gcc/ChangeLog

   1 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
   2
   3         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
   4         * configure: Regenerate.
   5         * config/sparc/sparc.opt (muser-mode): New option.
   6         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
   7         for LEON3.
   8         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
   9         * doc/invoke.texi (SPARC options): Document -muser-mode.
  10
  11 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
  12
  13         * cselib.c (find_slot_memmode): Delete.
  14         (cselib_hasher): Change compare_type to a struct.
  15         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
  16         constants.
  17         (preserve_constants_and_equivs): Adjust for new compare_type.
  18         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
  19         (wrap_constant): Delete.
  20         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
  21
  22 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
  23
  24         * doc/install.texi (Building with profile feedback): Remove
  25         outdated sentence.
  26
  27 2014-04-26  Tom de Vries  <tom@codesourcery.com>
  28
  29         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
  30         array accesses.
  31
  32 2014-04-25  Cary Coutant  <ccoutant@google.com>
  33
  34         PR debug/60929
  35         * dwarf2out.c (should_move_die_to_comdat): A type definition
  36         can contain a subprogram definition, but don't move it to a
  37         comdat unit.
  38         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
  39         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
  40         from original DIE.
  41         (clone_tree_hash): Rename to...
  42         (clone_tree_partial): ...this; change callers.  Copy
  43         DW_TAG_subprogram DIEs as declarations.
  44         (copy_decls_walk): Don't copy children of a declaration into a
  45         type unit.
  46
  47 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
  48
  49         PR target/60969
  50         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
  51         alternative 12.
  52
  53 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
  54
  55         * config/arm/predicates.md (call_insn_operand): Add long_call check.
  56         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
  57         reg for long_call.
  58         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
  59         restriction.
  60
  61 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  62
  63         * config/arm/arm.c (arm_cortex_a8_tune): Initialise
  64         T16-related fields.
  65
  66 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
  67
  68         PR tree-optimization/60930
  69         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
  70         creating a multiply candidate by folding two constant
  71         multiplicands when the result overflows.
  72
  73 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
  74
  75         PR tree-optimization/60960
  76         * tree-vect-generic.c (expand_vector_operation): Only call
  77         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
  78
  79 2014-04-25  Tom de Vries  <tom@codesourcery.com>
  80
  81         * expr.c (clobber_reg_mode): New function.
  82         * expr.h (clobber_reg): New function.
  83
  84 2014-04-25  Tom de Vries  <tom@codesourcery.com>
  85
  86         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
  87         clobbers.
  88
  89 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
  90             Tom de Vries  <tom@codesourcery.com>
  91
  92         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
  93         handle.
  94         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
  95         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
  96         new argument to find_all_hard_reg_sets call.
  97
  98 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  99
 100         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
 101         Use HOST_WIDE_INT_C for mask literal.
 102         (aarch_rev16_shleft_mask_imm_p): Likewise.
 103
 104 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
 105
 106         PR target/60941
 107         * config/sparc/sparc.md (ashlsi3_extend): Delete.
 108
 109 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
 110
 111         PR preprocessor/56540
 112         * config/i386/i386-c.c (ix86_target_macros): Define
 113         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
 114
 115 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 116
 117         * configure.ac (tga_func): Remove.
 118         (LIB_TLS_SPEC): Remove.
 119         * configure: Regenerate.
 120         * config.in: Regenerate.
 121         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
 122
 123 2014-04-25  Richard Biener  <rguenther@suse.de>
 124
 125         PR ipa/60912
 126         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
 127         call stmt use/clobber sets during stmt walk instead of
 128         walking the possibly incomplete set of caller edges.
 129
 130 2014-04-25  Richard Biener  <rguenther@suse.de>
 131
 132         PR ipa/60911
 133         * passes.c (apply_ipa_transforms): Inline into only caller ...
 134         (execute_one_pass): ... here.  Properly bring in function
 135         bodies for nodes we want to apply IPA transforms to.
 136
 137 2014-04-24  Cong Hou  <congh@google.com>
 138
 139         PR tree-optimization/60896
 140         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
 141         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
 142         (vect_mark_pattern_stmts): Set the def type of all statements in
 143         PATTERN_DEF_SEQ as vect_internal_def.
 144
 145 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
 146
 147         * doc/extend.texi (PowerPC Built-in Functions): Document new
 148         powerpc extended divide, bcd, pack/unpack 128-bit, builtin
 149         functions.
 150         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
 151
 152         * config/rs6000/predicates.md (const_0_to_3_operand): New
 153         predicate to match 0..3 integer constants.
 154
 155         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
 156         to support adding miscellaneous builtin functions.
 157         (BU_DFP_MISC_2): Likewise.
 158         (BU_P7_MISC_1): Likewise.
 159         (BU_P7_MISC_2): Likewise.
 160         (BU_P8V_MISC_3): Likewise.
 161         (BU_MISC_1): Likewise.
 162         (BU_MISC_2): Likewise.
 163         (DIVWE): Add extended divide builtin functions.
 164         (DIVWEO): Likewise.
 165         (DIVWEU): Likewise.
 166         (DIVWEUO): Likewise.
 167         (DIVDE): Likewise.
 168         (DIVDEO): Likewise.
 169         (DIVDEU): Likewise.
 170         (DIVDEUO): Likewise.
 171         (DXEX): Add decimal floating-point builtin functions.
 172         (DXEXQ): Likewise.
 173         (DDEDPD): Likewise.
 174         (DDEDPDQ): Likewise.
 175         (DENBCD): Likewise.
 176         (DENBCDQ): Likewise.
 177         (DIEX): Likewise.
 178         (DIEXQ): Likewise.
 179         (DSCLI): Likewise.
 180         (DSCLIQ): Likewise.
 181         (DSCRI): Likewise.
 182         (DSCRIQ): Likewise.
 183         (CDTBCD): Add new BCD builtin functions.
 184         (CBCDTD): Likewise.
 185         (ADDG6S): Likewise.
 186         (BCDADD): Likewise.
 187         (BCDADD_LT): Likewise.
 188         (BCDADD_EQ): Likewise.
 189         (BCDADD_GT): Likewise.
 190         (BCDADD_OV): Likewise.
 191         (BCDSUB): Likewise.
 192         (BCDSUB_LT): Likewise.
 193         (BCDSUB_EQ): Likewise.
 194         (BCDSUB_GT): Likewise.
 195         (BCDSUB_OV): Likewise.
 196         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
 197         (UNPACK_TD): Likewise.
 198         (PACK_TF): Likewise.
 199         (UNPACK_TF): Likewise.
 200         (UNPACK_TF_0): Likewise.
 201         (UNPACK_TF_1): Likewise.
 202         (PACK_V1TI): Likewise.
 203         (UNPACK_V1TI): Likewise.
 204
 205         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
 206         support for decimal floating point builtin functions.
 207         (rs6000_expand_ternop_builtin): Add checks for the new builtin
 208         functions that take constant arguments.
 209         (rs6000_invalid_builtin): Add decimal floating point builtin
 210         support.
 211         (rs6000_init_builtins): Setup long double, _Decimal64, and
 212         _Decimal128 types for new builtin functions.
 213         (builtin_function_type): Set the unsigned flags appropriately for
 214         the new builtin functions.
 215         (rs6000_opt_masks): Add support for decimal floating point builtin
 216         functions.
 217
 218         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
 219         floating point builtin functions.
 220         (RS6000_BTM_COMMON): Likewise.
 221         (RS6000_BTI_long_double): Likewise.
 222         (RS6000_BTI_dfloat64): Likewise.
 223         (RS6000_BTI_dfloat128): Likewise.
 224         (long_double_type_internal_node): Likewise.
 225         (dfloat64_type_internal_node): Likewise.
 226         (dfloat128_type_internal_node): Likewise.
 227
 228         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
 229         2.07 bcd arithmetic instructions.
 230         (UNSPEC_BCDSUB): Likewise.
 231         (UNSPEC_BCD_OVERFLOW): Likewise.
 232         (UNSPEC_BCD_ADD_SUB): Likewise.
 233         (bcd_add_sub): Likewise.
 234         (BCD_TEST): Likewise.
 235         (bcd<bcd_add_sub>): Likewise.
 236         (bcd<bcd_add_sub>_test): Likewise.
 237         (bcd<bcd_add_sub>_test2): Likewise.
 238         (bcd<bcd_add_sub>_<code>): Likewise.
 239         (peephole2 for combined bcd ops): Likewise.
 240
 241         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
 242         decimal floating point builtin functions.
 243         (UNSPEC_DENBCD): Likewise.
 244         (UNSPEC_DXEX): Likewise.
 245         (UNSPEC_DIEX): Likewise.
 246         (UNSPEC_DSCLI): Likewise.
 247         (UNSPEC_DSCRI): Likewise.
 248         (D64_D128): Likewise.
 249         (dfp_suffix): Likewise.
 250         (dfp_ddedpd_<mode>): Likewise.
 251         (dfp_denbcd_<mode>): Likewise.
 252         (dfp_dxex_<mode>): Likewise.
 253         (dfp_diex_<mode>): Likewise.
 254         (dfp_dscli_<mode>): Likewise.
 255         (dfp_dscri_<mode>): Likewise.
 256
 257         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
 258         builtin functions.
 259         (UNSPEC_CDTBCD): Likewise.
 260         (UNSPEC_CBCDTD): Likewise.
 261         (UNSPEC_DIVE): Add support for new extended divide builtin
 262         functions.
 263         (UNSPEC_DIVEO): Likewise.
 264         (UNSPEC_DIVEU): Likewise.
 265         (UNSPEC_DIVEUO): Likewise.
 266         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
 267         pack/unpack 128-bit types.
 268         (UNSPEC_PACK_128BIT): Likewise.
 269         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
 270         (udiv<mode>3): Use idiv_ldiv mode attribute.
 271         (div<mode>3): Likewise.
 272         (addg6s): Add new BCD builtin functions.
 273         (cdtbcd): Likewise.
 274         (cbcdtd): Likewise.
 275         (UNSPEC_DIV_EXTEND): Add support for new extended divide
 276         instructions.
 277         (div_extend): Likewise.
 278         (div<div_extend>_<mode>"): Likewise.
 279         (FP128_64): Add support for new builtin functions to pack/unpack
 280         128-bit types.
 281         (unpack<mode>): Likewise.
 282         (unpacktf_0): Likewise.
 283         (unpacktf_1): Likewise.
 284         (unpack<mode>_dm): Likewise.
 285         (unpack<mode>_nodm): Likewise.
 286         (pack<mode>): Likewise.
 287         (unpackv1ti): Likewise.
 288         (packv1ti): Likewise.
 289
 290 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
 291
 292         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
 293         is disabled.
 294
 295 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
 296
 297         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
 298         * gimplify.c (omp_is_private): Change last argument's type to int.
 299         Only diagnose lastprivate if the simd argument is 1, only diagnose
 300         linear if the simd argument is 2.
 301         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
 302         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
 303         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
 304         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
 305         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
 306         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 307         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
 308         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 309         * tree-nested.c (convert_nonlocal_omp_clauses,
 310         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
 311
 312 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
 313
 314         PR target/60822
 315         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
 316         operand 1.
 317
 318 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
 319
 320         * flag-types.h (enum ivar_visibility): Add.
 321
 322 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
 323
 324         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
 325         function * argument.
 326
 327 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
 328
 329         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
 330
 331 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 332             Tom de Vries  <tom@codesourcery.com>
 333
 334         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
 335         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
 336         reg-note.
 337         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
 338         * emit-rtl.c (try_split): Same.
 339
 340 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 341             Tom de Vries  <tom@codesourcery.com>
 342
 343         * common.opt (fuse-caller-save): New option.
 344
 345 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
 346
 347         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
 348         for big-endian.
 349
 350 2014-04-24  Richard Biener  <rguenther@suse.de>
 351
 352         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
 353         during TER and instead use the sepops interface for expanding
 354         non-GIMPLE_SINGLE_RHS.
 355
 356 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 357
 358         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
 359         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
 360
 361 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 362
 363         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
 364         assembler 64-bit option.
 365         * configure: Regenerate.
 366
 367 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 368
 369         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
 370         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
 371         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
 372         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
 373         (TARGET_CRYPTO): Take TARGET_SIMD into account.
 374
 375 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 376
 377         * config/aarch64/aarch64-builtins.c
 378         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
 379         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
 380         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
 381         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
 382         builtins.
 383         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
 384         (Vrevsuff): New mode attribute.
 385
 386 2014-04-24  Terry Guo  <terry.guo@arm.com>
 387
 388         * config/arm/arm.h (machine_function): Define variable
 389         after_arm_reorg here.
 390         * config/arm/arm.c (after_arm_reorg): Remove the definition.
 391         (arm_split_constant): Update the way to access variable
 392         after_arm_reorg.
 393         (arm_reorg): Ditto.
 394         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
 395
 396 2014-04-23  Tom de Vries  <tom@codesourcery.com>
 397
 398         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
 399
 400 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
 401
 402         * is-a.h: Update comments to reflect the following changes to the
 403         "pointerness" of the API, making the template parameter match the
 404         return type, allowing use of is-a.h with typedefs of pointers.
 405         (is_a_helper::cast): Return a T rather then a pointer to a T, so
 406         that the return type matches the parameter to the is_a_helper.
 407         (as_a): Likewise.
 408         (dyn_cast): Likewise.
 409
 410         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
 411         pointer from the is-a.h API.
 412
 413         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
 414         (is_a_helper <cgraph_node *>::test): ...this, matching change to
 415         is-a.h API.
 416         (is_a_helper <varpool_node>::test): Likewise, convert to...
 417         (is_a_helper <varpool_node *>::test): ...this.
 418
 419         (varpool_first_variable): Update for removal of implicit pointer
 420         from the is-a.h API.
 421         (varpool_next_variable): Likewise.
 422         (varpool_first_static_initializer): Likewise.
 423         (varpool_next_static_initializer): Likewise.
 424         (varpool_first_defined_variable): Likewise.
 425         (varpool_next_defined_variable): Likewise.
 426         (cgraph_first_defined_function): Likewise.
 427         (cgraph_next_defined_function): Likewise.
 428         (cgraph_first_function): Likewise.
 429         (cgraph_next_function): Likewise.
 430         (cgraph_first_function_with_gimple_body): Likewise.
 431         (cgraph_next_function_with_gimple_body): Likewise.
 432         (cgraph_alias_target): Likewise.
 433         (varpool_alias_target): Likewise.
 434         (cgraph_function_or_thunk_node): Likewise.
 435         (varpool_variable_node): Likewise.
 436         (symtab_real_symbol_p): Likewise.
 437         * cgraphunit.c (referred_to_p): Likewise.
 438         (analyze_functions): Likewise.
 439         (handle_alias_pairs): Likewise.
 440         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
 441         * gimple-ssa.h (gimple_vuse_op): Likewise.
 442         (gimple_vdef_op): Likewise.
 443         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
 444         * gimple.c (gimple_build_asm_1): Likewise.
 445         (gimple_build_try): Likewise.
 446         (gimple_build_resx): Likewise.
 447         (gimple_build_eh_dispatch): Likewise.
 448         (gimple_build_omp_for): Likewise.
 449         (gimple_omp_for_set_clauses): Likewise.
 450
 451         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
 452         (is_a_helper <gimple_statement_asm *>::test): ...this.
 453         (is_a_helper <gimple_statement_bind>::test): Convert to...
 454         (is_a_helper <gimple_statement_bind *>::test): ...this.
 455         (is_a_helper <gimple_statement_call>::test): Convert to...
 456         (is_a_helper <gimple_statement_call *>::test): ...this.
 457         (is_a_helper <gimple_statement_catch>::test): Convert to...
 458         (is_a_helper <gimple_statement_catch *>::test): ...this.
 459         (is_a_helper <gimple_statement_resx>::test): Convert to...
 460         (is_a_helper <gimple_statement_resx *>::test): ...this.
 461         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
 462         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
 463         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
 464         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
 465         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
 466         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
 467         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
 468         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
 469         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
 470         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
 471         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
 472         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
 473         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
 474         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
 475         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
 476         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
 477         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
 478         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
 479         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
 480         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
 481         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
 482         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
 483         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
 484         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
 485         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
 486         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
 487         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
 488         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
 489         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
 490         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
 491         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
 492         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
 493         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
 494         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
 495         (is_a_helper <gimple_statement_phi>::test): Convert to...
 496         (is_a_helper <gimple_statement_phi *>::test): ...this.
 497         (is_a_helper <gimple_statement_transaction>::test): Convert to...
 498         (is_a_helper <gimple_statement_transaction *>::test): ...this.
 499         (is_a_helper <gimple_statement_try>::test): Convert to...
 500         (is_a_helper <gimple_statement_try *>::test): ...this.
 501         (is_a_helper <gimple_statement_wce>::test): Convert to...
 502         (is_a_helper <gimple_statement_wce *>::test): ...this.
 503         (is_a_helper <const gimple_statement_asm>::test): Convert to...
 504         (is_a_helper <const gimple_statement_asm *>::test): ...this.
 505         (is_a_helper <const gimple_statement_bind>::test): Convert to...
 506         (is_a_helper <const gimple_statement_bind *>::test): ...this.
 507         (is_a_helper <const gimple_statement_call>::test): Convert to...
 508         (is_a_helper <const gimple_statement_call *>::test): ...this.
 509         (is_a_helper <const gimple_statement_catch>::test): Convert to...
 510         (is_a_helper <const gimple_statement_catch *>::test): ...this.
 511         (is_a_helper <const gimple_statement_resx>::test): Convert to...
 512         (is_a_helper <const gimple_statement_resx *>::test): ...this.
 513         (is_a_helper <const gimple_statement_eh_dispatch>::test):
 514         Convert to...
 515         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
 516         ...this.
 517         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
 518         to...
 519         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
 520         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
 521         Convert to...
 522         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
 523         ...this.
 524         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
 525         Convert to...
 526         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
 527         ...this.
 528         (is_a_helper <const gimple_statement_omp_return>::test): Convert
 529         to...
 530         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
 531         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
 532         to...
 533         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
 534         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
 535         to...
 536         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
 537         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
 538         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
 539         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
 540         to...
 541         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
 542         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
 543         to...
 544         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
 545         (is_a_helper <const gimple_statement_omp_target>::test): Convert
 546         to...
 547         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
 548         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
 549         to...
 550         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
 551         (is_a_helper <const gimple_statement_omp_single>::test): Convert
 552         to...
 553         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
 554         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
 555         to...
 556         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
 557         (is_a_helper <const gimple_statement_omp_task>::test): Convert
 558         to...
 559         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
 560         (is_a_helper <const gimple_statement_phi>::test): Convert to...
 561         (is_a_helper <const gimple_statement_phi *>::test): ...this.
 562         (is_a_helper <const gimple_statement_transaction>::test): Convert
 563         to...
 564         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
 565         (is_a_helper <const gimple_statement_with_ops>::test): Convert
 566         to...
 567         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
 568         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
 569         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
 570         (is_a_helper <const gimple_statement_with_memory_ops>::test):
 571         Convert to...
 572         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
 573         ...this.
 574         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
 575         to...
 576         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
 577
 578         (gimple_use_ops): Update for removal of implicit pointer from the
 579         is-a.h API.
 580         (gimple_set_use_ops): Likewise.
 581         (gimple_vuse): Likewise.
 582         (gimple_vdef): Likewise.
 583         (gimple_vuse_ptr): Likewise.
 584         (gimple_vdef_ptr): Likewise.
 585         (gimple_set_vuse): Likewise.
 586         (gimple_set_vdef): Likewise.
 587         (gimple_omp_return_set_lhs): Likewise.
 588         (gimple_omp_return_lhs): Likewise.
 589         (gimple_omp_return_lhs_ptr): Likewise.
 590         (gimple_call_fntype): Likewise.
 591         (gimple_call_set_fntype): Likewise.
 592         (gimple_call_set_internal_fn): Likewise.
 593         (gimple_call_use_set): Likewise.
 594         (gimple_call_clobber_set): Likewise.
 595         (gimple_bind_vars): Likewise.
 596         (gimple_bind_set_vars): Likewise.
 597         (gimple_bind_body_ptr): Likewise.
 598         (gimple_bind_set_body): Likewise.
 599         (gimple_bind_add_stmt): Likewise.
 600         (gimple_bind_block): Likewise.
 601         (gimple_bind_set_block): Likewise.
 602         (gimple_asm_ninputs): Likewise.
 603         (gimple_asm_noutputs): Likewise.
 604         (gimple_asm_nclobbers): Likewise.
 605         (gimple_asm_nlabels): Likewise.
 606         (gimple_asm_input_op): Likewise.
 607         (gimple_asm_input_op_ptr): Likewise.
 608         (gimple_asm_output_op): Likewise.
 609         (gimple_asm_output_op_ptr): Likewise.
 610         (gimple_asm_set_output_op): Likewise.
 611         (gimple_asm_clobber_op): Likewise.
 612         (gimple_asm_set_clobber_op): Likewise.
 613         (gimple_asm_label_op): Likewise.
 614         (gimple_asm_set_label_op): Likewise.
 615         (gimple_asm_string): Likewise.
 616         (gimple_catch_types): Likewise.
 617         (gimple_catch_types_ptr): Likewise.
 618         (gimple_catch_handler_ptr): Likewise.
 619         (gimple_catch_set_types): Likewise.
 620         (gimple_catch_set_handler): Likewise.
 621         (gimple_eh_filter_types): Likewise.
 622         (gimple_eh_filter_types_ptr): Likewise.
 623         (gimple_eh_filter_failure_ptr): Likewise.
 624         (gimple_eh_filter_set_types): Likewise.
 625         (gimple_eh_filter_set_failure): Likewise.
 626         (gimple_eh_must_not_throw_fndecl): Likewise.
 627         (gimple_eh_must_not_throw_set_fndecl): Likewise.
 628         (gimple_eh_else_n_body_ptr): Likewise.
 629         (gimple_eh_else_e_body_ptr): Likewise.
 630         (gimple_eh_else_set_n_body): Likewise.
 631         (gimple_eh_else_set_e_body): Likewise.
 632         (gimple_try_eval_ptr): Likewise.
 633         (gimple_try_cleanup_ptr): Likewise.
 634         (gimple_try_set_eval): Likewise.
 635         (gimple_try_set_cleanup): Likewise.
 636         (gimple_wce_cleanup_ptr): Likewise.
 637         (gimple_wce_set_cleanup): Likewise.
 638         (gimple_phi_capacity): Likewise.
 639         (gimple_phi_num_args): Likewise.
 640         (gimple_phi_result): Likewise.
 641         (gimple_phi_result_ptr): Likewise.
 642         (gimple_phi_set_result): Likewise.
 643         (gimple_phi_arg): Likewise.
 644         (gimple_phi_set_arg): Likewise.
 645         (gimple_resx_region): Likewise.
 646         (gimple_resx_set_region): Likewise.
 647         (gimple_eh_dispatch_region): Likewise.
 648         (gimple_eh_dispatch_set_region): Likewise.
 649         (gimple_omp_critical_name): Likewise.
 650         (gimple_omp_critical_name_ptr): Likewise.
 651         (gimple_omp_critical_set_name): Likewise.
 652         (gimple_omp_for_clauses): Likewise.
 653         (gimple_omp_for_clauses_ptr): Likewise.
 654         (gimple_omp_for_set_clauses): Likewise.
 655         (gimple_omp_for_collapse): Likewise.
 656         (gimple_omp_for_index): Likewise.
 657         (gimple_omp_for_index_ptr): Likewise.
 658         (gimple_omp_for_set_index): Likewise.
 659         (gimple_omp_for_initial): Likewise.
 660         (gimple_omp_for_initial_ptr): Likewise.
 661         (gimple_omp_for_set_initial): Likewise.
 662         (gimple_omp_for_final): Likewise.
 663         (gimple_omp_for_final_ptr): Likewise.
 664         (gimple_omp_for_set_final): Likewise.
 665         (gimple_omp_for_incr): Likewise.
 666         (gimple_omp_for_incr_ptr): Likewise.
 667         (gimple_omp_for_set_incr): Likewise.
 668         (gimple_omp_for_pre_body_ptr): Likewise.
 669         (gimple_omp_for_set_pre_body): Likewise.
 670         (gimple_omp_parallel_clauses): Likewise.
 671         (gimple_omp_parallel_clauses_ptr): Likewise.
 672         (gimple_omp_parallel_set_clauses): Likewise.
 673         (gimple_omp_parallel_child_fn): Likewise.
 674         (gimple_omp_parallel_child_fn_ptr): Likewise.
 675         (gimple_omp_parallel_set_child_fn): Likewise.
 676         (gimple_omp_parallel_data_arg): Likewise.
 677         (gimple_omp_parallel_data_arg_ptr): Likewise.
 678         (gimple_omp_parallel_set_data_arg): Likewise.
 679         (gimple_omp_task_clauses): Likewise.
 680         (gimple_omp_task_clauses_ptr): Likewise.
 681         (gimple_omp_task_set_clauses): Likewise.
 682         (gimple_omp_task_child_fn): Likewise.
 683         (gimple_omp_task_child_fn_ptr): Likewise.
 684         (gimple_omp_task_set_child_fn): Likewise.
 685         (gimple_omp_task_data_arg): Likewise.
 686         (gimple_omp_task_data_arg_ptr): Likewise.
 687         (gimple_omp_task_set_data_arg): Likewise.
 688         (gimple_omp_taskreg_clauses): Likewise.
 689         (gimple_omp_taskreg_clauses_ptr): Likewise.
 690         (gimple_omp_taskreg_set_clauses): Likewise.
 691         (gimple_omp_taskreg_child_fn): Likewise.
 692         (gimple_omp_taskreg_child_fn_ptr): Likewise.
 693         (gimple_omp_taskreg_set_child_fn): Likewise.
 694         (gimple_omp_taskreg_data_arg): Likewise.
 695         (gimple_omp_taskreg_data_arg_ptr): Likewise.
 696         (gimple_omp_taskreg_set_data_arg): Likewise.
 697         (gimple_omp_task_copy_fn): Likewise.
 698         (gimple_omp_task_copy_fn_ptr): Likewise.
 699         (gimple_omp_task_set_copy_fn): Likewise.
 700         (gimple_omp_task_arg_size): Likewise.
 701         (gimple_omp_task_arg_size_ptr): Likewise.
 702         (gimple_omp_task_set_arg_size): Likewise.
 703         (gimple_omp_task_arg_align): Likewise.
 704         (gimple_omp_task_arg_align_ptr): Likewise.
 705         (gimple_omp_task_set_arg_align): Likewise.
 706         (gimple_omp_single_clauses): Likewise.
 707         (gimple_omp_single_clauses_ptr): Likewise.
 708         (gimple_omp_single_set_clauses): Likewise.
 709         (gimple_omp_target_clauses): Likewise.
 710         (gimple_omp_target_clauses_ptr): Likewise.
 711         (gimple_omp_target_set_clauses): Likewise.
 712         (gimple_omp_target_child_fn): Likewise.
 713         (gimple_omp_target_child_fn_ptr): Likewise.
 714         (gimple_omp_target_set_child_fn): Likewise.
 715         (gimple_omp_target_data_arg): Likewise.
 716         (gimple_omp_target_data_arg_ptr): Likewise.
 717         (gimple_omp_target_set_data_arg): Likewise.
 718         (gimple_omp_teams_clauses): Likewise.
 719         (gimple_omp_teams_clauses_ptr): Likewise.
 720         (gimple_omp_teams_set_clauses): Likewise.
 721         (gimple_omp_sections_clauses): Likewise.
 722         (gimple_omp_sections_clauses_ptr): Likewise.
 723         (gimple_omp_sections_set_clauses): Likewise.
 724         (gimple_omp_sections_control): Likewise.
 725         (gimple_omp_sections_control_ptr): Likewise.
 726         (gimple_omp_sections_set_control): Likewise.
 727         (gimple_omp_for_set_cond): Likewise.
 728         (gimple_omp_for_cond): Likewise.
 729         (gimple_omp_atomic_store_set_val): Likewise.
 730         (gimple_omp_atomic_store_val): Likewise.
 731         (gimple_omp_atomic_store_val_ptr): Likewise.
 732         (gimple_omp_atomic_load_set_lhs): Likewise.
 733         (gimple_omp_atomic_load_lhs): Likewise.
 734         (gimple_omp_atomic_load_lhs_ptr): Likewise.
 735         (gimple_omp_atomic_load_set_rhs): Likewise.
 736         (gimple_omp_atomic_load_rhs): Likewise.
 737         (gimple_omp_atomic_load_rhs_ptr): Likewise.
 738         (gimple_omp_continue_control_def): Likewise.
 739         (gimple_omp_continue_control_def_ptr): Likewise.
 740         (gimple_omp_continue_set_control_def): Likewise.
 741         (gimple_omp_continue_control_use): Likewise.
 742         (gimple_omp_continue_control_use_ptr): Likewise.
 743         (gimple_omp_continue_set_control_use): Likewise.
 744         (gimple_transaction_body_ptr): Likewise.
 745         (gimple_transaction_label): Likewise.
 746         (gimple_transaction_label_ptr): Likewise.
 747         (gimple_transaction_set_body): Likewise.
 748         (gimple_transaction_set_label): Likewise.
 749
 750         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
 751         * ipa-inline-analysis.c (inline_write_summary): Likewise.
 752         * ipa-ref.c (ipa_record_reference): Likewise.
 753         * ipa-reference.c (analyze_function): Likewise.
 754         (ipa_reference_write_optimization_summary): Likewise.
 755         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
 756         (address_taken_from_non_vtable_p): Likewise.
 757         (comdat_can_be_unshared_p_1): Likewise.
 758         * lto-cgraph.c (lto_output_ref): Likewise.
 759         (add_references): Likewise.
 760         (compute_ltrans_boundary): Likewise.
 761         (output_symtab): Likewise.
 762         (input_ref): Likewise.
 763         (input_cgraph_1): Likewise.
 764         (output_cgraph_opt_summary): Likewise.
 765         * lto-streamer-out.c (lto_output): Likewise.
 766         (output_symbol_p): Likewise.
 767         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
 768         (lsei_start_function_in_partition): Likewise.
 769         (lsei_next_variable_in_partition): Likewise.
 770         (lsei_start_variable_in_partition): Likewise.
 771         * symtab.c (insert_to_assembler_name_hash): Likewise.
 772         (unlink_from_assembler_name_hash): Likewise.
 773         (symtab_unregister_node): Likewise.
 774         (symtab_remove_node): Likewise.
 775         (dump_symtab_node): Likewise.
 776         (verify_symtab_base): Likewise.
 777         (verify_symtab_node): Likewise.
 778         (symtab_make_decl_local): Likewise.
 779         (symtab_alias_ultimate_target): Likewise.
 780         (symtab_resolve_alias): Likewise.
 781         (symtab_get_symbol_partitioning_class): Likewise.
 782         * tree-phinodes.c (allocate_phi_node): Likewise.
 783         (reserve_phi_args_for_new_edge): Likewise.
 784         (remove_phi_args): Likewise.
 785         * varpool.c (varpool_node_for_asm): Likewise.
 786         (varpool_remove_unreferenced_decls): Likewise.
 787
 788 2014-04-23  Jeff Law  <law@redhat.com>
 789
 790         PR tree-optimization/60902
 791         * tree-ssa-threadedge.c
 792         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
 793         invalidate outputs from statements that do not produce useful
 794         outputs for threading.
 795
 796 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
 797
 798         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
 799         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
 800         machine descriptions for Stack Smashing Protector.
 801
 802 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
 803
 804         * aarch64.md (<optab>_rol<mode>3): New pattern.
 805         (<optab>_rolsi3_uxtw): Likewise.
 806         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
 807
 808 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
 809
 810         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
 811         (arm_cortex_a12_tune): Likewise.
 812
 813 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 814
 815         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
 816
 817 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 818
 819         * config/arm/arm.md (arm_rev16si2): New pattern.
 820         (arm_rev16si2_alt): Likewise.
 821         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
 822
 823 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 824
 825        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
 826        (rev16<mode>2_alt): Likewise.
 827        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
 828        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
 829        (aarch_rev16_shleft_mask_imm_p): Likewise.
 830        (aarch_rev16_p_1): Likewise.
 831        (aarch_rev16_p): Likewise.
 832        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
 833        (aarch_rev16_shright_mask_imm_p): Likewise.
 834        (aarch_rev16_shleft_mask_imm_p): Likewise.
 835
 836 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 837
 838        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
 839        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
 840        rev cost.
 841        (cortex_a53_extra_costs): Likewise.
 842        (cortex_a57_extra_costs): Likewise.
 843        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
 844        (cortexa7_extra_costs): Likewise.
 845        (cortexa8_extra_costs): Likewise.
 846        (cortexa12_extra_costs): Likewise.
 847        (cortexa15_extra_costs): Likewise.
 848        (v7m_extra_costs): Likewise.
 849        (arm_new_rtx_costs): Handle BSWAP.
 850
 851 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 852
 853        * config/arm/arm.c (cortexa8_extra_costs): New table.
 854        (arm_cortex_a8_tune): New tuning struct.
 855        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
 856
 857 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 858
 859        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
 860
 861 2014-04-23  Richard Biener  <rguenther@suse.de>
 862
 863         * Makefile.in (OBJS): Remove loop-unswitch.o.
 864         * loop-unswitch.c: Delete.
 865         * tree-pass.h (make_pass_rtl_unswitch): Remove.
 866         * passes.def (pass_rtl_unswitch): Likewise.
 867         * loop-init.c (gate_rtl_unswitch): Likewise.
 868         (rtl_unswitch): Likewise.
 869         (pass_data_rtl_unswitch): Likewise.
 870         (pass_rtl_unswitch): Likewise.
 871         (make_pass_rtl_unswitch): Likewise.
 872         * rtl.h (reversed_condition): Likewise.
 873         (compare_and_jump_seq): Likewise.
 874         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
 875         and make static.
 876         * loop-unroll.c (compare_and_jump_seq): Likewise.
 877
 878 2014-04-23  Richard Biener  <rguenther@suse.de>
 879
 880         PR tree-optimization/60903
 881         * tree-ssa-loop-im.c (analyze_memory_references): Remove
 882         commented code block.
 883         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
 884         loop flags to newly created BBs and edges.
 885
 886 2014-04-23  Nick Clifton  <nickc@redhat.com>
 887
 888         * config/msp430/msp430.c (msp430_handle_option): Move function
 889         to msp430-common.c
 890         (msp430_option_override): Simplify mcu and mcpu option handling.
 891         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
 892         support for -mhwmult command line option.
 893         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
 894         -mhwmult command line option.
 895         (msp430_hwmult_enabled): Delete.
 896         (msp43o_output_labelref): Add support for -mhwmult command line
 897         option.
 898         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
 899         (umulsidi3): Likewise.
 900         * config/msp430/msp430.opt (mmcu): Add Report attribute.
 901         (mcpu, mlarge, msmall): Likewise.
 902         (mhwmult): New option.
 903         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
 904         prototype.
 905         (msp430_is_f5_mcu): Remove prototype.
 906         (msp430_use_f5_series_hwmult): Add prototype.
 907         * config/msp430/msp430-opts.h: New file.
 908         * common/config/msp430: New directory.
 909         * common/config/msp430/msp430-common.c: New file.
 910         * config.gcc (msp430): Remove target_has_targetm_common.
 911         * doc/invoke.texi: Document -mhwmult command line option.
 912
 913 2014-04-23  Nick Clifton  <nickc@redhat.com>
 914
 915         * config/i386/cygwin.h (ENDFILE_SPEC): Include
 916         default-manifest.o if it can be found in the search path.
 917         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
 918
 919 2014-04-23  Terry Guo  <terry.guo@arm.com>
 920
 921         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
 922
 923 2014-04-23  Richard Biener  <rguenther@suse.de>
 924
 925         PR middle-end/60895
 926         * tree-inline.c (declare_return_variable): Use mark_addressable.
 927
 928 2014-04-23  Richard Biener  <rguenther@suse.de>
 929
 930         PR middle-end/60891
 931         * loop-init.c (loop_optimizer_init): Make sure to apply
 932         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
 933
 934 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
 935
 936         PR sanitizer/60275
 937         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
 938         New options.
 939         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
 940         if flag_sanitize_undefined_trap_on_error.
 941         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
 942         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
 943         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
 944         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
 945         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
 946         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
 947         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
 948         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
 949         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
 950         * ubsan.c (ubsan_instrument_unreachable): Return
 951         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
 952         (ubsan_expand_null_ifn): Emit __builtin_trap ()
 953         if flag_sanitize_undefined_trap_on_error and
 954         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
 955         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
 956         instrument_bool_enum_load): Emit __builtin_trap () if
 957         flag_sanitize_undefined_trap_on_error and
 958         __builtin_handle_*_abort () if !flag_sanitize_recover.
 959         * doc/invoke.texi (-fsanitize-recover,
 960         -fsanitize-undefined-trap-on-error): Document.
 961
 962 2014-04-22  Christian Bruel  <christian.bruel@st.com>
 963
 964         * config/sh/sh.md (mov<mode>): Replace movQIHI.
 965          Force immediates to SImode.
 966
 967 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
 968
 969         * config/nios2/nios2.md (UNSPEC_ROUND): New.
 970         (lroundsfsi2): New.
 971         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
 972         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
 973         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
 974         (nios2_fpu_insn): Add entry for round.
 975         (N2FPU_NO_ERRNO_P): Define.
 976         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
 977         flag_errno_math.
 978         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
 979
 980 2014-04-22  Richard Henderson  <rth@redhat.com>
 981
 982         * config/aarch64/aarch64 (addti3, subti3): New expanders.
 983         (add<GPI>3_compare0): Remove leading * from name.
 984         (add<GPI>3_carryin): Likewise.
 985         (sub<GPI>3_compare0): Likewise.
 986         (sub<GPI>3_carryin): Likewise.
 987         (<su_optab>mulditi3): New expander.
 988         (multi3): New expander.
 989         (madd<GPI>): Remove leading * from name.
 990
 991 2014-04-22  Martin Jambor  <mjambor@suse.cz>
 992
 993         * cgraphclones.c (cgraph_function_versioning): Copy
 994         ipa_transforms_to_apply instead of asserting it is empty.
 995
 996 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
 997
 998         PR target/60868
 999         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
1000         on count_exp to get mode.
1001
1002 2014-04-22  Andrew Pinski  <apinski@cavium.com>
1003
1004         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1005         Handle TLS for ILP32.
1006         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
1007         (tlsie_small_<mode>): this and handle PTR.
1008         (tlsie_small_sidi): New pattern.
1009         (tlsle_small): Change to an expand to handle ILP32.
1010         (tlsle_small_<mode>): New pattern.
1011         (tlsdesc_small): Rename to ...
1012         (tlsdesc_small_<mode>): this and handle PTR.
1013
1014 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1015
1016         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
1017
1018 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1019
1020         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1021         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
1022         (aarch64_types_signed_poly_qualifiers): Likewise.
1023         (aarch64_types_unsigned_signed_qualifiers): Likewise.
1024         (aarch64_types_poly_signed_qualifiers): Likewise.
1025         (TYPES_REINTERP_SS): Type macro added.
1026         (TYPES_REINTERP_SU): Likewise.
1027         (TYPES_REINTERP_SP): Likewise.
1028         (TYPES_REINTERP_US): Likewise.
1029         (TYPES_REINTERP_PS): Likewise.
1030         (aarch64_fold_builtin): New expression folding added.
1031         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
1032         Declarations removed.
1033         (REINTERP_SS): Declarations added.
1034         (REINTERP_US): Likewise.
1035         (REINTERP_PS): Likewise.
1036         (REINTERP_SU): Likewise.
1037         (REINTERP_SP): Likewise.
1038         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
1039         (vreinterpretq_p8_f64): Likewise.
1040         (vreinterpret_p16_f64): Likewise.
1041         (vreinterpretq_p16_f64): Likewise.
1042         (vreinterpret_f32_f64): Likewise.
1043         (vreinterpretq_f32_f64): Likewise.
1044         (vreinterpret_f64_f32): Likewise.
1045         (vreinterpret_f64_p8): Likewise.
1046         (vreinterpret_f64_p16): Likewise.
1047         (vreinterpret_f64_s8): Likewise.
1048         (vreinterpret_f64_s16): Likewise.
1049         (vreinterpret_f64_s32): Likewise.
1050         (vreinterpret_f64_s64): Likewise.
1051         (vreinterpret_f64_u8): Likewise.
1052         (vreinterpret_f64_u16): Likewise.
1053         (vreinterpret_f64_u32): Likewise.
1054         (vreinterpret_f64_u64): Likewise.
1055         (vreinterpretq_f64_f32): Likewise.
1056         (vreinterpretq_f64_p8): Likewise.
1057         (vreinterpretq_f64_p16): Likewise.
1058         (vreinterpretq_f64_s8): Likewise.
1059         (vreinterpretq_f64_s16): Likewise.
1060         (vreinterpretq_f64_s32): Likewise.
1061         (vreinterpretq_f64_s64): Likewise.
1062         (vreinterpretq_f64_u8): Likewise.
1063         (vreinterpretq_f64_u16): Likewise.
1064         (vreinterpretq_f64_u32): Likewise.
1065         (vreinterpretq_f64_u64): Likewise.
1066         (vreinterpret_s64_f64): Likewise.
1067         (vreinterpretq_s64_f64): Likewise.
1068         (vreinterpret_u64_f64): Likewise.
1069         (vreinterpretq_u64_f64): Likewise.
1070         (vreinterpret_s8_f64): Likewise.
1071         (vreinterpretq_s8_f64): Likewise.
1072         (vreinterpret_s16_f64): Likewise.
1073         (vreinterpretq_s16_f64): Likewise.
1074         (vreinterpret_s32_f64): Likewise.
1075         (vreinterpretq_s32_f64): Likewise.
1076         (vreinterpret_u8_f64): Likewise.
1077         (vreinterpretq_u8_f64): Likewise.
1078         (vreinterpret_u16_f64): Likewise.
1079         (vreinterpretq_u16_f64): Likewise.
1080         (vreinterpret_u32_f64): Likewise.
1081         (vreinterpretq_u32_f64): Likewise.
1082
1083 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1084
1085         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1086         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
1087         (vreinterpret_p8_s8): Likewise.
1088         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
1089         (vreinterpret_p8_s16): Likewise.
1090         (vreinterpret_p8_s32): Likewise.
1091         (vreinterpret_p8_s64): Likewise.
1092         (vreinterpret_p8_f32): Likewise.
1093         (vreinterpret_p8_u8): Likewise.
1094         (vreinterpret_p8_u16): Likewise.
1095         (vreinterpret_p8_u32): Likewise.
1096         (vreinterpret_p8_u64): Likewise.
1097         (vreinterpret_p8_p16): Likewise.
1098         (vreinterpretq_p8_s8): Likewise.
1099         (vreinterpretq_p8_s16): Likewise.
1100         (vreinterpretq_p8_s32): Likewise.
1101         (vreinterpretq_p8_s64): Likewise.
1102         (vreinterpretq_p8_f32): Likewise.
1103         (vreinterpretq_p8_u8): Likewise.
1104         (vreinterpretq_p8_u16): Likewise.
1105         (vreinterpretq_p8_u32): Likewise.
1106         (vreinterpretq_p8_u64): Likewise.
1107         (vreinterpretq_p8_p16): Likewise.
1108         (vreinterpret_p16_s8): Likewise.
1109         (vreinterpret_p16_s16): Likewise.
1110         (vreinterpret_p16_s32): Likewise.
1111         (vreinterpret_p16_s64): Likewise.
1112         (vreinterpret_p16_f32): Likewise.
1113         (vreinterpret_p16_u8): Likewise.
1114         (vreinterpret_p16_u16): Likewise.
1115         (vreinterpret_p16_u32): Likewise.
1116         (vreinterpret_p16_u64): Likewise.
1117         (vreinterpret_p16_p8): Likewise.
1118         (vreinterpretq_p16_s8): Likewise.
1119         (vreinterpretq_p16_s16): Likewise.
1120         (vreinterpretq_p16_s32): Likewise.
1121         (vreinterpretq_p16_s64): Likewise.
1122         (vreinterpretq_p16_f32): Likewise.
1123         (vreinterpretq_p16_u8): Likewise.
1124         (vreinterpretq_p16_u16): Likewise.
1125         (vreinterpretq_p16_u32): Likewise.
1126         (vreinterpretq_p16_u64): Likewise.
1127         (vreinterpretq_p16_p8): Likewise.
1128         (vreinterpret_f32_s8): Likewise.
1129         (vreinterpret_f32_s16): Likewise.
1130         (vreinterpret_f32_s32): Likewise.
1131         (vreinterpret_f32_s64): Likewise.
1132         (vreinterpret_f32_u8): Likewise.
1133         (vreinterpret_f32_u16): Likewise.
1134         (vreinterpret_f32_u32): Likewise.
1135         (vreinterpret_f32_u64): Likewise.
1136         (vreinterpret_f32_p8): Likewise.
1137         (vreinterpret_f32_p16): Likewise.
1138         (vreinterpretq_f32_s8): Likewise.
1139         (vreinterpretq_f32_s16): Likewise.
1140         (vreinterpretq_f32_s32): Likewise.
1141         (vreinterpretq_f32_s64): Likewise.
1142         (vreinterpretq_f32_u8): Likewise.
1143         (vreinterpretq_f32_u16): Likewise.
1144         (vreinterpretq_f32_u32): Likewise.
1145         (vreinterpretq_f32_u64): Likewise.
1146         (vreinterpretq_f32_p8): Likewise.
1147         (vreinterpretq_f32_p16): Likewise.
1148         (vreinterpret_s64_s8): Likewise.
1149         (vreinterpret_s64_s16): Likewise.
1150         (vreinterpret_s64_s32): Likewise.
1151         (vreinterpret_s64_f32): Likewise.
1152         (vreinterpret_s64_u8): Likewise.
1153         (vreinterpret_s64_u16): Likewise.
1154         (vreinterpret_s64_u32): Likewise.
1155         (vreinterpret_s64_u64): Likewise.
1156         (vreinterpret_s64_p8): Likewise.
1157         (vreinterpret_s64_p16): Likewise.
1158         (vreinterpretq_s64_s8): Likewise.
1159         (vreinterpretq_s64_s16): Likewise.
1160         (vreinterpretq_s64_s32): Likewise.
1161         (vreinterpretq_s64_f32): Likewise.
1162         (vreinterpretq_s64_u8): Likewise.
1163         (vreinterpretq_s64_u16): Likewise.
1164         (vreinterpretq_s64_u32): Likewise.
1165         (vreinterpretq_s64_u64): Likewise.
1166         (vreinterpretq_s64_p8): Likewise.
1167         (vreinterpretq_s64_p16): Likewise.
1168         (vreinterpret_u64_s8): Likewise.
1169         (vreinterpret_u64_s16): Likewise.
1170         (vreinterpret_u64_s32): Likewise.
1171         (vreinterpret_u64_s64): Likewise.
1172         (vreinterpret_u64_f32): Likewise.
1173         (vreinterpret_u64_u8): Likewise.
1174         (vreinterpret_u64_u16): Likewise.
1175         (vreinterpret_u64_u32): Likewise.
1176         (vreinterpret_u64_p8): Likewise.
1177         (vreinterpret_u64_p16): Likewise.
1178         (vreinterpretq_u64_s8): Likewise.
1179         (vreinterpretq_u64_s16): Likewise.
1180         (vreinterpretq_u64_s32): Likewise.
1181         (vreinterpretq_u64_s64): Likewise.
1182         (vreinterpretq_u64_f32): Likewise.
1183         (vreinterpretq_u64_u8): Likewise.
1184         (vreinterpretq_u64_u16): Likewise.
1185         (vreinterpretq_u64_u32): Likewise.
1186         (vreinterpretq_u64_p8): Likewise.
1187         (vreinterpretq_u64_p16): Likewise.
1188         (vreinterpret_s8_s16): Likewise.
1189         (vreinterpret_s8_s32): Likewise.
1190         (vreinterpret_s8_s64): Likewise.
1191         (vreinterpret_s8_f32): Likewise.
1192         (vreinterpret_s8_u8): Likewise.
1193         (vreinterpret_s8_u16): Likewise.
1194         (vreinterpret_s8_u32): Likewise.
1195         (vreinterpret_s8_u64): Likewise.
1196         (vreinterpret_s8_p8): Likewise.
1197         (vreinterpret_s8_p16): Likewise.
1198         (vreinterpretq_s8_s16): Likewise.
1199         (vreinterpretq_s8_s32): Likewise.
1200         (vreinterpretq_s8_s64): Likewise.
1201         (vreinterpretq_s8_f32): Likewise.
1202         (vreinterpretq_s8_u8): Likewise.
1203         (vreinterpretq_s8_u16): Likewise.
1204         (vreinterpretq_s8_u32): Likewise.
1205         (vreinterpretq_s8_u64): Likewise.
1206         (vreinterpretq_s8_p8): Likewise.
1207         (vreinterpretq_s8_p16): Likewise.
1208         (vreinterpret_s16_s8): Likewise.
1209         (vreinterpret_s16_s32): Likewise.
1210         (vreinterpret_s16_s64): Likewise.
1211         (vreinterpret_s16_f32): Likewise.
1212         (vreinterpret_s16_u8): Likewise.
1213         (vreinterpret_s16_u16): Likewise.
1214         (vreinterpret_s16_u32): Likewise.
1215         (vreinterpret_s16_u64): Likewise.
1216         (vreinterpret_s16_p8): Likewise.
1217         (vreinterpret_s16_p16): Likewise.
1218         (vreinterpretq_s16_s8): Likewise.
1219         (vreinterpretq_s16_s32): Likewise.
1220         (vreinterpretq_s16_s64): Likewise.
1221         (vreinterpretq_s16_f32): Likewise.
1222         (vreinterpretq_s16_u8): Likewise.
1223         (vreinterpretq_s16_u16): Likewise.
1224         (vreinterpretq_s16_u32): Likewise.
1225         (vreinterpretq_s16_u64): Likewise.
1226         (vreinterpretq_s16_p8): Likewise.
1227         (vreinterpretq_s16_p16): Likewise.
1228         (vreinterpret_s32_s8): Likewise.
1229         (vreinterpret_s32_s16): Likewise.
1230         (vreinterpret_s32_s64): Likewise.
1231         (vreinterpret_s32_f32): Likewise.
1232         (vreinterpret_s32_u8): Likewise.
1233         (vreinterpret_s32_u16): Likewise.
1234         (vreinterpret_s32_u32): Likewise.
1235         (vreinterpret_s32_u64): Likewise.
1236         (vreinterpret_s32_p8): Likewise.
1237         (vreinterpret_s32_p16): Likewise.
1238         (vreinterpretq_s32_s8): Likewise.
1239         (vreinterpretq_s32_s16): Likewise.
1240         (vreinterpretq_s32_s64): Likewise.
1241         (vreinterpretq_s32_f32): Likewise.
1242         (vreinterpretq_s32_u8): Likewise.
1243         (vreinterpretq_s32_u16): Likewise.
1244         (vreinterpretq_s32_u32): Likewise.
1245         (vreinterpretq_s32_u64): Likewise.
1246         (vreinterpretq_s32_p8): Likewise.
1247         (vreinterpretq_s32_p16): Likewise.
1248         (vreinterpret_u8_s8): Likewise.
1249         (vreinterpret_u8_s16): Likewise.
1250         (vreinterpret_u8_s32): Likewise.
1251         (vreinterpret_u8_s64): Likewise.
1252         (vreinterpret_u8_f32): Likewise.
1253         (vreinterpret_u8_u16): Likewise.
1254         (vreinterpret_u8_u32): Likewise.
1255         (vreinterpret_u8_u64): Likewise.
1256         (vreinterpret_u8_p8): Likewise.
1257         (vreinterpret_u8_p16): Likewise.
1258         (vreinterpretq_u8_s8): Likewise.
1259         (vreinterpretq_u8_s16): Likewise.
1260         (vreinterpretq_u8_s32): Likewise.
1261         (vreinterpretq_u8_s64): Likewise.
1262         (vreinterpretq_u8_f32): Likewise.
1263         (vreinterpretq_u8_u16): Likewise.
1264         (vreinterpretq_u8_u32): Likewise.
1265         (vreinterpretq_u8_u64): Likewise.
1266         (vreinterpretq_u8_p8): Likewise.
1267         (vreinterpretq_u8_p16): Likewise.
1268         (vreinterpret_u16_s8): Likewise.
1269         (vreinterpret_u16_s16): Likewise.
1270         (vreinterpret_u16_s32): Likewise.
1271         (vreinterpret_u16_s64): Likewise.
1272         (vreinterpret_u16_f32): Likewise.
1273         (vreinterpret_u16_u8): Likewise.
1274         (vreinterpret_u16_u32): Likewise.
1275         (vreinterpret_u16_u64): Likewise.
1276         (vreinterpret_u16_p8): Likewise.
1277         (vreinterpret_u16_p16): Likewise.
1278         (vreinterpretq_u16_s8): Likewise.
1279         (vreinterpretq_u16_s16): Likewise.
1280         (vreinterpretq_u16_s32): Likewise.
1281         (vreinterpretq_u16_s64): Likewise.
1282         (vreinterpretq_u16_f32): Likewise.
1283         (vreinterpretq_u16_u8): Likewise.
1284         (vreinterpretq_u16_u32): Likewise.
1285         (vreinterpretq_u16_u64): Likewise.
1286         (vreinterpretq_u16_p8): Likewise.
1287         (vreinterpretq_u16_p16): Likewise.
1288         (vreinterpret_u32_s8): Likewise.
1289         (vreinterpret_u32_s16): Likewise.
1290         (vreinterpret_u32_s32): Likewise.
1291         (vreinterpret_u32_s64): Likewise.
1292         (vreinterpret_u32_f32): Likewise.
1293         (vreinterpret_u32_u8): Likewise.
1294         (vreinterpret_u32_u16): Likewise.
1295         (vreinterpret_u32_u64): Likewise.
1296         (vreinterpret_u32_p8): Likewise.
1297         (vreinterpret_u32_p16): Likewise.
1298         (vreinterpretq_u32_s8): Likewise.
1299         (vreinterpretq_u32_s16): Likewise.
1300         (vreinterpretq_u32_s32): Likewise.
1301         (vreinterpretq_u32_s64): Likewise.
1302         (vreinterpretq_u32_f32): Likewise.
1303         (vreinterpretq_u32_u8): Likewise.
1304         (vreinterpretq_u32_u16): Likewise.
1305         (vreinterpretq_u32_u64): Likewise.
1306         (vreinterpretq_u32_p8): Likewise.
1307         (vreinterpretq_u32_p16): Likewise.
1308
1309 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1310
1311         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
1312         Pattern extended.
1313         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
1314         extended.
1315         (sqabs): Likewise.
1316         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
1317         (vqnegd_s64): Likewise.
1318         (vqabs_s64): Likewise.
1319         (vqabsd_s64): Likewise.
1320
1321 2014-04-22  Richard Henderson  <rth@redhat.com>
1322
1323         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
1324         computation to the top of the loop.
1325
1326 2014-04-22  Renlin  <renlin.li@arm.com>
1327             Jiong Wang  <jiong.wang@arm.com>
1328
1329         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
1330         * config/aarch64/aarch64.c (aarch64_layout_frame)
1331         (aarch64_initial_elimination_offset): Likewise.
1332
1333 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1334
1335         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
1336         Fix indentation.
1337
1338 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
1339
1340         * machmode.h (bitwise_mode_for_mode): Declare.
1341         * stor-layout.h (bitwise_type_for_mode): Likewise.
1342         * stor-layout.c (bitwise_mode_for_mode): New function.
1343         (bitwise_type_for_mode): Likewise.
1344         * builtins.c (fold_builtin_memory_op): Use it instead of
1345         int_mode_for_mode and build_nonstandard_integer_type.
1346
1347 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1348
1349         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
1350         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
1351         (*-*-solaris2*): Simplify.
1352         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
1353         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
1354         *-*-solaris2.9* handling.
1355
1356         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
1357         as bug.
1358         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
1359         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
1360         handling, simplify.
1361         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
1362         * configure: Regenerate.
1363
1364         * config/i386/sol2-9.h: Remove.
1365
1366         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
1367         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
1368         Remove Solaris 9 references.
1369
1370 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
1371
1372         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
1373         (floatuns<GPI:mode><GPF:mode>2): Remove.
1374         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
1375         and floatuns conversions.
1376         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
1377         and floatuns conversions.
1378         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
1379         (w1,w2): New mode attributes for inequal width conversions.
1380
1381 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
1382
1383         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
1384         the output asm format.
1385
1386 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
1387
1388         * config/aarch64/aarch64-simd.md
1389         (aarch64_cm<optab>di): Always split.
1390         (*aarch64_cm<optab>di): New.
1391         (aarch64_cmtstdi): Always split.
1392         (*aarch64_cmtstdi): New.
1393
1394 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1395
1396         PR tree-optimization/60823
1397         * omp-low.c (ipa_simd_modify_function_body): Go through
1398         all SSA_NAMEs and for those refering to vector arguments
1399         which are going to be replaced adjust SSA_NAME_VAR and,
1400         if it is a default definition, change it into a non-default
1401         definition assigned at the beginning of function from new_decl.
1402         (ipa_simd_modify_stmt_ops): Rewritten.
1403         * tree-dfa.c (set_ssa_default_def): When removing default def,
1404         check for NULL loc instead of NULL *loc.
1405
1406 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1407
1408         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
1409         restrictions on core registers for DImode values in Thumb2.
1410
1411 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1412
1413         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
1414         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
1415
1416 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1417
1418         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
1419         (*iordi_notzesidi_di): Likewise.
1420         (*iordi_notsesidi_di): Likewise.
1421
1422 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1423
1424         * config/arm/arm-protos.h (tune_params): New struct members.
1425         * config/arm/arm.c: Initialise tune_params per processor.
1426         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
1427         for speed, based on new tune_params.
1428
1429 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1430
1431         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
1432         added.
1433         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
1434         macro.
1435         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
1436         corrected.
1437         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
1438         * config/aarch64/arm_neon.h (vrnd_f64): Added.
1439         (vrnda_f64): Likewise.
1440         (vrndi_f64): Likewise.
1441         (vrndm_f64): Likewise.
1442         (vrndn_f64): Likewise.
1443         (vrndp_f64): Likewise.
1444         (vrndx_f64): Likewise.
1445
1446 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1447
1448         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
1449         GET_MODE_SIZE argument is enum machine_mode.
1450
1451 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1452
1453         PR target/60910
1454         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
1455         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
1456
1457 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
1458
1459         PR middle-end/60281
1460         * asan.c (asan_emit_stack_protection): Force the base to align to
1461         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
1462         appropriate bits if STRICT_ALIGNMENT.
1463         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
1464         when asan is on.
1465         (expand_used_vars): Leave a space in the stack frame for alignment
1466         if STRICT_ALIGNMENT.
1467
1468 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
1469
1470         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
1471         than a gimple.
1472         (gimple_store_p): Likewise.
1473         (gimple_assign_load_p): Likewise.
1474         (gimple_assign_cast_p): Likewise.
1475         (gimple_clobber_p): Likewise.
1476
1477         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
1478         rather than a gimple.
1479         (gimple_assign_cast_p): Likewise.
1480
1481 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
1482
1483         PR target/60735
1484         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
1485         If mode is DDmode and TARGET_E500_DOUBLE allow move.
1486
1487         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
1488         more debug information for E500 if -mdebug=reg.
1489
1490 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
1491
1492         PR target/60909
1493         * config/i386/i386.c (ix86_expand_builtin)
1494         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
1495         register for target RTX.
1496         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
1497
1498 2014-04-18  Cong Hou  <congh@google.com>
1499
1500         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
1501         the widen-mult pattern by handling two operands with different sizes,
1502         and operands whose size is smaller than half of the result type.
1503
1504 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1505
1506         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
1507         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
1508         (do_estimate_edge_time): Compute it.
1509         * ipa-inline.c (want_inline_small_function_p): Bypass
1510         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
1511
1512 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1513
1514         * ipa-inline.c (spec_rem): New static variable.
1515         (dump_overall_stats): New function.
1516         (dump_inline_stats): New function.
1517
1518 2014-04-18  Richard Henderson  <rth@redhat.com>
1519
1520         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
1521         to GET_MODE_SIZE, not a reg_class_t.
1522
1523 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1524
1525         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
1526         (vsx_xxmrglw_<mode>): Likewise.
1527
1528 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
1529
1530         PR target/60876
1531         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
1532         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
1533         (rs6000_init_hard_regno_mode_ok): Likewise.
1534
1535 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
1536
1537         * ipa-inline.c (inline_small_functions): Account only non-cold
1538         functions.
1539         * doc/invoke.texi (inline-unit-growth): Update documentation.
1540
1541 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
1542
1543         * config/rs6000/rs6000.md (addti3, subti3): New.
1544
1545 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
1546
1547         PR target/60863
1548         * config/i386/i386.c (ix86_expand_clear): Remove outdated
1549         comment.  Check optimize_insn_for_size_p instead of
1550         optimize_insn_for_speed_p.
1551
1552 2014-04-17  Martin Jambor  <mjambor@suse.cz>
1553
1554         * gimple-iterator.c (gsi_start_edge): New function.
1555         * gimple-iterator.h (gsi_start_edge): Declare.
1556         * tree-sra.c (single_non_eh_succ): New function.
1557         (disqualify_ops_if_throwing_stmt): Renamed to
1558         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
1559         having one non-EH successor BB.
1560         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
1561         generate loads into replacements.
1562         (sra_modify_assign): Likewise and and also use the simple path for
1563         such statements.
1564         (sra_modify_function_body): Commit statements on edges.
1565
1566 2014-04-17  Richard Biener  <rguenther@suse.de>
1567
1568         PR middle-end/60849
1569         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
1570         comparison results and add clarifying comment.
1571
1572 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1573
1574         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
1575         (blank_mode): Initialize it.
1576         (emit_mode_size_inline, emit_mode_nunits_inline,
1577         emit_mode_inner_inline): New functions.
1578         (emit_insn_modes_h): Call them and surround their output with
1579         #if GCC_VERSION >= 4001 ... #endif.
1580         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
1581         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
1582         mode_* arrays if the argument is __builtin_constant_p.
1583         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
1584         is enum machine_mode.
1585
1586 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1587
1588         * passes.c (opt_pass::execute): Adjust.
1589         (pass_manager::execute_pass_mode_switching): Likewise.
1590         (early_local_passes::execute): Likewise.
1591         (execute_one_pass): Pass cfun to the pass's execute method.
1592         * tree-pass.h (opt_pass::execute): Add function * argument.
1593         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1594         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1595         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1596         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1597         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1598         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
1599         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
1600         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1601         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1602         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
1603         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
1604         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
1605         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
1606         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
1607         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1608         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1609         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1610         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
1611         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
1612         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1613         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1614         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1615         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1616         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1617         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1618         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1619         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1620         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1621         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1622         Adjust.
1623
1624 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1625
1626         * passes.c (opt_pass::gate): Take function * argument.
1627         (gate_all_early_local_passes): Merge into
1628         (early_local_passes::gate): this.
1629         (gate_all_early_optimizations): Merge into
1630         (all_early_optimizations::gate): this.
1631         (gate_all_optimizations): Mege into
1632         (all_optimizations::gate): this.
1633         (gate_all_optimizations_g): Merge into
1634         (all_optimizations_g::gate): this.
1635         (gate_rest_of_compilation): Mege into
1636         (rest_of_compilation::gate): this.
1637         (gate_postreload): Merge into
1638         (postreload::gate): this.
1639         (dump_one_pass): Pass cfun to the pass's gate method.
1640         (execute_ipa_summary_passes): Likewise.
1641         (execute_one_pass): Likewise.
1642         (ipa_write_summaries_2): Likewise.
1643         (ipa_write_optimization_summaries_1): Likewise.
1644         (ipa_read_summaries_1): Likewise.
1645         (ipa_read_optimization_summaries_1): Likewise.
1646         (execute_ipa_stmt_fixups): Likewise.
1647         * tree-pass.h (opt_pass::gate): Add function * argument.
1648         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
1649         combine-stack-adj.c, combine.c, compare-elim.c,
1650         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1651         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
1652         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
1653         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
1654         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1655         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1656         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1657         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
1658         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
1659         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
1660         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1661         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1662         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
1663         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1664         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1665         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1666         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1667         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1668         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1669         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1670         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1671         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1672         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
1673         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
1674         var-tracking.c, vtable-verify.c, web.c: Adjust.
1675
1676 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1677
1678         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
1679         * configure: Regenerate.
1680
1681 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1682
1683         * passes.c (dump_one_pass): don't check pass->has_gate.
1684         (execute_ipa_summary_passes): Likewise.
1685         (execute_one_pass): Likewise.
1686         (ipa_write_summaries_2): Likewise.
1687         (ipa_write_optimization_summaries_1): Likewise.
1688         (ipa_read_optimization_summaries_1): Likewise.
1689         (execute_ipa_stmt_fixups): Likewise.
1690         * tree-pass.h (pass_data::has_gate): Remove.
1691         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1692         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1693         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1694         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1695         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1696         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
1697         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
1698         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
1699         gimple-low.c, gimple-ssa-isolate-paths.c,
1700         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
1701         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
1702         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1703         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
1704         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
1705         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
1706         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
1707         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1708         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
1709         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1710         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1711         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1712         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1713         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1714         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1715         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1716         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1717         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1718         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1719         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1720         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1721         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1722         Adjust.
1723
1724 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1725
1726         * pass_manager.h (pass_manager::register_dump_files_1): Remove
1727         declaration.
1728         * passes.c (pass_manager::register_dump_files_1): Merge into
1729         (pass_manager::register_dump_files): this, and remove its handling of
1730         properties since the pass always has the properties anyway.
1731         (pass_manager::pass_manager): Adjust.
1732
1733 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1734
1735         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
1736         * passes.c (pass_manager::register_dump_files_1): Remove dead code
1737         dealing with properties.
1738         (pass_manager::register_dump_files): Adjust.
1739
1740 2014-03-20  Mark Wielaard  <mjw@redhat.com>
1741
1742         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
1743         then represent the bound as normal constant value.
1744
1745 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1746
1747         PR target/60847
1748         Forward port from 4.8 branch
1749         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
1750
1751         * config/i386/bmiintrin.h (_blsi_u32): New.
1752         (_blsi_u64): Ditto.
1753         (_blsr_u32): Ditto.
1754         (_blsr_u64): Ditto.
1755         (_blsmsk_u32): Ditto.
1756         (_blsmsk_u64): Ditto.
1757         (_tzcnt_u32): Ditto.
1758         (_tzcnt_u64): Ditto.
1759
1760 2014-04-17  Kito Cheng  <kito@0xlab.org>
1761
1762         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
1763
1764 2014-04-17  Richard Biener  <rguenther@suse.de>
1765
1766         PR middle-end/60849
1767         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
1768         boolean results for comparisons.
1769
1770 2014-04-17  Richard Biener  <rguenther@suse.de>
1771
1772         PR tree-optimization/60836
1773         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
1774         initial PHI args to be gimple values.
1775
1776 2014-04-17   Richard Biener  <rguenther@suse.de>
1777
1778         PR tree-optimization/60841
1779         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
1780         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
1781         of stmts to SLP build.
1782         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
1783         (vect_analyze_slp): Likewise.
1784         (vect_analyze_slp_instance): Likewise.
1785         (vect_build_slp_tree): Limit overall SLP tree growth.
1786         * tree-vectorizer.h (vect_analyze_data_refs,
1787         vect_analyze_slp): Adjust prototypes.
1788
1789 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1790
1791         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
1792         Silvermont.
1793
1794 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1795
1796         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
1797         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
1798         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
1799         for TARGET_SLOW_PSHUFB
1800
1801 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1802
1803         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
1804         * config/i386/i386.c (intel_cost): Ditto.
1805
1806 2014-04-17  Joey Ye  <joey.ye@arm.com>
1807
1808         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
1809
1810 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1811
1812         * opts.c (common_handle_option): Disable -fipa-reference coorectly
1813         with -fuse-profile.
1814
1815 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1816
1817         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
1818         (type_all_derivations_known_p): New predicate.
1819         (type_all_ctors_visible_p): New predicate.
1820         (type_possibly_instantiated_p): New predicate.
1821         (get_odr_type): Compute all_derivations_known.
1822         (dump_odr_type): Dump the flag.
1823         (maybe_record_type): Cleanup.
1824         (record_target_from_binfo): Add bases_to_consider array;
1825         record bases for types w/o instances and skip CXX destructor.
1826         (possible_polymorphic_call_targets_1): Add bases_to_consider
1827         and consider_construction parameters; check if type may have instance.
1828         (get_polymorphic_call_info): Set maybe_in_construction to true
1829         when we know nothing.
1830         (record_targets_from_bases): Skip CXX destructors; they are
1831         never called for types in construction.
1832         (possible_polymorphic_call_targets): Do not record target when
1833         type may not have instance.
1834
1835 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1836
1837         PR ipa/60854
1838         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
1839         external aliases alive, too.
1840
1841 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
1842
1843         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
1844         definition.
1845
1846 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1847
1848         * final.c (compute_alignments): Do not apply loop alignment to a block
1849         falling through to the exit.
1850
1851 2014-04-16  Catherine Moore  <clm@codesourcery.com>
1852
1853         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
1854         Adjust constraints for microMIPS store patterns.
1855
1856 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
1857
1858         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
1859
1860 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1861
1862         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
1863         (append_use): Run at -O0.
1864         (append_vdef): Likewise.
1865         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
1866         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
1867
1868 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
1869
1870         PR tree-optimization/60844
1871         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
1872         (propagate_op_to_single_use, remove_visited_stmt_chain,
1873         linearize_expr, repropagate_negates, reassociate_bb): Use it
1874         instead of gsi_remove.
1875
1876 2014-04-16  Martin Jambor  <mjambor@suse.cz>
1877
1878         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
1879         ipa_transforms_to_apply.
1880         (cgraph_function_versioning): Assert that old_node has empty
1881         ipa_transforms_to_apply.
1882         * trans-mem.c (ipa_tm_create_version): Likewise.
1883         * tree-inline.c (tree_function_versioning): Do not duplicate
1884         ipa_transforms_to_apply.
1885
1886 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1887
1888         PR target/60817
1889         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
1890         x86_64-*-* cases.
1891         Pass necessary as flags on 64-bit Solaris/x86.
1892         Use lowercase relocs for x86_64-*-*.
1893         * configure: Regenerate.
1894
1895 2014-04-15  Jan Hubicka  <jh@suse.cz>
1896
1897         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
1898         (maybe_record_node, likely_target_p): Use it.
1899
1900 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1901
1902         PR target/60839
1903         Revert following patch
1904
1905         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
1906
1907         PR target/60735
1908         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1909         software floating point or no floating point registers, do not
1910         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
1911         in GPRs that occurs after we tested for GPRs that would never be
1912         true.
1913
1914         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1915         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1916         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
1917         specifically allow DDmode, since that does not use the SPE SIMD
1918         instructions.
1919
1920 2014-03-21  Mark Wielaard  <mjw@redhat.com>
1921
1922         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
1923         as unsigned or int depending on type and value used.
1924
1925 2014-04-15  Richard Biener  <rguenther@suse.de>
1926
1927         PR rtl-optimization/56965
1928         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
1929         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
1930         ... here.
1931         * alias.c (true_dependence_1): Do not call
1932         nonoverlapping_component_refs_p.
1933         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
1934         nonoverlapping_component_refs_p.
1935         (indirect_refs_may_alias_p): Likewise.
1936
1937 2014-04-15  Teresa Johnson  <tejohnson@google.com>
1938
1939         * cfg.c (dump_bb_info): Fix flags check.
1940         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
1941
1942 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1943
1944         PR rtl-optimization/60663
1945         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
1946         avoid 0 cost.
1947
1948 2014-04-15  Richard Biener  <rguenther@suse.de>
1949
1950         * lto-streamer.h (LTO_major_version): Bump to 4.
1951
1952 2014-04-15  Richard Biener  <rguenther@suse.de>
1953
1954         * common.opt (lto_partition_model): New enum.
1955         (flto-partition=): Merge separate options with a single with argument,
1956         add -flto-partition=one support.
1957         * flag-types.h (enum lto_partition_model): Declare.
1958         * opts.c (finish_options): Remove duplicate -flto-partition=
1959         option check.
1960         * lto-wrapper.c (run_gcc): Adjust.
1961
1962 2014-04-15  Richard Biener  <rguenther@suse.de>
1963
1964         * alias.c (ncr_compar): New function.
1965         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1966
1967 2014-04-15  Richard Biener  <rguenther@suse.de>
1968
1969         * alias.c (record_component_aliases): Do not walk BINFOs.
1970
1971 2014-04-15  Richard Biener  <rguenther@suse.de>
1972
1973         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1974         Add struct function argument and adjust.
1975         (find_func_aliases_for_call): Likewise.
1976         (find_func_aliases): Likewise.
1977         (find_func_clobbers): Likewise.
1978         (intra_create_variable_infos): Likewise.
1979         (compute_points_to_sets): Likewise.
1980         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
1981
1982 2014-04-15  Richard Biener  <rguenther@suse.de>
1983
1984         * tree.c (iterative_hash_expr): Use enum tree_code_class
1985         to store TREE_CODE_CLASS.
1986         (tree_block): Likewise.
1987         (tree_set_block): Likewise.
1988         * tree.h (fold_build_pointer_plus_loc): Use
1989         convert_to_ptrofftype_loc.
1990
1991 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
1992
1993         PR plugins/59335
1994         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1995         added in 4.9.
1996
1997 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
1998
1999         * cfgloop.h (struct loop): Move force_vectorize down.
2000         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
2001         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
2002         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
2003         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
2004         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
2005         * tree-core.h (enum annot_expr_kind): Add new kind values.
2006         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
2007         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
2008         kinds.
2009         * tree.def (ANNOTATE_EXPR): Tweak comment.
2010
2011 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2012
2013         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
2014         cxa_pure_virtual).
2015
2016 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
2017
2018         * tree.h (TYPE_IDENTIFIER): Declare.
2019         * tree.c (subrange_type_for_debug_p): Use it.
2020         * godump.c (go_format_type): Likewise.
2021         * dwarf2out.c (is_cxx_auto, modified_type_die,
2022         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
2023         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
2024
2025 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
2026
2027         PR lto/60820
2028         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
2029
2030 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
2031
2032         * config/i386/i386.c (examine_argument): Return bool.  Return true if
2033         parameter should be passed in memory.
2034         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
2035         (construct_container): Update calls to examine_argument.
2036         (function_arg_advance_64): Ditto.
2037         (return_in_memory_32): Merge with ix86_return_in_memory.
2038         (return_in_memory_64): Ditto.
2039         (return_in_memory_ms_64): Ditto.
2040
2041 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2042
2043         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
2044         * coverage.c (coverage_compute_profile_id): Handle externally visible
2045         symbols.
2046
2047 2014-04-14  Martin Jambor  <mjambor@suse.cz>
2048
2049         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
2050         DECL_DISREGARD_INLINE_LIMITS functions.
2051
2052 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2053
2054         PR target/60827
2055         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
2056
2057 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2058
2059         PR target/60827
2060         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
2061         optimize_insn_for_speed_p instead of
2062         optimize_function_for_speed_p.
2063
2064 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
2065
2066         * doc/invoke.texi (free): Document AArch64.
2067
2068 2014-04-14  Richard Biener  <rguenther@suse.de>
2069
2070         PR tree-optimization/60042
2071         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
2072         (insert_into_preds_of_block): Do not prevent PHI insertion
2073         for REFERENCE exprs here ...
2074         (eliminate_dom_walker::before_dom_children): ... but prevent
2075         their use here under similar conditions when applied to the
2076         IL after PRE optimizations.
2077
2078 2014-04-14  Richard Biener  <rguenther@suse.de>
2079
2080         * passes.def: Move early points-to after early SRA.
2081
2082 2014-04-14  Richard Biener  <rguenther@suse.de>
2083
2084         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
2085         check for which sign-changes we allow when forwarding
2086         a converted value into a switch.
2087
2088 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2089
2090         * stor-layout.c (place_field): Finalize non-constant offset for the
2091         field, if any.
2092
2093 2014-04-14  Richard Biener  <rguenther@suse.de>
2094
2095         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
2096         as argument.
2097         (expand_switch_using_bit_tests_p): Likewise.
2098         (process_switch): Compute and pass on speed_p based on the
2099         switch stmt.
2100         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
2101         optimize_bb_for_speed_p.
2102
2103 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2104
2105         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
2106         * function.h (struct function): Rename has_force_vect_loops into
2107         has_force_vectorize_loops.
2108         * lto-streamer-in.c (input_cfg): Adjust for renaming.
2109         (input_struct_function_base): Likewise.
2110         * lto-streamer-out.c (output_cfg): Likewise.
2111         (output_struct_function_base): Likewise.
2112         * omp-low.c (expand_omp_simd): Likewise.
2113         * tree-cfg.c (move_sese_region_to_fn): Likewise.
2114         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2115         (version_loop_for_if_conversion): Likewise.
2116         (tree_if_conversion): Likewise.
2117         (main_tree_if_conversion): Likewise.
2118         (gate_tree_if_conversion): Likewise.
2119         * tree-inline.c (copy_loops): Likewise.
2120         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
2121         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
2122         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
2123         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
2124         * tree-vectorizer.c (vectorize_loops): Likewise.
2125         * tree-vectorizer.h (unlimited_cost_model): Likewise.
2126
2127 2014-04-14  Richard Biener  <rguenther@suse.de>
2128
2129         PR lto/60720
2130         * lto-streamer-out.c (wrap_refs): New function.
2131         (lto_output): Wrap symbol references in global initializes in
2132         type-preserving MEM_REFs.
2133
2134 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2135
2136         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
2137
2138 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2139
2140         * config/sh/sh.md (setmemqi): New expand pattern.
2141         * config/sh/sh.h (CLEAR_RATIO): Define.
2142         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
2143         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
2144
2145 2014-04-14  Richard Biener  <rguenther@suse.de>
2146
2147         PR middle-end/55022
2148         * fold-const.c (negate_expr_p): Don't negate directional rounding
2149         division.
2150         (fold_negate_expr): Likewise.
2151
2152 2014-04-14  Richard Biener  <rguenther@suse.de>
2153
2154         PR tree-optimization/59817
2155         PR tree-optimization/60453
2156         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
2157         recursion to catch all CHRECs in the scalar evolution and restrict
2158         the predicate for the remains appropriately.
2159
2160 2014-04-12  Catherine Moore  <clm@codesourcery.com>
2161
2162         * config/mips/constraints.md: Add new register constraint "kb".
2163         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
2164         (*movhi_internal): Likewise.
2165         (*movqi_internal): Likewise.
2166         * config/mips/mips.h (M16_STORE_REGS): New register class.
2167         (REG_CLASS_NAMES): Add M16_STORE_REGS.
2168         (REG_CLASS_CONTENTS): Likewise.
2169         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
2170
2171 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2172
2173         PR c/60194
2174         * doc/invoke.texi (-Wformat-signedness): Document it.
2175         (Wformat=2): Mention that this enables -Wformat-signedness.
2176
2177 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2178
2179         * common/config/epiphany/epiphany-common.c
2180         (epiphany_option_optimization_table): Enable section anchors by
2181         default at -O1 or higher.
2182         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
2183         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
2184         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
2185         carries no extra cost.
2186         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
2187         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
2188         * config/epiphany/predicates.md (memclob_operand): New predicate.
2189         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
2190         Use memclob_operand predicate and X constraint for operand 3.
2191
2192 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2193
2194         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
2195         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
2196         its operands.
2197
2198 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2199
2200         PR rtl-optimization/60651
2201         * mode-switching.c (optimize_mode_switching): Make sure to emit
2202         sets of a lower numbered entity before sets of a higher numbered
2203         entity to a mode of the same or lower priority.
2204         When creating a seginfo for a basic block that starts with a code
2205         label, move the insertion point past the code label.
2206         (new_seginfo): Document and enforce requirement that
2207         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
2208         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
2209         * doc/tm.texi: Regenerate.
2210
2211 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2212
2213         PR target/60811
2214         * config/arc/arc.c (arc_save_restore): Fix assert typo.
2215
2216 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
2217
2218         * BASE-VER: Set to 4.10.0.
2219
2220 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2221
2222         PR other/59055
2223         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
2224         * doc/gcc.texi (Service): Update description in the @menu
2225         * doc/invoke.texi (Option Summary): Remove misplaced and
2226         duplicated @menu.
2227
2228 2014-04-11  Steve Ellcey  <sellcey@mips.com>
2229             Jakub Jelinek  <jakub@redhat.com>
2230
2231         PR middle-end/60556
2232         * expr.c (convert_move): Use emit_store_flag_force instead of
2233         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
2234         argument to it.
2235
2236 2014-04-11  Richard Biener  <rguenther@suse.de>
2237
2238         PR middle-end/60797
2239         * varasm.c (assemble_alias): Avoid endless error reporting
2240         recursion by setting TREE_ASM_WRITTEN.
2241
2242 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2243
2244         * config/s390/s390.md: Add a splitter for NOT rtx.
2245
2246 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
2247
2248         PR rtl-optimization/60663
2249         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
2250
2251 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
2252             Jakub Jelinek  <jakub@redhat.com>
2253
2254         PR lto/60567
2255         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
2256         flag from decl_node to node.
2257
2258 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2259
2260         PR debug/60655
2261         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
2262         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
2263         ameliorating the cases where it can be.
2264
2265 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
2266
2267         Revert
2268         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2269
2270         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2271         (loadsync_<mode>): Change mode.
2272         (load_quadpti, store_quadpti): New.
2273         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2274         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2275         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
2276
2277 2014-04-09  Cong Hou  <congh@google.com>
2278
2279         PR testsuite/60773
2280         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
2281         documentation.
2282
2283 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2284
2285         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
2286         instead of vnor to exploit possible fusion opportunity in the
2287         future.
2288         (altivec_expand_vec_perm_const_le): Likewise.
2289
2290 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2291
2292         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2293         (loadsync_<mode>): Change mode.
2294         (load_quadpti, store_quadpti): New.
2295         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2296         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2297
2298 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
2299
2300         PR target/60763
2301         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
2302         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
2303         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
2304
2305 2014-04-08  Richard Biener  <rguenther@suse.de>
2306
2307         PR middle-end/60706
2308         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
2309         a 64bit widest int print double-int similar to on HWI64 hosts.
2310
2311 2014-04-08  Richard Biener  <rguenther@suse.de>
2312
2313         PR tree-optimization/60785
2314         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
2315         default defs properly.
2316
2317 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
2318
2319         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
2320         (Weffc++): Likewise.
2321
2322 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
2323
2324         * ipa-devirt.c (maybe_record_node): When node is not recorded,
2325         set completep to false rather than true.
2326
2327 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
2328
2329         PR target/60504
2330         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
2331         ARM_TARGET2_DWARF_FORMAT.
2332
2333 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
2334
2335         PR target/60609
2336         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
2337         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
2338         ADDR_DIFF_VEC.
2339
2340 2014-04-07  Richard Biener  <rguenther@suse.de>
2341
2342         PR tree-optimization/60766
2343         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
2344         (may_eliminate_iv): Convert cand_value_at result to desired type.
2345
2346 2014-04-07  Jason Merrill  <jason@redhat.com>
2347
2348         PR c++/60731
2349         * common.opt (-fno-gnu-unique): Add.
2350         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
2351
2352 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2353
2354         * haifa-sched.c: Fix outdated function reference and minor
2355         grammar errors in introductory comment.
2356
2357 2014-04-07  Richard Biener  <rguenther@suse.de>
2358
2359         PR middle-end/60750
2360         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
2361         for noreturn calls.
2362         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
2363
2364 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
2365
2366         PR debug/55794
2367         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
2368         size accounting for thunks.
2369         (pa_asm_output_mi_thunk): Use final_start_function() and
2370         final_end_function() to output function start and end directives.
2371
2372 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2373
2374         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
2375         device specific ISA/ feature information. Remove short_sp and
2376         errata_skip ds.  Add avr_device_specific_features enum to have device
2377         specific info.
2378         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
2379         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
2380         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
2381         updated device specific info.
2382         * config/avr/avr-mcus.def: Merge device specific details to
2383         dev_attribute field.
2384         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
2385         errata_skip.
2386         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
2387         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
2388         assembler if RMW isa supported by current device.
2389         * config/avr/genmultilib.awk: Update as device info structure changed.
2390         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
2391
2392 2014-04-04  Cong Hou  <congh@google.com>
2393
2394         PR tree-optimization/60656
2395         * tree-vect-stmts.c (supportable_widening_operation):
2396         Fix a bug that elements in a vector with vect_used_by_reduction
2397         property are incorrectly reordered when the operation on it is not
2398         consistant with the one in reduction operation.
2399
2400 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
2401
2402         PR rtl-optimization/60155
2403         * gcse.c (record_set_data): New function.
2404         (single_set_gcse): New function.
2405         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
2406         (hoist_code): Likewise.
2407         (get_pressure_class_and_nregs): Likewise.
2408
2409 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
2410
2411         * explow.c (probe_stack_range): Emit a final optimization blockage.
2412
2413 2014-04-04  Anthony Green  <green@moxielogic.com>
2414
2415         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
2416         typos.
2417
2418 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
2419
2420         PR ipa/59626
2421         * lto-cgraph.c (input_overwrite_node): Check that partitioning
2422         flags are set only during streaming.
2423         * ipa.c (process_references, walk_polymorphic_call_targets,
2424         symtab_remove_unreachable_nodes): Drop bodies of always inline
2425         after early inlining.
2426         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
2427
2428 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
2429         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2430
2431         PR debug/60655
2432         * dwarf2out.c (const_ok_for_output_1): Reject expressions
2433         containing a NOT.
2434
2435 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2436
2437         PR bootstrap/60743
2438         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
2439         duration.
2440         (cortex_a53_fdivd): Likewise.
2441
2442 2014-04-04  Martin Jambor  <mjambor@suse.cz>
2443
2444         PR ipa/60640
2445         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
2446         Adjust all callers.
2447         * cgraph.c (clone_of_p): Also return true if thunks match.
2448         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
2449         cgraph_function_or_thunk_node and an obsolete comment.
2450         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
2451         file.
2452         (build_function_decl_skip_args): Likewise.
2453         (set_new_clone_decl_and_node_flags): New function.
2454         (duplicate_thunk_for_node): Likewise.
2455         (redirect_edge_duplicating_thunks): Likewise.
2456         (cgraph_clone_node): New parameter args_to_skip, pass it to
2457         redirect_edge_duplicating_thunks which is called instead of
2458         cgraph_redirect_edge_callee.
2459         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
2460         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
2461
2462 2014-04-04  Jeff Law  <law@redhat.com>
2463
2464         PR target/60657
2465         * config/arm/predicates.md (const_int_I_operand): New predicate.
2466         (const_int_M_operand): Similarly.
2467         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
2468         const_int_operand.
2469         (insv_t2, extv_reg, extzv_t2): Likewise.
2470         (load_multiple_with_writeback): Similarly for const_int_I_operand.
2471         (pop_multiple_with_writeback_and_return): Likewise.
2472         (vfp_pop_multiple_with_writeback): Likewise
2473
2474 2014-04-04  Richard Biener  <rguenther@suse.de>
2475
2476         PR ipa/60746
2477         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
2478         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
2479         non-GIMPLE_LABELs.
2480         * gimplify.h (gimple_add_tmp_var_fn): Declare.
2481         * gimplify.c (gimple_add_tmp_var_fn): New function.
2482         * gimple-expr.h (create_tmp_reg_fn): Declare.
2483         * gimple-expr.c (create_tmp_reg_fn): New function.
2484         * gimple-low.c (record_vars_into): Don't change cfun.
2485         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
2486         code generation without cfun.
2487
2488 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
2489
2490         PR bootstrap/60719
2491         * Makefile.in (install-driver): Fix shell scripting.
2492
2493 2014-04-03  Cong Hou  <congh@google.com>
2494
2495         PR tree-optimization/60505
2496         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
2497         threshold of number of iterations below which no vectorization
2498         will be done.
2499         * tree-vect-loop.c (new_loop_vec_info):
2500         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
2501         * tree-vect-loop.c (vect_analyze_loop_operations):
2502         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
2503         * tree-vect-loop.c (vect_transform_loop):
2504         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
2505         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
2506         of iterations of the loop and see if we should build the epilogue.
2507
2508 2014-04-03  Richard Biener  <rguenther@suse.de>
2509
2510         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
2511         (streamer_tree_cache_create): Adjust.
2512         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
2513         to allow optional nodes array.
2514         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
2515         (streamer_tree_cache_append): Likewise.
2516         (streamer_tree_cache_create): Create nodes array optionally
2517         as specified by parameter.
2518         * lto-streamer-out.c (create_output_block): Avoid maintaining
2519         the node array in the writer cache.
2520         (DFS_write_tree): Remove assertion.
2521         (produce_asm_for_decls): Free the out decl state hash table early.
2522         * lto-streamer-in.c (lto_data_in_create): Adjust for
2523         streamer_tree_cache_create prototype change.
2524
2525 2014-04-03  Richard Biener  <rguenther@suse.de>
2526
2527         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
2528         set TREE_CHAIN to NULL_TREE.
2529
2530 2014-04-03  Richard Biener  <rguenther@suse.de>
2531
2532         PR tree-optimization/60740
2533         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
2534         over all GIMPLE_COND operands.
2535
2536 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
2537
2538         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
2539         (Weffc++): Remove Scott's numbering, merge lists and reference
2540         Wnon-virtual-dtor.
2541
2542 2014-04-03  Nick Clifton  <nickc@redhat.com>
2543
2544         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
2545         properly.
2546
2547 2014-04-03  Martin Jambor  <mjambor@suse.cz>
2548
2549         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
2550         mention gcc_unreachable before failing.
2551         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
2552         removed symbols.
2553
2554 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
2555
2556         PR ipa/60659
2557         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
2558         inconsistent code and instead mark the context inconsistent.
2559         (possible_polymorphic_call_targets): For inconsistent contexts
2560         return empty complete list.
2561
2562 2014-04-02  Anthony Green  <green@moxielogic.com>
2563
2564         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
2565         (extendqisi2, extendhisi2): Define.
2566         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
2567         (WCHAR_TYPE): Change to unsigned int.
2568
2569 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2570
2571         PR tree-optimization/60733
2572         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
2573         insertion point for PHI candidates to be the end of the feeding
2574         block for the PHI argument.
2575
2576 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
2577
2578         PR rtl-optimization/60650
2579         * lra-constraints.c (process_alt_operands): Decrease reject for
2580         earlyclobber matching.
2581
2582 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2583
2584         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
2585
2586 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2587
2588         * config/spu/spu.c (pad_bb): Do not crash when the last
2589         insn is CODE_FOR_blockage.
2590
2591 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2592
2593         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
2594         lies outside the target mode.
2595
2596 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2597
2598         PR target/60735
2599         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2600         software floating point or no floating point registers, do not
2601         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2602         in GPRs that occurs after we tested for GPRs that would never be
2603         true.
2604
2605         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2606         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2607         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2608         specifically allow DDmode, since that does not use the SPE SIMD
2609         instructions.
2610
2611 2014-04-02  Richard Biener  <rguenther@suse.de>
2612
2613         PR middle-end/60729
2614         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
2615         MODE_INTs.  Properly use negv_optab.
2616         (expand_abs): Likewise.
2617
2618 2014-04-02  Richard Biener  <rguenther@suse.de>
2619
2620         PR bootstrap/60719
2621         * Makefile.in (install-driver): Guard extra installs with special
2622         names properly.
2623
2624 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
2625
2626         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2627         Document vec_vgbbd.
2628
2629 2014-04-01  Richard Henderson  <rth@redhat.com>
2630
2631         PR target/60704
2632         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
2633         alternative enabled before register allocation.
2634
2635 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
2636
2637         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
2638         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
2639         typo.
2640         (nios2_large_got_address): Remove unneeded 'sym' parameter.
2641         (nios2_got_address): Update nios2_large_got_address call site.
2642         (nios2_delegitimize_address): New function.
2643         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
2644         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
2645         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
2646
2647 2014-04-01  Martin Husemann  <martin@duskware.de>
2648
2649         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
2650         for -mabi=32.
2651
2652 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
2653
2654         PR rtl-optimization/60604
2655         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
2656         check from register_operand.
2657         (register_operand): Redefine in terms of general_operand.
2658         (nonmemory_operand): Use register_operand for the non-constant cases.
2659
2660 2014-04-01  Richard Biener  <rguenther@suse.de>
2661
2662         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
2663
2664 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2665
2666         * doc/invoke.texi (mapp-regs): Clarify.
2667
2668 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
2669
2670         * config/i386/avx512fintrin.h (__v32hi): Define type.
2671         (__v64qi): Likewise.
2672         (_mm512_set1_epi8): Define.
2673         (_mm512_set1_epi16): Define.
2674         (_mm512_set4_epi32): Define.
2675         (_mm512_set4_epi64): Define.
2676         (_mm512_set4_pd): Define.
2677         (_mm512_set4_ps): Define.
2678         (_mm512_setr4_epi64): Define.
2679         (_mm512_setr4_epi32): Define.
2680         (_mm512_setr4_pd): Define.
2681         (_mm512_setr4_ps): Define.
2682         (_mm512_setzero_epi32): Define.
2683
2684 2014-03-31  Martin Jambor  <mjambor@suse.cz>
2685
2686         PR middle-end/60647
2687         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
2688         callsite_arguments_match_p.  Updated all callers.  Also check types of
2689         corresponding formal parameters and actual arguments.
2690         (not_all_callers_have_enough_arguments_p) Renamed to
2691         some_callers_have_mismatched_arguments_p.
2692
2693 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
2694
2695         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
2696
2697 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
2698
2699         PR target/60034
2700         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
2701         section anchor.
2702
2703 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
2704
2705         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
2706         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
2707         Split out
2708         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
2709         Use FMAMODE_NOVF512 mode iterator.
2710         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
2711         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
2712         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
2713         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
2714         Split out
2715         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
2716         Use VF_128_256 mode iterator.
2717         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
2718         Ditto.
2719
2720 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2721
2722         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
2723         static chain if needed.
2724
2725 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2726
2727         PR target/60697
2728         * lra-constraints.c (index_part_to_reg): New.
2729         (process_address): Use it.
2730
2731 2014-03-27  Jeff Law  <law@redhat.com>
2732             Jakub Jelinek  <jakub@redhat.com>
2733
2734         PR target/60648
2735         * expr.c (do_tablejump): Use simplify_gen_binary rather than
2736         gen_rtx_{PLUS,MULT} to build up the address expression.
2737
2738         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
2739         creating non-canonical RTL.
2740
2741 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2742
2743         PR ipa/60243
2744         * ipa-inline.c (want_inline_small_function_p): Short circuit large
2745         functions; reorganize to make cheap checks first.
2746         (inline_small_functions): Do not estimate growth when dumping;
2747         it is expensive.
2748         * ipa-inline.h (inline_summary): Add min_size.
2749         (growth_likely_positive): New function.
2750         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
2751         (set_cond_stmt_execution_predicate): Cleanup.
2752         (estimate_edge_size_and_time): Compute min_size.
2753         (estimate_calls_size_and_time): Likewise.
2754         (estimate_node_size_and_time): Likewise.
2755         (inline_update_overall_summary): Update min_size.
2756         (do_estimate_edge_time): Likewise.
2757         (do_estimate_edge_size): Update.
2758         (do_estimate_edge_hints): Update.
2759         (growth_likely_positive): New function.
2760
2761 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
2762
2763         PR target/60693
2764         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
2765         also if addr has VOIDmode.
2766
2767 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2768
2769         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
2770         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
2771         Declare extern.
2772         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
2773         instructions as well as AdvancedSIMD loads.
2774
2775 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2776
2777         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2778         Use crypto_aese type.
2779         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
2780         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
2781         crypto_aese, crypto_aesmc.  Move to types.md.
2782         * config/arm/types.md (crypto_aes): Split into crypto_aese,
2783         crypto_aesmc.
2784         * config/arm/iterators.md (crypto_type): Likewise.
2785
2786 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2787
2788         * cgraph.c: Include expr.h and tree-dfa.h.
2789         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
2790         remove LHS.
2791
2792 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2793
2794         PR target/60675
2795         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
2796         regs from checking multi-reg pseudos.
2797
2798 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2799
2800         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
2801
2802 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2803
2804         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
2805         if it would clobber the stack pointer, even temporarily.
2806
2807 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
2808
2809         * mode-switching.c: Make small adjustments to the top comment.
2810
2811 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
2812
2813         * config/rs6000/constraints.md (wD constraint): New constraint to
2814         match the constant integer to get the top DImode/DFmode out of a
2815         vector in a VSX register.
2816
2817         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
2818         match the constant integer to get the top DImode/DFmode out of a
2819         vector in a VSX register.
2820
2821         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
2822         for ISA 2.07.
2823
2824         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2825         vbpermq builtins.
2826
2827         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
2828         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
2829
2830         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
2831         Optimize vec_extract of 64-bit values, where the value being
2832         extracted is in the top word, where we can use scalar
2833         instructions.  Add direct move and store support.  Combine the big
2834         endian/little endian vector select load support into a single insn.
2835         (vsx_extract_<mode>_internal1): Likewise.
2836         (vsx_extract_<mode>_internal2): Likewise.
2837         (vsx_extract_<mode>_load): Likewise.
2838         (vsx_extract_<mode>_store): Likewise.
2839         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
2840         combined into vsx_extract_<mode>_load.
2841         (vsx_extract_<mode>_one_le): Likewise.
2842
2843         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
2844         define the top 64-bit vector element.
2845
2846         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
2847         constraint.
2848
2849         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2850         Document vec_vbpermq builtin.
2851
2852         PR target/60672
2853         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
2854         enable use of xxsldwi and xxpermdi builtin functions.
2855         (vec_xxpermdi): Likewise.
2856
2857         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2858         Document use of vec_xxsldwi and vec_xxpermdi builtins.
2859
2860 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
2861
2862         PR rtl-optimization/60650
2863         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
2864         first_p.  Use it.
2865         (find_spills_for): New.
2866         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
2867         Spill all pseudos on the second iteration.
2868
2869 2014-03-27  Marek Polacek  <polacek@redhat.com>
2870
2871         PR c/50347
2872         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
2873         types.
2874
2875 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2876
2877         * config/s390/s390.c (s390_can_use_return_insn): Check for
2878         call-saved FPRs on 31 bit.
2879
2880 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
2881
2882         PR middle-end/60682
2883         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
2884         if they need regimplification, just drop them instead of
2885         calling gimple_regimplify_operands on them.
2886
2887 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2888
2889         PR target/60580
2890         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
2891         (aarch64_frame_pointer_required): Adjust logic.
2892         (aarch64_can_eliminate): Adjust logic.
2893         (aarch64_override_options_after_change): Adjust logic.
2894
2895 2014-03-27  Dehao Chen  <dehao@google.com>
2896
2897         * ipa-inline.c (early_inliner): Update node's inline info.
2898
2899 2014-03-26  Dehao Chen  <dehao@google.com>
2900
2901         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
2902         compiler inserted conditional jumps for NAN float check.
2903
2904 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2905
2906         * ubsan.h (ubsan_create_data): Change second argument's type
2907         to const location_t *.
2908         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
2909         _("<unknown>").
2910         (ubsan_create_data): Change second argument to const location_t *PLOC.
2911         Create Loc field whenever PLOC is non-NULL.
2912         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
2913         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
2914         callers.
2915
2916         PR other/59545
2917         * real.c (real_to_integer2): Change type of low to UHWI.
2918
2919 2014-03-26  Tobias Burnus  <burnus@net-b.de>
2920
2921         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
2922         (CILK_SELF_SPECS): New define.
2923         (driver_self_specs): Use it.
2924
2925 2014-03-26  Richard Biener  <rguenther@suse.de>
2926
2927         * tree-pretty-print.c (percent_K_format): Implement special
2928         case for LTO and its stripped down BLOCK tree.
2929
2930 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2931
2932         PR sanitizer/60636
2933         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
2934
2935         * tree-vrp.c (simplify_internal_call_using_ranges): If only
2936         one range is range_int_cst_p, but not both, at least optimize
2937         addition/subtraction of 0 and multiplication by 0 or 1.
2938         * gimple-fold.c (gimple_fold_call): Fold
2939         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
2940         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
2941         INTEGER_CSTs, try to fold at least x * 0 and y - y.
2942
2943 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
2944
2945         PR rtl-optimization/60452
2946         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
2947         <case REG>: Return 1 for invalid offsets from the frame pointer.
2948
2949 2014-03-26  Marek Polacek  <polacek@redhat.com>
2950
2951         PR c/37428
2952         * doc/extend.texi (C Extensions): Mention variable-length arrays in
2953         a structure/union.
2954
2955 2014-03-26  Marek Polacek  <polacek@redhat.com>
2956
2957         PR c/39525
2958         * doc/extend.texi (Designated Inits): Describe what happens to omitted
2959         field members.
2960
2961 2014-03-26  Marek Polacek  <polacek@redhat.com>
2962
2963         PR other/59545
2964         * ira-color.c (update_conflict_hard_regno_costs): Perform the
2965         multiplication in unsigned type.
2966
2967 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2968
2969         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2970
2971 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2972
2973         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2974
2975 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2976
2977         PR ipa/60315
2978         * cif-code.def (UNREACHABLE) New code.
2979         * ipa-inline.c (inline_small_functions): Skip edges to
2980         __builtlin_unreachable.
2981         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2982         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2983         predicate to __bulitin_unreachable.
2984         (set_cond_stmt_execution_predicate): Fix issue when
2985         invert_tree_comparison returns ERROR_MARK.
2986         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2987         propagate to inline clones.
2988         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2989         to unreachable.
2990         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2991         * cgraphclones.c (cgraph_clone_node): If call destination is already
2992         ureachable, do not redirect it back.
2993         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2994         unreachable.
2995
2996 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2997
2998         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2999         Do not modify inline clones.
3000
3001 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3002
3003         * config/i386/i386.md (general_sext_operand): New mode attr.
3004         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
3005         don't generate (sign_extend (const_int)).
3006         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
3007         operands[2].  Use We constraint instead of <i> and
3008         <general_sext_operand> predicate instead of <general_operand>.
3009         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
3010         * config/i386/constraints.md (We): New constraint.
3011         * config/i386/predicates.md (x86_64_sext_operand,
3012         sext_operand): New predicates.
3013
3014 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3015
3016         PR ipa/60600
3017         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
3018         inconsistent devirtualizations to __builtin_unreachable.
3019
3020 2014-03-25  Marek Polacek  <polacek@redhat.com>
3021
3022         PR c/35449
3023         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
3024
3025 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
3026
3027         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
3028         order of elements for big-endian.
3029
3030 2014-03-25  Richard Biener  <rguenther@suse.de>
3031
3032         PR middle-end/60635
3033         * gimplify-me.c (gimple_regimplify_operands): Update the
3034         re-gimplifed stmt.
3035
3036 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3037
3038         PR ipa/59176
3039         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
3040         (lto_output_varpool_node): Likewise.
3041         (input_overwrite_node): Likewise.
3042         (input_varpool_node): Likewise.
3043
3044 2014-03-25  Richard Biener  <rguenther@suse.de>
3045
3046         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
3047         (run_gcc): Likewise.
3048
3049 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3050
3051         * combine.c (simplify_compare_const): Add MODE argument.
3052         Handle mode_width 0 as very large mode_width.
3053         (try_combine, simplify_comparison): Adjust callers.
3054
3055         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
3056         type to avoid signed integer overflow.
3057         * explow.c (plus_constant): Likewise.
3058
3059 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3060
3061         * doc/generic.texi: Correct typos.
3062
3063 2014-03-24  Tobias Burnus  <burnus@net-b.de>
3064
3065         * doc/invoke.texi (-flto): Expand section about
3066         using static libraries with LTO.
3067
3068 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3069
3070         PR rtl-optimization/60501
3071         * optabs.def (addptr3_optab): New optab.
3072         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
3073         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
3074         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
3075
3076         * lra.c (emit_add3_insn): Use the addptr pattern if available.
3077
3078         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
3079
3080 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
3081
3082         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
3083         _mm512_set1_pd.
3084
3085         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
3086         (_mm256_undefined_ps): Define.
3087         (_mm256_undefined_pd): Define.
3088         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
3089         (_mm_undefined_pd): Define.
3090         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
3091         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
3092         (_mm512_undefined_ps): Define.
3093         (_mm512_undefined_pd): Define.
3094         Use _mm*_undefined_*.
3095         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
3096
3097 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
3098
3099         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
3100         (lshr_simd): DI mode added.
3101         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
3102         (aarch64_ushr_simddi): Likewise.
3103         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
3104         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
3105         (vshrd_n_u64): Likewise.
3106
3107 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3108
3109         * Makefile.in (s-macro_list): Depend on cc1.
3110
3111 2014-03-23  Teresa Johnson  <tejohnson@google.com>
3112
3113         * ipa-utils.c (ipa_print_order): Use specified dump file.
3114
3115 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
3116
3117         PR rtl-optimization/60601
3118         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
3119
3120         * gcc.c (eval_spec_function): Initialize save_growing_value.
3121
3122 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
3123
3124         PR sanitizer/60613
3125         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
3126         code == MINUS_EXPR, never swap op0 with op1.
3127
3128         * toplev.c (init_local_tick): Avoid signed integer multiplication
3129         overflow.
3130         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
3131         shift by first operand's bitsize.
3132
3133 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
3134
3135         PR target/60610
3136         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
3137         redefine to 1 or 0.
3138         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
3139         TARGET_ISA_64BIT_P(x).
3140
3141 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3142
3143         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
3144         pattern for vector nor instead of subtract from splat(-1).
3145         (altivec_expand_vec_perm_const_le): Likewise.
3146
3147 2014-03-21  Richard Henderson  <rth@twiddle.net>
3148
3149         PR target/60598
3150         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
3151         related insns after epilogue_completed.
3152
3153 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3154
3155         PR ipa/59176
3156         * cgraph.h (symtab_node): New flag body_removed.
3157         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
3158         when removing bodies.
3159         * symtab.c (dump_symtab_base): Dump body_removed flag.
3160         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
3161         had their bodies removed.
3162
3163 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3164
3165         PR ipa/60419
3166         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
3167         in the border.
3168
3169 2014-03-21  Richard Biener  <rguenther@suse.de>
3170
3171         PR tree-optimization/60577
3172         * tree-core.h (struct tree_base): Document nothrow_flag use
3173         in VAR_DECL_NONALIASED.
3174         * tree.h (VAR_DECL_NONALIASED): New.
3175         (may_be_aliased): Adjust.
3176         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
3177
3178 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3179
3180         * expr.c (expand_expr_real_1): Remove outdated comment.
3181
3182 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
3183
3184         PR middle-end/60597
3185         * ira.c (adjust_cleared_regs): Call copy_rtx on
3186         *reg_equiv[REGNO (loc)].src_p before passing it to
3187         simplify_replace_fn_rtx.
3188
3189         PR target/60568
3190         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
3191         into CONST, put pic register as first operand of PLUS.  Use
3192         gen_const_mem for both 32-bit and 64-bit PIC got loads.
3193
3194 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3195
3196         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
3197
3198 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3199
3200         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
3201         around for store forwarding issue in the FPU on the UT699.
3202         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
3203         loads and operations if -mfix-ut699 is specified.
3204         (divtf3_hq): Tweak attribute.
3205         (sqrttf2_hq): Likewise.
3206
3207 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3208
3209         * calls.c (store_one_arg): Remove incorrect const qualification on the
3210         type of the temporary.
3211         * cfgexpand.c (expand_return): Likewise.
3212         * expr.c (expand_constructor): Likewise.
3213         (expand_expr_real_1): Likewise.
3214
3215 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3216
3217         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
3218         of parts.
3219
3220 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
3221
3222         PR target/60039
3223         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
3224
3225 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
3226
3227         * config/arm/aarch-common-protos.h
3228         (alu_cost_table): Fix spelling of "extend".
3229         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
3230
3231 2014-03-19  Richard Biener  <rguenther@suse.de>
3232
3233         PR middle-end/60553
3234         * tree-core.h (tree_type_common): Re-order pointer members
3235         to reduce recursion depth during GC walks.
3236
3237 2014-03-19  Marek Polacek  <polacek@redhat.com>
3238
3239         PR sanitizer/60569
3240         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
3241         before accessing it.
3242
3243 2014-03-19  Richard Biener  <rguenther@suse.de>
3244
3245         PR lto/59543
3246         * lto-streamer-in.c (input_function): In WPA stage do not drop
3247         debug stmts.
3248
3249 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
3250
3251         PR tree-optimization/60559
3252         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
3253         with build_zero_cst assignment.
3254
3255 2014-03-18  Kai Tietz  <ktietz@redhat.com>
3256
3257         PR rtl-optimization/56356
3258         * sdbout.c (sdbout_parms): Verify that parms'
3259         incoming argument is valid.
3260         (sdbout_reg_parms): Likewise.
3261
3262 2014-03-18  Richard Henderson  <rth@redhat.com>
3263
3264         PR target/60562
3265         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
3266         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
3267         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
3268
3269 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
3270
3271         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
3272         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
3273         Italicize plugin event names in description.  Explain that
3274         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
3275         Remind that no GCC functions should be called after PLUGIN_FINISH.
3276         Explain what pragmas with expansion are.
3277
3278 2014-03-18  Martin Liska  <mliska@suse.cz>
3279
3280         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
3281         gimple call statement is update.
3282         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
3283         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
3284
3285 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3286
3287         PR sanitizer/60557
3288         * ubsan.c (ubsan_instrument_unreachable): Call
3289         initialize_sanitizer_builtins.
3290         (ubsan_pass): Likewise.
3291
3292         PR sanitizer/60535
3293         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
3294         varpool_finalize_decl instead of rest_of_decl_compilation.
3295
3296 2014-03-18  Richard Biener  <rguenther@suse.de>
3297
3298         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
3299         by using bitmap_and_compl instead of bitmap_and_compl_into.
3300         (df_rd_transfer_function): Likewise.
3301
3302 2014-03-18  Richard Biener  <rguenther@suse.de>
3303
3304         * doc/lto.texi (fresolution): Fix typo.
3305
3306 2014-03-18  Richard Biener  <rguenther@suse.de>
3307
3308         * doc/invoke.texi (flto): Update for changes in 4.9.
3309
3310 2014-03-18  Richard Biener  <rguenther@suse.de>
3311
3312         * doc/loop.texi: Remove section on the removed lambda framework.
3313         Update loop docs with recent changes in preserving loop structure.
3314
3315 2014-03-18  Richard Biener  <rguenther@suse.de>
3316
3317         * doc/lto.texi (-fresolution): Document.
3318
3319 2014-03-18  Richard Biener  <rguenther@suse.de>
3320
3321         * doc/contrib.texi: Adjust my name.
3322
3323 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3324
3325         PR ipa/58721
3326         * internal-fn.c: Include diagnostic-core.h.
3327         (expand_BUILTIN_EXPECT): New function.
3328         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
3329         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
3330         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
3331         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
3332         IFN_BUILTIN_EXPECT.
3333         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
3334         Revert 3 argument __builtin_expect code.
3335         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
3336         * gimple-fold.c (gimple_fold_call): Likewise.
3337         * tree.h (fold_builtin_expect): New prototype.
3338         * builtins.c (build_builtin_expect_predicate): Add predictor
3339         argument, if non-NULL, create 3 argument __builtin_expect.
3340         (fold_builtin_expect): No longer static.  Add ARG2 argument,
3341         pass it through to build_builtin_expect_predicate.
3342         (fold_builtin_2): Adjust caller.
3343         (fold_builtin_3): Handle BUILT_IN_EXPECT.
3344         * internal-fn.def (BUILTIN_EXPECT): New.
3345
3346 2014-03-18  Tobias Burnus  <burnus@net-b.de>
3347
3348         PR ipa/58721
3349         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
3350         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
3351         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
3352
3353 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
3354
3355         PR ipa/58721
3356         * predict.c (combine_predictions_for_bb): Fix up formatting.
3357         (expr_expected_value_1, expr_expected_value): Add predictor argument,
3358         fill what it points to if non-NULL.
3359         (tree_predict_by_opcode): Adjust caller, use the predictor.
3360         * predict.def (PRED_COMPARE_AND_SWAP): Add.
3361
3362 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
3363
3364         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
3365         proper constant for the store mode.
3366
3367 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
3368
3369         * symtab.c (change_decl_assembler_name): Fix transparent alias
3370         chain construction.
3371
3372 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
3373
3374         * config/aarch64/aarch64.c: Correct the comments about the
3375         aarch64 stack layout.
3376
3377 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
3378
3379         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
3380         check for GF_OMP_FOR_KIND_FOR.
3381
3382 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
3383
3384         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
3385         ymm and zmm register names.
3386
3387 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
3388
3389         PR target/60516
3390         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
3391         note creation for the 2010-08-31 changes.
3392
3393 2014-03-17  Marek Polacek  <polacek@redhat.com>
3394
3395         PR middle-end/60534
3396         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
3397         as -fno-tree-loop-vectorize.
3398         (expand_omp_simd): Likewise.
3399
3400 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
3401
3402         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
3403         (eligible_for_call_delay): New prototype.
3404         * config/sparc/sparc.c (tls_call_delay): Rename into...
3405         (eligible_for_call_delay): ...this.  Return false if the instruction
3406         cannot be put in the delay slot of a branch.
3407         (eligible_for_restore_insn): Simplify.
3408         (eligible_for_return_delay): Return false if the instruction cannot be
3409         put in the delay slot of a branch and simplify.
3410         (eligible_for_sibcall_delay): Return false if the instruction cannot be
3411         put in the delay slot of a branch.
3412         * config/sparc/sparc.md (fix_ut699): New attribute.
3413         (tls_call_delay): Delete.
3414         (in_call_delay): Reimplement.
3415         (eligible_for_sibcall_delay): Rename into...
3416         (in_sibcall_delay): ...this.
3417         (eligible_for_return_delay): Rename into...
3418         (in_return_delay): ...this.
3419         (in_branch_delay): Reimplement.
3420         (in_uncond_branch_delay): Delete.
3421         (in_annul_branch_delay): Delete.
3422
3423 2014-03-14  Richard Henderson  <rth@redhat.com>
3424
3425         PR target/60525
3426         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
3427         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
3428         (*floathi<X87MODEF>2_i387_with_temp): Remove.
3429         (floathi splitters): Remove.
3430         (float<SWI48x>xf2): New pattern.
3431         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
3432         code that tried to handle DImode for 32-bit, but which was excluded
3433         by the pattern's condition.  Drop allocation of stack temporary.
3434         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
3435         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
3436         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
3437         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
3438         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
3439         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
3440         (*float<SWI48><MODEF>2_sse_interunit): Remove.
3441         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
3442         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
3443         (*float<SWI48x><X87MODEF>2_i387): Remove.
3444         (all float _with_temp splitters): Remove.
3445         (*float<SWI48x><MODEF>2_i387): New pattern.
3446         (*float<SWI48><MODEF>2_sse): New pattern.
3447         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
3448         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
3449
3450 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
3451             Marek Polacek  <polacek@redhat.com>
3452
3453         PR middle-end/60484
3454         * common.opt (dump_base_name_prefixed): New Variable.
3455         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
3456         if x_dump_base_name_prefixed is already set, set it at the end.
3457
3458 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
3459
3460         PR rtl-optimization/60508
3461         * lra-constraints.c (get_reload_reg): Add new parameter
3462         in_subreg_p.
3463         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
3464         Pass the new parameter values.
3465
3466 2014-03-14  Richard Biener  <rguenther@suse.de>
3467
3468         * common.opt: Revert unintented changes from r205065.
3469         * opts.c: Likewise.
3470
3471 2014-03-14  Richard Biener  <rguenther@suse.de>
3472
3473         PR middle-end/60518
3474         * cfghooks.c (split_block): Properly adjust all loops the
3475         block was a latch of.
3476
3477 2014-03-14  Martin Jambor  <mjambor@suse.cz>
3478
3479         PR lto/60461
3480         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
3481         and simplify it.
3482
3483 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
3484
3485         PR target/59396
3486         * config/avr/avr.c (avr_set_current_function): Pass function name
3487         through default_strip_name_encoding before sanity checking instead
3488         of skipping the first char of the assembler name.
3489
3490 2014-03-13  Richard Henderson  <rth@redhat.com>
3491
3492         PR debug/60438
3493         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
3494         (ix86_force_to_memory, ix86_free_from_memory): Remove.
3495         * config/i386/i386-protos.h: Likewise.
3496         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
3497         in the expander instead of a splitter.
3498         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
3499         any possibility of requiring a memory.
3500         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
3501         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
3502         (fp branch splitters): Update for ix86_split_fp_branch.
3503         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
3504         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
3505         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
3506         (*fop_<MODEF>_2_i387): Remove f/r alternative.
3507         (*fop_<MODEF>_3_i387): Likewise.
3508         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
3509         (splitters for the fop_* register patterns): Remove.
3510         (fscalexf4_i387): Rename from *fscalexf4_i387.
3511         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
3512
3513 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3514
3515         PR tree-optimization/59779
3516         * tree-dfa.c (get_ref_base_and_extent): Use double_int
3517         type for bitsize and maxsize instead of HOST_WIDE_INT.
3518
3519 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
3520
3521         PR rtl-optimization/57320
3522         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
3523         the CFG after thread_prologue_and_epilogue_insns.
3524
3525 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
3526
3527         PR rtl-optimization/57189
3528         * lra-constraints.c (process_alt_operands): Disfavor spilling
3529         vector pseudos.
3530
3531 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
3532
3533         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
3534
3535 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3536
3537         PR tree-optimization/59025
3538         PR middle-end/60418
3539         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
3540         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
3541
3542 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
3543
3544         PR target/60486
3545         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
3546         calls of avr_out_plus_1.
3547
3548 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
3549
3550         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
3551         BB's single pred and update the father loop's latch info later.
3552
3553 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
3554
3555         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
3556         (VEC_M): Likewise.
3557         (VEC_N): Likewise.
3558         (VEC_R): Likewise.
3559         (VEC_base): Likewise.
3560         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
3561         registers, we need to swap double words in little endian mode.
3562
3563         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
3564         to be a container mode for 128-bit integer operations added in ISA
3565         2.07.  Unlike TImode and PTImode, the preferred register set is
3566         the Altivec/VMX registers for the 128-bit operations.
3567
3568         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
3569         declarations.
3570         (rs6000_split_128bit_ok_p): Likewise.
3571
3572         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
3573         macros for creating ISA 2.07 normal and overloaded builtin
3574         functions with 3 arguments.
3575         (BU_P8V_OVERLOAD_3): Likewise.
3576         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
3577         for use as overloaded functions.
3578         (VPERM_1TI_UNS): Likewise.
3579         (VSEL_1TI): Likewise.
3580         (VSEL_1TI_UNS): Likewise.
3581         (ST_INTERNAL_1ti): Likewise.
3582         (LD_INTERNAL_1ti): Likewise.
3583         (XXSEL_1TI): Likewise.
3584         (XXSEL_1TI_UNS): Likewise.
3585         (VPERM_1TI): Likewise.
3586         (VPERM_1TI_UNS): Likewise.
3587         (XXPERMDI_1TI): Likewise.
3588         (SET_1TI): Likewise.
3589         (LXVD2X_V1TI): Likewise.
3590         (STXVD2X_V1TI): Likewise.
3591         (VEC_INIT_V1TI): Likewise.
3592         (VEC_SET_V1TI): Likewise.
3593         (VEC_EXT_V1TI): Likewise.
3594         (EQV_V1TI): Likewise.
3595         (NAND_V1TI): Likewise.
3596         (ORC_V1TI): Likewise.
3597         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
3598         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
3599         overloaded builtin.
3600         (VADDUQM): Likewise.
3601         (VSUBCUQ): Likewise.
3602         (VADDEUQM): Likewise.
3603         (VADDECUQ): Likewise.
3604         (VSUBEUQM): Likewise.
3605         (VSUBECUQ): Likewise.
3606
3607         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
3608         __int128_t and __uint128_t types.
3609         (__uint128_type): Likewise.
3610         (altivec_categorize_keyword): Add support for vector __int128_t,
3611         vector __uint128_t, vector __int128, and vector unsigned __int128
3612         as a container type for TImode operations that need to be done in
3613         VSX/Altivec registers.
3614         (rs6000_macro_to_expand): Likewise.
3615         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
3616         to support 128-bit integer instructions vaddcuq, vadduqm,
3617         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
3618         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
3619
3620         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
3621         for V1TImode, and set up preferences to use VSX/Altivec registers.
3622         Setup VSX reload handlers.
3623         (rs6000_debug_reg_global): Likewise.
3624         (rs6000_init_hard_regno_mode_ok): Likewise.
3625         (rs6000_preferred_simd_mode): Likewise.
3626         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
3627         (easy_altivec_constant): Likewise.
3628         (output_vec_const_move): Likewise.
3629         (rs6000_expand_vector_set): Convert V1TImode set and extract to
3630         simple move.
3631         (rs6000_expand_vector_extract): Likewise.
3632         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
3633         addressing.
3634         (rs6000_const_vec): Add support for V1TImode.
3635         (rs6000_emit_le_vsx_load): Swap double words when loading or
3636         storing TImode/V1TImode.
3637         (rs6000_emit_le_vsx_store): Likewise.
3638         (rs6000_emit_le_vsx_move): Likewise.
3639         (rs6000_emit_move): Add support for V1TImode.
3640         (altivec_expand_ld_builtin): Likewise.
3641         (altivec_expand_st_builtin): Likewise.
3642         (altivec_expand_vec_init_builtin): Likewise.
3643         (altivec_expand_builtin): Likewise.
3644         (rs6000_init_builtins): Add support for V1TImode type.  Add
3645         support for ISA 2.07 128-bit integer builtins.  Define type names
3646         for the VSX/Altivec vector types.
3647         (altivec_init_builtins): Add support for overloaded vector
3648         functions with V1TImode type.
3649         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
3650         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
3651         external function.
3652         (rs6000_split_128bit_ok_p): Likewise.
3653         (rs6000_handle_altivec_attribute): Create V1TImode from vector
3654         __int128_t and vector __uint128_t.
3655
3656         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
3657         and mode attributes.
3658         (VSX_M): Likewise.
3659         (VSX_M2): Likewise.
3660         (VSm): Likewise.
3661         (VSs): Likewise.
3662         (VSr): Likewise.
3663         (VSv): Likewise.
3664         (VS_scalar): Likewise.
3665         (VS_double): Likewise.
3666         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
3667
3668         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
3669         we support the ISA 2.07 128-bit integer arithmetic instructions.
3670         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
3671         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
3672         and TImode types for use with the builtin functions.
3673         (V1TI_type_node): Likewise.
3674         (unsigned_V1TI_type_node): Likewise.
3675         (intTI_type_internal_node): Likewise.
3676         (uintTI_type_internal_node): Likewise.
3677
3678         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
3679         128-bit builtin functions.
3680         (UNSPEC_VADDEUQM): Likewise.
3681         (UNSPEC_VADDECUQ): Likewise.
3682         (UNSPEC_VSUBCUQ): Likewise.
3683         (UNSPEC_VSUBEUQM): Likewise.
3684         (UNSPEC_VSUBECUQ): Likewise.
3685         (VM): Add V1TImode to vector mode iterators.
3686         (VM2): Likewise.
3687         (VI_unit): Likewise.
3688         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
3689         (altivec_vaddcuq): Likewise.
3690         (altivec_vsubuqm): Likewise.
3691         (altivec_vsubcuq): Likewise.
3692         (altivec_vaddeuqm): Likewise.
3693         (altivec_vaddecuq): Likewise.
3694         (altivec_vsubeuqm): Likewise.
3695         (altivec_vsubecuq): Likewise.
3696
3697         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
3698         mode iterators.
3699         (BOOL_128): Likewise.
3700         (BOOL_REGS_OUTPUT): Likewise.
3701         (BOOL_REGS_OP1): Likewise.
3702         (BOOL_REGS_OP2): Likewise.
3703         (BOOL_REGS_UNARY): Likewise.
3704         (BOOL_REGS_AND_CR0): Likewise.
3705
3706         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
3707         128-bit integer builtin support.
3708         (vec_vadduqm): Likewise.
3709         (vec_vaddecuq): Likewise.
3710         (vec_vaddeuqm): Likewise.
3711         (vec_vsubecuq): Likewise.
3712         (vec_vsubeuqm): Likewise.
3713         (vec_vsubcuq): Likewise.
3714         (vec_vsubuqm): Likewise.
3715
3716         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3717         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
3718         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
3719         128-bit integer add/subtract to ISA 2.07.
3720
3721 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
3722
3723         * config/arc/arc.c (arc_predicate_delay_insns):
3724         Fix third argument passed to conditionalize_nonjump.
3725
3726 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
3727
3728         * config/aarch64/aarch64-builtins.c
3729         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
3730         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
3731         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
3732         instead of __builtin_lfloor.
3733         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
3734
3735 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3736
3737         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
3738         (tree_ssa_ifcombine_bb_1): New function.
3739         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
3740         is an empty forwarder block to then_bb or vice versa and then_bb
3741         and else_bb are effectively swapped.
3742
3743 2014-03-12  Christian Bruel  <christian.bruel@st.com>
3744
3745         PR target/60264
3746         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
3747         REG_CFA_DEF_CFA note.
3748         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
3749         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
3750
3751 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
3752
3753         PR tree-optimization/60454
3754         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
3755
3756 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3757
3758         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
3759         Do not define target_cpu_default2 to generic.
3760         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
3761         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
3762         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
3763
3764 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3765             Marc Glisse  <marc.glisse@inria.fr>
3766
3767         PR tree-optimization/60502
3768         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
3769         instead of build_low_bits_mask.
3770
3771 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3772
3773         PR middle-end/60482
3774         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
3775         if there are multiple uses, but op doesn't live on E edge.
3776         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
3777         clobber stmts before __builtin_unreachable.
3778
3779 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
3780
3781         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
3782         hard_frame_pointer_rtx.
3783         * cse.c (cse_insn): Remove volatile check.
3784         * cselib.c (cselib_process_insn): Likewise.
3785         * dse.c (scan_insn): Likewise.
3786
3787 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3788
3789         * config/arc/arc.c (conditionalize_nonjump): New function,
3790         broken out of ...
3791         (arc_ifcvt): ... this.
3792         (arc_predicate_delay_insns): Use it.
3793
3794 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3795
3796         * config/arc/predicates.md (extend_operand): During/after reload,
3797         allow const_int_operand.
3798         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
3799         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
3800         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
3801         to "i".
3802         (umulsi3_highpart_i): Likewise.
3803
3804 2014-03-11  Richard Biener  <rguenther@suse.de>
3805
3806         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
3807         Add asserts to guard possible wrong-code bugs.
3808
3809 2014-03-11  Richard Biener  <rguenther@suse.de>
3810
3811         PR tree-optimization/60429
3812         PR tree-optimization/60485
3813         * tree-ssa-structalias.c (set_union_with_increment): Properly
3814         take into account all fields that overlap the shifted vars.
3815         (do_sd_constraint): Likewise.
3816         (do_ds_constraint): Likewise.
3817         (get_constraint_for_ptr_offset): Likewise.
3818
3819 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
3820
3821         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
3822         (nios2_compute_frame_layout):
3823         Add calculation of cfun->machine->fp_save_offset.
3824         (nios2_expand_prologue): Correct setting of frame pointer register
3825         in prologue.
3826         (nios2_expand_epilogue): Update recovery of stack pointer from
3827         frame pointer accordingly.
3828         (nios2_initial_elimination_offset): Update calculation of offset
3829         for eliminating to HARD_FRAME_POINTER_REGNUM.
3830
3831 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
3832
3833         PR ipa/60457
3834         * ipa.c (symtab_remove_unreachable_nodes): Don't call
3835         cgraph_get_create_node on VAR_DECLs.
3836
3837 2014-03-10  Richard Biener  <rguenther@suse.de>
3838
3839         PR middle-end/60474
3840         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
3841
3842 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
3843
3844         * config/vms/vms.opt (vms_float_format): New variable.
3845
3846 2014-03-08  Tobias Burnus  <burnus@net-b.de>
3847
3848         * doc/invoke.texi (-fcilkplus): Update implementation status.
3849
3850 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
3851             Richard Biener  <rguenther@suse.de>
3852
3853         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
3854         consistently accross all TUs.
3855         (run_gcc): Enable -fshort-double automatically at link at link-time
3856         and disallow override.
3857
3858 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
3859
3860         PR target/58271
3861         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
3862         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
3863         if they can't be used.
3864
3865 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3866
3867         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
3868         for Solaris 11/x86 ld.
3869         * configure: Regenerate.
3870
3871 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3872
3873         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
3874         (LIB_TLS_SPEC): Save as ld_tls_libs.
3875         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
3876         (HAVE_AS_IX86_TLSLDM): New test.
3877         * configure, config.in: Regenerate.
3878         * config/i386/i386.c (legitimize_tls_address): Fall back to
3879         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
3880         cannot support TLS_MODEL_LOCAL_DYNAMIC.
3881         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
3882         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
3883
3884 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
3885
3886         * common.opt (fira-loop-pressure): Mark as optimization.
3887
3888 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
3889
3890         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
3891         an OpenMP mappable type.
3892
3893 2014-03-06  Matthias Klose  <doko@ubuntu.com>
3894
3895         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
3896         MULTILIB_OSDIRNAMES is not defined.
3897
3898 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
3899             Meador Inge  <meadori@codesourcery.com>
3900
3901         PR target/58595
3902         * config/arm/arm.c (arm_tls_symbol_p): Remove.
3903         (arm_legitimize_address): Call legitimize_tls_address for any
3904         arm_tls_referenced_p expression, handle constant addend.  Call it
3905         before testing for !TARGET_ARM.
3906         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
3907
3908 2014-03-06  Richard Biener  <rguenther@suse.de>
3909
3910         PR middle-end/60445
3911         PR lto/60424
3912         PR lto/60427
3913         Revert
3914         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3915
3916         * tree-streamer.c (record_common_node): Assert we don't record
3917         nodes with type double.
3918         (preload_common_node): Skip type double, complex double and double
3919         pointer since it is now frontend dependent due to fshort-double option.
3920
3921 2014-03-06  Richard Biener  <rguenther@suse.de>
3922
3923         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
3924         or -fno-lto is specified and the linker has full plugin support.
3925         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
3926         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
3927         * lto-wrapper.c (merge_and_complain): Merge compile-time
3928         optimization levels.
3929         (run_gcc): And pass it through to the link options.
3930
3931 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
3932
3933         PR debug/60381
3934         Revert:
3935         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3936         PR debug/59992
3937         * cselib.c (remove_useless_values): Skip to avoid quadratic
3938         behavior if the condition moved from...
3939         (cselib_process_insn): ... here holds.
3940
3941 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3942
3943         PR plugins/59335
3944         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
3945         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
3946
3947         PR plugins/59335
3948         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
3949         (TM_H): Add x86-tune.def.
3950
3951 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3952
3953         * config/aarch64/aarch64.c (generic_tunings):
3954         Use cortexa57_extra_costs.
3955
3956 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3957
3958         PR lto/60404
3959         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
3960         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
3961         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
3962         cost for in_lto_p.
3963
3964 2014-03-04  Heiher  <r@hev.cc>
3965
3966         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3967         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3968
3969 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
3970
3971         * config/i386/predicates.md (const2356_operand): Change to ...
3972         (const2367_operand): ... this.
3973         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3974         const2367_operand.
3975         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3976         (*avx512pf_scatterpf<mode>sf): Ditto.
3977         (avx512pf_scatterpf<mode>df): Ditto.
3978         (*avx512pf_scatterpf<mode>df_mask): Ditto.
3979         (*avx512pf_scatterpf<mode>df): Ditto.
3980         * config/i386/i386.c (ix86_expand_builtin): Update
3981         incorrect hint operand error message.
3982
3983 2014-03-04  Richard Biener  <rguenther@suse.de>
3984
3985         * lto-section-in.c (lto_get_section_data): Fix const cast.
3986
3987 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3988
3989         * tree-streamer.c (record_common_node): Assert we don't record
3990         nodes with type double.
3991         (preload_common_node): Skip type double, complex double and double
3992         pointer since it is now frontend dependent due to fshort-double option.
3993
3994 2014-03-04  Richard Biener  <rguenther@suse.de>
3995
3996         PR lto/60405
3997         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3998         (lto_input_toplevel_asms): Likewise.
3999         * lto-section-in.c (lto_get_section_data): Instead do it here
4000         for every section.
4001
4002 2014-03-04  Richard Biener  <rguenther@suse.de>
4003
4004         PR tree-optimization/60382
4005         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
4006         dead PHIs a reduction.
4007
4008 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
4009
4010         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
4011         hint value.
4012         (_mm_prefetch): Move out of GCC target("sse") pragma.
4013         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
4014         GCC target("prfchw") pragma.
4015         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
4016         for locality <= 2.
4017         * config/i386/i386.c (ix86_option_override_internal): Enable
4018         -mprfchw with -mprefetchwt1.
4019
4020 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4021
4022         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
4023         Mark as varying.
4024
4025 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4026
4027         * opts.h (CL_PCH_IGNORE): Define.
4028         * targhooks.c (option_affects_pch_p):
4029         Return false for options that have CL_PCH_IGNORE set.
4030         * opt-functions.awk: Process PchIgnore.
4031         * doc/options.texi: Document PchIgnore.
4032
4033         * config/arc/arc.opt (misize): Add PchIgnore property.
4034
4035 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4036
4037         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
4038         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
4039         constraint on constants to permit them being loaded into
4040         GENERAL_REGS or BASE_REGS.
4041
4042 2014-03-03  Nick Clifton  <nickc@redhat.com>
4043
4044         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
4045         anti-cacnonical alternatives.
4046         (negandhi3_real): New pattern.
4047         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
4048
4049 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
4050
4051         * config/avr/avr-mcus.def: Remove atxmega16x1.
4052         * config/avr/avr-tables.opt: Regenerate.
4053         * config/avr/t-multilib: Regenerate.
4054         * doc/avr-mmcu.texi: Regenerate.
4055
4056 2014-03-03  Tobias Grosser  <tobias@grosser.es>
4057             Mircea Namolaru  <mircea.namolaru@inria.fr>
4058
4059         PR tree-optimization/58028
4060         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
4061         scalar dimensions.
4062
4063 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4064
4065         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
4066         not handled by recognizers.
4067
4068 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
4069
4070         PR middle-end/60175
4071         * function.c (expand_function_end): Don't emit
4072         clobber_return_register sequence if clobber_after is a BARRIER.
4073         * cfgexpand.c (construct_exit_block): Append instructions before
4074         return_label to prev_bb.
4075
4076 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4077
4078         * config/rs6000/constraints.md: Document reserved use of "wc".
4079
4080 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4081
4082         PR ipa/60150
4083         * ipa.c (function_and_variable_visibility): When dissolving comdat
4084         group, also set all symbols to local.
4085
4086 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4087
4088         PR ipa/60306
4089
4090         Revert:
4091         2013-12-14   Jan Hubicka  <jh@suse.cz>
4092         PR middle-end/58477
4093         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
4094
4095 2014-03-02  Jon Beniston  <jon@beniston.com>
4096
4097         PR bootstrap/48230
4098         PR bootstrap/50927
4099         PR bootstrap/52466
4100         PR target/46898
4101         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
4102           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
4103         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
4104         (simple_return, *simple_return): New patterns
4105         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
4106         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
4107
4108 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
4109
4110         * dwarf2out.c (gen_subprogram_die): Tidy.
4111
4112 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
4113
4114         PR target/60071
4115         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
4116         (*mov_t_msb_neg_negc): ... this new insn.
4117
4118 2014-02-28  Jason Merrill  <jason@redhat.com>
4119
4120         PR c++/58678
4121         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
4122         function.
4123
4124 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
4125
4126         PR c++/60314
4127         * dwarf2out.c (decltype_auto_die): New static.
4128         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
4129         (gen_type_die_with_usage): Handle 'decltype(auto)'.
4130         (is_cxx_auto): Likewise.
4131
4132 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
4133
4134         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
4135         we are not using general regs only.
4136
4137 2014-02-28  Richard Biener  <rguenther@suse.de>
4138
4139         PR target/60280
4140         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
4141         previous fix and only allow to remove trivial pre-headers
4142         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
4143         (remove_forwarder_block): Properly update the latch of a loop.
4144
4145 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4146
4147         PR debug/59992
4148         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
4149         (cselib_preserved_hash_table): New.
4150         (preserve_constants_and_equivs): Move preserved vals to it.
4151         (cselib_find_slot): Look it up first.
4152         (cselib_init): Initialize it.
4153         (cselib_finish): Release it.
4154         (dump_cselib_table): Dump it.
4155
4156 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4157
4158         PR debug/59992
4159         * cselib.c (remove_useless_values): Skip to avoid quadratic
4160         behavior if the condition moved from...
4161         (cselib_process_insn): ... here holds.
4162
4163 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4164
4165         PR debug/57232
4166         * var-tracking.c (vt_initialize): Apply the same condition to
4167         preserve the CFA base value.
4168
4169 2014-02-28  Joey Ye  <joey.ye@arm.com>
4170
4171         PR target/PR60169
4172         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
4173         if reload in progress or completed.
4174
4175 2014-02-28  Tobias Burnus  <burnus@net-b.de>
4176
4177         PR middle-end/60147
4178         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
4179         NAMELIST_DECL.
4180
4181 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
4182
4183         * doc/tm.texi.in (Condition Code Status): Update documention for
4184         relative locations of cc0-setter and cc0-user.
4185
4186 2014-02-27  Jeff Law  <law@redhat.com>
4187
4188         PR rtl-optimization/52714
4189         * combine.c (try_combine): When splitting an unrecognized PARALLEL
4190         into two independent simple sets, if I3 is a jump, ensure the
4191         pattern we place into I3 is a (set (pc) ...).
4192
4193 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
4194             Jeff Law  <law@redhat.com>
4195
4196         PR rtl-optimization/49847
4197         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
4198         are in different blocks.
4199         * doc/tm.texi (Condition Code Status): Update documention for
4200         relative locations of cc0-setter and cc0-user.
4201
4202 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
4203
4204         PR target/59222
4205         * lra.c (lra_emit_add): Check SUBREG too.
4206
4207 2014-02-27  Andreas Schwab  <schwab@suse.de>
4208
4209         * config/m68k/m68k.c (m68k_option_override): Disable
4210         -flive-range-shrinkage for classic m68k.
4211         (m68k_override_options_after_change): Likewise.
4212
4213 2014-02-27  Marek Polacek  <polacek@redhat.com>
4214
4215         PR middle-end/59223
4216         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
4217         -Wmaybe-uninitialized.
4218
4219 2014-02-27  Alan Modra  <amodra@gmail.com>
4220
4221         PR target/57936
4222         * reload1.c (emit_input_reload_insns): When reload_override_in,
4223         set old to rl->in_reg when rl->in_reg is a subreg.
4224
4225 2014-02-26  Richard Biener  <rguenther@suse.de>
4226
4227         PR bootstrap/60343
4228         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
4229
4230 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4231
4232         * common/config/i386/predicates.md (const1256_operand): Remove.
4233         (const2356_operand): New.
4234         (const_1_to_2_operand): Remove.
4235         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4236         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4237         (*avx512pf_gatherpf<mode>sf): Ditto.
4238         (avx512pf_gatherpf<mode>df): Ditto.
4239         (*avx512pf_gatherpf<mode>df_mask): Ditto.
4240         (*avx512pf_gatherpf<mode>df): Ditto.
4241         (avx512pf_scatterpf<mode>sf): Ditto.
4242         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4243         (*avx512pf_scatterpf<mode>sf): Ditto.
4244         (avx512pf_scatterpf<mode>df): Ditto.
4245         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4246         (*avx512pf_scatterpf<mode>df): Ditto.
4247         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
4248
4249 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
4250
4251         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
4252         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
4253         (_mm512_mask_testn_epi64_mask): Move to ...
4254         * config/i386/avx512cdintrin.h: Here.
4255         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
4256         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
4257         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
4258         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
4259         TARGET_AVX512F from TARGET_AVX512CD.
4260
4261 2014-02-26  Richard Biener  <rguenther@suse.de>
4262
4263         PR ipa/60327
4264         * ipa.c (walk_polymorphic_call_targets): Properly guard
4265         call to inline_update_overall_summary.
4266
4267 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
4268
4269         PR target/60280
4270         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
4271         and latches only if requested.  Fix latch if it is removed.
4272         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
4273         LOOPS_HAVE_PREHEADERS.
4274
4275 2014-02-25  Andrew Pinski  <apinski@cavium.com>
4276
4277         * builtins.c (expand_builtin_thread_pointer): Create a new target
4278         when the target is NULL.
4279
4280 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
4281
4282         PR rtl-optimization/60317
4283         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4284         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4285         * lra-assigns.c: Include params.h.
4286         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
4287         other reload pseudos considerations.
4288
4289 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4290
4291         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
4292         to use canonical form for nor<mode>3.
4293
4294 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4295
4296         PR target/55426
4297         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
4298         conversions.
4299
4300 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4301
4302         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
4303         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
4304         (ix86_handle_option): Handle OPT_mprefetchwt1.
4305         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
4306         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4307         PREFETCHWT1 CPUID.
4308         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4309         OPTION_MASK_ISA_PREFETCHWT1.
4310         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
4311         (PTA_PREFETCHWT1): New.
4312         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
4313         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
4314         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
4315         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
4316         (*prefetch_avx512pf_<mode>_: Change into ...
4317         (*prefetch_prefetchwt1_<mode>: This.
4318         * config/i386/i386.opt (mprefetchwt1): New.
4319         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
4320         (_mm_prefetch): Handle intent to write.
4321         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
4322
4323 2014-02-25  Richard Biener  <rguenther@suse.de>
4324
4325         PR middle-end/60291
4326         * emit-rtl.c (mem_attrs_htab): Remove.
4327         (mem_attrs_htab_hash): Likewise.
4328         (mem_attrs_htab_eq): Likewise.
4329         (set_mem_attrs): Always allocate new mem-attrs when something changed.
4330         (init_emit_once): Do not allocate mem_attrs_htab.
4331
4332 2014-02-25  Richard Biener  <rguenther@suse.de>
4333
4334         PR lto/60319
4335         * lto-opts.c (lto_write_options): Output non-explicit conservative
4336         -fwrapv, -fno-trapv and -fno-strict-overflow.
4337         * lto-wrapper.c (merge_and_complain): Handle merging those options.
4338         (run_gcc): And pass them through.
4339
4340 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4341
4342         * sel-sched.c (calculate_new_fences): New parameter ptime.
4343         Calculate it as a maximum over all fence cycles.
4344         (sel_sched_region_2): Adjust the call to calculate_new_fences.
4345         Print the final schedule timing when sched_verbose.
4346
4347 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4348
4349         PR rtl-optimization/60292
4350         * sel-sched.c (fill_vec_av_set): Do not reset target availability
4351         bit fot the fence instruction.
4352
4353 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
4354
4355         * calls.h: Fix typo in comment.
4356
4357 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
4358
4359         * config/pa/pa.c (pa_output_move_double): Don't valididate when
4360         adjusting offsetable addresses.
4361
4362 2014-02-24  Guozhi Wei  <carrot@google.com>
4363
4364         * sparseset.h (sparseset_pop): Fix the wrong index.
4365
4366 2014-02-24  Walter Lee  <walt@tilera.com>
4367
4368         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
4369         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
4370         triplet.
4371         * common/config/tilegx/tilegx-common.c
4372         (TARGET_DEFAULT_TARGET_FLAGS): Define.
4373         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
4374         (LINK_SPEC): Ditto.
4375         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
4376         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
4377         (tilegx_gimplify_va_arg_expr): Handle big endian.
4378         (tilegx_expand_unaligned_load): Ditto.
4379         (tilegx_expand_unaligned_store): Ditto.
4380         (TARGET_RETURN_IN_MSB): New.
4381         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
4382         (TARGET_ENDIAN_DEFAULT): New.
4383         (TARGET_BIG_ENDIAN): Handle big endian.
4384         (BYTES_BIG_ENDIAN): Ditto.
4385         (WORDS_BIG_ENDIAN): Ditto.
4386         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
4387         (ENDIAN_SPEC): New.
4388         (EXTRA_SPECS): New.
4389         * config/tilegx/tilegx.md (extv): Handle big endian.
4390         (extzv): Ditto.
4391         (insn_st<n>): Ditto.
4392         (insn_st<n>_add<bitsuffix>): Ditto.
4393         (insn_stnt<n>): Ditto.
4394         (insn_stnt<n>_add<bitsuffix>):Ditto.
4395         (vec_interleave_highv8qi): Handle big endian.
4396         (vec_interleave_highv8qi_be): New.
4397         (vec_interleave_highv8qi_le): New.
4398         (insn_v1int_h): Handle big endian.
4399         (vec_interleave_lowv8qi): Handle big endian.
4400         (vec_interleave_lowv8qi_be): New.
4401         (vec_interleave_lowv8qi_le): New.
4402         (insn_v1int_l): Handle big endian.
4403         (vec_interleave_highv4hi): Handle big endian.
4404         (vec_interleave_highv4hi_be): New.
4405         (vec_interleave_highv4hi_le): New.
4406         (insn_v2int_h): Handle big endian.
4407         (vec_interleave_lowv4hi): Handle big endian.
4408         (vec_interleave_lowv4hi_be): New.
4409         (vec_interleave_lowv4hi_le): New.
4410         (insn_v2int_l): Handle big endian.
4411         (vec_interleave_highv2si): Handle big endian.
4412         (vec_interleave_highv2si_be): New.
4413         (vec_interleave_highv2si_le): New.
4414         (insn_v4int_h): Handle big endian.
4415         (vec_interleave_lowv2si): Handle big endian.
4416         (vec_interleave_lowv2si_be): New.
4417         (vec_interleave_lowv2si_le): New.
4418         (insn_v4int_l): Handle big endian.
4419         * config/tilegx/tilegx.opt (mbig-endian): New option.
4420         (mlittle-endian): New option.
4421         * doc/install.texi: Document tilegxbe-linux.
4422         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
4423
4424 2014-02-24  Martin Jambor  <mjambor@suse.cz>
4425
4426         PR ipa/60266
4427         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
4428         there are no parameter descriptors.
4429
4430 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
4431
4432         PR rtl-optimization/60268
4433         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
4434         initialization to ...
4435         (sched_rgn_init): ... here.
4436         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
4437
4438 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4439
4440         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
4441         names.
4442
4443 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
4444
4445         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
4446         definition.
4447
4448 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4449
4450         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
4451         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
4452
4453 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4454
4455         * config/microblaze/predicates.md: Add cmp_op predicate.
4456         * config/microblaze/microblaze.md: Add branch_compare instruction
4457         which uses cmp_op predicate and emits cmp insn before branch.
4458         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
4459         to microblaze_expand_conditional_branch and consolidate logic.
4460         (microblaze_expand_conditional_branch): emit branch_compare
4461         insn instead of handling cmp op separate from branch insn.
4462
4463 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4464
4465         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
4466         to permit subregs.
4467
4468 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4469
4470         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
4471         define_insn with define_expand and new define_insn
4472         *altivec_lve<VI_char>x_internal.
4473         (altivec_stve<VI_char>x): Replace define_insn with define_expand
4474         and new define_insn *altivec_stve<VI_char>x_internal.
4475         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
4476         prototype.
4477         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
4478         lve*x built-ins.
4479         (altivec_expand_stvex_be): New function.
4480
4481 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
4482
4483         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
4484         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
4485         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
4486         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
4487
4488 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
4489
4490         PR target/60298
4491         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
4492         instead of emit_move_insn.
4493
4494 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4495
4496         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
4497         vspltw with vsldoi.
4498         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
4499         gen_altivec_vsumsws.
4500
4501 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4502
4503         * config/rs6000/altivec.md (altivec_lvxl): Rename as
4504         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
4505         (altivec_lvxl_<mode>): New define_expand incorporating
4506         -maltivec=be semantics where needed.
4507         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
4508         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
4509         semantics where needed.
4510         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
4511         (altivec_stvx_<mode>): New define_expand incorporating
4512         -maltivec=be semantics where needed.
4513         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
4514         VM2 iterator instead of V4SI.
4515         (altivec_stvxl_<mode>): New define_expand incorporating
4516         -maltivec=be semantics where needed.
4517         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
4518         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
4519         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
4520         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
4521         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
4522         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
4523         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
4524         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
4525         ALTIVEC_BUILTIN_STVXL.
4526         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
4527         (altivec_expand_stvx_be): Likewise.
4528         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
4529         (altivec_expand_lvx_be): Likewise.
4530         (altivec_expand_stvx_be): Likewise.
4531         (altivec_expand_builtin): Add cases for
4532         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
4533         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
4534         (altivec_init_builtins): Add definitions for
4535         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
4536         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
4537
4538 2014-02-21  Catherine Moore  <clm@codesourcery.com>
4539
4540         * doc/invoke.texi (mvirt, mno-virt): Document.
4541         * config/mips/mips.opt (mvirt): New option.
4542         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
4543
4544 2014-02-21  Richard Biener  <rguenther@suse.de>
4545
4546         PR tree-optimization/60276
4547         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
4548         (STMT_VINFO_MIN_NEG_DIST): New macro.
4549         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
4550         STMT_VINFO_MIN_NEG_DIST.
4551         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
4552         made for negative dependence distances still hold.
4553
4554 2014-02-21  Richard Biener  <rguenther@suse.de>
4555
4556         PR middle-end/60291
4557         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
4558         DECL_INITIAL for globals not in the current function context.
4559
4560 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4561
4562         PR tree-optimization/56490
4563         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
4564         * tree-ssa-uninit.c: Include params.h.
4565         (compute_control_dep_chain): Add num_calls argument, return false
4566         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
4567         num_calls to recursive call.
4568         (find_predicates): Change dep_chain into normal array,
4569         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
4570         variable and adjust compute_control_dep_chain caller.
4571         (find_def_preds): Likewise.
4572
4573 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
4574
4575         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4576         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
4577
4578 2014-02-21  Nick Clifton  <nickc@redhat.com>
4579
4580         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
4581         (pushhi1): Likewise.
4582         (popqi1): Add mode to pre_dec.
4583         (pophi1): Likewise.
4584
4585 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4586
4587         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
4588         mode for mask of V8SFmode permutation.
4589
4590 2014-02-20  Richard Henderson  <rth@redhat.com>
4591
4592         PR c++/60272
4593         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
4594         a new pseudo for OLDVAL.
4595
4596 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
4597
4598         PR target/57896
4599         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
4600         gen_reg_rtx if d->testing_p.
4601         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
4602         if d->testing_p and we will certainly return true.
4603         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
4604         if d->testing_p.
4605
4606 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
4607
4608         * emit-rtl.c (gen_reg_rtx): Assert that
4609         crtl->emit.regno_pointer_align_length is non-zero.
4610
4611 2014-02-20  Richard Henderson  <rth@redhat.com>
4612
4613         PR c++/60272
4614         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
4615         on failure the store back into EXPECT.
4616
4617 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
4618             Sandra Loosemore  <sandra@codesourcery.com>
4619
4620         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
4621         * config/nios2/nios2.c (nios2_function_profiler): Add
4622         -fPIC (flag_pic == 2) support.
4623         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
4624         (nios2_large_offset_p): New function.
4625         (nios2_unspec_reloc_p): Move up position, update to use
4626         nios2_large_offset_p.
4627         (nios2_unspec_address): Remove function.
4628         (nios2_unspec_offset): New function.
4629         (nios2_large_got_address): New function.
4630         (nios2_got_address): Add large offset support.
4631         (nios2_legitimize_tls_address): Update usage of removed and new
4632         functions.
4633         (nios2_symbol_binds_local_p): New function.
4634         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
4635         (nios2_legitimize_address): Update to use nios2_large_offset_p.
4636         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
4637         (nios2_print_operand): Merge H/L processing, add hiadj/lo
4638         processing for (const (unspec ...)).
4639         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
4640
4641 2014-02-20  Richard Biener  <rguenther@suse.de>
4642
4643         * tree-cfg.c (replace_uses_by): Mark altered BBs before
4644         doing the substitution.
4645         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
4646
4647 2014-02-20  Martin Jambor  <mjambor@suse.cz>
4648
4649         PR ipa/55260
4650         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
4651         info when checking whether lattices are bottom.
4652
4653 2014-02-20  Richard Biener  <rguenther@suse.de>
4654
4655         PR middle-end/60221
4656         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
4657         regions at -O0.
4658
4659 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
4660
4661         PR ipa/58555
4662         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
4663         parameter specifying the scaling.
4664         (inline_call): Update.
4665         (want_inline_recursively): Guard division by zero.
4666         (recursive_inlining): Update.
4667         * ipa-inline.h (clone_inlined_nodes): Update.
4668
4669 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4670
4671         PR target/60204
4672         * config/i386/i386.c (classify_argument): Pass structures of size
4673         64 bytes or less in register.
4674
4675 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4676             Kirill Yukhin  <kirill.yukhin@intel.com>
4677
4678         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
4679         (_mm_rcp28_round_ss): Ditto.
4680         (_mm_rsqrt28_round_sd): Ditto.
4681         (_mm_rsqrt28_round_ss): Ditto.
4682         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
4683         (_mm_rcp14_round_ss): Ditto.
4684         (_mm_rsqrt14_round_sd): Ditto.
4685         (_mm_rsqrt14_round_ss): Ditto.
4686         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
4687         the first input operand, get rid of match_dup.
4688         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
4689         attribute to sse.
4690         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4691         Ditto.
4692         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
4693         operand as the first input operand, set type attribute.
4694         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4695         Set type attribute.
4696         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
4697         operand as the first input operand, set type attribute.
4698
4699 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4700
4701         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
4702         bit of zero.
4703
4704 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
4705
4706         PR target/60207
4707         * config/i386/i386.c (construct_container): Remove TFmode check
4708         for X86_64_INTEGER_CLASS.
4709
4710 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
4711
4712         PR target/59794
4713         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
4714         only when -Wpsabi is enabled.
4715
4716 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
4717
4718          PR target/59799
4719         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
4720         passing arrays in registers are the same as for structs, so remove the
4721         special case for them.
4722
4723 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
4724
4725         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
4726         destination type, extract only the valid bits if the source type is not
4727         integral and has a different mode.
4728
4729 2014-02-19  Richard Biener  <rguenther@suse.de>
4730
4731         PR ipa/60243
4732         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
4733         for all calls.
4734
4735 2014-02-19  Richard Biener  <rguenther@suse.de>
4736
4737         PR ipa/60243
4738         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
4739         (ipa_modify_call_arguments): Emit an argument load explicitely and
4740         preserve virtual SSA form there and for the replacement call.
4741         Do not update SSA form nor free dominance info.
4742
4743 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4744
4745         * ipa.c (function_and_variable_visibility): Also clear WEAK
4746         flag when disolving COMDAT_GROUP.
4747
4748 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4749
4750         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
4751         * ipa-prop.c (ipa_set_jf_known_type): Return early when
4752         not devirtualizing.
4753         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
4754         do more sanity checks.
4755         (detect_type_change): Return true when giving up early.
4756         (compute_complex_assign_jump_func): Fix type parameter of
4757         ipa_set_ancestor_jf.
4758         (compute_complex_ancestor_jump_func): Likewise.
4759         (update_jump_functions_after_inlining): Fix updating of
4760         ancestor function.
4761         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
4762
4763 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4764
4765         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
4766         inline clones when edge disappears.
4767
4768 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4769
4770         PR target/60203
4771         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
4772         Split 64-bit moves into 2 patterns.  Do not allow the use of
4773         direct move for TDmode in little endian, since the decimal value
4774         has little endian bytes within a word, but the 64-bit pieces are
4775         ordered in a big endian fashion, and normal subreg's of TDmode are
4776         not allowed.
4777         (mov<mode>_64bit_dm): Likewise.
4778         (movtd_64bit_nodm): Likewise.
4779
4780 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4781
4782         PR tree-optimization/60174
4783         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
4784         statement of an SSA_NAME that occurs in an abnormal PHI node.
4785
4786 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4787
4788         PR sanitizer/60142
4789         * final.c (SEEN_BB): Remove.
4790         (SEEN_NOTE, SEEN_EMITTED): Renumber.
4791         (final_scan_insn): Don't force_source_line on second
4792         NOTE_INSN_BASIC_BLOCK.
4793
4794 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
4795
4796         PR target/60205
4797         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
4798         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
4799         (type_natural_mode): Warn ABI change when %zmm register is not
4800         available for AVX512F vector value passing.
4801
4802 2014-02-18  Kai Tietz  <ktietz@redhat.com>
4803
4804         PR target/60193
4805         * config/i386/i386.c (ix86_expand_prologue): Use value in
4806         rax register as displacement when restoring %r10 or %rax.
4807         Fix wrong offset when restoring both registers.
4808
4809 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4810
4811         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
4812         assertion with conditional return.
4813
4814 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4815             Uros Bizjak  <ubizjak@gmail.com>
4816
4817         PR driver/60233
4818         * config/i386/driver-i386.c (host_detect_local_cpu): If
4819         YMM state is not saved by the OS, also clear has_f16c.  Move
4820         CPUID 0x80000001 handling before YMM state saving checking.
4821
4822 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
4823
4824         PR rtl-optimization/58960
4825         * haifa-sched.c (alloc_global_sched_pressure_data): New,
4826         factored out from ...
4827         (sched_init): ... here.
4828         (free_global_sched_pressure_data): New, factored out from ...
4829         (sched_finish): ... here.
4830         * sched-int.h (free_global_sched_pressure_data): Declare.
4831         * sched-rgn.c (nr_regions_initial): New static global.
4832         (haifa_find_rgns): Initialize it.
4833         (schedule_region): Disable sched-pressure for the newly
4834         generated regions.
4835
4836 2014-02-17  Richard Biener  <rguenther@suse.de>
4837
4838         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
4839         release SSA defs of pattern stmts.
4840
4841 2014-02-17  Richard Biener  <rguenther@suse.de>
4842
4843         * tree-inline.c (expand_call_inline): Release the virtual
4844         operand defined by the call we are about to inline.
4845
4846 2014-02-17  Richard Biener  <rguenther@suse.de>
4847
4848         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
4849
4850 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
4851             Ilya Tocar  <ilya.tocar@intel.com>
4852
4853         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
4854         arguments order in builtin.
4855         (_mm512_permutexvar_epi64): Ditto.
4856         (_mm512_mask_permutexvar_epi64): Ditto
4857         (_mm512_maskz_permutexvar_epi32): Ditto
4858         (_mm512_permutexvar_epi32): Ditto
4859         (_mm512_mask_permutexvar_epi32): Ditto
4860
4861 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4862
4863         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
4864         (p8_vmrgow): Likewise.
4865
4866 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4867
4868         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
4869         endian targets.
4870
4871 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
4872
4873         PR target/60203
4874         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
4875         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
4876         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
4877         using direct move instructions on ISA 2.07.  Also adjust
4878         instruction length for 64-bit.
4879         (mov<mode>_64bit, TFmode/TDmode): Likewise.
4880         (mov<mode>_32bit, TFmode/TDmode): Likewise.
4881
4882 2014-02-15  Alan Modra  <amodra@gmail.com>
4883
4884         PR target/58675
4885         PR target/57935
4886         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
4887         find_replacement on parts of insn rtl that might be reloaded.
4888
4889 2014-02-15  Richard Biener  <rguenther@suse.de>
4890
4891         PR tree-optimization/60183
4892         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
4893         (tree_ssa_phiprop): Calculate and free post-dominators.
4894
4895 2014-02-14  Jeff Law  <law@redhat.com>
4896
4897         PR rtl-optimization/60131
4898         * ree.c (get_extended_src_reg): New function.
4899         (combine_reaching_defs): Use it rather than assuming location of REG.
4900         (find_and_remove_re): Verify first operand of extension is
4901         a REG before adding the insns to the copy list.
4902
4903 2014-02-14  Roland McGrath  <mcgrathr@google.com>
4904
4905         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
4906         * configure: Regenerated.
4907         * config.in: Regenerated.
4908         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
4909         instead of ASM_SHORT.
4910
4911 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
4912             Richard Earnshaw  <rearnsha@arm.com>
4913
4914         PR rtl-optimization/59535
4915         * lra-constraints.c (process_alt_operands): Encourage alternative
4916         when unassigned pseudo class is superset of the alternative class.
4917         (inherit_reload_reg): Don't inherit when optimizing for code size.
4918         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
4919         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
4920         modes not less than 4 for Thumb1.
4921
4922 2014-02-14  Kyle McMartin  <kyle@redhat.com>
4923
4924         PR pch/60010
4925         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
4926
4927 2014-02-14  Richard Biener  <rguenther@suse.de>
4928
4929         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4930         (get_frame_arg): Drop the assert with langhook types_compatible_p.
4931         Do not strip INDIRECT_REFs.
4932
4933 2014-02-14  Richard Biener  <rguenther@suse.de>
4934
4935         PR lto/60179
4936         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
4937         DECL_FUNCTION_SPECIFIC_TARGET.
4938         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
4939         * tree-streamer-out.c (pack_ts_target_option): Remove.
4940         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
4941         (write_ts_function_decl_tree_pointers): Do not stream
4942         DECL_FUNCTION_SPECIFIC_TARGET.
4943         * tree-streamer-in.c (unpack_ts_target_option): Remove.
4944         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
4945         (lto_input_ts_function_decl_tree_pointers): Do not stream
4946         DECL_FUNCTION_SPECIFIC_TARGET.
4947
4948 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
4949
4950         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
4951         (get_initial_def_for_induction, vectorizable_induction): Ignore
4952         debug stmts when looking for exit_phi.
4953         (vectorizable_live_operation): Fix up condition.
4954
4955 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4956
4957         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4958         nreverse() because it changes the content of original tree list.
4959
4960 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4961
4962         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
4963         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
4964
4965 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4966
4967         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4968         GNU coding standards.
4969
4970 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4971
4972         PR debug/60152
4973         * dwarf2out.c (gen_subprogram_die): Don't call
4974         add_calling_convention_attribute if subr_die is old_die.
4975
4976 2014-02-13  Sharad Singhai  <singhai@google.com>
4977
4978         * doc/optinfo.texi: Fix order of nodes.
4979
4980 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
4981
4982         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4983         operands[2], not operands[3].
4984
4985 2014-02-13  Richard Biener  <rguenther@suse.de>
4986
4987         PR bootstrap/59878
4988         * doc/install.texi (ISL): Update recommended version to 0.12.2,
4989         mention the possibility of an in-tree build.
4990         (CLooG): Update recommended version to 0.18.1, mention the
4991         possibility of an in-tree build and clarify that the ISL
4992         bundled with CLooG does not work.
4993
4994 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4995
4996         PR target/43546
4997         * expr.c (compress_float_constant): If x is a hard register,
4998         extend into a pseudo and then move to x.
4999
5000 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5001
5002         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
5003         caused by bad second argument to warning_at() with -mhotpatch and
5004         nested functions (e.g. with gfortran).
5005
5006 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
5007
5008         * opts.c (option_name): Remove "enabled by default" rider.
5009
5010 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
5011
5012         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
5013
5014 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
5015             Uros Bizjak  <ubizjak@gmail.com>
5016
5017         PR target/60151
5018         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
5019         * configure: Regenerated.
5020
5021 2014-02-12  Richard Biener  <rguenther@suse.de>
5022
5023         * vec.c (vec_prefix::calculate_allocation): Move as
5024         inline variant to vec.h.
5025         (vec_prefix::calculate_allocation_1): New out-of-line version.
5026         * vec.h (vec_prefix::calculate_allocation_1): Declare.
5027         (vec_prefix::m_has_auto_buf): Rename to ...
5028         (vec_prefix::m_using_auto_storage): ... this.
5029         (vec_prefix::calculate_allocation): Inline the easy cases
5030         and dispatch to calculate_allocation_1 which doesn't need the
5031         prefix address.
5032         (va_heap::reserve): Use gcc_checking_assert.
5033         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
5034         m_using_auto_storage.
5035         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
5036         member and adjust.
5037         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
5038         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
5039         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
5040
5041 2014-02-12  Richard Biener  <rguenther@suse.de>
5042
5043         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
5044         when we found a dependence.
5045
5046 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
5047
5048         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
5049         common code...
5050         (maybe_fold_stmt): ... into this new function.
5051         * omp-low.c (lower_omp): Update comment.
5052
5053         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
5054         last use.
5055
5056         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
5057         dereference.
5058
5059 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
5060
5061         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
5062         identifiers in comments.
5063         (cortexa53_extra_costs): Likewise.
5064         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
5065         (cortexa7_extra_costs): Likewise.
5066         (cortexa12_extra_costs): Likewise.
5067         (cortexa15_extra_costs): Likewise.
5068         (v7m_extra_costs): Likewise.
5069
5070 2014-02-12  Richard Biener  <rguenther@suse.de>
5071
5072         PR middle-end/60092
5073         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
5074         of posix_memalign being successful.
5075         (lower_stmt): Restrict lowering of posix_memalign to when
5076         -ftree-bit-ccp is enabled.
5077
5078 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5079
5080         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
5081         arg_loc.
5082         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
5083
5084 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
5085
5086         PR rtl-optimization/60116
5087         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
5088         other_insn once the combination has been validated.
5089
5090 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
5091
5092         PR lto/59468
5093         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
5094         and wrapper.
5095         * ipa-devirt.c: Include demangle.h
5096         (odr_violation_reported): New static variable.
5097         (add_type_duplicate): Update odr_violations.
5098         (maybe_record_node): Add completep parameter; update it.
5099         (record_target_from_binfo): Add COMPLETEP parameter;
5100         update it as needed.
5101         (possible_polymorphic_call_targets_1): Likewise.
5102         (struct polymorphic_call_target_d): Add nonconstruction_targets;
5103         rename FINAL to COMPLETE.
5104         (record_targets_from_bases): Sanity check we found the binfo;
5105         fix COMPLETEP updating.
5106         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
5107         parameter, fix computing of COMPLETEP.
5108         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
5109         at LTO time do demangling.
5110         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
5111         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
5112         parameter.
5113         (gimple_get_virt_method_for_binfo): Likewise.
5114         * gimple-fold.h (gimple_get_virt_method_for_binfo,
5115         gimple_get_virt_method_for_vtable): Update prototypes.
5116
5117 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
5118
5119         PR target/49008
5120         * genautomata.c (add_presence_absence): Fix typo with
5121         {final_}presence_list.
5122
5123 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5124
5125         PR target/60137
5126         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
5127         for VSX/Altivec vectors that land in GPR registers.
5128
5129 2014-02-11  Richard Henderson  <rth@redhat.com>
5130             Jakub Jelinek  <jakub@redhat.com>
5131
5132         PR debug/59776
5133         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
5134         around drhs if type conversion to lacc->type is not useless.
5135
5136 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5137
5138         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
5139         tuning struct.
5140         (cortex-a57.cortex-a53): Likewise.
5141         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
5142
5143 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5144
5145         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
5146         arm_restrict_it.
5147
5148 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
5149
5150         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
5151         add_options_for_arm_vfp3.
5152
5153 2014-02-11  Jeff Law  <law@redhat.com>
5154
5155         PR middle-end/54041
5156         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
5157         object with an undesirable mode.
5158
5159 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5160
5161         PR libgomp/60107
5162         * config/i386/sol2-9.h: New file.
5163         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
5164         *-*-solaris2.9*): Use it.
5165
5166 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5167
5168         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
5169         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
5170
5171 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5172
5173         * config/microblaze/microblaze.c: Extend mcpu version format
5174
5175 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
5176
5177         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
5178
5179 2014-02-10  Richard Henderson  <rth@redhat.com>
5180
5181         PR target/59927
5182         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
5183         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
5184         ms-abi vs -mno-accumulate-outgoing-args.
5185         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
5186         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
5187         respect to ms-abi.
5188
5189 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5190
5191         PR middle-end/60080
5192         * cfgexpand.c (expand_asm_operands): Attach source location to
5193         ASM_INPUT rtx objects.
5194         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
5195
5196 2014-02-10  Nick Clifton  <nickc@redhat.com>
5197
5198         * config/mn10300/mn10300.c (popcount): New function.
5199         (mn10300_expand_prologue): Include saved registers in stack usage
5200         count.
5201
5202 2014-02-10  Jeff Law  <law@redhat.com>
5203
5204         PR middle-end/52306
5205         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
5206         when changing the SET_DEST of a prior insn to avoid an input reload.
5207
5208 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5209
5210         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
5211         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
5212         -mcall-openbsd, or -mcall-linux.
5213         (CC1_ENDIAN_BIG_SPEC): Remove.
5214         (CC1_ENDIAN_LITTLE_SPEC): Remove.
5215         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5216         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
5217         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
5218         and %cc1_endian_default.
5219         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5220
5221 2014-02-10  Richard Biener  <rguenther@suse.de>
5222
5223         PR tree-optimization/60115
5224         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
5225         MEM_REF handling.  Properly verify that the accesses are not
5226         out of the objects bound.
5227
5228 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5229
5230         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
5231         coretex to cortex.
5232
5233 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
5234
5235         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
5236         proper constants and fix formatting.
5237         (possible_polymorphic_call_targets): Fix formatting.
5238
5239 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
5240             Ilya Tocar  <ilya.tocar@intel.com>
5241
5242         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
5243         (_mm512_loadu_epi32): Renamed into...
5244         (_mm512_loadu_si512): This.
5245         (_mm512_storeu_epi32): Renamed into...
5246         (_mm512_storeu_si512): This.
5247         (_mm512_maskz_ceil_ps): Removed.
5248         (_mm512_maskz_ceil_pd): Ditto.
5249         (_mm512_maskz_floor_ps): Ditto.
5250         (_mm512_maskz_floor_pd): Ditto.
5251         (_mm512_floor_round_ps): Ditto.
5252         (_mm512_floor_round_pd): Ditto.
5253         (_mm512_ceil_round_ps): Ditto.
5254         (_mm512_ceil_round_pd): Ditto.
5255         (_mm512_mask_floor_round_ps): Ditto.
5256         (_mm512_mask_floor_round_pd): Ditto.
5257         (_mm512_mask_ceil_round_ps): Ditto.
5258         (_mm512_mask_ceil_round_pd): Ditto.
5259         (_mm512_maskz_floor_round_ps): Ditto.
5260         (_mm512_maskz_floor_round_pd): Ditto.
5261         (_mm512_maskz_ceil_round_ps): Ditto.
5262         (_mm512_maskz_ceil_round_pd): Ditto.
5263         (_mm512_expand_pd): Ditto.
5264         (_mm512_expand_ps): Ditto.
5265         * config/i386/i386.c (ix86_builtins): Remove
5266         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
5267         (bdesc_args): Ditto.
5268         * config/i386/predicates.md (const1256_operand): New.
5269         (const_1_to_2_operand): Ditto.
5270         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
5271         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5272         (*avx512pf_gatherpf<mode>sf): Ditto.
5273         (avx512pf_gatherpf<mode>df): Ditto.
5274         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5275         (*avx512pf_gatherpf<mode>df): Ditto.
5276         (avx512pf_scatterpf<mode>sf): Ditto.
5277         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5278         (*avx512pf_scatterpf<mode>sf): Ditto.
5279         (avx512pf_scatterpf<mode>df): Ditto.
5280         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5281         (*avx512pf_scatterpf<mode>df): Ditto.
5282         (avx512f_expand<mode>): Removed.
5283         (<shift_insn><mode>3<mask_name>): Change predicate type.
5284
5285 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5286
5287         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
5288         not at the end of datarefs vector use ordered_remove to avoid
5289         reordering datarefs vector.
5290
5291         PR c/59984
5292         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
5293         mark local addressable non-static vars as GOVD_PRIVATE
5294         instead of GOVD_LOCAL.
5295         * omp-low.c (lower_omp_for): Move gimple_bind_vars
5296         and BLOCK_VARS of gimple_bind_block to new_stmt rather
5297         than copying them.
5298
5299         PR middle-end/60092
5300         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
5301         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
5302         assume_aligned or alloc_align attributes.
5303         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
5304         arguments.  Handle also assume_aligned and alloc_align attributes.
5305         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
5306         calls to functions with assume_aligned or alloc_align attributes.
5307         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
5308
5309 2014-02-08  Terry Guo  <terry.guo@arm.com>
5310
5311         * doc/invoke.texi: Document ARM -march=armv7e-m.
5312
5313 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5314
5315         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
5316         flag on __cilkrts_rethrow builtin.
5317
5318         PR ipa/60026
5319         * ipa-cp.c (determine_versionability): Fail at -O0
5320         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
5321         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
5322
5323         Revert:
5324         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5325
5326         PR ipa/60026
5327         * tree-inline.c (copy_forbidden): Fail for
5328         __attribute__((optimize (0))) functions.
5329
5330 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5331
5332         * varpool.c: Include pointer-set.h.
5333         (varpool_remove_unreferenced_decls): Variables in other partitions
5334         will not be output; be however careful to not lose information
5335         about partitioning.
5336
5337 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5338
5339         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
5340         lookup in the vtable constructor.
5341
5342 2014-02-07  Jeff Law  <law@redhat.com>
5343
5344         PR target/40977
5345         * config/m68k/m68k.md (ashldi_extsi): Turn into a
5346         define_insn_and_split.
5347
5348         * ipa-inline.c (inline_small_functions): Fix typos.
5349
5350 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5351
5352         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
5353         (s390_can_use_return_insn): Declare.
5354         * config/s390/s390.h (EPILOGUE_USES): Define.
5355         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
5356         instructions.
5357         (s390_chunkify_start): Handle return JUMP_LABELs.
5358         (s390_early_mach): Emit a main_pool instruction on the entry edge.
5359         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
5360         (s390_can_use_return_insn): New functions.
5361         (s390_fix_long_loop_prediction): Handle conditional returns.
5362         (TARGET_SET_UP_BY_PROLOGUE): Define.
5363         * config/s390/s390.md (ANY_RETURN): New code iterator.
5364         (*creturn, *csimple_return, return, simple_return): New patterns.
5365
5366 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5367
5368         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
5369         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
5370         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
5371         REG_CFA_RESTORE list when deciding not to restore a register.
5372
5373 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5374
5375         * config/s390/s390.c: Include tree-pass.h and context.h.
5376         (s390_early_mach): New function, split out from...
5377         (s390_emit_prologue): ...here.
5378         (pass_data_s390_early_mach): New pass structure.
5379         (pass_s390_early_mach): New class.
5380         (s390_option_override): Create and register early_mach pass.
5381         Move to end of file.
5382
5383 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5384
5385         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
5386         to match for the exit block.
5387
5388 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5389
5390         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
5391         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
5392         Reject misaligned operands.
5393
5394 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5395
5396         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
5397
5398 2014-02-07  Richard Biener  <rguenther@suse.de>
5399
5400         PR middle-end/60092
5401         * gimple-low.c (lower_builtin_posix_memalign): New function.
5402         (lower_stmt): Call it to lower posix_memalign in a way
5403         to make alignment info accessible.
5404
5405 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
5406
5407         PR c++/60082
5408         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
5409         __builtin_setjmp_receiver.
5410
5411 2014-02-07  Richard Biener  <rguenther@suse.de>
5412
5413         PR middle-end/60092
5414         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
5415         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
5416         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5417         Handle BUILT_IN_POSIX_MEMALIGN.
5418         (find_func_clobbers): Likewise.
5419         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5420         (call_may_clobber_ref_p_1): Likewise.
5421
5422 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5423
5424         PR ipa/59918
5425         * ipa-devirt.c (record_target_from_binfo): Remove overactive
5426         sanity check.
5427
5428 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5429
5430         PR ipa/59469
5431         * lto-cgraph.c (lto_output_node): Use
5432         symtab_get_symbol_partitioning_class.
5433         (lto_output_varpool_node): likewise.
5434         (symtab_get_symbol_partitioning_class): Move here from
5435         lto/lto-partition.c
5436         * cgraph.h (symbol_partitioning_class): Likewise.
5437         (symtab_get_symbol_partitioning_class): Declare.
5438
5439 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5440
5441         * ggc.h (ggc_internal_cleared_alloc): New macro.
5442         * vec.h (vec_safe_copy): Handle memory stats.
5443         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
5444         * target-globals.c (save_target_globals): Likewise.
5445
5446 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5447
5448         PR target/60077
5449         * expr.c (emit_move_resolve_push): Export; be bit more selective
5450         on when to clear alias set.
5451         * expr.h (emit_move_resolve_push): Declare.
5452         * function.h (struct function): Add tail_call_marked.
5453         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
5454         * config/i386/i386-protos.h (ix86_expand_push): Remove.
5455         * config/i386/i386.md (TImode move expander): De not call
5456         ix86_expand_push.
5457         (FP push expanders): Preserve memory attributes.
5458         * config/i386/sse.md (push<mode>1): Remove.
5459         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
5460         (ix86_expand_push): Remove.
5461         * config/i386/mmx.md (push<mode>1): Remove.
5462
5463 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5464
5465         PR rtl-optimization/60030
5466         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
5467         lopart with paradoxical subreg before shifting it up by hprec.
5468
5469 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5470
5471         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
5472         Remove extra newline at end of file.
5473         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
5474         (arm_issue_rate): Handle cortexa57.
5475         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
5476         (cortex-a57.cortex-a53): Likewise.
5477
5478 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5479
5480         PR target/59575
5481         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
5482         don't record in REG_FRAME_RELATED_EXPR registers not set in that
5483         bitmask.
5484         (arm_expand_prologue): Adjust all callers.
5485         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
5486         info, registers also at the lowest numbered registers side.  Use
5487         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
5488         XEXP.
5489
5490         PR debug/59992
5491         * var-tracking.c (adjust_mems): Before adding a SET to
5492         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
5493
5494 2014-02-06  Alan Modra  <amodra@gmail.com>
5495
5496         PR target/60032
5497         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
5498         change SDmode to DDmode when lra_in_progress.
5499
5500 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5501
5502         PR middle-end/59150
5503         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
5504         free_data_ref on the dr first, and before goto again also set dr
5505         to the next dr.  For simd_lane_access, free old datarefs[i] before
5506         overwriting it.  For get_vectype_for_scalar_type failure, don't
5507         free_data_ref if simd_lane_access.
5508
5509         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
5510
5511         PR target/60062
5512         * tree.h (opts_for_fn): New inline function.
5513         (opt_for_fn): Define.
5514         * config/i386/i386.c (ix86_function_regparm): Use
5515         opt_for_fn (decl, optimize) instead of optimize.
5516
5517 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5518
5519         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
5520         for SYMBOL_REF in large memory model.
5521
5522 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5523
5524         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
5525         and crypto support.
5526         (cortex-a57): Likewise.
5527         (cortex-a57.cortex-a53): Likewise.
5528
5529 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
5530             Kugan Vivekanandarajah  <kuganv@linaro.org>
5531
5532         * config/arm/arm.c (arm_vector_alignment_reachable): Check
5533         unaligned_access.
5534         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
5535
5536 2014-02-06  Richard Biener  <rguenther@suse.de>
5537
5538         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
5539         set_loop_copy and initialize_original_copy_tables.
5540
5541 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
5542
5543         * config/aarch64/aarch64-simd.md
5544         (aarch64_ashr_simddi): Change QI to SI.
5545
5546 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5547             Jakub Jelinek  <jakub@redhat.com>
5548
5549         PR middle-end/60013
5550         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
5551         of the dataflow.
5552
5553 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5554
5555         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
5556         CODE_FOR_altivec_vpku[hw]um to
5557         CODE_FOR_altivec_vpku[hw]um_direct.
5558         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
5559         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
5560         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
5561         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
5562
5563 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5564
5565         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
5566         generation for -maltivec=be.
5567         (altivec_vsumsws): Simplify redundant test.
5568
5569 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5570
5571         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
5572         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
5573         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
5574         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
5575         gen_altivec_vpkuwum.
5576         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
5577         BYTES_BIG_ENDIAN.
5578         (altivec_vpks<VI_char>ss): Likewise.
5579         (altivec_vpks<VI_char>us): Likewise.
5580         (altivec_vpku<VI_char>us): Likewise.
5581         (altivec_vpku<VI_char>um): Likewise.
5582         (altivec_vpku<VI_char>um_direct): New (copy of
5583         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
5584         internal use).
5585         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
5586         target is little endian and -maltivec=be is not specified.
5587         (*altivec_vupkhs<VU_char>_direct): New (copy of
5588         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
5589         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
5590         target is little endian and -maltivec=be is not specified.
5591         (*altivec_vupkls<VU_char>_direct): New (copy of
5592         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
5593         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
5594         little endian and -maltivec=be is not specified.
5595         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
5596         little endian and -maltivec=be is not specified.
5597
5598 2014-02-05  Richard Henderson  <rth@redhat.com>
5599
5600         PR debug/52727
5601         * combine-stack-adj.c: Revert r206943.
5602         * sched-int.h (struct deps_desc): Add last_args_size.
5603         * sched-deps.c (init_deps): Initialize it.
5604         (sched_analyze_insn): Add OUTPUT dependencies between insns that
5605         contain REG_ARGS_SIZE notes.
5606
5607 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5608
5609         * lto-cgraph.c (asm_nodes_output): Make global.
5610         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
5611         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
5612         (driver_handle_option): Handle OPT_fwpa.
5613
5614 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
5615
5616         PR ipa/59947
5617         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
5618         a comment typo and formatting issue.  If odr_hash hasn't been
5619         created, return vNULL and set *completep to false.
5620
5621         PR middle-end/57499
5622         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
5623         bb with no successors.
5624
5625 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
5626
5627         PR target/59718
5628         * doc/invoke.texi (-march): Clarify documentation for ARM.
5629         (-mtune): Likewise.
5630         (-mcpu): Likewise.
5631
5632 2014-02-05  Richard Biener  <rguenther@suse.de>
5633
5634         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
5635         when not vectorizing because of too many alias checks.
5636         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
5637         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
5638
5639 2014-02-05  Nick Clifton  <nickc@redhat.com>
5640
5641         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
5642         accept extended registers in any mode when compiling for the MN10300.
5643
5644 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
5645
5646         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
5647         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
5648         sanitization attributes.
5649         (can_inline_edge_p): Likewise.
5650         (sanitize_attrs_match_for_inline_p): New function.
5651
5652 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5653
5654         * ipa-prop.c (detect_type_change): Shor circuit testing of
5655         type changes on THIS pointer.
5656
5657 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
5658
5659         PR target/59777
5660         * config/pa/pa.c (legitimize_tls_address): Return original address
5661         if not passed a SYMBOL_REF rtx.
5662         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
5663         addresses.
5664         (pa_emit_move_sequence): Simplify TLS source operands.
5665         (pa_legitimate_constant_p): Reject all TLS constants.
5666         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
5667         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
5668
5669 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5670
5671         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
5672         groups when we know they are controlled by LTO.
5673         * varasm.c (default_binds_local_p_1): If object is in other partition,
5674         it will be resolved locally.
5675
5676 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5677
5678         * config/host-linux.c (linux_gt_pch_use_address): Don't
5679         use SSIZE_MAX because it is not always defined.
5680
5681 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
5682
5683         PR bootstrap/59913
5684         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
5685         threshold for pseudo splitting.
5686         (update_ebb_live_info): Process call argument hard registers and
5687         hard registers from insn definition too.
5688         (max_small_class_regs_num): New constant.
5689         (inherit_in_ebb): Update live hard regs through EBBs.  Update
5690         reloads_num only for small register classes.  Don't split for
5691         outputs of jumps.
5692
5693 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
5694
5695         PR ipa/60058
5696         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
5697         is non-null.
5698
5699 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5700
5701         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
5702         visibility is safe.
5703
5704 2014-02-04  Marek Polacek  <polacek@redhat.com>
5705
5706         * gdbinit.in (pel): Define.
5707
5708 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5709
5710         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
5711         behavior.
5712
5713 2014-02-04  Richard Biener  <rguenther@suse.de>
5714
5715         PR lto/59723
5716         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
5717         in function context local.
5718         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
5719         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
5720         similar to LTO_imported_decl_ref.
5721
5722 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5723
5724         PR tree-optimization/60002
5725         * cgraphclones.c (build_function_decl_skip_args): Clear
5726         DECL_LANG_SPECIFIC.
5727
5728         PR tree-optimization/60023
5729         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
5730         false to gsi_replace.
5731         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
5732         has been in some EH region and vec_stmt could throw, add
5733         vec_stmt into the same EH region.
5734         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
5735         has no lhs, ignore it.
5736         * internal-fn.c (expand_MASK_LOAD): Likewise.
5737
5738         PR ipa/60026
5739         * tree-inline.c (copy_forbidden): Fail for
5740         __attribute__((optimize (0))) functions.
5741
5742         PR other/58712
5743         * omp-low.c (simd_clone_struct_copy): If from->inbranch
5744         is set, copy one less argument.
5745         (expand_simd_clones): Don't subtract clone_info->inbranch
5746         from simd_clone_struct_alloc argument.
5747
5748         PR rtl-optimization/57915
5749         * recog.c (simplify_while_replacing): If all unary/binary/relational
5750         operation arguments are constant, attempt to simplify those.
5751
5752         PR middle-end/59261
5753         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
5754         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
5755
5756 2014-02-04  Richard Biener  <rguenther@suse.de>
5757
5758         PR tree-optimization/60012
5759         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
5760         TBAA disambiguation to all DDRs.
5761
5762 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5763
5764         PR target/59788
5765         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
5766         (LINK_SPEC): Use it for -shared, -shared-libgcc.
5767
5768 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5769
5770         PR ipa/59882
5771         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
5772
5773 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5774
5775         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
5776         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
5777
5778 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5779
5780         PR ipa/59831
5781         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
5782         to figure out targets of polymorphic calls with known decl.
5783         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5784         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
5785         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
5786         (get_polymorphic_call_info): ... here.
5787         (get_polymorphic_call_info_from_invariant): New function.
5788
5789 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5790
5791         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
5792         lookup via vtable pointer; check for type consistency
5793         and turn inconsitent facts into UNREACHABLE.
5794         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5795         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
5796         type inconsistent querries; return UNREACHABLE instead.
5797
5798 2014-02-03  Richard Henderson  <rth@twiddle.net>
5799
5800         PR tree-opt/59924
5801         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
5802         already processed this node.
5803         (normalize_one_pred_1): Pass along mark_set.
5804         (normalize_one_pred): Create and destroy a pointer_set_t.
5805         (normalize_one_pred_chain): Likewise.
5806
5807 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
5808
5809         PR gcov-profile/58602
5810         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
5811
5812 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5813
5814         PR ipa/59831
5815         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
5816         -fno-devirtualize; try to devirtualize by the knowledge of
5817         virtual table pointer given by aggregate propagation.
5818         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5819         (ipa_print_node_jump_functions): Dump also offset that
5820         is relevant for polymorphic calls.
5821         (determine_known_aggregate_parts): Add arg_type parameter; use it
5822         instead of determining the type from pointer type.
5823         (ipa_compute_jump_functions_for_edge): Update call of
5824         determine_known_aggregate_parts.
5825         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
5826         (gimple_get_virt_method_for_binfo): ... here; simplify using
5827         vtable_pointer_value_to_vtable.
5828         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
5829         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
5830         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
5831         (vtable_pointer_value_to_vtable): Break out from ...; handle also
5832         POINTER_PLUS_EXPR.
5833         (vtable_pointer_value_to_binfo): ... here.
5834         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
5835
5836 2014-02-03  Teresa Johnson  <tejohnson@google.com>
5837
5838         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
5839         redef of outer loop index variable.
5840
5841 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
5842
5843         PR c++/53017
5844         PR c++/59211
5845         * doc/extend.texi (Function Attributes): Typo.
5846
5847 2014-02-03  Cong Hou  <congh@google.com>
5848
5849         PR tree-optimization/60000
5850         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
5851         if the vectorized statement is a store.  A store statement can only
5852         appear at the end of pattern statements.
5853
5854 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5855
5856         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
5857         (ix86_option_override_internal): Default long double to 64-bit for
5858         32-bit Bionic and to 128-bit for 64-bit Bionic.
5859
5860         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
5861         TARGET_LONG_DOUBLE_128 is true.
5862         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
5863
5864         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
5865         (mlong-double-64): Negate -mlong-double-128.
5866         (mlong-double-128): New option.
5867
5868         * config/i386/i386-c.c (ix86_target_macros): Define
5869         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
5870
5871         * doc/invoke.texi: Document -mlong-double-128.
5872
5873 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5874
5875         PR rtl-optimization/60024
5876         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
5877
5878 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
5879
5880         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
5881
5882 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5883
5884         PR rtl-optimization/57662
5885         * sel-sched.c (code_motion_path_driver): Do not mark already not
5886         existing blocks in the visiting bitmap.
5887
5888 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5889
5890         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
5891         on the insn being emitted.
5892
5893 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
5894             Will Deacon  <will.deacon@arm.com>
5895
5896         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
5897
5898 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5899
5900         * config/arm/arm-tables.opt: Regenerate.
5901
5902 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5903
5904         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
5905         for vector types other than V16QImode.
5906         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
5907         define_expand, and call altivec_expand_vec_perm_le when producing
5908         code with little endian element order.
5909         (*altivec_vperm_<mode>_internal): New insn having previous
5910         behavior of altivec_vperm_<mode>.
5911         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
5912         altivec_expand_vec_perm_le when producing code with little endian
5913         element order.
5914         (*altivec_vperm_<mode>_uns_internal): New insn having previous
5915         behavior of altivec_vperm_<mode>_uns.
5916
5917 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5918
5919         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
5920         (altivec_vsumsws): Add handling for -maltivec=be with a little
5921         endian target.
5922         (altivec_vsumsws_direct): New.
5923         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
5924         gen_altivec_vsumsws.
5925
5926 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5927
5928         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
5929         vtable_pointer_value_to_binfo): New functions.
5930         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
5931         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
5932
5933 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
5934
5935         * config/nios2/nios2.md (load_got_register): Initialize GOT
5936         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
5937         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
5938
5939 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5940
5941         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
5942         preserverd by passthrough, do not propagate the type.
5943
5944 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5945
5946         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
5947         (mips_atomic_assign_expand_fenv): New function.
5948         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
5949
5950 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5951
5952         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
5953         (__builtin_mips_set_fcsr): Likewise.
5954         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
5955         MIPS_USI_FTYPE_VOID.
5956         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
5957         (mips16_expand_set_fcsr): Likewise.
5958         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
5959         (mips16_set_fcsr_stub): Likewise.
5960         (mips16_get_fcsr_one_only_stub): New class.
5961         (mips16_set_fcsr_one_only_stub): Likewise.
5962         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
5963         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
5964         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
5965         (hard_float): New availability predicate.
5966         (mips_builtins): Add get_fcsr and set_fcsr.
5967         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5968         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5969         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5970         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5971         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5972         patterns.
5973
5974 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5975
5976         * config/mips/mips.c (mips_one_only_stub): New class.
5977         (mips_need_mips16_rdhwr_p): Replace with...
5978         (mips16_rdhwr_stub): ...this new variable.
5979         (mips16_stub_call_address): New function.
5980         (mips16_rdhwr_one_only_stub): New class.
5981         (mips_expand_thread_pointer): Use mips16_stub_call_address.
5982         (mips_output_mips16_rdhwr): Delete.
5983         (mips_finish_stub): New function.
5984         (mips_code_end): Use it to handle rdhwr stubs.
5985
5986 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
5987
5988         PR target/60017
5989         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5990         when calculating size of integer atomic types.
5991
5992 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
5993
5994         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5995
5996 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
5997
5998         PR tree-optimization/60003
5999         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
6000         * profile.c (branch_prob): Use gimple_call_builtin_p
6001         to check for BUILT_IN_SETJMP_RECEIVER.
6002         * tree-inline.c (copy_bb): Call notice_special_calls.
6003
6004 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
6005
6006         PR bootstrap/59985
6007         * lra-constraints.c (process_alt_operands): Update reload_sum only
6008         on the first pass.
6009
6010 2014-01-31  Richard Henderson  <rth@redhat.com>
6011
6012         PR middle-end/60004
6013         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
6014         until after else_eh is processed.
6015
6016 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6017
6018         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
6019         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
6020         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
6021         in smmintrin.h, remove them.
6022         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
6023         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
6024         * config/i386/i386.md (ROUND_SAE): Fix value.
6025         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
6026         (const48_operand): New.
6027         * config/i386/subst.md (round), (round_expand): Use
6028         const_4_or_8_to_11_operand.
6029         (round_saeonly), (round_saeonly_expand): Use const48_operand.
6030
6031 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6032
6033         * config/i386/constraints.md (Yk): Swap meaning with k.
6034         * config/i386/i386.md (movhi_internal): Change Yk to k.
6035         (movqi_internal): Ditto.
6036         (*k<logic><mode>): Ditto.
6037         (*andhi_1): Ditto.
6038         (*andqi_1): Ditto.
6039         (kandn<mode>): Ditto.
6040         (*<code>hi_1): Ditto.
6041         (*<code>qi_1): Ditto.
6042         (kxnor<mode>): Ditto.
6043         (kortestzhi): Ditto.
6044         (kortestchi): Ditto.
6045         (kunpckhi): Ditto.
6046         (*one_cmplhi2_1): Ditto.
6047         (*one_cmplqi2_1): Ditto.
6048         * config/i386/sse.md (): Change k to Yk.
6049         (avx512f_load<mode>_mask): Ditto.
6050         (avx512f_blendm<mode>): Ditto.
6051         (avx512f_store<mode>_mask): Ditto.
6052         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
6053         (avx512f_storedqu<mode>_mask): Ditto.
6054         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
6055         Ditto.
6056         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
6057         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
6058         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
6059         (avx512f_maskcmp<mode>3): Ditto.
6060         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
6061         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
6062         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
6063         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
6064         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
6065         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
6066         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
6067         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
6068         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
6069         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
6070         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
6071         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
6072         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
6073         (vec_extract_lo_<mode>_maskm): Ditto.
6074         (vec_extract_hi_<mode>_maskm): Ditto.
6075         (avx512f_vternlog<mode>_mask): Ditto.
6076         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
6077         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
6078         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
6079         (avx512f_<code>v8div16qi2_mask): Ditto.
6080         (avx512f_<code>v8div16qi2_mask_store): Ditto.
6081         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
6082         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
6083         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
6084         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
6085         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6086         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6087         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6088         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6089         (avx512cd_maskb_vec_dupv8di): Ditto.
6090         (avx512cd_maskw_vec_dupv16si): Ditto.
6091         (avx512f_vpermi2var<mode>3_maskz): Ditto.
6092         (avx512f_vpermi2var<mode>3_mask): Ditto.
6093         (avx512f_vpermi2var<mode>3_mask): Ditto.
6094         (avx512f_vpermt2var<mode>3_maskz): Ditto.
6095         (*avx512f_gathersi<mode>): Ditto.
6096         (*avx512f_gathersi<mode>_2): Ditto.
6097         (*avx512f_gatherdi<mode>): Ditto.
6098         (*avx512f_gatherdi<mode>_2): Ditto.
6099         (*avx512f_scattersi<mode>): Ditto.
6100         (*avx512f_scatterdi<mode>): Ditto.
6101         (avx512f_compress<mode>_mask): Ditto.
6102         (avx512f_compressstore<mode>_mask): Ditto.
6103         (avx512f_expand<mode>_mask): Ditto.
6104         * config/i386/subst.md (mask): Change k to Yk.
6105         (mask_scalar_merge): Ditto.
6106         (sd): Ditto.
6107
6108 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
6109
6110         * doc/extend.texi (Vector Extensions): Document ?: in C++.
6111
6112 2014-01-31  Richard Biener  <rguenther@suse.de>
6113
6114         PR middle-end/59990
6115         * builtins.c (fold_builtin_memory_op): Make sure to not
6116         use a floating-point mode or a boolean or enumeral type for
6117         the copy operation.
6118
6119 2014-01-30  DJ Delorie  <dj@redhat.com>
6120
6121         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
6122         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
6123         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6124         whenever main() has an epilogue.
6125
6126 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6127
6128         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
6129         unused variable "field".
6130         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
6131         (vsx_mergeh_<mode>): Likewise.
6132         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
6133         (altivec_vmrghh): Likewise.
6134         (altivec_vmrghw): Likewise.
6135         (altivec_vmrglb): Likewise.
6136         (altivec_vmrglh): Likewise.
6137         (altivec_vmrglw): Likewise.
6138         (altivec_vspltb): Add missing uses.
6139         (altivec_vsplth): Likewise.
6140         (altivec_vspltw): Likewise.
6141         (altivec_vspltsf): Likewise.
6142
6143 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6144
6145         PR target/59923
6146         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
6147         frame related instructions.
6148
6149 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
6150
6151         PR rtl-optimization/59959
6152         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
6153         any reload of register whose subreg is invalid.
6154
6155 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6156
6157         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
6158         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
6159         Add missing return type - void.
6160
6161 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6162
6163         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
6164         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
6165         remove element index adjustment for endian (now handled in vsx.md
6166         and altivec.md).
6167         (altivec_expand_vec_perm_const): Use
6168         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
6169         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
6170         (vsx_xxspltw_<mode>): Adjust element index for little endian.
6171         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
6172         define_expand and a new define_insn *altivec_vspltb_internal;
6173         adjust for -maltivec=be on a little endian target.
6174         (altivec_vspltb_direct): New.
6175         (altivec_vsplth): Divide into a define_expand and a new
6176         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
6177         little endian target.
6178         (altivec_vsplth_direct): New.
6179         (altivec_vspltw): Divide into a define_expand and a new
6180         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
6181         little endian target.
6182         (altivec_vspltw_direct): New.
6183         (altivec_vspltsf): Divide into a define_expand and a new
6184         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
6185         a little endian target.
6186
6187 2014-01-30  Richard Biener  <rguenther@suse.de>
6188
6189         PR tree-optimization/59993
6190         * tree-ssa-forwprop.c (associate_pointerplus): Check we
6191         can propagate form the earlier stmt and avoid the transform
6192         when the intermediate result is needed.
6193
6194 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
6195
6196         * README.Portability: Fix typo.
6197
6198 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
6199
6200         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
6201         comparison_operator with ordered_comparison_operator.
6202
6203 2014-01-30  Nick Clifton  <nickc@redhat.com>
6204
6205         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
6206         Rename to mn10300_store_multiple_regs.
6207         * config/mn10300/mn10300.c: Likewise.
6208         * config/mn10300/mn10300.md (store_movm): Fix typo: call
6209         store_multiple_regs.
6210         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
6211         Call mn10300_store_multiple_regs.
6212
6213 2014-01-30  Nick Clifton  <nickc@redhat.com>
6214             DJ Delorie  <dj@redhat.com>
6215
6216         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
6217         %fp 2 to keep registers after it properly word-aligned.
6218         (rl78_alloc_physical_registers_umul): Handle the case where both
6219         input operands are the same.
6220
6221 2014-01-30  Richard Biener  <rguenther@suse.de>
6222
6223         PR tree-optimization/59903
6224         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
6225         check properly.
6226
6227 2014-01-30  Jason Merrill  <jason@redhat.com>
6228
6229         PR c++/59633
6230         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
6231
6232         PR c++/59645
6233         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
6234
6235 2014-01-30  Richard Biener  <rguenther@suse.de>
6236
6237         PR tree-optimization/59951
6238         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
6239
6240 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
6241
6242         PR target/59784
6243         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
6244         SFmode to DFmode case.
6245
6246 2014-01-29  DJ Delorie  <dj@redhat.com>
6247
6248         * config/msp430/msp430.opt (-minrt): New.
6249         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
6250         if -minrt given.
6251         (ENDFILE_SPEC): Likewise.
6252
6253 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
6254
6255         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
6256         (estimate_function_body_sizes): Use it.
6257
6258 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
6259
6260         PR c++/58561
6261         * dwarf2out.c (is_cxx_auto): New.
6262         (is_base_type): Use it.
6263         (gen_type_die_with_usage): Likewise.
6264
6265 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6266
6267         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
6268         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
6269         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
6270         -maltivec=be with LE targets.
6271         (vsx_mergeh_<mode>): Likewise.
6272         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
6273         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
6274         (altivec_vmrghb): Replace with define_expand and new
6275         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
6276         (altivec_vmrghb_direct): New define_insn.
6277         (altivec_vmrghh): Replace with define_expand and new
6278         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
6279         (altivec_vmrghh_direct): New define_insn.
6280         (altivec_vmrghw): Replace with define_expand and new
6281         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
6282         (altivec_vmrghw_direct): New define_insn.
6283         (*altivec_vmrghsf): Adjust for endianness.
6284         (altivec_vmrglb): Replace with define_expand and new
6285         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
6286         (altivec_vmrglb_direct): New define_insn.
6287         (altivec_vmrglh): Replace with define_expand and new
6288         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
6289         (altivec_vmrglh_direct): New define_insn.
6290         (altivec_vmrglw): Replace with define_expand and new
6291         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
6292         (altivec_vmrglw_direct): New define_insn.
6293         (*altivec_vmrglsf): Adjust for endianness.
6294         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6295         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6296         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6297         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6298         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6299         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6300         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6301         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6302
6303 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6304
6305         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
6306         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
6307         whitespace.
6308
6309 2014-01-29  Richard Biener  <rguenther@suse.de>
6310
6311         PR tree-optimization/58742
6312         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
6313         associate_pointerplus_align.
6314         (associate_pointerplus_diff): New function.
6315         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
6316         and associate_pointerplus_diff.
6317
6318 2014-01-29  Richard Biener  <rguenther@suse.de>
6319
6320         * lto-streamer.h (LTO_major_version): Bump to 3.
6321         (LTO_minor_version): Reset to 0.
6322
6323 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
6324
6325         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
6326         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
6327         (arm_file_start): Generate correct asm header for armv7ve.
6328         * config/arm/bpabi.h: Add multilib support for armv7ve.
6329         * config/arm/driver-arm.c: Change the architectures of cortex-a7
6330         and cortex-a15 to armv7ve.
6331         * config/arm/t-aprofile: Add multilib support for armv7ve.
6332         * doc/invoke.texi: Document -march=armv7ve.
6333
6334 2014-01-29  Richard Biener  <rguenther@suse.de>
6335
6336         PR tree-optimization/58742
6337         * tree-ssa-forwprop.c (associate_plusminus): Return true
6338         if we changed sth, defer EH cleanup to ...
6339         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
6340         (simplify_mult): New function.
6341
6342 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6343
6344         PR middle-end/59917
6345         PR tree-optimization/59920
6346         * tree.c (build_common_builtin_nodes): Remove
6347         __builtin_setjmp_dispatcher initialization.
6348         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
6349         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
6350         instead of gsi_after_labels + manually skipping debug stmts.
6351         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
6352         ignore bbs with IFN_ABNORMAL_DISPATCHER.
6353         * tree-inline.c (copy_edges_for_bb): Remove
6354         can_make_abnormal_goto argument, instead add abnormal_goto_dest
6355         argument.  Ignore computed_goto_p stmts.  Don't call
6356         make_abnormal_goto_edges.  If a call might need abnormal edges
6357         for non-local gotos, see if it already has an edge to
6358         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
6359         with true argument, don't do anything then, otherwise add
6360         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
6361         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
6362         caller.
6363         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
6364         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
6365         (lower_stmt): Don't set data->calls_builtin_setjmp.
6366         (lower_builtin_setjmp): Adjust comment.
6367         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
6368         * tree-cfg.c (found_computed_goto): Remove.
6369         (factor_computed_gotos): Remove.
6370         (make_goto_expr_edges): Return bool, true for computed gotos.
6371         Don't call make_abnormal_goto_edges.
6372         (build_gimple_cfg): Don't set found_computed_goto, don't call
6373         factor_computed_gotos.
6374         (computed_goto_p): No longer static.
6375         (make_blocks): Don't set found_computed_goto.
6376         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
6377         (make_edges): If make_goto_expr_edges returns true, push bb
6378         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
6379         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
6380         vector.  Record mapping between bbs and OpenMP regions if there
6381         are any, adjust make_gimple_omp_edges caller.  Call
6382         handle_abnormal_edges.
6383         (make_abnormal_goto_edges): Remove.
6384         * tree-cfg.h (make_abnormal_goto_edges): Remove.
6385         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
6386         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
6387         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
6388         * internal-fn.def (ABNORMAL_DISPATCHER): New.
6389         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
6390         filling *region also set *region_idx to (*region)->entry->index.
6391
6392         PR other/58712
6393         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
6394         For REGs set ORIGINAL_REGNO.
6395
6396 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
6397
6398         * doc/md.texi: Mention that a target shouldn't implement
6399         vec_widen_(s|u)mul_even/odd pair if it is less efficient
6400         than hi/lo pair.
6401
6402 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6403
6404         PR tree-optimization/59594
6405         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
6406         a copy of the datarefs vector rather than the vector itself.
6407
6408 2014-01-28  Jason Merrill  <jason@redhat.com>
6409
6410         PR c++/53756
6411         * dwarf2out.c (auto_die): New static.
6412         (gen_type_die_with_usage): Handle C++1y 'auto'.
6413         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
6414         on definition.
6415
6416 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
6417
6418         PR target/59672
6419         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
6420         (SPEC_X32): Likewise.
6421         (SPEC_64): Likewise.
6422         * config/i386/i386.c (ix86_option_override_internal): Turn off
6423         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
6424         for TARGET_16BIT.
6425         (x86_file_start): Output .code16gcc for TARGET_16BIT.
6426         * config/i386/i386.h (TARGET_16BIT): New macro.
6427         (TARGET_16BIT_P): Likewise.
6428         * config/i386/i386.opt: Add m16.
6429         * doc/invoke.texi: Document -m16.
6430
6431 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
6432
6433         PR preprocessor/59935
6434         * input.c (location_get_source_line): Bail out on when line number
6435         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
6436
6437 2014-01-28  Richard Biener  <rguenther@suse.de>
6438
6439         PR tree-optimization/58742
6440         * tree-ssa-forwprop.c (associate_plusminus): Handle
6441         pointer subtraction of the form (T)(P + A) - (T)P.
6442
6443 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6444
6445         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
6446         at const_int_cost.
6447
6448 2014-01-28  Richard Biener  <rguenther@suse.de>
6449
6450         Revert
6451         2014-01-28  Richard Biener  <rguenther@suse.de>
6452
6453         PR rtl-optimization/45364
6454         PR rtl-optimization/59890
6455         * var-tracking.c (local_get_addr_clear_given_value): Handle
6456         already cleared slot.
6457         (val_reset): Handle not allocated local_get_addr_cache.
6458         (vt_find_locations): Use post-order on the inverted CFG.
6459
6460 2014-01-28  Richard Biener  <rguenther@suse.de>
6461
6462         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
6463
6464 2014-01-28  Richard Biener  <rguenther@suse.de>
6465
6466         PR rtl-optimization/45364
6467         PR rtl-optimization/59890
6468         * var-tracking.c (local_get_addr_clear_given_value): Handle
6469         already cleared slot.
6470         (val_reset): Handle not allocated local_get_addr_cache.
6471         (vt_find_locations): Use post-order on the inverted CFG.
6472
6473 2014-01-28  Alan Modra  <amodra@gmail.com>
6474
6475         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
6476         * configure.ac <recursive call for build != host>: Define
6477         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
6478         and LD_FOR_BUILD too.
6479         * configure: Regenerate.
6480
6481 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
6482
6483         * config/i386/i386.c (get_builtin_code_for_version): Separate
6484         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
6485         Broadwell from Haswell.
6486
6487 2014-01-27  Steve Ellcey  <sellcey@mips.com>
6488
6489         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6490         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
6491         * config/mips/mips.c (mips_option_override): Change setting
6492         of TARGET_DSP.
6493         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
6494         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
6495         Change from Mask to Var.
6496
6497 2014-01-27  Jeff Law  <law@redhat.com>
6498
6499         * ipa-inline.c (inline_small_functions): Fix typo.
6500
6501 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
6502
6503         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
6504         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
6505         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
6506         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
6507         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
6508         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
6509         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
6510         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
6511         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
6512         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
6513         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
6514         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
6515         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
6516         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
6517         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
6518         (_mm512_storeu_epi64): Ditto.
6519         (_mm512_cmpge_epi32_mask): Ditto.
6520         (_mm512_cmpge_epu32_mask): Ditto.
6521         (_mm512_cmpge_epi64_mask): Ditto.
6522         (_mm512_cmpge_epu64_mask): Ditto.
6523         (_mm512_cmple_epi32_mask): Ditto.
6524         (_mm512_cmple_epu32_mask): Ditto.
6525         (_mm512_cmple_epi64_mask): Ditto.
6526         (_mm512_cmple_epu64_mask): Ditto.
6527         (_mm512_cmplt_epi32_mask): Ditto.
6528         (_mm512_cmplt_epu32_mask): Ditto.
6529         (_mm512_cmplt_epi64_mask): Ditto.
6530         (_mm512_cmplt_epu64_mask): Ditto.
6531         (_mm512_cmpneq_epi32_mask): Ditto.
6532         (_mm512_cmpneq_epu32_mask): Ditto.
6533         (_mm512_cmpneq_epi64_mask): Ditto.
6534         (_mm512_cmpneq_epu64_mask): Ditto.
6535         (_mm512_expand_pd): Ditto.
6536         (_mm512_expand_ps): Ditto.
6537         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
6538         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
6539         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
6540         * config/i386/i386.c (ix86_builtins): Add
6541         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
6542         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
6543         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
6544         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
6545         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
6546         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
6547         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
6548         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
6549         IX86_BUILTIN_PMOVUSQW512_MEM.
6550         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
6551         __builtin_ia32_pmovsqd512mem_mask,
6552         __builtin_ia32_pmovqd512mem_mask,
6553         __builtin_ia32_pmovusqw512mem_mask,
6554         __builtin_ia32_pmovsqw512mem_mask,
6555         __builtin_ia32_pmovqw512mem_mask,
6556         __builtin_ia32_pmovusdw512mem_mask,
6557         __builtin_ia32_pmovsdw512mem_mask,
6558         __builtin_ia32_pmovdw512mem_mask,
6559         __builtin_ia32_pmovqb512mem_mask,
6560         __builtin_ia32_pmovusqb512mem_mask,
6561         __builtin_ia32_pmovsqb512mem_mask,
6562         __builtin_ia32_pmovusdb512mem_mask,
6563         __builtin_ia32_pmovsdb512mem_mask,
6564         __builtin_ia32_pmovdb512mem_mask.
6565         (bdesc_args): Add __builtin_ia32_expanddf512,
6566         __builtin_ia32_expandsf512.
6567         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
6568         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
6569         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
6570         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
6571         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
6572         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
6573         (avx512f_<code>v8div16qi2_mask_store): This.
6574         (avx512f_expand<mode>): New.
6575
6576 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
6577
6578         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
6579         New.
6580         (_mm512_mask_prefetch_i64gather_pd): Ditto.
6581         (_mm512_prefetch_i32scatter_pd): Ditto.
6582         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6583         (_mm512_prefetch_i64scatter_pd): Ditto.
6584         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6585         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
6586         (_mm512_mask_prefetch_i64gather_ps): Ditto.
6587         (_mm512_prefetch_i32scatter_ps): Ditto.
6588         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6589         (_mm512_prefetch_i64scatter_ps): Ditto.
6590         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6591         * config/i386/i386-builtin-types.def: Define
6592         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
6593         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
6594         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
6595         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
6596         IX86_BUILTIN_SCATTERPFQPD.
6597         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
6598         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
6599         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
6600         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
6601         __builtin_ia32_scatterpfqps.
6602         (ix86_expand_builtin): Expand new built-ins.
6603         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
6604         fix memory access data type.
6605         (*avx512pf_gatherpf<mode>_mask): Ditto.
6606         (*avx512pf_gatherpf<mode>): Ditto.
6607         (avx512pf_scatterpf<mode>): Ditto.
6608         (*avx512pf_scatterpf<mode>_mask): Ditto.
6609         (*avx512pf_scatterpf<mode>): Ditto.
6610         (GATHER_SCATTER_SF_MEM_MODE): New.
6611         (avx512pf_gatherpf<mode>df): Ditto.
6612         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6613         (*avx512pf_scatterpf<mode>df): Ditto.
6614
6615 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
6616
6617         PR bootstrap/59934
6618         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
6619         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
6620         reached.
6621
6622 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6623
6624         * common/config/arm/arm-common.c
6625         (arm_rewrite_mcpu): Handle multiple names.
6626         * config/arm/arm.h
6627         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6628
6629 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6630
6631         * gimple-builder.h (create_gimple_tmp): Delete.
6632
6633 2014-01-27  Christian Bruel  <christian.bruel@st.com>
6634
6635         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
6636         words comparisons.
6637
6638 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
6639
6640         * config/pa/pa.md (call): Generate indirect long calls to non-local
6641         functions when outputing 32-bit code.
6642         (call_value): Likewise except for special call to buggy powf function.
6643
6644         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
6645         portable runtime and PIC indirect calls.
6646         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
6647         and PIC call sequences.  Use ldo instead of blr to set return register
6648         in PIC call sequence.
6649
6650 2014-01-25  Walter Lee  <walt@tilera.com>
6651
6652         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
6653         avoid clobbering a live register.
6654
6655 2014-01-25  Walter Lee  <walt@tilera.com>
6656
6657         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
6658         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
6659         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
6660         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
6661
6662 2014-01-25  Walter Lee  <walt@tilera.com>
6663
6664         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
6665         arguments on even registers.
6666         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
6667         STACK_BOUNDARY.
6668         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
6669         (BIGGEST_ALIGNMENT): Ditto.
6670         (BIGGEST_FIELD_ALIGNMENT): Ditto.
6671
6672 2014-01-25  Walter Lee  <walt@tilera.com>
6673
6674         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
6675         insns before bundling.
6676         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
6677
6678 2014-01-25  Walter Lee  <walt@tilera.com>
6679
6680         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
6681         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
6682         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
6683
6684 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6685
6686         * config/mips/constraints.md (kl): Delete.
6687         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
6688         define expands, using...
6689         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
6690         instructions for MIPS16.
6691         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
6692         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
6693
6694 2014-01-25  Walter Lee  <walt@tilera.com>
6695
6696         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
6697         (clzdi2): Ditto.
6698         (ffsdi2): Ditto.
6699
6700 2014-01-25  Walter Lee  <walt@tilera.com>
6701
6702         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
6703         (TARGET_EXPAND_TO_RTL_HOOK): Define.
6704
6705 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6706
6707         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
6708         Handle XOR.
6709
6710 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
6711
6712         * print-rtl.c (in_call_function_usage): New var.
6713         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
6714         EXPR_LIST mode as mode and not as reg note name.
6715
6716         PR middle-end/59561
6717         * cfgloopmanip.c (copy_loop_info): If
6718         loop->warned_aggressive_loop_optimizations, make sure
6719         the flag is set in target loop too.
6720
6721 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6722
6723         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
6724         flag_cilkplus.
6725         * builtins.def: Likewise.
6726         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
6727         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
6728         * ira.c (ira_setup_eliminable_regset): Likewise.
6729         * omp-low.c (gate_expand_omp): Likewise.
6730         (execute_lower_omp): Likewise.
6731         (diagnose_sb_0): Likewise.
6732         (gate_diagnose_omp_blocks): Likewise.
6733         (simd_clone_clauses_extract): Likewise.
6734         (gate): Likewise.
6735
6736 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6737
6738         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
6739         correction for little endian...
6740         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
6741         here.
6742
6743 2014-01-24  Jeff Law  <law@redhat.com>
6744
6745         PR tree-optimization/59919
6746         * tree-vrp.c (find_assert_locations_1): Do not register asserts
6747         for non-returning calls.
6748
6749 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
6750
6751         * common/config/aarch64/aarch64-common.c
6752         (aarch64_rewrite_mcpu): Handle multiple names.
6753         * config/aarch64/aarch64.h
6754         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6755
6756 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
6757
6758         * input.c (add_file_to_cache_tab): Handle the case where fopen
6759         returns NULL.
6760
6761 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
6762
6763         PR target/59929
6764         * config/i386/i386.md (pushsf splitter): Get stack adjustment
6765         from push operand if code of push isn't PRE_DEC.
6766
6767 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6768
6769         PR target/59909
6770         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
6771         -mquad-memory-atomic.  Update -mquad-memory documentation to say
6772         it is only used for non-atomic loads/stores.
6773
6774         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
6775         -mquad-memory or -mquad-memory-atomic switches.
6776
6777         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
6778         -mquad-memory-atomic to ISA 2.07 support.
6779
6780         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
6781         to separate support of normal quad word memory operations (ldq, stq)
6782         from the atomic quad word memory operations.
6783
6784         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6785         support to separate non-atomic quad word operations from atomic
6786         quad word operations.  Disable non-atomic quad word operations in
6787         little endian mode so that we don't have to swap words after the
6788         load and before the store.
6789         (quad_load_store_p): Add comment about atomic quad word support.
6790         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
6791         options printed with -mdebug=reg.
6792
6793         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
6794         -mquad-memory-atomic as the test for whether we have quad word
6795         atomic instructions.
6796         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
6797         or -mp8-vector are used, allow byte/half-word atomic operations.
6798
6799         * config/rs6000/sync.md (load_lockedti): Insure that the address
6800         is a proper indexed or indirect address for the lqarx instruction.
6801         On little endian systems, swap the hi/lo registers after the lqarx
6802         instruction.
6803         (load_lockedpti): Use indexed_or_indirect_operand predicate to
6804         insure the address is valid for the lqarx instruction.
6805         (store_conditionalti): Insure that the address is a proper indexed
6806         or indirect address for the stqcrx. instruction.  On little endian
6807         systems, swap the hi/lo registers before doing the stqcrx.
6808         instruction.
6809         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
6810         insure the address is valid for the stqcrx. instruction.
6811
6812         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
6813         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
6814         type of quad memory support is available.
6815
6816 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
6817
6818         PR regression/59915
6819         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
6820         there is a danger of looping.
6821
6822 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
6823
6824         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6825         force flag_ira_loop_pressure if set via command line.
6826
6827 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6828
6829         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
6830         (ashr_simd): New builtin handling DI mode.
6831         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
6832         (aarch64_sshr_simddi): New match pattern.
6833         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
6834         (vshrd_n_s64): Likewise.
6835         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
6836
6837 2014-01-23  Nick Clifton  <nickc@redhat.com>
6838
6839         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
6840         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
6841         favour of mcu specific scripts.
6842         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
6843         430x multilibs.
6844
6845 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6846             Alex Velenko  <Alex.Velenko@arm.com>
6847
6848         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
6849         (vaddv_s16): Likewise.
6850         (vaddv_s32): Likewise.
6851         (vaddv_u8): Likewise.
6852         (vaddv_u16): Likewise.
6853         (vaddv_u32): Likewise.
6854         (vaddvq_s8): Likewise.
6855         (vaddvq_s16): Likewise.
6856         (vaddvq_s32): Likewise.
6857         (vaddvq_s64): Likewise.
6858         (vaddvq_u8): Likewise.
6859         (vaddvq_u16): Likewise.
6860         (vaddvq_u32): Likewise.
6861         (vaddvq_u64): Likewise.
6862         (vaddv_f32): Likewise.
6863         (vaddvq_f32): Likewise.
6864         (vaddvq_f64): Likewise.
6865         (vmaxv_f32): Likewise.
6866         (vmaxv_s8): Likewise.
6867         (vmaxv_s16): Likewise.
6868         (vmaxv_s32): Likewise.
6869         (vmaxv_u8): Likewise.
6870         (vmaxv_u16): Likewise.
6871         (vmaxv_u32): Likewise.
6872         (vmaxvq_f32): Likewise.
6873         (vmaxvq_f64): Likewise.
6874         (vmaxvq_s8): Likewise.
6875         (vmaxvq_s16): Likewise.
6876         (vmaxvq_s32): Likewise.
6877         (vmaxvq_u8): Likewise.
6878         (vmaxvq_u16): Likewise.
6879         (vmaxvq_u32): Likewise.
6880         (vmaxnmv_f32): Likewise.
6881         (vmaxnmvq_f32): Likewise.
6882         (vmaxnmvq_f64): Likewise.
6883         (vminv_f32): Likewise.
6884         (vminv_s8): Likewise.
6885         (vminv_s16): Likewise.
6886         (vminv_s32): Likewise.
6887         (vminv_u8): Likewise.
6888         (vminv_u16): Likewise.
6889         (vminv_u32): Likewise.
6890         (vminvq_f32): Likewise.
6891         (vminvq_f64): Likewise.
6892         (vminvq_s8): Likewise.
6893         (vminvq_s16): Likewise.
6894         (vminvq_s32): Likewise.
6895         (vminvq_u8): Likewise.
6896         (vminvq_u16): Likewise.
6897         (vminvq_u32): Likewise.
6898         (vminnmv_f32): Likewise.
6899         (vminnmvq_f32): Likewise.
6900         (vminnmvq_f64): Likewise.
6901
6902 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6903
6904         * config/aarch64/aarch64-simd.md
6905         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
6906         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
6907         (*aarch64_mul3_elt<mode>): Likewise.
6908         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
6909         (*aarch64_mul3_elt_to_64v2df): Likewise.
6910         (*aarch64_mla_elt<mode>): Likewise.
6911         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
6912         (*aarch64_mls_elt<mode>): Likewise.
6913         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
6914         (*aarch64_fma4_elt<mode>): Likewise.
6915         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
6916         (*aarch64_fma4_elt_to_64v2df): Likewise.
6917         (*aarch64_fnma4_elt<mode>): Likewise.
6918         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6919         (*aarch64_fnma4_elt_to_64v2df): Likewise.
6920         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
6921         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
6922         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6923         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6924         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6925         (aarch64_sqdmull_lane<mode>_internal): Likewise.
6926         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6927
6928 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6929
6930         * config/aarch64/aarch64-simd.md
6931         (aarch64_be_checked_get_lane<mode>): New define_expand.
6932         * config/aarch64/aarch64-simd-builtins.def
6933         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
6934         New builtin definition.
6935         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
6936         Use new safe be builtin.
6937
6938 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6939
6940         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
6941         New define_insn.
6942         (aarch64_be_st1<mode>): Likewise.
6943         (aarch_ld1<VALL:mode>): Define_expand modified.
6944         (aarch_st1<VALL:mode>): Likewise.
6945         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
6946         (UNSPEC_ST1): Likewise.
6947
6948 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
6949
6950         * config/microblaze/microblaze.md: Add trap insn and attribute
6951
6952 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
6953
6954         PR preprocessor/58580
6955         * input.h (location_get_source_line): Take an additional line_size
6956         parameter.
6957         (void diagnostics_file_cache_fini): Declare new function.
6958         * input.c (struct fcache): New type.
6959         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
6960         New static constants.
6961         (diagnostic_file_cache_init, total_lines_num)
6962         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
6963         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
6964         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
6965         (get_next_line, read_next_line, goto_next_line, read_line_num):
6966         New static function definitions.
6967         (diagnostic_file_cache_fini): New function.
6968         (location_get_source_line): Take an additional output line_len
6969         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
6970         read_line_num.
6971         * diagnostic.c (diagnostic_finish): Call
6972         diagnostic_file_cache_fini.
6973         (adjust_line): Take an additional input parameter for the length
6974         of the line, rather than calculating it with strlen.
6975         (diagnostic_show_locus): Adjust the use of
6976         location_get_source_line and adjust_line with respect to their new
6977         signature.  While displaying a line now, do not stop at the first
6978         null byte.  Rather, display the zero byte as a space and keep
6979         going until we reach the size of the line.
6980         * Makefile.in: Add vec.o to OBJS-libcommon
6981
6982 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6983             Ilya Tocar     <ilya.tocar@intel.com>
6984
6985         * config/i386/avx512fintrin.h (_mm512_kmov): New.
6986         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6987         (__builtin_ia32_kmov16): Ditto.
6988         * config/i386/i386.md (UNSPEC_KMOV): New.
6989         (kmovw): Ditto.
6990
6991 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6992
6993         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6994         (_mm512_storeu_si512): Ditto.
6995
6996 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
6997
6998         PR target/52125
6999         * rtl.h (get_referenced_operands): Declare.
7000         * recog.c (get_referenced_operands): New function.
7001         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
7002         operands have been referenced when recording LO_SUM references.
7003
7004 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
7005
7006         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
7007
7008 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7009
7010         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
7011         Enable for generic and recent AMD targets.
7012
7013 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7014
7015         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
7016         ARG_SIZE note when adjustment was eliminated.
7017
7018 2014-01-22  Jeff Law  <law@redhat.com>
7019
7020         PR tree-optimization/59597
7021         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
7022         in file.  Accept new argument REGISTERING and use it to modify
7023         dump output appropriately.
7024         (register_jump_thread): Corresponding changes.
7025         (mark_threaded_blocks): Reinstate code to cancel unprofitable
7026         thread paths involving joiner blocks.  Add code to dump cancelled
7027         jump threading paths.
7028
7029 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
7030
7031         PR rtl-optimization/59477
7032         * lra-constraints.c (inherit_in_ebb): Process call for living hard
7033         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
7034
7035 2014-01-22  Tom Tromey  <tromey@redhat.com>
7036
7037         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
7038         PARAMS.
7039         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
7040
7041 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7042
7043         PR rtl-optimization/59896
7044         * lra-constraints.c (process_alt_operands): Check unused note for
7045         matched operands of insn with no output reloads.
7046
7047 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
7048
7049         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
7050         (mips_move_from_gpr_cost): Likewise.
7051
7052 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7053
7054         PR rtl-optimization/59858
7055         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
7056         ira_class_hard_regs_num.
7057         (process_alt_operands): Increase reject for dying matched operand.
7058
7059 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
7060
7061         PR target/59003
7062         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
7063         smaller than size, perform several stores or loads and stores
7064         at dst + count - size to store or copy all of size bytes, rather
7065         than just last modesize bytes.
7066
7067 2014-01-20  DJ Delorie  <dj@redhat.com>
7068
7069         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
7070         that CLOBBERs are REGs before propogating their values.
7071
7072 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
7073
7074         PR middle-end/59789
7075         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
7076         (cgraph_inline_failed_type): New function.
7077         * cgraph.h (DEFCIFCODE): Add type.
7078         (cgraph_inline_failed_type_t): New enum.
7079         (cgraph_inline_failed_type): New prototype.
7080         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
7081         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
7082         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
7083         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
7084         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
7085         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
7086         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
7087         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
7088         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
7089         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
7090         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
7091         OPTIMIZATION_MISMATCH.
7092         * tree-inline.c (expand_call_inline): Emit errors during
7093         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
7094
7095 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7096
7097         PR target/59685
7098         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
7099         mode attribute in insn output.
7100
7101 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
7102
7103         * output.h (output_constant): Delete.
7104         * varasm.c (output_constant): Make private.
7105
7106 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
7107
7108         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
7109
7110 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7111
7112         PR middle-end/59860
7113         * tree.h (fold_builtin_strcat): New prototype.
7114         * builtins.c (fold_builtin_strcat): No longer static.  Add len
7115         argument, if non-NULL, don't call c_strlen.  Optimize
7116         directly into __builtin_memcpy instead of __builtin_strcpy.
7117         (fold_builtin_2): Adjust fold_builtin_strcat caller.
7118         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
7119
7120 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7121
7122         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7123         for SImode_address_operand operands, having only a REG argument.
7124
7125 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7126
7127         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
7128         loader name using mbig-endian.
7129         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
7130
7131 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7132
7133         * doc/invoke.texi (-march): Clarify documentation for AArch64.
7134         (-mtune): Likewise.
7135         (-mcpu): Likewise.
7136
7137 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
7138
7139         * config/aarch64/aarch64-protos.h
7140         (aarch64_cannot_change_mode_class_ptr): Declare.
7141         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
7142         aarch64_cannot_change_mode_class_ptr): New.
7143         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
7144         backend hook aarch64_cannot_change_mode_class.
7145
7146 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7147
7148         * common/config/aarch64/aarch64-common.c
7149         (aarch64_handle_option): Don't handle any option order logic here.
7150         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
7151         selected_cpu, warn on architecture version mismatch.
7152         (aarch64_override_options): Fix parsing order for option strings.
7153
7154 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7155             Iain Sandoe  <iain@codesourcery.com>
7156
7157         PR bootstrap/59496
7158         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
7159         warning.  Amend comment to reflect current functionality.
7160
7161 2014-01-20  Richard Biener  <rguenther@suse.de>
7162
7163         PR middle-end/59860
7164         * builtins.c (fold_builtin_strcat): Remove case better handled
7165         by tree-ssa-strlen.c.
7166
7167 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
7168
7169         * config/aarch64/aarch64.opt
7170         (mcpu, march, mtune): Make case-insensitive.
7171
7172 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7173
7174         PR target/59880
7175         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7176         if operands[1] is a REG or ZERO_EXTEND of a REG.
7177
7178 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
7179
7180         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
7181
7182 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
7183
7184         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
7185         long non-pic millicode calls.
7186
7187 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7188
7189         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
7190
7191 2014-01-19  Kito Cheng  <kito@0xlab.org>
7192
7193         * builtins.c (expand_movstr): Check movstr expand done or fail.
7194
7195 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7196             H.J. Lu  <hongjiu.lu@intel.com>
7197
7198         PR target/59379
7199         * config/i386/i386.md (*lea<mode>): Zero-extend return register
7200         to DImode for zero-extended addresses.
7201
7202 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
7203
7204         PR rtl-optimization/57763
7205         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
7206         on the new indirect jump_insn and increment LABEL_NUSES (label).
7207
7208 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
7209
7210         PR bootstrap/59580
7211         PR bootstrap/59583
7212         * config.gcc (x86_archs): New variable.
7213         (x86_64_archs): Likewise.
7214         (x86_cpus): Likewise.
7215         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
7216         --with-arch/--with-cpu= options.
7217         Support --with-arch=/--with-cpu={nehalem,westmere,
7218         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
7219
7220 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7221
7222         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
7223         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
7224
7225 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7226
7227         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
7228
7229 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
7230
7231         PR target/58944
7232         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
7233         clear cpp_get_options (parse_in)->warn_unused_macros for
7234         ix86_target_macros_internal with cpp_define.
7235
7236 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
7237
7238         * jump.c (delete_related_insns): Keep (use (insn))s.
7239         * reorg.c (redundant_insn): Check for barriers too.
7240
7241 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7242
7243         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
7244
7245 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
7246
7247         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
7248         call to $$dyncall when TARGET_LONG_CALLS is true.
7249
7250 2014-01-17  Jeff Law  <law@redhat.com>
7251
7252         * ree.c (combine_set_extension): Temporarily disable test for
7253         changing number of hard registers.
7254
7255 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7256
7257         PR middle-end/58125
7258         * ipa-inline-analysis.c (inline_free_summary):
7259         Do not free summary of aliases.
7260
7261 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7262
7263         PR middle-end/59706
7264         * gimplify.c (gimplify_expr): Use create_tmp_var
7265         instead of create_tmp_var_raw.  If cond doesn't have
7266         integral type, don't add the IFN_ANNOTATE builtin at all.
7267
7268 2014-01-17  Martin Jambor  <mjambor@suse.cz>
7269
7270         PR ipa/59736
7271         * ipa-cp.c (prev_edge_clone): New variable.
7272         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
7273         Also resize prev_edge_clone vector.
7274         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
7275         (ipcp_edge_removal_hook): New function.
7276         (ipcp_driver): Register ipcp_edge_removal_hook.
7277
7278 2014-01-17  Andrew Pinski  <apinski@cavium.com>
7279             Steve Ellcey  <sellcey@mips.com>
7280
7281         PR target/59462
7282         * config/mips/mips.c (mips_print_operand): Check operand mode instead
7283         of operator mode.
7284
7285 2014-01-17  Jeff Law  <law@redhat.com>
7286
7287         PR middle-end/57904
7288         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
7289         so that pass_ccp runs first.
7290
7291 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7292
7293         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
7294         (ix86_adjust_cost): Use !TARGET_XXX.
7295         (do_reorder_for_imul): Likewise.
7296         (swap_top_of_ready_list): Likewise.
7297         (ix86_sched_reorder): Likewise.
7298
7299 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7300
7301         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7302         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
7303         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
7304         (intel_memset): New.  Duplicate slm_memset.
7305         (intel_cost): New.  Duplicate slm_cost.
7306         (m_INTEL): New macro.
7307         (processor_target_table): Add "intel".
7308         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
7309         with PROCESSOR_INTEL for "intel".
7310         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
7311         PROCESSOR_SILVERMONT.
7312         (ix86_issue_rate): Likewise.
7313         (ix86_adjust_cost): Likewise.
7314         (ia32_multipass_dfa_lookahead): Likewise.
7315         (swap_top_of_ready_list): Likewise.
7316         (ix86_sched_reorder): Likewise.
7317         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
7318         instead of TARGET_OPT_AGU.
7319         * config/i386/i386.h (TARGET_INTEL): New.
7320         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
7321         (processor_type): Add PROCESSOR_INTEL.
7322         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
7323         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
7324
7325 2014-01-17  Marek Polacek  <polacek@redhat.com>
7326
7327         PR c/58346
7328         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
7329         size is zero.
7330
7331 2014-01-17  Richard Biener  <rguenther@suse.de>
7332
7333         PR tree-optimization/46590
7334         * opts.c (default_options_table): Add entries for
7335         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
7336         all enabled at -O1 but not for -Og.
7337         * common.opt (fbranch-count-reg): Remove Init(1).
7338         (fmove-loop-invariants): Likewise.
7339         (ftree-pta): Likewise.
7340
7341 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7342
7343         * config/i386/i386.c (ix86_data_alignment): For compatibility with
7344         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
7345         decls to at least the GCC 4.8 used alignments.
7346
7347         PR fortran/59440
7348         * tree-nested.c (convert_nonlocal_reference_stmt,
7349         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
7350         of GIMPLE_BIND stmts, adjust associated decls.
7351
7352 2014-01-17  Richard Biener  <rguenther@suse.de>
7353
7354         PR tree-optimization/46590
7355         * vec.h (vec<>::bseach): New member function implementing
7356         binary search according to C89 bsearch.
7357         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
7358         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
7359         bitmap pointer again.  Make accesses_in_loop a flat array.
7360         (mem_ref_obstack): New global.
7361         (outermost_indep_loop): Adjust for mem_ref->stored changes.
7362         (mark_ref_stored): Likewise.
7363         (ref_indep_loop_p_2): Likewise.
7364         (set_ref_stored_in_loop): New helper function.
7365         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
7366         (memref_free): Adjust.
7367         (record_mem_ref_loc): Simplify.
7368         (gather_mem_refs_stmt): Adjust.
7369         (sort_locs_in_loop_postorder_cmp): New function.
7370         (analyze_memory_references): Sort accesses_in_loop after
7371         loop postorder number.
7372         (find_ref_loc_in_loop_cmp): New function.
7373         (for_all_locs_in_loop): Find relevant cluster of locs in
7374         accesses_in_loop and iterate without recursion.
7375         (execute_sm): Avoid uninit warning.
7376         (struct ref_always_accessed): Simplify.
7377         (ref_always_accessed::operator ()): Likewise.
7378         (ref_always_accessed_p): Likewise.
7379         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
7380         loop postorder numbers here.
7381         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
7382         numbers.
7383
7384 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7385
7386         PR c++/57945
7387         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
7388         on decls for which assemble_alias has been called.
7389
7390 2014-01-17  Nick Clifton  <nickc@redhat.com>
7391
7392         * config/msp430/msp430.opt: (mcpu): New option.
7393         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
7394         (msp430_option_override): Parse target_cpu.  If the MCU name
7395         matches a generic string, clear target_mcu.
7396         (msp430_attr): Allow numeric interrupt values up to 63.
7397         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
7398         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
7399         option.
7400         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
7401         Add mcpu matches.
7402         * config/msp430/msp430.md (popm): Use %J rather than %I.
7403         (addsi3): Use msp430_nonimmediate_operand for operand 2.
7404         (addhi_cy_i): Use immediate_operand for operand 2.
7405         * doc/invoke.texi: Document -mcpu option.
7406
7407 2014-01-17  Richard Biener  <rguenther@suse.de>
7408
7409         PR rtl-optimization/38518
7410         * df.h (df_analyze_loop): Declare.
7411         * df-core.c: Include cfgloop.h.
7412         (df_analyze_1): Split out main part of df_analyze.
7413         (df_analyze): Adjust.
7414         (loop_inverted_post_order_compute): New function.
7415         (loop_post_order_compute): Likewise.
7416         (df_analyze_loop): New function avoiding whole-function
7417         postorder computes.
7418         * loop-invariant.c (find_defs): Use df_analyze_loop.
7419         (find_invariants): Adjust.
7420         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
7421
7422 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
7423
7424         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
7425         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
7426
7427 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
7428
7429         * ipa-ref.c (ipa_remove_stmt_references): Fix references
7430         traversal when removing references.
7431
7432 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
7433
7434         PR ipa/59775
7435         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
7436
7437 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
7438
7439         PR middle-end/56791
7440         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
7441         pushing a reload for an autoinc when we had previously reloaded an
7442         inner part of the address.
7443
7444 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
7445
7446         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
7447         field.
7448         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
7449         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
7450         when not giving up or versioning for alias only because of
7451         loop->safelen.
7452         (vect_analyze_data_ref_dependences): Set to true.
7453         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
7454         is a GIMPLE_PHI.
7455         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
7456         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
7457         to the condition.
7458
7459         PR middle-end/58344
7460         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
7461
7462         PR target/59839
7463         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
7464         operand 0 predicate for gathers, use a new pseudo as subtarget.
7465
7466 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7467
7468         PR middle-end/59609
7469         * lra-constraints.c (process_alt_operands): Add printing debug info.
7470         Check absence of input/output reloads for matched operands too.
7471
7472 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7473
7474         PR rtl-optimization/59835
7475         * ira.c (ira_init_register_move_cost): Increase cost for
7476         impossible modes.
7477
7478 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
7479
7480         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
7481
7482 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
7483
7484         PR target/59780
7485         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
7486         non-register objects.  Use gen_(high/low)part more consistently.
7487         Fix assertions.
7488
7489 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
7490
7491         PR target/59844
7492         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
7493         endian support, remove tests for WORDS_BIG_ENDIAN.
7494         (p8_mfvsrd_3_<mode>): Likewise.
7495         (reload_gpr_from_vsx<mode>): Likewise.
7496         (reload_gpr_from_vsxsf): Likewise.
7497         (p8_mfvsrd_4_disf): Likewise.
7498
7499 2014-01-16  Richard Biener  <rguenther@suse.de>
7500
7501         PR rtl-optimization/46590
7502         * lcm.c (compute_antinout_edge): Use postorder iteration.
7503         (compute_laterin): Use inverted postorder iteration.
7504
7505 2014-01-16  Nick Clifton  <nickc@redhat.com>
7506
7507         PR middle-end/28865
7508         * varasm.c (output_constant): Return the number of bytes actually
7509         emitted.
7510         (output_constructor_array_range): Update the field size with the
7511         number of bytes emitted by output_constant.
7512         (output_constructor_regular_field): Likewise.  Also do not
7513         complain if the total number of bytes emitted is now greater
7514         than the expected fieldpos.
7515         * output.h (output_constant): Update prototype and descriptive comment.
7516
7517 2014-01-16  Marek Polacek  <polacek@redhat.com>
7518
7519         PR middle-end/59827
7520         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
7521         it is error_mark_node.
7522
7523 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
7524
7525         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
7526         VALID_AVX256_REG_OR_OI_MODE.
7527
7528 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
7529
7530         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
7531         current procedure should be profiled.
7532
7533 2014-01-15  Andrew Pinski  <apinski@cavium.com>
7534
7535         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
7536         of moving from/to the STACK_REG register class.
7537
7538 2014-01-15  Richard Henderson  <rth@redhat.com>
7539
7540         PR debug/54694
7541         * reginfo.c (global_regs_decl): Globalize.
7542         * rtl.h (global_regs_decl): Declare.
7543         * ira.c (do_reload): Diagnose frame_pointer_needed and it
7544         reserved via global_regs.
7545
7546 2014-01-15  Teresa Johnson  <tejohnson@google.com>
7547
7548         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
7549
7550 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7551
7552         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
7553         and vmulosh rather than call gen_vec_widen_smult_*.
7554         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
7555         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
7556         (vec_widen_smult_even_v16qi): Likewise.
7557         (vec_widen_umult_even_v8hi): Likewise.
7558         (vec_widen_smult_even_v8hi): Likewise.
7559         (vec_widen_umult_odd_v16qi): Likewise.
7560         (vec_widen_smult_odd_v16qi): Likewise.
7561         (vec_widen_umult_odd_v8hi): Likewise.
7562         (vec_widen_smult_odd_v8hi): Likewise.
7563         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
7564         vmuloub rather than call gen_vec_widen_umult_*.
7565         (vec_widen_umult_lo_v16qi): Likewise.
7566         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
7567         vmulosb rather than call gen_vec_widen_smult_*.
7568         (vec_widen_smult_lo_v16qi): Likewise.
7569         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
7570         rather than call gen_vec_widen_umult_*.
7571         (vec_widen_umult_lo_v8hi): Likewise.
7572         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
7573         rather than call gen_vec_widen_smult_*.
7574         (vec_widen_smult_lo_v8hi): Likewise.
7575
7576 2014-01-15  Jeff Law  <law@redhat.com>
7577
7578         PR tree-optimization/59747
7579         * ree.c (find_and_remove_re): Properly handle case where a second
7580         eliminated extension requires widening a copy created for elimination
7581         of a prior extension.
7582         (combine_set_extension): Ensure that the number of hard regs needed
7583         for a destination register does not change when we widen it.
7584
7585 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7586
7587         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
7588         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
7589         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
7590         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
7591         (avr-*-rtems*): Likewise.
7592         (bfin*-rtems*): Likewise.
7593         (moxie-*-rtems*): Likewise.
7594         (h8300-*-rtems*): Likewise.
7595         (i[34567]86-*-rtems*): Likewise.
7596         (lm32-*-rtems*): Likewise.
7597         (m32r-*-rtems*): Likewise.
7598         (m68k-*-rtems*): Likewise.
7599         (microblaze*-*-rtems*): Likewise.
7600         (mips*-*-rtems*): Likewise.
7601         (powerpc-*-rtems*): Likewise.
7602         (sh-*-rtems*): Likewise.
7603         (sparc-*-rtems*): Likewise.
7604         (sparc64-*-rtems*): Likewise.
7605         (v850-*-rtems*): Likewise.
7606         (m32c-*-rtems*): Likewise.
7607
7608 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
7609
7610         PR rtl-optimization/59511
7611         * ira.c (ira_init_register_move_cost): Use memory costs for some
7612         cases of register move cost calculations.
7613         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
7614         instead of BB frequency.
7615         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
7616         * lra-assigns.c (find_hard_regno_for): Ditto.
7617
7618 2014-01-15  Richard Biener  <rguenther@suse.de>
7619
7620         PR tree-optimization/59822
7621         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
7622         (vectorizable_load): Use it to hoist defs of uses of invariant
7623         loads out of the loop.
7624
7625 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
7626             Kugan Vivekanandarajah  <kuganv@linaro.org>
7627
7628         PR target/59695
7629         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
7630         truncation.
7631
7632 2014-01-15  Richard Biener  <rguenther@suse.de>
7633
7634         PR rtl-optimization/59802
7635         * lcm.c (compute_available): Use inverted postorder to seed
7636         the initial worklist.
7637
7638 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7639
7640         PR target/59803
7641         * config/s390/s390.c (s390_preferred_reload_class): Don't return
7642         ADDR_REGS for invalid symrefs in non-PIC code.
7643
7644 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
7645
7646         PR other/58712
7647         * builtins.c (determine_block_size): Initialize *probable_max_size
7648         even if len_rtx is CONST_INT.
7649
7650 2014-01-14  Andrew Pinski  <apinski@cavium.com>
7651
7652         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
7653         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
7654         (cortexa53_tunings): Likewise.
7655         (aarch64_sched_issue_rate): New function.
7656         (TARGET_SCHED_ISSUE_RATE): Define.
7657
7658 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7659
7660         * ira-costs.c (find_costs_and_classes): Add missed
7661         ira_init_register_move_cost_if_necessary.
7662
7663 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7664
7665         PR target/59787
7666         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
7667
7668 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
7669
7670         PR target/59794
7671         * config/i386/i386.c (type_natural_mode): Add a bool parameter
7672         to indicate if type is used for function return value.  Warn ABI
7673         change if the vector mode isn't available for function return value.
7674         (ix86_function_arg_advance): Pass false to type_natural_mode.
7675         (ix86_function_arg): Likewise.
7676         (ix86_gimplify_va_arg): Likewise.
7677         (function_arg_32): Don't warn ABI change.
7678         (ix86_function_value): Pass true to type_natural_mode.
7679         (ix86_return_in_memory): Likewise.
7680         (ix86_struct_value_rtx): Removed.
7681         (TARGET_STRUCT_VALUE_RTX): Likewise.
7682
7683 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7684
7685         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
7686         converting a conditional jump into a conditional return.
7687
7688 2014-01-14  Richard Biener  <rguenther@suse.de>
7689
7690         PR tree-optimization/58921
7691         PR tree-optimization/59006
7692         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
7693         hoisting invariant stmts.
7694         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
7695         invariant loads on the preheader edge if possible.
7696
7697 2014-01-14  Joey Ye  <joey.ye@arm.com>
7698
7699         * doc/plugin.texi (Building GCC plugins): Update to C++.
7700
7701 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
7702
7703         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
7704         (_mm_rcp28_round_ss): Ditto.
7705         (_mm_rsqrt28_round_sd): Ditto.
7706         (_mm_rsqrt28_round_ss): Ditto.
7707         (_mm_rcp28_sd): Ditto.
7708         (_mm_rcp28_ss): Ditto.
7709         (_mm_rsqrt28_sd): Ditto.
7710         (_mm_rsqrt28_ss): Ditto.
7711         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
7712         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
7713         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
7714         (IX86_BUILTIN_RCP28SD): Ditto.
7715         (IX86_BUILTIN_RCP28SS): Ditto.
7716         (IX86_BUILTIN_RSQRT28SD): Ditto.
7717         (IX86_BUILTIN_RSQRT28SS): Ditto.
7718         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
7719         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
7720         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
7721         (ix86_expand_special_args_builtin): Expand new FTYPE.
7722         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
7723         (srcp14<mode>): Make insn unary.
7724         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
7725         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
7726         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
7727         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
7728         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
7729         Fix rounding: make it SAE only.
7730         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7731         Ditto.
7732         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7733         Ditto.
7734         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
7735         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
7736         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
7737         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
7738         (round_saeonly_mask_scalar_operand4): Ditto.
7739         (round_saeonly_mask_scalar_op3): Ditto.
7740         (round_saeonly_mask_scalar_op4): Ditto.
7741
7742 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7743
7744         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7745         Implement -maltivec=be for vec_insert and vec_extract.
7746
7747 2014-01-10  DJ Delorie  <dj@redhat.com>
7748
7749         * config/msp430/msp430.md (call_internal): Don't allow memory
7750         references with SP as the base register.
7751         (call_value_internal): Likewise.
7752         * config/msp430/constraints.md (Yc): New.  For memory references
7753         that don't use SP as a base register.
7754
7755         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
7756         "an integer without a # prefix"
7757         * config/msp430/msp430.md (epilogue_helper): Use it.
7758
7759 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7760
7761         PR target/59617
7762         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
7763         AVX512F gather builtins.
7764         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
7765         on gather decls with INTEGER_TYPE masktype.
7766         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
7767         directly into the builtin rather than hoisting it before loop.
7768
7769         PR tree-optimization/59387
7770         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
7771         (scev_const_prop): If folded_casts and type has undefined overflow,
7772         use force_gimple_operand instead of force_gimple_operand_gsi and
7773         for each added stmt if it is assign with
7774         arith_code_with_undefined_signed_overflow, call
7775         rewrite_to_defined_overflow.
7776         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
7777         gimple-fold.h instead.
7778         (arith_code_with_undefined_signed_overflow,
7779         rewrite_to_defined_overflow): Moved to ...
7780         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
7781         rewrite_to_defined_overflow): ... here.  No longer static.
7782         Include gimplify-me.h.
7783         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
7784         rewrite_to_defined_overflow): New prototypes.
7785
7786 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7787
7788         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
7789
7790 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
7791
7792         * builtins.c (get_object_alignment_2): Minor tweak.
7793         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
7794
7795 2014-01-13  Christian Bruel  <christian.bruel@st.com>
7796
7797         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
7798         optimized non constant lengths.
7799
7800 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7801
7802         PR libgomp/59194
7803         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
7804         load as __atomic_load_N if possible.
7805
7806 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7807
7808         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
7809         target parameter.
7810         (rs6000_expand_builtin): Adjust call.
7811
7812 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7813
7814         PR target/58115
7815         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
7816         * config/rs6000/rs6000.c: Include target-globals.h.
7817         (rs6000_set_current_function): Instead of doing target_reinit
7818         unconditionally, use save_target_globals_default_opts and
7819         restore_target_globals.
7820
7821         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
7822         FPSCR.
7823         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
7824         (rs6000_expand_builtin): Handle mffs and mtfsf.
7825         (rs6000_init_builtins): Define mffs and mtfsf.
7826         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
7827         (rs6000_mffs): New pattern.
7828         (rs6000_mtfsf): New pattern.
7829
7830 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
7831
7832         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
7833         Start narrowing with START.  Apply candidate-use pair
7834         and check overall cost in narrowing.
7835         (iv_ca_prune): Pass new argument.
7836
7837 2014-01-10  Jeff Law  <law@redhat.com>
7838
7839         PR middle-end/59743
7840         * ree.c (combine_reaching_defs): Ensure the defining statement
7841         occurs before the extension when optimizing extensions with
7842         different source and destination hard registers.
7843
7844 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7845
7846         PR ipa/58585
7847         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
7848         vtables into the type inheritance graph.
7849
7850 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7851
7852         PR rtl-optimization/59754
7853         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
7854         modes in the REGNO != REGNO case.
7855
7856 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7857
7858         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
7859
7860 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7861
7862         PR tree-optimization/59745
7863         * tree-predcom.c (tree_predictive_commoning_loop): Call
7864         free_affine_expand_cache if giving up because components is NULL.
7865
7866         * target-globals.c (save_target_globals): Allocate < 4KB structs using
7867         GC in payload of target_globals struct instead of allocating them on
7868         the heap and the larger structs separately using GC.
7869         * target-globals.h (struct target_globals): Make regs, hard_regs,
7870         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
7871         of GTY((skip)) and change type to void *.
7872         (reset_target_globals): Cast loads from those fields to corresponding
7873         types.
7874
7875 2014-01-10  Steve Ellcey  <sellcey@mips.com>
7876
7877         PR plugins/59335
7878         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
7879         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
7880         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
7881
7882 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
7883
7884         PR target/59744
7885         * aarch64-modes.def (CC_Zmode): New flags mode.
7886         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
7887         represents an equality.
7888         (aarch64_get_condition_code): Handle CC_Zmode.
7889         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
7890
7891 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7892
7893         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
7894         extraction in good case.
7895
7896 2014-01-10  Richard Biener  <rguenther@suse.de>
7897
7898         PR tree-optimization/59374
7899         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
7900         checking after SLP discovery.  Mark stmts not participating
7901         in any SLP instance properly.
7902
7903 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7904
7905         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
7906         when handling a SET rtx.
7907
7908 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7909
7910         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
7911         (cortex-a57): Likewise.
7912         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
7913
7914 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7915
7916         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
7917         non-iwmmxt builtins.
7918
7919 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7920
7921         PR ipa/58252
7922         PR ipa/59226
7923         * ipa-devirt.c record_target_from_binfo): Take as argument
7924         stack of binfos and lookup matching one for virtual inheritance.
7925         (possible_polymorphic_call_targets_1): Update.
7926
7927 2014-01-10  Huacai Chen  <chenhc@lemote.com>
7928
7929         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
7930         kernel strings for Loongson-2E/2F/3A.
7931
7932 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7933
7934         PR middle-end/59670
7935         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
7936         is_gimple_call before calling gimple_call_internal_p.
7937
7938 2014-01-09  Steve Ellcey  <sellcey@mips.com>
7939
7940         * Makefile.in (TREE_FLOW_H): Remove.
7941         (TREE_SSA_H): Add file names from tree-flow.h.
7942         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
7943         * tree.h: Remove tree-flow.h reference.
7944         * hash-table.h: Remove tree-flow.h reference.
7945         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
7946         reference with tree-ssa-loop.h.
7947
7948 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7949
7950         * doc/invoke.texi: Add -maltivec={be,le} options, and document
7951         default element-order behavior for -maltivec.
7952         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
7953         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
7954         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
7955         when targeting big endian, at least for now.
7956         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
7957
7958 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7959
7960         PR middle-end/47735
7961         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
7962         var satisfies use_register_for_decl, just take into account type
7963         alignment, rather than decl alignment.
7964
7965         PR tree-optimization/59622
7966         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
7967         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7968         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7969         Don't devirtualize for inplace at all.  For targets.length () == 1,
7970         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7971
7972 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7973
7974         * config/i386/i386.md (cpu): Remove the unused btver1.
7975
7976 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7977
7978         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7979
7980 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7981
7982         PR target/58115
7983         * tree-core.h (struct target_globals): New forward declaration.
7984         (struct tree_target_option): Add globals field.
7985         * tree.h (TREE_TARGET_GLOBALS): Define.
7986         (prepare_target_option_nodes_for_pch): New prototype.
7987         * target-globals.h (struct target_globals): Define even if
7988         !SWITCHABLE_TARGET.
7989         * tree.c (prepare_target_option_node_for_pch,
7990         prepare_target_option_nodes_for_pch): New functions.
7991         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7992         * config/i386/i386.c: Include target-globals.h.
7993         (ix86_set_current_function): Instead of doing target_reinit
7994         unconditionally, use save_target_globals_default_opts and
7995         restore_target_globals.
7996
7997 2014-01-09  Richard Biener  <rguenther@suse.de>
7998
7999         PR tree-optimization/59715
8000         * tree-cfg.h (split_critical_edges): Declare.
8001         * tree-cfg.c (split_critical_edges): Export.
8002         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
8003
8004 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
8005
8006         * cfgexpand.c (expand_stack_vars): Optionally disable
8007         asan stack protection.
8008         (expand_used_vars): Likewise.
8009         (partition_stack_vars): Likewise.
8010         * asan.c (asan_emit_stack_protection): Optionally disable
8011         after return stack usage.
8012         (instrument_derefs): Optionally disable memory access instrumentation.
8013         (instrument_builtin_call): Likewise.
8014         (instrument_strlen_call): Likewise.
8015         (asan_protect_global): Optionally disable global variables protection.
8016         * doc/invoke.texi: Added doc for new options.
8017         * params.def: Added new options.
8018         * params.h: Likewise.
8019
8020 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8021
8022         PR rtl-optimization/59724
8023         * ifcvt.c (cond_exec_process_if_block): Don't call
8024         flow_find_head_matching_sequence with 0 longest_match.
8025         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
8026         non-active insns if !stop_after.
8027         (try_head_merge_bb): Revert 2014-01-07 changes.
8028
8029 2014-01-08  Jeff Law  <law@redhat.com>
8030
8031         * ree.c (get_sub_rtx): New function, extracted from...
8032         (merge_def_and_ext): Here.
8033         (combine_reaching_defs): Use get_sub_rtx.
8034
8035 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
8036
8037         * cgraph.h (varpool_variable_node): Do not choke on null node.
8038
8039 2014-01-08  Catherine Moore  <clm@codesourcery.com>
8040
8041         * config/mips/mips.md (simple_return): Attempt to use JRC
8042         for microMIPS.
8043         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
8044
8045 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8046
8047         PR rtl-optimization/59137
8048         * reorg.c (steal_delay_list_from_target): Call update_block for
8049         elided insns.
8050         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
8051
8052 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8053
8054         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8055         two duplicate entries.
8056
8057 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8058
8059         Revert:
8060         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
8061
8062         * config/mips/mips.c (mips_truncated_op_cost): New function.
8063         (mips_rtx_costs): Adjust test for BADDU.
8064         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
8065
8066         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
8067
8068         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
8069         (*baddu_si): ...this new pattern.
8070
8071 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8072
8073         PR ipa/59722
8074         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
8075
8076 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8077
8078         PR middle-end/57748
8079         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
8080         inner_reference_p.
8081         (expand_expr, expand_normal): Adjust.
8082         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
8083         inner_reference_p. Use inner_reference_p to expand inner references.
8084         (store_expr): Adjust.
8085         * cfgexpand.c (expand_call_stmt): Adjust.
8086
8087 2014-01-08  Rong Xu  <xur@google.com>
8088
8089         * gcov-io.c (gcov_var): Move from gcov-io.h.
8090         (gcov_position): Ditto.
8091         (gcov_is_error): Ditto.
8092         (gcov_rewrite): Ditto.
8093         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
8094         only part to libgcc/libgcov.h.
8095
8096 2014-01-08  Marek Polacek  <polacek@redhat.com>
8097
8098         PR middle-end/59669
8099         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
8100
8101 2014-01-08  Marek Polacek  <polacek@redhat.com>
8102
8103         PR sanitizer/59667
8104         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
8105
8106 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8107
8108         PR rtl-optimization/59649
8109         * stor-layout.c (get_mode_bounds): For BImode return
8110         0 and STORE_FLAG_VALUE.
8111
8112 2014-01-08  Richard Biener  <rguenther@suse.de>
8113
8114         PR middle-end/59630
8115         * gimple.h (is_gimple_builtin_call): Remove.
8116         (gimple_builtin_call_types_compatible_p): New.
8117         (gimple_call_builtin_p): New overload.
8118         * gimple.c (is_gimple_builtin_call): Remove.
8119         (validate_call): Rename to ...
8120         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
8121         check return types.
8122         (validate_type): New static function.
8123         (gimple_call_builtin_p): New overload and adjust.
8124         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
8125         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
8126         (gimple_fold_stmt_to_constant_1): Likewise.
8127         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
8128
8129 2014-01-08  Richard Biener  <rguenther@suse.de>
8130
8131         PR middle-end/59471
8132         * gimplify.c (gimplify_expr): Gimplify register-register type
8133         VIEW_CONVERT_EXPRs to separate stmts.
8134
8135 2014-01-07  Jeff Law  <law@redhat.com>
8136
8137         PR middle-end/53623
8138         * ree.c (combine_set_extension): Handle case where source
8139         and destination registers in an extension insn are different.
8140         (combine_reaching_defs): Allow source and destination registers
8141         in extension to be different under limited circumstances.
8142         (add_removable_extension): Remove restriction that the
8143         source and destination registers in the extension are the same.
8144         (find_and_remove_re): Emit a copy from the extension's
8145         destination to its source after the defining insn if
8146         the source and destination registers are different.
8147
8148         PR middle-end/59285
8149         * ifcvt.c (merge_if_block): If we are merging a block with more than
8150         one successor with a block with no successors, remove any BARRIER
8151         after the second block.
8152
8153 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
8154
8155         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
8156
8157 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
8158
8159         PR target/59652
8160         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
8161         for 14-bit register offsets when INT14_OK_STRICT is false.
8162
8163 2014-01-07  Roland Stigge  <stigge@antcom.de>
8164             Michael Meissner  <meissner@linux.vnet.ibm.com>
8165
8166         PR 57386/target
8167         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
8168         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
8169
8170 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
8171
8172         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
8173         -mcpu.
8174
8175 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
8176
8177         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
8178         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
8179         rtx is const0_rtx or not.
8180
8181 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
8182
8183         PR target/58115
8184         * target-globals.c (save_target_globals): Remove this_fn_optab
8185         handling.
8186         * toplev.c: Include optabs.h.
8187         (target_reinit): Temporarily restore the global options if another
8188         set of options are in force.
8189
8190 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
8191
8192         PR rtl-optimization/58668
8193         * cfgcleanup.c (flow_find_cross_jump): Don't count
8194         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
8195         to determine what is counted.
8196         (flow_find_head_matching_sequence): Use active_insn_p to determine
8197         what is counted.
8198         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
8199         counting change.
8200         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
8201         determine what is counted.
8202
8203         PR tree-optimization/59643
8204         * tree-predcom.c (split_data_refs_to_components): If one dr is
8205         read and one write, determine_offset fails and the write isn't
8206         in the bad component, just put the read into the bad component.
8207
8208 2014-01-07  Mike Stump  <mikestump@comcast.net>
8209             Jakub Jelinek  <jakub@redhat.com>
8210
8211         PR pch/59436
8212         * tree-core.h (struct tree_optimization_option): Change optabs
8213         type from unsigned char * to void *.
8214         * optabs.c (init_tree_optimization_optabs): Adjust
8215         TREE_OPTIMIZATION_OPTABS initialization.
8216
8217 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
8218
8219         PR target/59644
8220         * config/i386/i386.h (struct machine_function): Add
8221         no_drap_save_restore field.
8222         * config/i386/i386.c (ix86_save_reg): Use
8223         !cfun->machine->no_drap_save_restore instead of
8224         crtl->stack_realign_needed.
8225         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
8226         this function clears frame_pointer_needed.  Set
8227         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
8228         and DRAP reg is needed.
8229
8230 2014-01-06  Marek Polacek  <polacek@redhat.com>
8231
8232         PR c/57773
8233         * doc/implement-c.texi: Mention that other integer types are
8234         permitted as bit-field types in strictly conforming mode.
8235
8236 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
8237
8238         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
8239         is newly allocated.
8240
8241 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
8242
8243         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
8244
8245 2014-01-06  Martin Jambor  <mjambor@suse.cz>
8246
8247         PR ipa/59008
8248         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
8249         to int.
8250         * ipa-prop.c (ipa_print_node_params): Fix indentation.
8251
8252 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
8253
8254         PR debug/59350
8255         PR debug/59510
8256         * var-tracking.c (add_stores): Preserve the value of the source even if
8257         we don't record the store.
8258
8259 2014-01-06  Terry Guo  <terry.guo@arm.com>
8260
8261         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
8262
8263 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
8264
8265         PR bootstrap/59541
8266         * config/darwin.c (darwin_function_section): Adjust return values to
8267         correspond to optimisation changes made in r206070.
8268
8269 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
8270
8271         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
8272         from prefetch_block tune setting.
8273         (nocona_cost): Correct size of prefetch block to 64.
8274
8275 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
8276
8277         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
8278         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
8279         used to save the static chain register in the computation of the offset
8280         from which the FP registers need to be restored.
8281
8282 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
8283
8284         PR tree-optimization/59519
8285         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
8286         ICE if get_current_def (current_new_name) is already non-NULL, as long
8287         as it is a phi result of some other phi in *new_exit_bb that has
8288         the same argument.
8289
8290         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
8291         or vmovdqu* for misaligned_operand.
8292         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
8293         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
8294         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
8295         aligned_mem for AVX512F masked aligned load and store builtins and for
8296         non-temporal moves.
8297
8298 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
8299
8300         PR tree-optimization/59651
8301         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
8302         Address range for negative step should be added by TYPE_SIZE_UNIT.
8303
8304 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
8305
8306         * config/m68k/m68k.c (handle_move_double): Handle pushes with
8307         overlapping registers also for registers other than the stack pointer.
8308
8309 2014-01-03  Marek Polacek  <polacek@redhat.com>
8310
8311         PR other/59661
8312         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
8313         __builtin_FILE.
8314
8315 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
8316
8317         PR target/59625
8318         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
8319         asm goto as jump.
8320
8321         * config/i386/i386.md (MODE_SIZE): New mode attribute.
8322         (push splitter): Use <P:MODE_SIZE> instead of
8323         GET_MODE_SIZE (<P:MODE>mode).
8324         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
8325         (mov -1, reg peephole2): Likewise.
8326         * config/i386/sse.md (*mov<mode>_internal,
8327         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
8328         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
8329         *<code><mode>3, *andnot<mode>3<mask_name>,
8330         <mask_codefor><code><mode>3<mask_name>): Likewise.
8331         * config/i386/subst.md (mask_mode512bit_condition,
8332         sd_mask_mode512bit_condition): Likewise.
8333
8334 2014-01-02  Xinliang David Li  <davidxl@google.com>
8335
8336         PR tree-optimization/59303
8337         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
8338         (dump_predicates): Better output format.
8339         (pred_equal_p): New function.
8340         (is_neq_relop_p): Ditto.
8341         (is_neq_zero_form_p): Ditto.
8342         (pred_expr_equal_p): Ditto.
8343         (pred_neg_p): Ditto.
8344         (simplify_pred): Ditto.
8345         (simplify_preds_2): Ditto.
8346         (simplify_preds_3): Ditto.
8347         (simplify_preds_4): Ditto.
8348         (simplify_preds): Ditto.
8349         (push_pred): Ditto.
8350         (push_to_worklist): Ditto.
8351         (get_pred_info_from_cmp): Ditto.
8352         (is_degenerated_phi): Ditto.
8353         (normalize_one_pred_1): Ditto.
8354         (normalize_one_pred): Ditto.
8355         (normalize_one_pred_chain): Ditto.
8356         (normalize_preds): Ditto.
8357         (normalize_cond_1): Remove function.
8358         (normalize_cond): Ditto.
8359         (is_gcond_subset_of): Ditto.
8360         (is_subset_of_any): Ditto.
8361         (is_or_set_subset_of): Ditto.
8362         (is_and_set_subset_of): Ditto.
8363         (is_norm_cond_subset_of): Ditto.
8364         (pred_chain_length_cmp): Ditto.
8365         (convert_control_dep_chain_into_preds): Type change.
8366         (find_predicates): Ditto.
8367         (find_def_preds): Ditto.
8368         (destroy_predicates_vecs): Ditto.
8369         (find_matching_predicates_in_rest_chains): Ditto.
8370         (use_pred_not_overlap_with_undef_path_pred): Ditto.
8371         (is_pred_expr_subset): Ditto.
8372         (is_pred_chain_subset_of): Ditto.
8373         (is_included_in): Ditto.
8374         (is_superset_of): Ditto.
8375
8376 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8377
8378         Update copyright years.
8379
8380 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8381
8382         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
8383         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
8384         config/arc/arc.md, config/arc/arc.opt,
8385         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
8386         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
8387         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
8388         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
8389         config/linux-protos.h, config/linux.c, config/winnt-c.c,
8390         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
8391         vtable-verify.c, vtable-verify.h: Use the standard form for the
8392         copyright notice.
8393
8394 2014-01-02  Tobias Burnus  <burnus@net-b.de>
8395
8396         * gcc.c (process_command): Update copyright notice dates.
8397         * gcov-dump.c: Ditto.
8398         * gcov.c: Ditto.
8399         * doc/cpp.texi: Bump @copying's copyright year.
8400         * doc/cppinternals.texi: Ditto.
8401         * doc/gcc.texi: Ditto.
8402         * doc/gccint.texi: Ditto.
8403         * doc/gcov.texi: Ditto.
8404         * doc/install.texi: Ditto.
8405         * doc/invoke.texi: Ditto.
8406
8407 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8408
8409         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
8410
8411 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
8412
8413         * config/i386/sse.md (*mov<mode>_internal): Guard
8414         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
8415
8416         PR rtl-optimization/59647
8417         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
8418         new_rtx into UNSIGNED_FLOAT rtxes.
8419 \f
8420 Copyright (C) 2014 Free Software Foundation, Inc.
8421
8422 Copying and distribution of this file, with or without modification,
8423 are permitted in any medium without royalty provided the copyright
8424 notice and this notice are preserved.