_DEC.
3402 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3405 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3406 -mquad-memory-atomic. Update -mquad-memory documentation to say
3407 it is only used for non-atomic loads/stores.
3409 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3410 -mquad-memory or -mquad-memory-atomic switches.
3412 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3413 -mquad-memory-atomic to ISA 2.07 support.
3415 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3416 to separate support of normal quad word memory operations (ldq, stq)
3417 from the atomic quad word memory operations.
3419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3420 support to separate non-atomic quad word operations from atomic
3421 quad word operations. Disable non-atomic quad word operations in
3422 little endian mode so that we don't have to swap words after the
3423 load and before the store.
3424 (quad_load_store_p): Add comment about atomic quad word support.
3425 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3426 options printed with -mdebug=reg.
3428 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3429 -mquad-memory-atomic as the test for whether we have quad word
3430 atomic instructions.
3431 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3432 or -mp8-vector are used, allow byte/half-word atomic operations.
3434 * config/rs6000/sync.md (load_lockedti): Insure that the address
3435 is a proper indexed or indirect address for the lqarx instruction.
3436 On little endian systems, swap the hi/lo registers after the lqarx
3438 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3439 insure the address is valid for the lqarx instruction.
3440 (store_conditionalti): Insure that the address is a proper indexed
3441 or indirect address for the stqcrx. instruction. On little endian
3442 systems, swap the hi/lo registers before doing the stqcrx.
3444 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3445 insure the address is valid for the stqcrx. instruction.
3447 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3448 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3449 type of quad memory support is available.
3451 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3454 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3455 there is a danger of looping.
3457 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3459 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3460 force flag_ira_loop_pressure if set via command line.
3462 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3464 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3465 (ashr_simd): New builtin handling DI mode.
3466 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3467 (aarch64_sshr_simddi): New match pattern.
3468 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3469 (vshrd_n_s64): Likewise.
3470 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3472 2014-01-23 Nick Clifton <nickc@redhat.com>
3474 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3475 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3476 favour of mcu specific scripts.
3477 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3480 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3481 Alex Velenko <Alex.Velenko@arm.com>
3483 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3484 (vaddv_s16): Likewise.
3485 (vaddv_s32): Likewise.
3486 (vaddv_u8): Likewise.
3487 (vaddv_u16): Likewise.
3488 (vaddv_u32): Likewise.
3489 (vaddvq_s8): Likewise.
3490 (vaddvq_s16): Likewise.
3491 (vaddvq_s32): Likewise.
3492 (vaddvq_s64): Likewise.
3493 (vaddvq_u8): Likewise.
3494 (vaddvq_u16): Likewise.
3495 (vaddvq_u32): Likewise.
3496 (vaddvq_u64): Likewise.
3497 (vaddv_f32): Likewise.
3498 (vaddvq_f32): Likewise.
3499 (vaddvq_f64): Likewise.
3500 (vmaxv_f32): Likewise.
3501 (vmaxv_s8): Likewise.
3502 (vmaxv_s16): Likewise.
3503 (vmaxv_s32): Likewise.
3504 (vmaxv_u8): Likewise.
3505 (vmaxv_u16): Likewise.
3506 (vmaxv_u32): Likewise.
3507 (vmaxvq_f32): Likewise.
3508 (vmaxvq_f64): Likewise.
3509 (vmaxvq_s8): Likewise.
3510 (vmaxvq_s16): Likewise.
3511 (vmaxvq_s32): Likewise.
3512 (vmaxvq_u8): Likewise.
3513 (vmaxvq_u16): Likewise.
3514 (vmaxvq_u32): Likewise.
3515 (vmaxnmv_f32): Likewise.
3516 (vmaxnmvq_f32): Likewise.
3517 (vmaxnmvq_f64): Likewise.
3518 (vminv_f32): Likewise.
3519 (vminv_s8): Likewise.
3520 (vminv_s16): Likewise.
3521 (vminv_s32): Likewise.
3522 (vminv_u8): Likewise.
3523 (vminv_u16): Likewise.
3524 (vminv_u32): Likewise.
3525 (vminvq_f32): Likewise.
3526 (vminvq_f64): Likewise.
3527 (vminvq_s8): Likewise.
3528 (vminvq_s16): Likewise.
3529 (vminvq_s32): Likewise.
3530 (vminvq_u8): Likewise.
3531 (vminvq_u16): Likewise.
3532 (vminvq_u32): Likewise.
3533 (vminnmv_f32): Likewise.
3534 (vminnmvq_f32): Likewise.
3535 (vminnmvq_f64): Likewise.
3537 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3539 * config/aarch64/aarch64-simd.md
3540 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3541 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3542 (*aarch64_mul3_elt<mode>): Likewise.
3543 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3544 (*aarch64_mul3_elt_to_64v2df): Likewise.
3545 (*aarch64_mla_elt<mode>): Likewise.
3546 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3547 (*aarch64_mls_elt<mode>): Likewise.
3548 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3549 (*aarch64_fma4_elt<mode>): Likewise.
3550 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3551 (*aarch64_fma4_elt_to_64v2df): Likewise.
3552 (*aarch64_fnma4_elt<mode>): Likewise.
3553 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3554 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3555 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3556 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3557 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3558 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3559 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3560 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3561 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3563 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3565 * config/aarch64/aarch64-simd.md
3566 (aarch64_be_checked_get_lane<mode>): New define_expand.
3567 * config/aarch64/aarch64-simd-builtins.def
3568 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3569 New builtin definition.
3570 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3571 Use new safe be builtin.
3573 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3575 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3577 (aarch64_be_st1<mode>): Likewise.
3578 (aarch_ld1<VALL:mode>): Define_expand modified.
3579 (aarch_st1<VALL:mode>): Likewise.
3580 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3581 (UNSPEC_ST1): Likewise.
3583 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3585 * config/microblaze/microblaze.md: Add trap insn and attribute
3587 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3589 PR preprocessor/58580
3590 * input.h (location_get_source_line): Take an additional line_size
3592 (void diagnostics_file_cache_fini): Declare new function.
3593 * input.c (struct fcache): New type.
3594 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3595 New static constants.
3596 (diagnostic_file_cache_init, total_lines_num)
3597 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3598 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3599 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3600 (get_next_line, read_next_line, goto_next_line, read_line_num):
3601 New static function definitions.
3602 (diagnostic_file_cache_fini): New function.
3603 (location_get_source_line): Take an additional output line_len
3604 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3606 * diagnostic.c (diagnostic_finish): Call
3607 diagnostic_file_cache_fini.
3608 (adjust_line): Take an additional input parameter for the length
3609 of the line, rather than calculating it with strlen.
3610 (diagnostic_show_locus): Adjust the use of
3611 location_get_source_line and adjust_line with respect to their new
3612 signature. While displaying a line now, do not stop at the first
3613 null byte. Rather, display the zero byte as a space and keep
3614 going until we reach the size of the line.
3615 * Makefile.in: Add vec.o to OBJS-libcommon
3617 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3618 Ilya Tocar <ilya.tocar@intel.com>
3620 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3621 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3622 (__builtin_ia32_kmov16): Ditto.
3623 * config/i386/i386.md (UNSPEC_KMOV): New.
3626 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3628 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3629 (_mm512_storeu_si512): Ditto.
3631 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3634 * rtl.h (get_referenced_operands): Declare.
3635 * recog.c (get_referenced_operands): New function.
3636 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3637 operands have been referenced when recording LO_SUM references.
3639 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3641 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3643 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3645 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3646 Enable for generic and recent AMD targets.
3648 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3650 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3651 ARG_SIZE note when adjustment was eliminated.
3653 2014-01-22 Jeff Law <law@redhat.com>
3655 PR tree-optimization/59597
3656 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3657 in file. Accept new argument REGISTERING and use it to modify
3658 dump output appropriately.
3659 (register_jump_thread): Corresponding changes.
3660 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3661 thread paths involving joiner blocks. Add code to dump cancelled
3662 jump threading paths.
3664 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3666 PR rtl-optimization/59477
3667 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3668 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3670 2014-01-22 Tom Tromey <tromey@redhat.com>
3672 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3674 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3676 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3678 PR rtl-optimization/59896
3679 * lra-constraints.c (process_alt_operands): Check unused note for
3680 matched operands of insn with no output reloads.
3682 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3684 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3685 (mips_move_from_gpr_cost): Likewise.
3687 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3689 PR rtl-optimization/59858
3690 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3691 ira_class_hard_regs_num.
3692 (process_alt_operands): Increase reject for dying matched operand.
3694 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3697 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3698 smaller than size, perform several stores or loads and stores
3699 at dst + count - size to store or copy all of size bytes, rather
3700 than just last modesize bytes.
3702 2014-01-20 DJ Delorie <dj@redhat.com>
3704 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3705 that CLOBBERs are REGs before propogating their values.
3707 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3710 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3711 (cgraph_inline_failed_type): New function.
3712 * cgraph.h (DEFCIFCODE): Add type.
3713 (cgraph_inline_failed_type_t): New enum.
3714 (cgraph_inline_failed_type): New prototype.
3715 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3716 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3717 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3718 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3719 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3720 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3721 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3722 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3723 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3724 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3725 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3726 OPTIMIZATION_MISMATCH.
3727 * tree-inline.c (expand_call_inline): Emit errors during
3728 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3730 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3733 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3734 mode attribute in insn output.
3736 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3738 * output.h (output_constant): Delete.
3739 * varasm.c (output_constant): Make private.
3741 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3743 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3745 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3748 * tree.h (fold_builtin_strcat): New prototype.
3749 * builtins.c (fold_builtin_strcat): No longer static. Add len
3750 argument, if non-NULL, don't call c_strlen. Optimize
3751 directly into __builtin_memcpy instead of __builtin_strcpy.
3752 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3753 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3755 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3757 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3758 for SImode_address_operand operands, having only a REG argument.
3760 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3762 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3763 loader name using mbig-endian.
3764 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3766 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3768 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3772 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3774 * config/aarch64/aarch64-protos.h
3775 (aarch64_cannot_change_mode_class_ptr): Declare.
3776 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3777 aarch64_cannot_change_mode_class_ptr): New.
3778 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3779 backend hook aarch64_cannot_change_mode_class.
3781 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3783 * common/config/aarch64/aarch64-common.c
3784 (aarch64_handle_option): Don't handle any option order logic here.
3785 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3786 selected_cpu, warn on architecture version mismatch.
3787 (aarch64_override_options): Fix parsing order for option strings.
3789 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3790 Iain Sandoe <iain@codesourcery.com>
3793 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3794 warning. Amend comment to reflect current functionality.
3796 2014-01-20 Richard Biener <rguenther@suse.de>
3799 * builtins.c (fold_builtin_strcat): Remove case better handled
3800 by tree-ssa-strlen.c.
3802 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3804 * config/aarch64/aarch64.opt
3805 (mcpu, march, mtune): Make case-insensitive.
3807 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3810 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3811 if operands[1] is a REG or ZERO_EXTEND of a REG.
3813 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3815 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3817 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3819 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3820 long non-pic millicode calls.
3822 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3824 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3826 2014-01-19 Kito Cheng <kito@0xlab.org>
3828 * builtins.c (expand_movstr): Check movstr expand done or fail.
3830 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3831 H.J. Lu <hongjiu.lu@intel.com>
3834 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3835 to DImode for zero-extended addresses.
3837 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3839 PR rtl-optimization/57763
3840 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3841 on the new indirect jump_insn and increment LABEL_NUSES (label).
3843 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3847 * config.gcc (x86_archs): New variable.
3848 (x86_64_archs): Likewise.
3849 (x86_cpus): Likewise.
3850 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3851 --with-arch/--with-cpu= options.
3852 Support --with-arch=/--with-cpu={nehalem,westmere,
3853 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3855 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3857 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3858 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3860 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3862 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3864 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3867 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3868 clear cpp_get_options (parse_in)->warn_unused_macros for
3869 ix86_target_macros_internal with cpp_define.
3871 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3873 * jump.c (delete_related_insns): Keep (use (insn))s.
3874 * reorg.c (redundant_insn): Check for barriers too.
3876 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3878 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3880 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3882 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3883 call to $$dyncall when TARGET_LONG_CALLS is true.
3885 2014-01-17 Jeff Law <law@redhat.com>
3887 * ree.c (combine_set_extension): Temporarily disable test for
3888 changing number of hard registers.
3890 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3893 * ipa-inline-analysis.c (inline_free_summary):
3894 Do not free summary of aliases.
3896 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3899 * gimplify.c (gimplify_expr): Use create_tmp_var
3900 instead of create_tmp_var_raw. If cond doesn't have
3901 integral type, don't add the IFN_ANNOTATE builtin at all.
3903 2014-01-17 Martin Jambor <mjambor@suse.cz>
3906 * ipa-cp.c (prev_edge_clone): New variable.
3907 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3908 Also resize prev_edge_clone vector.
3909 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3910 (ipcp_edge_removal_hook): New function.
3911 (ipcp_driver): Register ipcp_edge_removal_hook.
3913 2014-01-17 Andrew Pinski <apinski@cavium.com>
3914 Steve Ellcey <sellcey@mips.com>
3917 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3920 2014-01-17 Jeff Law <law@redhat.com>
3923 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3924 so that pass_ccp runs first.
3926 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3928 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3929 (ix86_adjust_cost): Use !TARGET_XXX.
3930 (do_reorder_for_imul): Likewise.
3931 (swap_top_of_ready_list): Likewise.
3932 (ix86_sched_reorder): Likewise.
3934 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3936 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3937 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3938 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3939 (intel_memset): New. Duplicate slm_memset.
3940 (intel_cost): New. Duplicate slm_cost.
3941 (m_INTEL): New macro.
3942 (processor_target_table): Add "intel".
3943 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3944 with PROCESSOR_INTEL for "intel".
3945 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3946 PROCESSOR_SILVERMONT.
3947 (ix86_issue_rate): Likewise.
3948 (ix86_adjust_cost): Likewise.
3949 (ia32_multipass_dfa_lookahead): Likewise.
3950 (swap_top_of_ready_list): Likewise.
3951 (ix86_sched_reorder): Likewise.
3952 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3953 instead of TARGET_OPT_AGU.
3954 * config/i386/i386.h (TARGET_INTEL): New.
3955 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3956 (processor_type): Add PROCESSOR_INTEL.
3957 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3958 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3960 2014-01-17 Marek Polacek <polacek@redhat.com>
3963 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3966 2014-01-17 Richard Biener <rguenther@suse.de>
3968 PR tree-optimization/46590
3969 * opts.c (default_options_table): Add entries for
3970 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3971 all enabled at -O1 but not for -Og.
3972 * common.opt (fbranch-count-reg): Remove Init(1).
3973 (fmove-loop-invariants): Likewise.
3974 (ftree-pta): Likewise.
3976 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3978 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3979 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3980 decls to at least the GCC 4.8 used alignments.
3983 * tree-nested.c (convert_nonlocal_reference_stmt,
3984 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3985 of GIMPLE_BIND stmts, adjust associated decls.
3987 2014-01-17 Richard Biener <rguenther@suse.de>
3989 PR tree-optimization/46590
3990 * vec.h (vec<>::bseach): New member function implementing
3991 binary search according to C89 bsearch.
3992 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3993 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3994 bitmap pointer again. Make accesses_in_loop a flat array.
3995 (mem_ref_obstack): New global.
3996 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3997 (mark_ref_stored): Likewise.
3998 (ref_indep_loop_p_2): Likewise.
3999 (set_ref_stored_in_loop): New helper function.
4000 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4001 (memref_free): Adjust.
4002 (record_mem_ref_loc): Simplify.
4003 (gather_mem_refs_stmt): Adjust.
4004 (sort_locs_in_loop_postorder_cmp): New function.
4005 (analyze_memory_references): Sort accesses_in_loop after
4006 loop postorder number.
4007 (find_ref_loc_in_loop_cmp): New function.
4008 (for_all_locs_in_loop): Find relevant cluster of locs in
4009 accesses_in_loop and iterate without recursion.
4010 (execute_sm): Avoid uninit warning.
4011 (struct ref_always_accessed): Simplify.
4012 (ref_always_accessed::operator ()): Likewise.
4013 (ref_always_accessed_p): Likewise.
4014 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4015 loop postorder numbers here.
4016 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4019 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4022 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4023 on decls for which assemble_alias has been called.
4025 2014-01-17 Nick Clifton <nickc@redhat.com>
4027 * config/msp430/msp430.opt: (mcpu): New option.
4028 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4029 (msp430_option_override): Parse target_cpu. If the MCU name
4030 matches a generic string, clear target_mcu.
4031 (msp430_attr): Allow numeric interrupt values up to 63.
4032 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4033 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4035 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4037 * config/msp430/msp430.md (popm): Use %J rather than %I.
4038 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4039 (addhi_cy_i): Use immediate_operand for operand 2.
4040 * doc/invoke.texi: Document -mcpu option.
4042 2014-01-17 Richard Biener <rguenther@suse.de>
4044 PR rtl-optimization/38518
4045 * df.h (df_analyze_loop): Declare.
4046 * df-core.c: Include cfgloop.h.
4047 (df_analyze_1): Split out main part of df_analyze.
4048 (df_analyze): Adjust.
4049 (loop_inverted_post_order_compute): New function.
4050 (loop_post_order_compute): Likewise.
4051 (df_analyze_loop): New function avoiding whole-function
4053 * loop-invariant.c (find_defs): Use df_analyze_loop.
4054 (find_invariants): Adjust.
4055 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4057 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4059 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4060 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4062 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4064 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4065 traversal when removing references.
4067 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4070 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4072 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4075 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4076 pushing a reload for an autoinc when we had previously reloaded an
4077 inner part of the address.
4079 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4081 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4083 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4084 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4085 when not giving up or versioning for alias only because of
4087 (vect_analyze_data_ref_dependences): Set to true.
4088 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4090 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4091 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4095 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4098 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4099 operand 0 predicate for gathers, use a new pseudo as subtarget.
4101 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4104 * lra-constraints.c (process_alt_operands): Add printing debug info.
4105 Check absence of input/output reloads for matched operands too.
4107 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4109 PR rtl-optimization/59835
4110 * ira.c (ira_init_register_move_cost): Increase cost for
4113 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4115 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4117 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4120 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4121 non-register objects. Use gen_(high/low)part more consistently.
4124 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4127 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4128 endian support, remove tests for WORDS_BIG_ENDIAN.
4129 (p8_mfvsrd_3_<mode>): Likewise.
4130 (reload_gpr_from_vsx<mode>): Likewise.
4131 (reload_gpr_from_vsxsf): Likewise.
4132 (p8_mfvsrd_4_disf): Likewise.
4134 2014-01-16 Richard Biener <rguenther@suse.de>
4136 PR rtl-optimization/46590
4137 * lcm.c (compute_antinout_edge): Use postorder iteration.
4138 (compute_laterin): Use inverted postorder iteration.
4140 2014-01-16 Nick Clifton <nickc@redhat.com>
4143 * varasm.c (output_constant): Return the number of bytes actually
4145 (output_constructor_array_range): Update the field size with the
4146 number of bytes emitted by output_constant.
4147 (output_constructor_regular_field): Likewise. Also do not
4148 complain if the total number of bytes emitted is now greater
4149 than the expected fieldpos.
4150 * output.h (output_constant): Update prototype and descriptive comment.
4152 2014-01-16 Marek Polacek <polacek@redhat.com>
4155 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4156 it is error_mark_node.
4158 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4160 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4161 VALID_AVX256_REG_OR_OI_MODE.
4163 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4165 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4166 current procedure should be profiled.
4168 2014-01-15 Andrew Pinski <apinski@cavium.com>
4170 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4171 of moving from/to the STACK_REG register class.
4173 2014-01-15 Richard Henderson <rth@redhat.com>
4176 * reginfo.c (global_regs_decl): Globalize.
4177 * rtl.h (global_regs_decl): Declare.
4178 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4179 reserved via global_regs.
4181 2014-01-15 Teresa Johnson <tejohnson@google.com>
4183 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4185 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4187 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4188 and vmulosh rather than call gen_vec_widen_smult_*.
4189 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4190 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4191 (vec_widen_smult_even_v16qi): Likewise.
4192 (vec_widen_umult_even_v8hi): Likewise.
4193 (vec_widen_smult_even_v8hi): Likewise.
4194 (vec_widen_umult_odd_v16qi): Likewise.
4195 (vec_widen_smult_odd_v16qi): Likewise.
4196 (vec_widen_umult_odd_v8hi): Likewise.
4197 (vec_widen_smult_odd_v8hi): Likewise.
4198 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4199 vmuloub rather than call gen_vec_widen_umult_*.
4200 (vec_widen_umult_lo_v16qi): Likewise.
4201 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4202 vmulosb rather than call gen_vec_widen_smult_*.
4203 (vec_widen_smult_lo_v16qi): Likewise.
4204 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4205 rather than call gen_vec_widen_umult_*.
4206 (vec_widen_umult_lo_v8hi): Likewise.
4207 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4208 rather than call gen_vec_widen_smult_*.
4209 (vec_widen_smult_lo_v8hi): Likewise.
4211 2014-01-15 Jeff Law <law@redhat.com>
4213 PR tree-optimization/59747
4214 * ree.c (find_and_remove_re): Properly handle case where a second
4215 eliminated extension requires widening a copy created for elimination
4216 of a prior extension.
4217 (combine_set_extension): Ensure that the number of hard regs needed
4218 for a destination register does not change when we widen it.
4220 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4222 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4223 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4224 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4225 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4226 (avr-*-rtems*): Likewise.
4227 (bfin*-rtems*): Likewise.
4228 (moxie-*-rtems*): Likewise.
4229 (h8300-*-rtems*): Likewise.
4230 (i[34567]86-*-rtems*): Likewise.
4231 (lm32-*-rtems*): Likewise.
4232 (m32r-*-rtems*): Likewise.
4233 (m68k-*-rtems*): Likewise.
4234 (microblaze*-*-rtems*): Likewise.
4235 (mips*-*-rtems*): Likewise.
4236 (powerpc-*-rtems*): Likewise.
4237 (sh-*-rtems*): Likewise.
4238 (sparc-*-rtems*): Likewise.
4239 (sparc64-*-rtems*): Likewise.
4240 (v850-*-rtems*): Likewise.
4241 (m32c-*-rtems*): Likewise.
4243 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4245 PR rtl-optimization/59511
4246 * ira.c (ira_init_register_move_cost): Use memory costs for some
4247 cases of register move cost calculations.
4248 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4249 instead of BB frequency.
4250 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4251 * lra-assigns.c (find_hard_regno_for): Ditto.
4253 2014-01-15 Richard Biener <rguenther@suse.de>
4255 PR tree-optimization/59822
4256 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4257 (vectorizable_load): Use it to hoist defs of uses of invariant
4258 loads out of the loop.
4260 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4261 Kugan Vivekanandarajah <kuganv@linaro.org>
4264 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4267 2014-01-15 Richard Biener <rguenther@suse.de>
4269 PR rtl-optimization/59802
4270 * lcm.c (compute_available): Use inverted postorder to seed
4271 the initial worklist.
4273 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4276 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4277 ADDR_REGS for invalid symrefs in non-PIC code.
4279 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4282 * builtins.c (determine_block_size): Initialize *probable_max_size
4283 even if len_rtx is CONST_INT.
4285 2014-01-14 Andrew Pinski <apinski@cavium.com>
4287 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4288 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4289 (cortexa53_tunings): Likewise.
4290 (aarch64_sched_issue_rate): New function.
4291 (TARGET_SCHED_ISSUE_RATE): Define.
4293 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4295 * ira-costs.c (find_costs_and_classes): Add missed
4296 ira_init_register_move_cost_if_necessary.
4298 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4301 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4303 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4306 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4307 to indicate if type is used for function return value. Warn ABI
4308 change if the vector mode isn't available for function return value.
4309 (ix86_function_arg_advance): Pass false to type_natural_mode.
4310 (ix86_function_arg): Likewise.
4311 (ix86_gimplify_va_arg): Likewise.
4312 (function_arg_32): Don't warn ABI change.
4313 (ix86_function_value): Pass true to type_natural_mode.
4314 (ix86_return_in_memory): Likewise.
4315 (ix86_struct_value_rtx): Removed.
4316 (TARGET_STRUCT_VALUE_RTX): Likewise.
4318 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4320 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4321 converting a conditional jump into a conditional return.
4323 2014-01-14 Richard Biener <rguenther@suse.de>
4325 PR tree-optimization/58921
4326 PR tree-optimization/59006
4327 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4328 hoisting invariant stmts.
4329 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4330 invariant loads on the preheader edge if possible.
4332 2014-01-14 Joey Ye <joey.ye@arm.com>
4334 * doc/plugin.texi (Building GCC plugins): Update to C++.
4336 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4338 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4339 (_mm_rcp28_round_ss): Ditto.
4340 (_mm_rsqrt28_round_sd): Ditto.
4341 (_mm_rsqrt28_round_ss): Ditto.
4342 (_mm_rcp28_sd): Ditto.
4343 (_mm_rcp28_ss): Ditto.
4344 (_mm_rsqrt28_sd): Ditto.
4345 (_mm_rsqrt28_ss): Ditto.
4346 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4347 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4348 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4349 (IX86_BUILTIN_RCP28SD): Ditto.
4350 (IX86_BUILTIN_RCP28SS): Ditto.
4351 (IX86_BUILTIN_RSQRT28SD): Ditto.
4352 (IX86_BUILTIN_RSQRT28SS): Ditto.
4353 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4354 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4355 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4356 (ix86_expand_special_args_builtin): Expand new FTYPE.
4357 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4358 (srcp14<mode>): Make insn unary.
4359 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4360 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4361 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4362 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4363 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4364 Fix rounding: make it SAE only.
4365 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4367 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4369 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4370 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4371 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4372 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4373 (round_saeonly_mask_scalar_operand4): Ditto.
4374 (round_saeonly_mask_scalar_op3): Ditto.
4375 (round_saeonly_mask_scalar_op4): Ditto.
4377 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4379 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4380 Implement -maltivec=be for vec_insert and vec_extract.
4382 2014-01-10 DJ Delorie <dj@redhat.com>
4384 * config/msp430/msp430.md (call_internal): Don't allow memory
4385 references with SP as the base register.
4386 (call_value_internal): Likewise.
4387 * config/msp430/constraints.md (Yc): New. For memory references
4388 that don't use SP as a base register.
4390 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4391 "an integer without a # prefix"
4392 * config/msp430/msp430.md (epilogue_helper): Use it.
4394 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4397 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4398 AVX512F gather builtins.
4399 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4400 on gather decls with INTEGER_TYPE masktype.
4401 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4402 directly into the builtin rather than hoisting it before loop.
4404 PR tree-optimization/59387
4405 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4406 (scev_const_prop): If folded_casts and type has undefined overflow,
4407 use force_gimple_operand instead of force_gimple_operand_gsi and
4408 for each added stmt if it is assign with
4409 arith_code_with_undefined_signed_overflow, call
4410 rewrite_to_defined_overflow.
4411 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4412 gimple-fold.h instead.
4413 (arith_code_with_undefined_signed_overflow,
4414 rewrite_to_defined_overflow): Moved to ...
4415 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4416 rewrite_to_defined_overflow): ... here. No longer static.
4417 Include gimplify-me.h.
4418 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4419 rewrite_to_defined_overflow): New prototypes.
4421 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4423 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4425 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4427 * builtins.c (get_object_alignment_2): Minor tweak.
4428 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4430 2014-01-13 Christian Bruel <christian.bruel@st.com>
4432 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4433 optimized non constant lengths.
4435 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4438 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4439 load as __atomic_load_N if possible.
4441 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4443 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4445 (rs6000_expand_builtin): Adjust call.
4447 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4450 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4451 * config/rs6000/rs6000.c: Include target-globals.h.
4452 (rs6000_set_current_function): Instead of doing target_reinit
4453 unconditionally, use save_target_globals_default_opts and
4454 restore_target_globals.
4456 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4458 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4459 (rs6000_expand_builtin): Handle mffs and mtfsf.
4460 (rs6000_init_builtins): Define mffs and mtfsf.
4461 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4462 (rs6000_mffs): New pattern.
4463 (rs6000_mtfsf): New pattern.
4465 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4467 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4468 Start narrowing with START. Apply candidate-use pair
4469 and check overall cost in narrowing.
4470 (iv_ca_prune): Pass new argument.
4472 2014-01-10 Jeff Law <law@redhat.com>
4475 * ree.c (combine_reaching_defs): Ensure the defining statement
4476 occurs before the extension when optimizing extensions with
4477 different source and destination hard registers.
4479 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4482 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4483 vtables into the type inheritance graph.
4485 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4487 PR rtl-optimization/59754
4488 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4489 modes in the REGNO != REGNO case.
4491 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4493 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4495 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4497 PR tree-optimization/59745
4498 * tree-predcom.c (tree_predictive_commoning_loop): Call
4499 free_affine_expand_cache if giving up because components is NULL.
4501 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4502 GC in payload of target_globals struct instead of allocating them on
4503 the heap and the larger structs separately using GC.
4504 * target-globals.h (struct target_globals): Make regs, hard_regs,
4505 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4506 of GTY((skip)) and change type to void *.
4507 (reset_target_globals): Cast loads from those fields to corresponding
4510 2014-01-10 Steve Ellcey <sellcey@mips.com>
4513 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4514 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4515 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4517 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4520 * aarch64-modes.def (CC_Zmode): New flags mode.
4521 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4522 represents an equality.
4523 (aarch64_get_condition_code): Handle CC_Zmode.
4524 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4526 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4528 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4529 extraction in good case.
4531 2014-01-10 Richard Biener <rguenther@suse.de>
4533 PR tree-optimization/59374
4534 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4535 checking after SLP discovery. Mark stmts not participating
4536 in any SLP instance properly.
4538 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4540 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4541 when handling a SET rtx.
4543 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4545 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4546 (cortex-a57): Likewise.
4547 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4549 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4551 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4552 non-iwmmxt builtins.
4554 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4558 * ipa-devirt.c record_target_from_binfo): Take as argument
4559 stack of binfos and lookup matching one for virtual inheritance.
4560 (possible_polymorphic_call_targets_1): Update.
4562 2014-01-10 Huacai Chen <chenhc@lemote.com>
4564 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4565 kernel strings for Loongson-2E/2F/3A.
4567 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4570 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4571 is_gimple_call before calling gimple_call_internal_p.
4573 2014-01-09 Steve Ellcey <sellcey@mips.com>
4575 * Makefile.in (TREE_FLOW_H): Remove.
4576 (TREE_SSA_H): Add file names from tree-flow.h.
4577 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4578 * tree.h: Remove tree-flow.h reference.
4579 * hash-table.h: Remove tree-flow.h reference.
4580 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4581 reference with tree-ssa-loop.h.
4583 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4585 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4586 default element-order behavior for -maltivec.
4587 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4589 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4590 when targeting big endian, at least for now.
4591 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4593 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4596 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4597 var satisfies use_register_for_decl, just take into account type
4598 alignment, rather than decl alignment.
4600 PR tree-optimization/59622
4601 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4602 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4603 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4604 Don't devirtualize for inplace at all. For targets.length () == 1,
4605 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4607 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4609 * config/i386/i386.md (cpu): Remove the unused btver1.
4611 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4613 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4615 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4618 * tree-core.h (struct target_globals): New forward declaration.
4619 (struct tree_target_option): Add globals field.
4620 * tree.h (TREE_TARGET_GLOBALS): Define.
4621 (prepare_target_option_nodes_for_pch): New prototype.
4622 * target-globals.h (struct target_globals): Define even if
4624 * tree.c (prepare_target_option_node_for_pch,
4625 prepare_target_option_nodes_for_pch): New functions.
4626 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4627 * config/i386/i386.c: Include target-globals.h.
4628 (ix86_set_current_function): Instead of doing target_reinit
4629 unconditionally, use save_target_globals_default_opts and
4630 restore_target_globals.
4632 2014-01-09 Richard Biener <rguenther@suse.de>
4634 PR tree-optimization/59715
4635 * tree-cfg.h (split_critical_edges): Declare.
4636 * tree-cfg.c (split_critical_edges): Export.
4637 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4639 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4641 * cfgexpand.c (expand_stack_vars): Optionally disable
4642 asan stack protection.
4643 (expand_used_vars): Likewise.
4644 (partition_stack_vars): Likewise.
4645 * asan.c (asan_emit_stack_protection): Optionally disable
4646 after return stack usage.
4647 (instrument_derefs): Optionally disable memory access instrumentation.
4648 (instrument_builtin_call): Likewise.
4649 (instrument_strlen_call): Likewise.
4650 (asan_protect_global): Optionally disable global variables protection.
4651 * doc/invoke.texi: Added doc for new options.
4652 * params.def: Added new options.
4653 * params.h: Likewise.
4655 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4657 PR rtl-optimization/59724
4658 * ifcvt.c (cond_exec_process_if_block): Don't call
4659 flow_find_head_matching_sequence with 0 longest_match.
4660 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4661 non-active insns if !stop_after.
4662 (try_head_merge_bb): Revert 2014-01-07 changes.
4664 2014-01-08 Jeff Law <law@redhat.com>
4666 * ree.c (get_sub_rtx): New function, extracted from...
4667 (merge_def_and_ext): Here.
4668 (combine_reaching_defs): Use get_sub_rtx.
4670 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4672 * cgraph.h (varpool_variable_node): Do not choke on null node.
4674 2014-01-08 Catherine Moore <clm@codesourcery.com>
4676 * config/mips/mips.md (simple_return): Attempt to use JRC
4678 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4680 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4682 PR rtl-optimization/59137
4683 * reorg.c (steal_delay_list_from_target): Call update_block for
4685 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4687 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4689 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4690 two duplicate entries.
4692 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4695 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4697 * config/mips/mips.c (mips_truncated_op_cost): New function.
4698 (mips_rtx_costs): Adjust test for BADDU.
4699 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4701 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4703 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4704 (*baddu_si): ...this new pattern.
4706 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4709 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4711 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4714 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4716 (expand_expr, expand_normal): Adjust.
4717 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4718 inner_reference_p. Use inner_reference_p to expand inner references.
4719 (store_expr): Adjust.
4720 * cfgexpand.c (expand_call_stmt): Adjust.
4722 2014-01-08 Rong Xu <xur@google.com>
4724 * gcov-io.c (gcov_var): Move from gcov-io.h.
4725 (gcov_position): Ditto.
4726 (gcov_is_error): Ditto.
4727 (gcov_rewrite): Ditto.
4728 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4729 only part to libgcc/libgcov.h.
4731 2014-01-08 Marek Polacek <polacek@redhat.com>
4734 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4736 2014-01-08 Marek Polacek <polacek@redhat.com>
4739 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4741 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4743 PR rtl-optimization/59649
4744 * stor-layout.c (get_mode_bounds): For BImode return
4745 0 and STORE_FLAG_VALUE.
4747 2014-01-08 Richard Biener <rguenther@suse.de>
4750 * gimple.h (is_gimple_builtin_call): Remove.
4751 (gimple_builtin_call_types_compatible_p): New.
4752 (gimple_call_builtin_p): New overload.
4753 * gimple.c (is_gimple_builtin_call): Remove.
4754 (validate_call): Rename to ...
4755 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4757 (validate_type): New static function.
4758 (gimple_call_builtin_p): New overload and adjust.
4759 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4760 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4761 (gimple_fold_stmt_to_constant_1): Likewise.
4762 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4764 2014-01-08 Richard Biener <rguenther@suse.de>
4767 * gimplify.c (gimplify_expr): Gimplify register-register type
4768 VIEW_CONVERT_EXPRs to separate stmts.
4770 2014-01-07 Jeff Law <law@redhat.com>
4773 * ree.c (combine_set_extension): Handle case where source
4774 and destination registers in an extension insn are different.
4775 (combine_reaching_defs): Allow source and destination registers
4776 in extension to be different under limited circumstances.
4777 (add_removable_extension): Remove restriction that the
4778 source and destination registers in the extension are the same.
4779 (find_and_remove_re): Emit a copy from the extension's
4780 destination to its source after the defining insn if
4781 the source and destination registers are different.
4784 * ifcvt.c (merge_if_block): If we are merging a block with more than
4785 one successor with a block with no successors, remove any BARRIER
4786 after the second block.
4788 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4790 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4792 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4795 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4796 for 14-bit register offsets when INT14_OK_STRICT is false.
4798 2014-01-07 Roland Stigge <stigge@antcom.de>
4799 Michael Meissner <meissner@linux.vnet.ibm.com>
4802 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4803 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4805 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4807 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4810 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4812 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4813 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4814 rtx is const0_rtx or not.
4816 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4819 * target-globals.c (save_target_globals): Remove this_fn_optab
4821 * toplev.c: Include optabs.h.
4822 (target_reinit): Temporarily restore the global options if another
4823 set of options are in force.
4825 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4827 PR rtl-optimization/58668
4828 * cfgcleanup.c (flow_find_cross_jump): Don't count
4829 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4830 to determine what is counted.
4831 (flow_find_head_matching_sequence): Use active_insn_p to determine
4833 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4835 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4836 determine what is counted.
4838 PR tree-optimization/59643
4839 * tree-predcom.c (split_data_refs_to_components): If one dr is
4840 read and one write, determine_offset fails and the write isn't
4841 in the bad component, just put the read into the bad component.
4843 2014-01-07 Mike Stump <mikestump@comcast.net>
4844 Jakub Jelinek <jakub@redhat.com>
4847 * tree-core.h (struct tree_optimization_option): Change optabs
4848 type from unsigned char * to void *.
4849 * optabs.c (init_tree_optimization_optabs): Adjust
4850 TREE_OPTIMIZATION_OPTABS initialization.
4852 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4855 * config/i386/i386.h (struct machine_function): Add
4856 no_drap_save_restore field.
4857 * config/i386/i386.c (ix86_save_reg): Use
4858 !cfun->machine->no_drap_save_restore instead of
4859 crtl->stack_realign_needed.
4860 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4861 this function clears frame_pointer_needed. Set
4862 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4863 and DRAP reg is needed.
4865 2014-01-06 Marek Polacek <polacek@redhat.com>
4868 * doc/implement-c.texi: Mention that other integer types are
4869 permitted as bit-field types in strictly conforming mode.
4871 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4873 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4876 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4878 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4880 2014-01-06 Martin Jambor <mjambor@suse.cz>
4883 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4885 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4887 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4891 * var-tracking.c (add_stores): Preserve the value of the source even if
4892 we don't record the store.
4894 2014-01-06 Terry Guo <terry.guo@arm.com>
4896 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4898 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4901 * config/darwin.c (darwin_function_section): Adjust return values to
4902 correspond to optimisation changes made in r206070.
4904 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4906 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4907 from prefetch_block tune setting.
4908 (nocona_cost): Correct size of prefetch block to 64.
4910 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4912 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4913 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4914 used to save the static chain register in the computation of the offset
4915 from which the FP registers need to be restored.
4917 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4919 PR tree-optimization/59519
4920 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4921 ICE if get_current_def (current_new_name) is already non-NULL, as long
4922 as it is a phi result of some other phi in *new_exit_bb that has
4925 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4926 or vmovdqu* for misaligned_operand.
4927 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4928 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4929 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4930 aligned_mem for AVX512F masked aligned load and store builtins and for
4933 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4935 PR tree-optimization/59651
4936 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4937 Address range for negative step should be added by TYPE_SIZE_UNIT.
4939 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4941 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4942 overlapping registers also for registers other than the stack pointer.
4944 2014-01-03 Marek Polacek <polacek@redhat.com>
4947 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4950 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4953 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4956 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4957 (push splitter): Use <P:MODE_SIZE> instead of
4958 GET_MODE_SIZE (<P:MODE>mode).
4959 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4960 (mov -1, reg peephole2): Likewise.
4961 * config/i386/sse.md (*mov<mode>_internal,
4962 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4963 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4964 *<code><mode>3, *andnot<mode>3<mask_name>,
4965 <mask_codefor><code><mode>3<mask_name>): Likewise.
4966 * config/i386/subst.md (mask_mode512bit_condition,
4967 sd_mask_mode512bit_condition): Likewise.
4969 2014-01-02 Xinliang David Li <davidxl@google.com>
4971 PR tree-optimization/59303
4972 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4973 (dump_predicates): Better output format.
4974 (pred_equal_p): New function.
4975 (is_neq_relop_p): Ditto.
4976 (is_neq_zero_form_p): Ditto.
4977 (pred_expr_equal_p): Ditto.
4978 (pred_neg_p): Ditto.
4979 (simplify_pred): Ditto.
4980 (simplify_preds_2): Ditto.
4981 (simplify_preds_3): Ditto.
4982 (simplify_preds_4): Ditto.
4983 (simplify_preds): Ditto.
4985 (push_to_worklist): Ditto.
4986 (get_pred_info_from_cmp): Ditto.
4987 (is_degenerated_phi): Ditto.
4988 (normalize_one_pred_1): Ditto.
4989 (normalize_one_pred): Ditto.
4990 (normalize_one_pred_chain): Ditto.
4991 (normalize_preds): Ditto.
4992 (normalize_cond_1): Remove function.
4993 (normalize_cond): Ditto.
4994 (is_gcond_subset_of): Ditto.
4995 (is_subset_of_any): Ditto.
4996 (is_or_set_subset_of): Ditto.
4997 (is_and_set_subset_of): Ditto.
4998 (is_norm_cond_subset_of): Ditto.
4999 (pred_chain_length_cmp): Ditto.
5000 (convert_control_dep_chain_into_preds): Type change.
5001 (find_predicates): Ditto.
5002 (find_def_preds): Ditto.
5003 (destroy_predicates_vecs): Ditto.
5004 (find_matching_predicates_in_rest_chains): Ditto.
5005 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5006 (is_pred_expr_subset): Ditto.
5007 (is_pred_chain_subset_of): Ditto.
5008 (is_included_in): Ditto.
5009 (is_superset_of): Ditto.
5011 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5013 Update copyright years.
5015 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5017 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5018 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5019 config/arc/arc.md, config/arc/arc.opt,
5020 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5021 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5022 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5023 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5024 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5025 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5026 vtable-verify.c, vtable-verify.h: Use the standard form for the
5029 2014-01-02 Tobias Burnus <burnus@net-b.de>
5031 * gcc.c (process_command): Update copyright notice dates.
5032 * gcov-dump.c: Ditto.
5034 * doc/cpp.texi: Bump @copying's copyright year.
5035 * doc/cppinternals.texi: Ditto.
5036 * doc/gcc.texi: Ditto.
5037 * doc/gccint.texi: Ditto.
5038 * doc/gcov.texi: Ditto.
5039 * doc/install.texi: Ditto.
5040 * doc/invoke.texi: Ditto.
5042 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5044 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5046 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5048 * config/i386/sse.md (*mov<mode>_internal): Guard
5049 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5051 PR rtl-optimization/59647
5052 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5053 new_rtx into UNSIGNED_FLOAT rtxes.
5055 Copyright (C) 2014 Free Software Foundation, Inc.
5057 Copying and distribution of this file, with or without modification,
5058 are permitted in any medium without royalty provided the copyright
5059 notice and this notice are preserved.