gcc/ChangeLog

   1 2014-04-29  Alan Lawrence  <alan.lawrence@arm.com>
   2
   3         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
   4         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
   5         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
   6         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
   7         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
   8         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
   9         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
  10         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
  11
  12 2014-04-29  David Malcolm  <dmalcolm@redhat.com>
  13
  14         * tree-cfg.c (dump_function_to_file): Dump the return type of
  15         functions, in a line to itself before the function body, mimicking
  16         the layout of a C function.
  17
  18 2014-04-29  Jakub Jelinek  <jakub@redhat.com>
  19
  20         PR tree-optimization/60971
  21         * tree-tailcall.c (process_assignment): Reject conversions which
  22         reduce precision.
  23
  24 2014-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
  25
  26         * calls.c (initialize_argument_information): Always treat
  27         PUSH_ARGS_REVERSED as 1, simplify code accordingly.
  28         (expand_call): Likewise.
  29         (emit_library_call_calue_1): Likewise.
  30         * expr.c (PUSH_ARGS_REVERSED): Do not define.
  31         (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
  32         code accordingly.
  33
  34 2014-04-29  Nick Clifton  <nickc@redhat.com>
  35
  36         * config/msp430/msp430.md (umulsidi): Fix typo.
  37         (mulhisi3): Enable even inside interrupt handlers.
  38         * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
  39         bigger return address pushed in large mode.
  40
  41 2014-04-29  Nick Clifton  <nickc@redhat.com>
  42
  43         * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
  44         (arc_init_reg_tables): Use a machine_mode enum to iterate over
  45         available modes.
  46         * config/m32r/m32r.c (init_reg_tables): Likewise.
  47         * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
  48         enum to hold the modes.
  49
  50 2014-04-29  Richard Biener  <rguenther@suse.de>
  51
  52         * dominance.c (free_dominance_info): Add overload with
  53         function parameter.
  54         (dom_info_state): Likewise.
  55         (dom_info_available_p): Likewise.
  56         * basic-block.h (free_dominance_info, dom_info_state,
  57         dom_info_available_p): Declare overloads.
  58         * passes.c (execute_function_todo): Verify that verifiers
  59         don't change dominator info state.  Drop dominator info
  60         for IPA pass invocations.
  61         * cgraph.c (release_function_body): Restore asserts that
  62         dominator information is released.
  63
  64 2014-04-29  Patrick Palka  <patrick@parcs.ath.cx>
  65
  66         * doc/invoke.texi: Fix typo.
  67         * tree-vrp.c: Fix typos.
  68         * gimple.c (infer_nonnull_range): Reorder operands of an &&
  69         condition.
  70
  71 2014-04-29  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
  72
  73         * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
  74
  75 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
  76
  77         * config/aarch64/aarch64-builtins.c
  78         (aarch64_types_storestruct_lane_qualifiers): New.
  79         (TYPES_STORESTRUCT_LANE): Likewise.
  80         * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
  81         (st3_lane): Likewise.
  82         (st4_lane): Likewise.
  83         * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
  84         (vec_store_lanesci_lane<mode>): Likewise.
  85         (vec_store_lanesxi_lane<mode>): Likewise.
  86         (aarch64_st2_lane<VQ:mode>): Likewise.
  87         (aarch64_st3_lane<VQ:mode>): Likewise.
  88         (aarch64_st4_lane<VQ:mode>): Likewise.
  89         * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
  90         * config/aarch64/arm_neon.h
  91         (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
  92         use new macro arguments.
  93         (__ST3_LANE_FUNC): Likewise.
  94         (__ST4_LANE_FUNC): Likewise.
  95         * config/aarch64/iterators.md (V_TWO_ELEM): New.
  96         (V_THREE_ELEM): Likewise.
  97         (V_FOUR_ELEM): Likewise.
  98
  99 2014-04-28  David Malcolm  <dmalcolm@redhat.com>
 100
 101         * doc/gimple.texi: Replace the description of the now-defunct
 102         union gimple_statement_d with a diagram showing the
 103         gimple_statement_base class hierarchy and its relationships to
 104         the GSS_ and GIMPLE_ enums.
 105
 106 2014-04-28  James Greenhalgh  <james.greenhalgh@arm.com>
 107
 108         * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
 109         * config/aarch64/aarch64.c
 110         (aarch64_cannot_change_mode_class): Weaken conditions.
 111         (aarch64_modes_tieable_p): New.
 112         * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
 113
 114 2014-04-28  Pat Haugen  <pthaugen@us.ibm.com>
 115
 116         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
 117         (loadsync_<mode>): Change mode.
 118         (load_quadpti, store_quadpti): New.
 119         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
 120         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
 121
 122 2014-04-28  Martin Jambor  <mjambor@suse.cz>
 123
 124         * tree-sra.c (sra_modify_expr): Generate new memory accesses with
 125         same alias type as the original statement.
 126         (subreplacement_assignment_data): New type.
 127         (handle_unscalarized_data_in_subtree): New type of parameter,
 128         generate new memory accesses with same alias type as the original
 129         statement.
 130         (load_assign_lhs_subreplacements): Likewise.
 131         (sra_modify_constructor_assign): Generate new memory accesses with
 132         same alias type as the original statement.
 133
 134 2014-04-28  Richard Biener  <rguenther@suse.de>
 135
 136         * tree-pass.h (TODO_verify_il): Define.
 137         (TODO_verify_all): Complete properly.
 138         * passes.c (execute_function_todo): Move existing loop-closed
 139         SSA verification under TODO_verify_il.
 140         (execute_one_pass): Trigger TODO_verify_il at todo-after time.
 141         * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
 142         Fix tree sharing issue.
 143
 144 2014-04-28  Richard Biener  <rguenther@suse.de>
 145
 146         PR middle-end/60092
 147         * builtins.def (DEF_C11_BUILTIN): Add.
 148         (BUILT_IN_ALIGNED_ALLOC): Likewise.
 149         * coretypes.h (enum function_class): Add function_c11_misc.
 150         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
 151         BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
 152         (call_may_clobber_ref_p_1): Likewise.
 153         * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
 154         (mark_all_reaching_defs_necessary_1): Likewise.
 155         (propagate_necessity): Likewise.
 156         (eliminate_unnecessary_stmts): Likewise.
 157         * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
 158
 159 2014-04-28  Richard Biener  <rguenther@suse.de>
 160
 161         * tree-vrp.c (vrp_var_may_overflow): Remove.
 162         (vrp_visit_phi_node): Rather than bumping to +-INF possibly
 163         with overflow immediately bump to one before that value and
 164         let iteration figure out overflow status.
 165
 166 2014-04-28  Richard Biener  <rguenther@suse.de>
 167
 168         * configure.ac: Do valgrind header checks unconditionally.
 169         Add --enable-valgrind-annotations.
 170         * system.h: Guard valgrind header inclusion with
 171         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
 172         * alloc-pool.c (pool_alloc, pool_free): Use
 173         ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
 174         to guard possibly dead code.
 175         * config.in: Regenerated.
 176         * configure: Likewise.
 177
 178 2014-04-28  Jeff Law  <law@redhat.com>
 179
 180         PR tree-optimization/60902
 181         * tree-ssa-threadedge.c
 182         (record_temporary_equivalences_from_stmts_at_dest): Only iterate
 183         over real defs when invalidating outputs from statements that do not
 184         produce useful outputs for threading.
 185
 186 2014-04-28  Richard Biener  <rguenther@suse.de>
 187
 188         PR tree-optimization/60979
 189         * graphite-scop-detection.c (scopdet_basic_block_info): Reject
 190         SCOPs that end in a block with a successor with abnormal
 191         predecessors.
 192
 193 2014-04-28  Richard Biener  <rguenther@suse.de>
 194
 195         * tree-pass.h (execute_pass_list): Adjust prototype.
 196         * passes.c (pass_manager::execute_early_local_passes):
 197         Adjust.
 198         (do_per_function): Change callback signature, push all actual
 199         work to the callbals.
 200         (do_per_function_toporder): Likewise.
 201         (execute_function_dump): Adjust.
 202         (execute_function_todo): Likewise.
 203         (clear_last_verified): Likewise.
 204         (verify_curr_properties): Likewise.
 205         (update_properties_after_pass): Likewise.
 206         (execute_pass_list_1): Split out from ...
 207         (execute_pass_list): ... here.  Adjust.
 208         (execute_ipa_pass_list): Likewise.
 209         * cgraphunit.c (cgraph_add_new_function): Adjust.
 210         (analyze_function): Likewise.
 211         (expand_function): Likewise.
 212         * cgraph.c (release_function_body): Free dominance info
 213         here instead of asserting it was magically freed elsewhere.
 214
 215 2014-04-28  Eric Botcazou  <ebotcazou@adacore.com>
 216
 217         * configure.ac: Tweak GAS check for LEON instructions on SPARC.
 218         * configure: Regenerate.
 219         * config/sparc/sparc.opt (muser-mode): New option.
 220         * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
 221         for LEON3.
 222         (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
 223         * doc/invoke.texi (SPARC options): Document -muser-mode.
 224
 225 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
 226
 227         * cselib.c (find_slot_memmode): Delete.
 228         (cselib_hasher): Change compare_type to a struct.
 229         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
 230         constants.
 231         (preserve_constants_and_equivs): Adjust for new compare_type.
 232         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
 233         (wrap_constant): Delete.
 234         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
 235
 236 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
 237
 238         * doc/install.texi (Building with profile feedback): Remove
 239         outdated sentence.
 240
 241 2014-04-26  Tom de Vries  <tom@codesourcery.com>
 242
 243         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
 244         array accesses.
 245
 246 2014-04-25  Cary Coutant  <ccoutant@google.com>
 247
 248         PR debug/60929
 249         * dwarf2out.c (should_move_die_to_comdat): A type definition
 250         can contain a subprogram definition, but don't move it to a
 251         comdat unit.
 252         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
 253         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
 254         from original DIE.
 255         (clone_tree_hash): Rename to...
 256         (clone_tree_partial): ...this; change callers.  Copy
 257         DW_TAG_subprogram DIEs as declarations.
 258         (copy_decls_walk): Don't copy children of a declaration into a
 259         type unit.
 260
 261 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
 262
 263         PR target/60969
 264         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
 265         alternative 12.
 266
 267 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
 268
 269         * config/arm/predicates.md (call_insn_operand): Add long_call check.
 270         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
 271         reg for long_call.
 272         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
 273         restriction.
 274
 275 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 276
 277         * config/arm/arm.c (arm_cortex_a8_tune): Initialise
 278         T16-related fields.
 279
 280 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 281
 282         PR tree-optimization/60930
 283         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
 284         creating a multiply candidate by folding two constant
 285         multiplicands when the result overflows.
 286
 287 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
 288
 289         PR tree-optimization/60960
 290         * tree-vect-generic.c (expand_vector_operation): Only call
 291         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
 292
 293 2014-04-25  Tom de Vries  <tom@codesourcery.com>
 294
 295         * expr.c (clobber_reg_mode): New function.
 296         * expr.h (clobber_reg): New function.
 297
 298 2014-04-25  Tom de Vries  <tom@codesourcery.com>
 299
 300         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
 301         clobbers.
 302
 303 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
 304             Tom de Vries  <tom@codesourcery.com>
 305
 306         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
 307         handle.
 308         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
 309         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
 310         new argument to find_all_hard_reg_sets call.
 311
 312 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 313
 314         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
 315         Use HOST_WIDE_INT_C for mask literal.
 316         (aarch_rev16_shleft_mask_imm_p): Likewise.
 317
 318 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
 319
 320         PR target/60941
 321         * config/sparc/sparc.md (ashlsi3_extend): Delete.
 322
 323 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
 324
 325         PR preprocessor/56540
 326         * config/i386/i386-c.c (ix86_target_macros): Define
 327         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
 328
 329 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 330
 331         * configure.ac (tga_func): Remove.
 332         (LIB_TLS_SPEC): Remove.
 333         * configure: Regenerate.
 334         * config.in: Regenerate.
 335         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
 336
 337 2014-04-25  Richard Biener  <rguenther@suse.de>
 338
 339         PR ipa/60912
 340         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
 341         call stmt use/clobber sets during stmt walk instead of
 342         walking the possibly incomplete set of caller edges.
 343
 344 2014-04-25  Richard Biener  <rguenther@suse.de>
 345
 346         PR ipa/60911
 347         * passes.c (apply_ipa_transforms): Inline into only caller ...
 348         (execute_one_pass): ... here.  Properly bring in function
 349         bodies for nodes we want to apply IPA transforms to.
 350
 351 2014-04-24  Cong Hou  <congh@google.com>
 352
 353         PR tree-optimization/60896
 354         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
 355         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
 356         (vect_mark_pattern_stmts): Set the def type of all statements in
 357         PATTERN_DEF_SEQ as vect_internal_def.
 358
 359 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
 360
 361         * doc/extend.texi (PowerPC Built-in Functions): Document new
 362         powerpc extended divide, bcd, pack/unpack 128-bit, builtin
 363         functions.
 364         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
 365
 366         * config/rs6000/predicates.md (const_0_to_3_operand): New
 367         predicate to match 0..3 integer constants.
 368
 369         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
 370         to support adding miscellaneous builtin functions.
 371         (BU_DFP_MISC_2): Likewise.
 372         (BU_P7_MISC_1): Likewise.
 373         (BU_P7_MISC_2): Likewise.
 374         (BU_P8V_MISC_3): Likewise.
 375         (BU_MISC_1): Likewise.
 376         (BU_MISC_2): Likewise.
 377         (DIVWE): Add extended divide builtin functions.
 378         (DIVWEO): Likewise.
 379         (DIVWEU): Likewise.
 380         (DIVWEUO): Likewise.
 381         (DIVDE): Likewise.
 382         (DIVDEO): Likewise.
 383         (DIVDEU): Likewise.
 384         (DIVDEUO): Likewise.
 385         (DXEX): Add decimal floating-point builtin functions.
 386         (DXEXQ): Likewise.
 387         (DDEDPD): Likewise.
 388         (DDEDPDQ): Likewise.
 389         (DENBCD): Likewise.
 390         (DENBCDQ): Likewise.
 391         (DIEX): Likewise.
 392         (DIEXQ): Likewise.
 393         (DSCLI): Likewise.
 394         (DSCLIQ): Likewise.
 395         (DSCRI): Likewise.
 396         (DSCRIQ): Likewise.
 397         (CDTBCD): Add new BCD builtin functions.
 398         (CBCDTD): Likewise.
 399         (ADDG6S): Likewise.
 400         (BCDADD): Likewise.
 401         (BCDADD_LT): Likewise.
 402         (BCDADD_EQ): Likewise.
 403         (BCDADD_GT): Likewise.
 404         (BCDADD_OV): Likewise.
 405         (BCDSUB): Likewise.
 406         (BCDSUB_LT): Likewise.
 407         (BCDSUB_EQ): Likewise.
 408         (BCDSUB_GT): Likewise.
 409         (BCDSUB_OV): Likewise.
 410         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
 411         (UNPACK_TD): Likewise.
 412         (PACK_TF): Likewise.
 413         (UNPACK_TF): Likewise.
 414         (UNPACK_TF_0): Likewise.
 415         (UNPACK_TF_1): Likewise.
 416         (PACK_V1TI): Likewise.
 417         (UNPACK_V1TI): Likewise.
 418
 419         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
 420         support for decimal floating point builtin functions.
 421         (rs6000_expand_ternop_builtin): Add checks for the new builtin
 422         functions that take constant arguments.
 423         (rs6000_invalid_builtin): Add decimal floating point builtin
 424         support.
 425         (rs6000_init_builtins): Setup long double, _Decimal64, and
 426         _Decimal128 types for new builtin functions.
 427         (builtin_function_type): Set the unsigned flags appropriately for
 428         the new builtin functions.
 429         (rs6000_opt_masks): Add support for decimal floating point builtin
 430         functions.
 431
 432         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
 433         floating point builtin functions.
 434         (RS6000_BTM_COMMON): Likewise.
 435         (RS6000_BTI_long_double): Likewise.
 436         (RS6000_BTI_dfloat64): Likewise.
 437         (RS6000_BTI_dfloat128): Likewise.
 438         (long_double_type_internal_node): Likewise.
 439         (dfloat64_type_internal_node): Likewise.
 440         (dfloat128_type_internal_node): Likewise.
 441
 442         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
 443         2.07 bcd arithmetic instructions.
 444         (UNSPEC_BCDSUB): Likewise.
 445         (UNSPEC_BCD_OVERFLOW): Likewise.
 446         (UNSPEC_BCD_ADD_SUB): Likewise.
 447         (bcd_add_sub): Likewise.
 448         (BCD_TEST): Likewise.
 449         (bcd<bcd_add_sub>): Likewise.
 450         (bcd<bcd_add_sub>_test): Likewise.
 451         (bcd<bcd_add_sub>_test2): Likewise.
 452         (bcd<bcd_add_sub>_<code>): Likewise.
 453         (peephole2 for combined bcd ops): Likewise.
 454
 455         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
 456         decimal floating point builtin functions.
 457         (UNSPEC_DENBCD): Likewise.
 458         (UNSPEC_DXEX): Likewise.
 459         (UNSPEC_DIEX): Likewise.
 460         (UNSPEC_DSCLI): Likewise.
 461         (UNSPEC_DSCRI): Likewise.
 462         (D64_D128): Likewise.
 463         (dfp_suffix): Likewise.
 464         (dfp_ddedpd_<mode>): Likewise.
 465         (dfp_denbcd_<mode>): Likewise.
 466         (dfp_dxex_<mode>): Likewise.
 467         (dfp_diex_<mode>): Likewise.
 468         (dfp_dscli_<mode>): Likewise.
 469         (dfp_dscri_<mode>): Likewise.
 470
 471         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
 472         builtin functions.
 473         (UNSPEC_CDTBCD): Likewise.
 474         (UNSPEC_CBCDTD): Likewise.
 475         (UNSPEC_DIVE): Add support for new extended divide builtin
 476         functions.
 477         (UNSPEC_DIVEO): Likewise.
 478         (UNSPEC_DIVEU): Likewise.
 479         (UNSPEC_DIVEUO): Likewise.
 480         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
 481         pack/unpack 128-bit types.
 482         (UNSPEC_PACK_128BIT): Likewise.
 483         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
 484         (udiv<mode>3): Use idiv_ldiv mode attribute.
 485         (div<mode>3): Likewise.
 486         (addg6s): Add new BCD builtin functions.
 487         (cdtbcd): Likewise.
 488         (cbcdtd): Likewise.
 489         (UNSPEC_DIV_EXTEND): Add support for new extended divide
 490         instructions.
 491         (div_extend): Likewise.
 492         (div<div_extend>_<mode>"): Likewise.
 493         (FP128_64): Add support for new builtin functions to pack/unpack
 494         128-bit types.
 495         (unpack<mode>): Likewise.
 496         (unpacktf_0): Likewise.
 497         (unpacktf_1): Likewise.
 498         (unpack<mode>_dm): Likewise.
 499         (unpack<mode>_nodm): Likewise.
 500         (pack<mode>): Likewise.
 501         (unpackv1ti): Likewise.
 502         (packv1ti): Likewise.
 503
 504 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
 505
 506         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
 507         is disabled.
 508
 509 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
 510
 511         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
 512         * gimplify.c (omp_is_private): Change last argument's type to int.
 513         Only diagnose lastprivate if the simd argument is 1, only diagnose
 514         linear if the simd argument is 2.
 515         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
 516         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
 517         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
 518         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
 519         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
 520         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 521         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
 522         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
 523         * tree-nested.c (convert_nonlocal_omp_clauses,
 524         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
 525
 526 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
 527
 528         PR target/60822
 529         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
 530         operand 1.
 531
 532 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
 533
 534         * flag-types.h (enum ivar_visibility): Add.
 535
 536 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
 537
 538         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
 539         function * argument.
 540
 541 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
 542
 543         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
 544
 545 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 546             Tom de Vries  <tom@codesourcery.com>
 547
 548         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
 549         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
 550         reg-note.
 551         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
 552         * emit-rtl.c (try_split): Same.
 553
 554 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
 555             Tom de Vries  <tom@codesourcery.com>
 556
 557         * common.opt (fuse-caller-save): New option.
 558
 559 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
 560
 561         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
 562         for big-endian.
 563
 564 2014-04-24  Richard Biener  <rguenther@suse.de>
 565
 566         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
 567         during TER and instead use the sepops interface for expanding
 568         non-GIMPLE_SINGLE_RHS.
 569
 570 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 571
 572         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
 573         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
 574
 575 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
 576
 577         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
 578         assembler 64-bit option.
 579         * configure: Regenerate.
 580
 581 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 582
 583         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
 584         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
 585         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
 586         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
 587         (TARGET_CRYPTO): Take TARGET_SIMD into account.
 588
 589 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
 590
 591         * config/aarch64/aarch64-builtins.c
 592         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
 593         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
 594         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
 595         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
 596         builtins.
 597         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
 598         (Vrevsuff): New mode attribute.
 599
 600 2014-04-24  Terry Guo  <terry.guo@arm.com>
 601
 602         * config/arm/arm.h (machine_function): Define variable
 603         after_arm_reorg here.
 604         * config/arm/arm.c (after_arm_reorg): Remove the definition.
 605         (arm_split_constant): Update the way to access variable
 606         after_arm_reorg.
 607         (arm_reorg): Ditto.
 608         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
 609
 610 2014-04-23  Tom de Vries  <tom@codesourcery.com>
 611
 612         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
 613
 614 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
 615
 616         * is-a.h: Update comments to reflect the following changes to the
 617         "pointerness" of the API, making the template parameter match the
 618         return type, allowing use of is-a.h with typedefs of pointers.
 619         (is_a_helper::cast): Return a T rather then a pointer to a T, so
 620         that the return type matches the parameter to the is_a_helper.
 621         (as_a): Likewise.
 622         (dyn_cast): Likewise.
 623
 624         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
 625         pointer from the is-a.h API.
 626
 627         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
 628         (is_a_helper <cgraph_node *>::test): ...this, matching change to
 629         is-a.h API.
 630         (is_a_helper <varpool_node>::test): Likewise, convert to...
 631         (is_a_helper <varpool_node *>::test): ...this.
 632
 633         (varpool_first_variable): Update for removal of implicit pointer
 634         from the is-a.h API.
 635         (varpool_next_variable): Likewise.
 636         (varpool_first_static_initializer): Likewise.
 637         (varpool_next_static_initializer): Likewise.
 638         (varpool_first_defined_variable): Likewise.
 639         (varpool_next_defined_variable): Likewise.
 640         (cgraph_first_defined_function): Likewise.
 641         (cgraph_next_defined_function): Likewise.
 642         (cgraph_first_function): Likewise.
 643         (cgraph_next_function): Likewise.
 644         (cgraph_first_function_with_gimple_body): Likewise.
 645         (cgraph_next_function_with_gimple_body): Likewise.
 646         (cgraph_alias_target): Likewise.
 647         (varpool_alias_target): Likewise.
 648         (cgraph_function_or_thunk_node): Likewise.
 649         (varpool_variable_node): Likewise.
 650         (symtab_real_symbol_p): Likewise.
 651         * cgraphunit.c (referred_to_p): Likewise.
 652         (analyze_functions): Likewise.
 653         (handle_alias_pairs): Likewise.
 654         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
 655         * gimple-ssa.h (gimple_vuse_op): Likewise.
 656         (gimple_vdef_op): Likewise.
 657         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
 658         * gimple.c (gimple_build_asm_1): Likewise.
 659         (gimple_build_try): Likewise.
 660         (gimple_build_resx): Likewise.
 661         (gimple_build_eh_dispatch): Likewise.
 662         (gimple_build_omp_for): Likewise.
 663         (gimple_omp_for_set_clauses): Likewise.
 664
 665         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
 666         (is_a_helper <gimple_statement_asm *>::test): ...this.
 667         (is_a_helper <gimple_statement_bind>::test): Convert to...
 668         (is_a_helper <gimple_statement_bind *>::test): ...this.
 669         (is_a_helper <gimple_statement_call>::test): Convert to...
 670         (is_a_helper <gimple_statement_call *>::test): ...this.
 671         (is_a_helper <gimple_statement_catch>::test): Convert to...
 672         (is_a_helper <gimple_statement_catch *>::test): ...this.
 673         (is_a_helper <gimple_statement_resx>::test): Convert to...
 674         (is_a_helper <gimple_statement_resx *>::test): ...this.
 675         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
 676         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
 677         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
 678         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
 679         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
 680         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
 681         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
 682         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
 683         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
 684         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
 685         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
 686         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
 687         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
 688         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
 689         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
 690         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
 691         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
 692         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
 693         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
 694         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
 695         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
 696         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
 697         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
 698         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
 699         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
 700         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
 701         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
 702         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
 703         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
 704         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
 705         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
 706         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
 707         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
 708         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
 709         (is_a_helper <gimple_statement_phi>::test): Convert to...
 710         (is_a_helper <gimple_statement_phi *>::test): ...this.
 711         (is_a_helper <gimple_statement_transaction>::test): Convert to...
 712         (is_a_helper <gimple_statement_transaction *>::test): ...this.
 713         (is_a_helper <gimple_statement_try>::test): Convert to...
 714         (is_a_helper <gimple_statement_try *>::test): ...this.
 715         (is_a_helper <gimple_statement_wce>::test): Convert to...
 716         (is_a_helper <gimple_statement_wce *>::test): ...this.
 717         (is_a_helper <const gimple_statement_asm>::test): Convert to...
 718         (is_a_helper <const gimple_statement_asm *>::test): ...this.
 719         (is_a_helper <const gimple_statement_bind>::test): Convert to...
 720         (is_a_helper <const gimple_statement_bind *>::test): ...this.
 721         (is_a_helper <const gimple_statement_call>::test): Convert to...
 722         (is_a_helper <const gimple_statement_call *>::test): ...this.
 723         (is_a_helper <const gimple_statement_catch>::test): Convert to...
 724         (is_a_helper <const gimple_statement_catch *>::test): ...this.
 725         (is_a_helper <const gimple_statement_resx>::test): Convert to...
 726         (is_a_helper <const gimple_statement_resx *>::test): ...this.
 727         (is_a_helper <const gimple_statement_eh_dispatch>::test):
 728         Convert to...
 729         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
 730         ...this.
 731         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
 732         to...
 733         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
 734         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
 735         Convert to...
 736         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
 737         ...this.
 738         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
 739         Convert to...
 740         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
 741         ...this.
 742         (is_a_helper <const gimple_statement_omp_return>::test): Convert
 743         to...
 744         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
 745         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
 746         to...
 747         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
 748         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
 749         to...
 750         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
 751         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
 752         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
 753         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
 754         to...
 755         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
 756         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
 757         to...
 758         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
 759         (is_a_helper <const gimple_statement_omp_target>::test): Convert
 760         to...
 761         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
 762         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
 763         to...
 764         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
 765         (is_a_helper <const gimple_statement_omp_single>::test): Convert
 766         to...
 767         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
 768         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
 769         to...
 770         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
 771         (is_a_helper <const gimple_statement_omp_task>::test): Convert
 772         to...
 773         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
 774         (is_a_helper <const gimple_statement_phi>::test): Convert to...
 775         (is_a_helper <const gimple_statement_phi *>::test): ...this.
 776         (is_a_helper <const gimple_statement_transaction>::test): Convert
 777         to...
 778         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
 779         (is_a_helper <const gimple_statement_with_ops>::test): Convert
 780         to...
 781         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
 782         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
 783         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
 784         (is_a_helper <const gimple_statement_with_memory_ops>::test):
 785         Convert to...
 786         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
 787         ...this.
 788         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
 789         to...
 790         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
 791
 792         (gimple_use_ops): Update for removal of implicit pointer from the
 793         is-a.h API.
 794         (gimple_set_use_ops): Likewise.
 795         (gimple_vuse): Likewise.
 796         (gimple_vdef): Likewise.
 797         (gimple_vuse_ptr): Likewise.
 798         (gimple_vdef_ptr): Likewise.
 799         (gimple_set_vuse): Likewise.
 800         (gimple_set_vdef): Likewise.
 801         (gimple_omp_return_set_lhs): Likewise.
 802         (gimple_omp_return_lhs): Likewise.
 803         (gimple_omp_return_lhs_ptr): Likewise.
 804         (gimple_call_fntype): Likewise.
 805         (gimple_call_set_fntype): Likewise.
 806         (gimple_call_set_internal_fn): Likewise.
 807         (gimple_call_use_set): Likewise.
 808         (gimple_call_clobber_set): Likewise.
 809         (gimple_bind_vars): Likewise.
 810         (gimple_bind_set_vars): Likewise.
 811         (gimple_bind_body_ptr): Likewise.
 812         (gimple_bind_set_body): Likewise.
 813         (gimple_bind_add_stmt): Likewise.
 814         (gimple_bind_block): Likewise.
 815         (gimple_bind_set_block): Likewise.
 816         (gimple_asm_ninputs): Likewise.
 817         (gimple_asm_noutputs): Likewise.
 818         (gimple_asm_nclobbers): Likewise.
 819         (gimple_asm_nlabels): Likewise.
 820         (gimple_asm_input_op): Likewise.
 821         (gimple_asm_input_op_ptr): Likewise.
 822         (gimple_asm_output_op): Likewise.
 823         (gimple_asm_output_op_ptr): Likewise.
 824         (gimple_asm_set_output_op): Likewise.
 825         (gimple_asm_clobber_op): Likewise.
 826         (gimple_asm_set_clobber_op): Likewise.
 827         (gimple_asm_label_op): Likewise.
 828         (gimple_asm_set_label_op): Likewise.
 829         (gimple_asm_string): Likewise.
 830         (gimple_catch_types): Likewise.
 831         (gimple_catch_types_ptr): Likewise.
 832         (gimple_catch_handler_ptr): Likewise.
 833         (gimple_catch_set_types): Likewise.
 834         (gimple_catch_set_handler): Likewise.
 835         (gimple_eh_filter_types): Likewise.
 836         (gimple_eh_filter_types_ptr): Likewise.
 837         (gimple_eh_filter_failure_ptr): Likewise.
 838         (gimple_eh_filter_set_types): Likewise.
 839         (gimple_eh_filter_set_failure): Likewise.
 840         (gimple_eh_must_not_throw_fndecl): Likewise.
 841         (gimple_eh_must_not_throw_set_fndecl): Likewise.
 842         (gimple_eh_else_n_body_ptr): Likewise.
 843         (gimple_eh_else_e_body_ptr): Likewise.
 844         (gimple_eh_else_set_n_body): Likewise.
 845         (gimple_eh_else_set_e_body): Likewise.
 846         (gimple_try_eval_ptr): Likewise.
 847         (gimple_try_cleanup_ptr): Likewise.
 848         (gimple_try_set_eval): Likewise.
 849         (gimple_try_set_cleanup): Likewise.
 850         (gimple_wce_cleanup_ptr): Likewise.
 851         (gimple_wce_set_cleanup): Likewise.
 852         (gimple_phi_capacity): Likewise.
 853         (gimple_phi_num_args): Likewise.
 854         (gimple_phi_result): Likewise.
 855         (gimple_phi_result_ptr): Likewise.
 856         (gimple_phi_set_result): Likewise.
 857         (gimple_phi_arg): Likewise.
 858         (gimple_phi_set_arg): Likewise.
 859         (gimple_resx_region): Likewise.
 860         (gimple_resx_set_region): Likewise.
 861         (gimple_eh_dispatch_region): Likewise.
 862         (gimple_eh_dispatch_set_region): Likewise.
 863         (gimple_omp_critical_name): Likewise.
 864         (gimple_omp_critical_name_ptr): Likewise.
 865         (gimple_omp_critical_set_name): Likewise.
 866         (gimple_omp_for_clauses): Likewise.
 867         (gimple_omp_for_clauses_ptr): Likewise.
 868         (gimple_omp_for_set_clauses): Likewise.
 869         (gimple_omp_for_collapse): Likewise.
 870         (gimple_omp_for_index): Likewise.
 871         (gimple_omp_for_index_ptr): Likewise.
 872         (gimple_omp_for_set_index): Likewise.
 873         (gimple_omp_for_initial): Likewise.
 874         (gimple_omp_for_initial_ptr): Likewise.
 875         (gimple_omp_for_set_initial): Likewise.
 876         (gimple_omp_for_final): Likewise.
 877         (gimple_omp_for_final_ptr): Likewise.
 878         (gimple_omp_for_set_final): Likewise.
 879         (gimple_omp_for_incr): Likewise.
 880         (gimple_omp_for_incr_ptr): Likewise.
 881         (gimple_omp_for_set_incr): Likewise.
 882         (gimple_omp_for_pre_body_ptr): Likewise.
 883         (gimple_omp_for_set_pre_body): Likewise.
 884         (gimple_omp_parallel_clauses): Likewise.
 885         (gimple_omp_parallel_clauses_ptr): Likewise.
 886         (gimple_omp_parallel_set_clauses): Likewise.
 887         (gimple_omp_parallel_child_fn): Likewise.
 888         (gimple_omp_parallel_child_fn_ptr): Likewise.
 889         (gimple_omp_parallel_set_child_fn): Likewise.
 890         (gimple_omp_parallel_data_arg): Likewise.
 891         (gimple_omp_parallel_data_arg_ptr): Likewise.
 892         (gimple_omp_parallel_set_data_arg): Likewise.
 893         (gimple_omp_task_clauses): Likewise.
 894         (gimple_omp_task_clauses_ptr): Likewise.
 895         (gimple_omp_task_set_clauses): Likewise.
 896         (gimple_omp_task_child_fn): Likewise.
 897         (gimple_omp_task_child_fn_ptr): Likewise.
 898         (gimple_omp_task_set_child_fn): Likewise.
 899         (gimple_omp_task_data_arg): Likewise.
 900         (gimple_omp_task_data_arg_ptr): Likewise.
 901         (gimple_omp_task_set_data_arg): Likewise.
 902         (gimple_omp_taskreg_clauses): Likewise.
 903         (gimple_omp_taskreg_clauses_ptr): Likewise.
 904         (gimple_omp_taskreg_set_clauses): Likewise.
 905         (gimple_omp_taskreg_child_fn): Likewise.
 906         (gimple_omp_taskreg_child_fn_ptr): Likewise.
 907         (gimple_omp_taskreg_set_child_fn): Likewise.
 908         (gimple_omp_taskreg_data_arg): Likewise.
 909         (gimple_omp_taskreg_data_arg_ptr): Likewise.
 910         (gimple_omp_taskreg_set_data_arg): Likewise.
 911         (gimple_omp_task_copy_fn): Likewise.
 912         (gimple_omp_task_copy_fn_ptr): Likewise.
 913         (gimple_omp_task_set_copy_fn): Likewise.
 914         (gimple_omp_task_arg_size): Likewise.
 915         (gimple_omp_task_arg_size_ptr): Likewise.
 916         (gimple_omp_task_set_arg_size): Likewise.
 917         (gimple_omp_task_arg_align): Likewise.
 918         (gimple_omp_task_arg_align_ptr): Likewise.
 919         (gimple_omp_task_set_arg_align): Likewise.
 920         (gimple_omp_single_clauses): Likewise.
 921         (gimple_omp_single_clauses_ptr): Likewise.
 922         (gimple_omp_single_set_clauses): Likewise.
 923         (gimple_omp_target_clauses): Likewise.
 924         (gimple_omp_target_clauses_ptr): Likewise.
 925         (gimple_omp_target_set_clauses): Likewise.
 926         (gimple_omp_target_child_fn): Likewise.
 927         (gimple_omp_target_child_fn_ptr): Likewise.
 928         (gimple_omp_target_set_child_fn): Likewise.
 929         (gimple_omp_target_data_arg): Likewise.
 930         (gimple_omp_target_data_arg_ptr): Likewise.
 931         (gimple_omp_target_set_data_arg): Likewise.
 932         (gimple_omp_teams_clauses): Likewise.
 933         (gimple_omp_teams_clauses_ptr): Likewise.
 934         (gimple_omp_teams_set_clauses): Likewise.
 935         (gimple_omp_sections_clauses): Likewise.
 936         (gimple_omp_sections_clauses_ptr): Likewise.
 937         (gimple_omp_sections_set_clauses): Likewise.
 938         (gimple_omp_sections_control): Likewise.
 939         (gimple_omp_sections_control_ptr): Likewise.
 940         (gimple_omp_sections_set_control): Likewise.
 941         (gimple_omp_for_set_cond): Likewise.
 942         (gimple_omp_for_cond): Likewise.
 943         (gimple_omp_atomic_store_set_val): Likewise.
 944         (gimple_omp_atomic_store_val): Likewise.
 945         (gimple_omp_atomic_store_val_ptr): Likewise.
 946         (gimple_omp_atomic_load_set_lhs): Likewise.
 947         (gimple_omp_atomic_load_lhs): Likewise.
 948         (gimple_omp_atomic_load_lhs_ptr): Likewise.
 949         (gimple_omp_atomic_load_set_rhs): Likewise.
 950         (gimple_omp_atomic_load_rhs): Likewise.
 951         (gimple_omp_atomic_load_rhs_ptr): Likewise.
 952         (gimple_omp_continue_control_def): Likewise.
 953         (gimple_omp_continue_control_def_ptr): Likewise.
 954         (gimple_omp_continue_set_control_def): Likewise.
 955         (gimple_omp_continue_control_use): Likewise.
 956         (gimple_omp_continue_control_use_ptr): Likewise.
 957         (gimple_omp_continue_set_control_use): Likewise.
 958         (gimple_transaction_body_ptr): Likewise.
 959         (gimple_transaction_label): Likewise.
 960         (gimple_transaction_label_ptr): Likewise.
 961         (gimple_transaction_set_body): Likewise.
 962         (gimple_transaction_set_label): Likewise.
 963
 964         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
 965         * ipa-inline-analysis.c (inline_write_summary): Likewise.
 966         * ipa-ref.c (ipa_record_reference): Likewise.
 967         * ipa-reference.c (analyze_function): Likewise.
 968         (ipa_reference_write_optimization_summary): Likewise.
 969         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
 970         (address_taken_from_non_vtable_p): Likewise.
 971         (comdat_can_be_unshared_p_1): Likewise.
 972         * lto-cgraph.c (lto_output_ref): Likewise.
 973         (add_references): Likewise.
 974         (compute_ltrans_boundary): Likewise.
 975         (output_symtab): Likewise.
 976         (input_ref): Likewise.
 977         (input_cgraph_1): Likewise.
 978         (output_cgraph_opt_summary): Likewise.
 979         * lto-streamer-out.c (lto_output): Likewise.
 980         (output_symbol_p): Likewise.
 981         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
 982         (lsei_start_function_in_partition): Likewise.
 983         (lsei_next_variable_in_partition): Likewise.
 984         (lsei_start_variable_in_partition): Likewise.
 985         * symtab.c (insert_to_assembler_name_hash): Likewise.
 986         (unlink_from_assembler_name_hash): Likewise.
 987         (symtab_unregister_node): Likewise.
 988         (symtab_remove_node): Likewise.
 989         (dump_symtab_node): Likewise.
 990         (verify_symtab_base): Likewise.
 991         (verify_symtab_node): Likewise.
 992         (symtab_make_decl_local): Likewise.
 993         (symtab_alias_ultimate_target): Likewise.
 994         (symtab_resolve_alias): Likewise.
 995         (symtab_get_symbol_partitioning_class): Likewise.
 996         * tree-phinodes.c (allocate_phi_node): Likewise.
 997         (reserve_phi_args_for_new_edge): Likewise.
 998         (remove_phi_args): Likewise.
 999         * varpool.c (varpool_node_for_asm): Likewise.
1000         (varpool_remove_unreferenced_decls): Likewise.
1001
1002 2014-04-23  Jeff Law  <law@redhat.com>
1003
1004         PR tree-optimization/60902
1005         * tree-ssa-threadedge.c
1006         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
1007         invalidate outputs from statements that do not produce useful
1008         outputs for threading.
1009
1010 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
1011
1012         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
1013         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
1014         machine descriptions for Stack Smashing Protector.
1015
1016 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
1017
1018         * aarch64.md (<optab>_rol<mode>3): New pattern.
1019         (<optab>_rolsi3_uxtw): Likewise.
1020         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
1021
1022 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
1023
1024         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
1025         (arm_cortex_a12_tune): Likewise.
1026
1027 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1028
1029         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
1030
1031 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1032
1033         * config/arm/arm.md (arm_rev16si2): New pattern.
1034         (arm_rev16si2_alt): Likewise.
1035         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
1036
1037 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1038
1039        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
1040        (rev16<mode>2_alt): Likewise.
1041        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
1042        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
1043        (aarch_rev16_shleft_mask_imm_p): Likewise.
1044        (aarch_rev16_p_1): Likewise.
1045        (aarch_rev16_p): Likewise.
1046        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
1047        (aarch_rev16_shright_mask_imm_p): Likewise.
1048        (aarch_rev16_shleft_mask_imm_p): Likewise.
1049
1050 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1051
1052        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
1053        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
1054        rev cost.
1055        (cortex_a53_extra_costs): Likewise.
1056        (cortex_a57_extra_costs): Likewise.
1057        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
1058        (cortexa7_extra_costs): Likewise.
1059        (cortexa8_extra_costs): Likewise.
1060        (cortexa12_extra_costs): Likewise.
1061        (cortexa15_extra_costs): Likewise.
1062        (v7m_extra_costs): Likewise.
1063        (arm_new_rtx_costs): Handle BSWAP.
1064
1065 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1066
1067        * config/arm/arm.c (cortexa8_extra_costs): New table.
1068        (arm_cortex_a8_tune): New tuning struct.
1069        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
1070
1071 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1072
1073        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
1074
1075 2014-04-23  Richard Biener  <rguenther@suse.de>
1076
1077         * Makefile.in (OBJS): Remove loop-unswitch.o.
1078         * loop-unswitch.c: Delete.
1079         * tree-pass.h (make_pass_rtl_unswitch): Remove.
1080         * passes.def (pass_rtl_unswitch): Likewise.
1081         * loop-init.c (gate_rtl_unswitch): Likewise.
1082         (rtl_unswitch): Likewise.
1083         (pass_data_rtl_unswitch): Likewise.
1084         (pass_rtl_unswitch): Likewise.
1085         (make_pass_rtl_unswitch): Likewise.
1086         * rtl.h (reversed_condition): Likewise.
1087         (compare_and_jump_seq): Likewise.
1088         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
1089         and make static.
1090         * loop-unroll.c (compare_and_jump_seq): Likewise.
1091
1092 2014-04-23  Richard Biener  <rguenther@suse.de>
1093
1094         PR tree-optimization/60903
1095         * tree-ssa-loop-im.c (analyze_memory_references): Remove
1096         commented code block.
1097         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
1098         loop flags to newly created BBs and edges.
1099
1100 2014-04-23  Nick Clifton  <nickc@redhat.com>
1101
1102         * config/msp430/msp430.c (msp430_handle_option): Move function
1103         to msp430-common.c
1104         (msp430_option_override): Simplify mcu and mcpu option handling.
1105         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
1106         support for -mhwmult command line option.
1107         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
1108         -mhwmult command line option.
1109         (msp430_hwmult_enabled): Delete.
1110         (msp43o_output_labelref): Add support for -mhwmult command line
1111         option.
1112         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
1113         (umulsidi3): Likewise.
1114         * config/msp430/msp430.opt (mmcu): Add Report attribute.
1115         (mcpu, mlarge, msmall): Likewise.
1116         (mhwmult): New option.
1117         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
1118         prototype.
1119         (msp430_is_f5_mcu): Remove prototype.
1120         (msp430_use_f5_series_hwmult): Add prototype.
1121         * config/msp430/msp430-opts.h: New file.
1122         * common/config/msp430: New directory.
1123         * common/config/msp430/msp430-common.c: New file.
1124         * config.gcc (msp430): Remove target_has_targetm_common.
1125         * doc/invoke.texi: Document -mhwmult command line option.
1126
1127 2014-04-23  Nick Clifton  <nickc@redhat.com>
1128
1129         * config/i386/cygwin.h (ENDFILE_SPEC): Include
1130         default-manifest.o if it can be found in the search path.
1131         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
1132
1133 2014-04-23  Terry Guo  <terry.guo@arm.com>
1134
1135         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
1136
1137 2014-04-23  Richard Biener  <rguenther@suse.de>
1138
1139         PR middle-end/60895
1140         * tree-inline.c (declare_return_variable): Use mark_addressable.
1141
1142 2014-04-23  Richard Biener  <rguenther@suse.de>
1143
1144         PR middle-end/60891
1145         * loop-init.c (loop_optimizer_init): Make sure to apply
1146         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
1147
1148 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1149
1150         PR sanitizer/60275
1151         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
1152         New options.
1153         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
1154         if flag_sanitize_undefined_trap_on_error.
1155         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
1156         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
1157         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
1158         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1159         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
1160         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
1161         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
1162         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
1163         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
1164         * ubsan.c (ubsan_instrument_unreachable): Return
1165         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
1166         (ubsan_expand_null_ifn): Emit __builtin_trap ()
1167         if flag_sanitize_undefined_trap_on_error and
1168         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
1169         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
1170         instrument_bool_enum_load): Emit __builtin_trap () if
1171         flag_sanitize_undefined_trap_on_error and
1172         __builtin_handle_*_abort () if !flag_sanitize_recover.
1173         * doc/invoke.texi (-fsanitize-recover,
1174         -fsanitize-undefined-trap-on-error): Document.
1175
1176 2014-04-22  Christian Bruel  <christian.bruel@st.com>
1177
1178         * config/sh/sh.md (mov<mode>): Replace movQIHI.
1179          Force immediates to SImode.
1180
1181 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
1182
1183         * config/nios2/nios2.md (UNSPEC_ROUND): New.
1184         (lroundsfsi2): New.
1185         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
1186         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
1187         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
1188         (nios2_fpu_insn): Add entry for round.
1189         (N2FPU_NO_ERRNO_P): Define.
1190         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
1191         flag_errno_math.
1192         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
1193
1194 2014-04-22  Richard Henderson  <rth@redhat.com>
1195
1196         * config/aarch64/aarch64 (addti3, subti3): New expanders.
1197         (add<GPI>3_compare0): Remove leading * from name.
1198         (add<GPI>3_carryin): Likewise.
1199         (sub<GPI>3_compare0): Likewise.
1200         (sub<GPI>3_carryin): Likewise.
1201         (<su_optab>mulditi3): New expander.
1202         (multi3): New expander.
1203         (madd<GPI>): Remove leading * from name.
1204
1205 2014-04-22  Martin Jambor  <mjambor@suse.cz>
1206
1207         * cgraphclones.c (cgraph_function_versioning): Copy
1208         ipa_transforms_to_apply instead of asserting it is empty.
1209
1210 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
1211
1212         PR target/60868
1213         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
1214         on count_exp to get mode.
1215
1216 2014-04-22  Andrew Pinski  <apinski@cavium.com>
1217
1218         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1219         Handle TLS for ILP32.
1220         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
1221         (tlsie_small_<mode>): this and handle PTR.
1222         (tlsie_small_sidi): New pattern.
1223         (tlsle_small): Change to an expand to handle ILP32.
1224         (tlsle_small_<mode>): New pattern.
1225         (tlsdesc_small): Rename to ...
1226         (tlsdesc_small_<mode>): this and handle PTR.
1227
1228 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1229
1230         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
1231
1232 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1233
1234         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1235         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
1236         (aarch64_types_signed_poly_qualifiers): Likewise.
1237         (aarch64_types_unsigned_signed_qualifiers): Likewise.
1238         (aarch64_types_poly_signed_qualifiers): Likewise.
1239         (TYPES_REINTERP_SS): Type macro added.
1240         (TYPES_REINTERP_SU): Likewise.
1241         (TYPES_REINTERP_SP): Likewise.
1242         (TYPES_REINTERP_US): Likewise.
1243         (TYPES_REINTERP_PS): Likewise.
1244         (aarch64_fold_builtin): New expression folding added.
1245         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
1246         Declarations removed.
1247         (REINTERP_SS): Declarations added.
1248         (REINTERP_US): Likewise.
1249         (REINTERP_PS): Likewise.
1250         (REINTERP_SU): Likewise.
1251         (REINTERP_SP): Likewise.
1252         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
1253         (vreinterpretq_p8_f64): Likewise.
1254         (vreinterpret_p16_f64): Likewise.
1255         (vreinterpretq_p16_f64): Likewise.
1256         (vreinterpret_f32_f64): Likewise.
1257         (vreinterpretq_f32_f64): Likewise.
1258         (vreinterpret_f64_f32): Likewise.
1259         (vreinterpret_f64_p8): Likewise.
1260         (vreinterpret_f64_p16): Likewise.
1261         (vreinterpret_f64_s8): Likewise.
1262         (vreinterpret_f64_s16): Likewise.
1263         (vreinterpret_f64_s32): Likewise.
1264         (vreinterpret_f64_s64): Likewise.
1265         (vreinterpret_f64_u8): Likewise.
1266         (vreinterpret_f64_u16): Likewise.
1267         (vreinterpret_f64_u32): Likewise.
1268         (vreinterpret_f64_u64): Likewise.
1269         (vreinterpretq_f64_f32): Likewise.
1270         (vreinterpretq_f64_p8): Likewise.
1271         (vreinterpretq_f64_p16): Likewise.
1272         (vreinterpretq_f64_s8): Likewise.
1273         (vreinterpretq_f64_s16): Likewise.
1274         (vreinterpretq_f64_s32): Likewise.
1275         (vreinterpretq_f64_s64): Likewise.
1276         (vreinterpretq_f64_u8): Likewise.
1277         (vreinterpretq_f64_u16): Likewise.
1278         (vreinterpretq_f64_u32): Likewise.
1279         (vreinterpretq_f64_u64): Likewise.
1280         (vreinterpret_s64_f64): Likewise.
1281         (vreinterpretq_s64_f64): Likewise.
1282         (vreinterpret_u64_f64): Likewise.
1283         (vreinterpretq_u64_f64): Likewise.
1284         (vreinterpret_s8_f64): Likewise.
1285         (vreinterpretq_s8_f64): Likewise.
1286         (vreinterpret_s16_f64): Likewise.
1287         (vreinterpretq_s16_f64): Likewise.
1288         (vreinterpret_s32_f64): Likewise.
1289         (vreinterpretq_s32_f64): Likewise.
1290         (vreinterpret_u8_f64): Likewise.
1291         (vreinterpretq_u8_f64): Likewise.
1292         (vreinterpret_u16_f64): Likewise.
1293         (vreinterpretq_u16_f64): Likewise.
1294         (vreinterpret_u32_f64): Likewise.
1295         (vreinterpretq_u32_f64): Likewise.
1296
1297 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1298
1299         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1300         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
1301         (vreinterpret_p8_s8): Likewise.
1302         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
1303         (vreinterpret_p8_s16): Likewise.
1304         (vreinterpret_p8_s32): Likewise.
1305         (vreinterpret_p8_s64): Likewise.
1306         (vreinterpret_p8_f32): Likewise.
1307         (vreinterpret_p8_u8): Likewise.
1308         (vreinterpret_p8_u16): Likewise.
1309         (vreinterpret_p8_u32): Likewise.
1310         (vreinterpret_p8_u64): Likewise.
1311         (vreinterpret_p8_p16): Likewise.
1312         (vreinterpretq_p8_s8): Likewise.
1313         (vreinterpretq_p8_s16): Likewise.
1314         (vreinterpretq_p8_s32): Likewise.
1315         (vreinterpretq_p8_s64): Likewise.
1316         (vreinterpretq_p8_f32): Likewise.
1317         (vreinterpretq_p8_u8): Likewise.
1318         (vreinterpretq_p8_u16): Likewise.
1319         (vreinterpretq_p8_u32): Likewise.
1320         (vreinterpretq_p8_u64): Likewise.
1321         (vreinterpretq_p8_p16): Likewise.
1322         (vreinterpret_p16_s8): Likewise.
1323         (vreinterpret_p16_s16): Likewise.
1324         (vreinterpret_p16_s32): Likewise.
1325         (vreinterpret_p16_s64): Likewise.
1326         (vreinterpret_p16_f32): Likewise.
1327         (vreinterpret_p16_u8): Likewise.
1328         (vreinterpret_p16_u16): Likewise.
1329         (vreinterpret_p16_u32): Likewise.
1330         (vreinterpret_p16_u64): Likewise.
1331         (vreinterpret_p16_p8): Likewise.
1332         (vreinterpretq_p16_s8): Likewise.
1333         (vreinterpretq_p16_s16): Likewise.
1334         (vreinterpretq_p16_s32): Likewise.
1335         (vreinterpretq_p16_s64): Likewise.
1336         (vreinterpretq_p16_f32): Likewise.
1337         (vreinterpretq_p16_u8): Likewise.
1338         (vreinterpretq_p16_u16): Likewise.
1339         (vreinterpretq_p16_u32): Likewise.
1340         (vreinterpretq_p16_u64): Likewise.
1341         (vreinterpretq_p16_p8): Likewise.
1342         (vreinterpret_f32_s8): Likewise.
1343         (vreinterpret_f32_s16): Likewise.
1344         (vreinterpret_f32_s32): Likewise.
1345         (vreinterpret_f32_s64): Likewise.
1346         (vreinterpret_f32_u8): Likewise.
1347         (vreinterpret_f32_u16): Likewise.
1348         (vreinterpret_f32_u32): Likewise.
1349         (vreinterpret_f32_u64): Likewise.
1350         (vreinterpret_f32_p8): Likewise.
1351         (vreinterpret_f32_p16): Likewise.
1352         (vreinterpretq_f32_s8): Likewise.
1353         (vreinterpretq_f32_s16): Likewise.
1354         (vreinterpretq_f32_s32): Likewise.
1355         (vreinterpretq_f32_s64): Likewise.
1356         (vreinterpretq_f32_u8): Likewise.
1357         (vreinterpretq_f32_u16): Likewise.
1358         (vreinterpretq_f32_u32): Likewise.
1359         (vreinterpretq_f32_u64): Likewise.
1360         (vreinterpretq_f32_p8): Likewise.
1361         (vreinterpretq_f32_p16): Likewise.
1362         (vreinterpret_s64_s8): Likewise.
1363         (vreinterpret_s64_s16): Likewise.
1364         (vreinterpret_s64_s32): Likewise.
1365         (vreinterpret_s64_f32): Likewise.
1366         (vreinterpret_s64_u8): Likewise.
1367         (vreinterpret_s64_u16): Likewise.
1368         (vreinterpret_s64_u32): Likewise.
1369         (vreinterpret_s64_u64): Likewise.
1370         (vreinterpret_s64_p8): Likewise.
1371         (vreinterpret_s64_p16): Likewise.
1372         (vreinterpretq_s64_s8): Likewise.
1373         (vreinterpretq_s64_s16): Likewise.
1374         (vreinterpretq_s64_s32): Likewise.
1375         (vreinterpretq_s64_f32): Likewise.
1376         (vreinterpretq_s64_u8): Likewise.
1377         (vreinterpretq_s64_u16): Likewise.
1378         (vreinterpretq_s64_u32): Likewise.
1379         (vreinterpretq_s64_u64): Likewise.
1380         (vreinterpretq_s64_p8): Likewise.
1381         (vreinterpretq_s64_p16): Likewise.
1382         (vreinterpret_u64_s8): Likewise.
1383         (vreinterpret_u64_s16): Likewise.
1384         (vreinterpret_u64_s32): Likewise.
1385         (vreinterpret_u64_s64): Likewise.
1386         (vreinterpret_u64_f32): Likewise.
1387         (vreinterpret_u64_u8): Likewise.
1388         (vreinterpret_u64_u16): Likewise.
1389         (vreinterpret_u64_u32): Likewise.
1390         (vreinterpret_u64_p8): Likewise.
1391         (vreinterpret_u64_p16): Likewise.
1392         (vreinterpretq_u64_s8): Likewise.
1393         (vreinterpretq_u64_s16): Likewise.
1394         (vreinterpretq_u64_s32): Likewise.
1395         (vreinterpretq_u64_s64): Likewise.
1396         (vreinterpretq_u64_f32): Likewise.
1397         (vreinterpretq_u64_u8): Likewise.
1398         (vreinterpretq_u64_u16): Likewise.
1399         (vreinterpretq_u64_u32): Likewise.
1400         (vreinterpretq_u64_p8): Likewise.
1401         (vreinterpretq_u64_p16): Likewise.
1402         (vreinterpret_s8_s16): Likewise.
1403         (vreinterpret_s8_s32): Likewise.
1404         (vreinterpret_s8_s64): Likewise.
1405         (vreinterpret_s8_f32): Likewise.
1406         (vreinterpret_s8_u8): Likewise.
1407         (vreinterpret_s8_u16): Likewise.
1408         (vreinterpret_s8_u32): Likewise.
1409         (vreinterpret_s8_u64): Likewise.
1410         (vreinterpret_s8_p8): Likewise.
1411         (vreinterpret_s8_p16): Likewise.
1412         (vreinterpretq_s8_s16): Likewise.
1413         (vreinterpretq_s8_s32): Likewise.
1414         (vreinterpretq_s8_s64): Likewise.
1415         (vreinterpretq_s8_f32): Likewise.
1416         (vreinterpretq_s8_u8): Likewise.
1417         (vreinterpretq_s8_u16): Likewise.
1418         (vreinterpretq_s8_u32): Likewise.
1419         (vreinterpretq_s8_u64): Likewise.
1420         (vreinterpretq_s8_p8): Likewise.
1421         (vreinterpretq_s8_p16): Likewise.
1422         (vreinterpret_s16_s8): Likewise.
1423         (vreinterpret_s16_s32): Likewise.
1424         (vreinterpret_s16_s64): Likewise.
1425         (vreinterpret_s16_f32): Likewise.
1426         (vreinterpret_s16_u8): Likewise.
1427         (vreinterpret_s16_u16): Likewise.
1428         (vreinterpret_s16_u32): Likewise.
1429         (vreinterpret_s16_u64): Likewise.
1430         (vreinterpret_s16_p8): Likewise.
1431         (vreinterpret_s16_p16): Likewise.
1432         (vreinterpretq_s16_s8): Likewise.
1433         (vreinterpretq_s16_s32): Likewise.
1434         (vreinterpretq_s16_s64): Likewise.
1435         (vreinterpretq_s16_f32): Likewise.
1436         (vreinterpretq_s16_u8): Likewise.
1437         (vreinterpretq_s16_u16): Likewise.
1438         (vreinterpretq_s16_u32): Likewise.
1439         (vreinterpretq_s16_u64): Likewise.
1440         (vreinterpretq_s16_p8): Likewise.
1441         (vreinterpretq_s16_p16): Likewise.
1442         (vreinterpret_s32_s8): Likewise.
1443         (vreinterpret_s32_s16): Likewise.
1444         (vreinterpret_s32_s64): Likewise.
1445         (vreinterpret_s32_f32): Likewise.
1446         (vreinterpret_s32_u8): Likewise.
1447         (vreinterpret_s32_u16): Likewise.
1448         (vreinterpret_s32_u32): Likewise.
1449         (vreinterpret_s32_u64): Likewise.
1450         (vreinterpret_s32_p8): Likewise.
1451         (vreinterpret_s32_p16): Likewise.
1452         (vreinterpretq_s32_s8): Likewise.
1453         (vreinterpretq_s32_s16): Likewise.
1454         (vreinterpretq_s32_s64): Likewise.
1455         (vreinterpretq_s32_f32): Likewise.
1456         (vreinterpretq_s32_u8): Likewise.
1457         (vreinterpretq_s32_u16): Likewise.
1458         (vreinterpretq_s32_u32): Likewise.
1459         (vreinterpretq_s32_u64): Likewise.
1460         (vreinterpretq_s32_p8): Likewise.
1461         (vreinterpretq_s32_p16): Likewise.
1462         (vreinterpret_u8_s8): Likewise.
1463         (vreinterpret_u8_s16): Likewise.
1464         (vreinterpret_u8_s32): Likewise.
1465         (vreinterpret_u8_s64): Likewise.
1466         (vreinterpret_u8_f32): Likewise.
1467         (vreinterpret_u8_u16): Likewise.
1468         (vreinterpret_u8_u32): Likewise.
1469         (vreinterpret_u8_u64): Likewise.
1470         (vreinterpret_u8_p8): Likewise.
1471         (vreinterpret_u8_p16): Likewise.
1472         (vreinterpretq_u8_s8): Likewise.
1473         (vreinterpretq_u8_s16): Likewise.
1474         (vreinterpretq_u8_s32): Likewise.
1475         (vreinterpretq_u8_s64): Likewise.
1476         (vreinterpretq_u8_f32): Likewise.
1477         (vreinterpretq_u8_u16): Likewise.
1478         (vreinterpretq_u8_u32): Likewise.
1479         (vreinterpretq_u8_u64): Likewise.
1480         (vreinterpretq_u8_p8): Likewise.
1481         (vreinterpretq_u8_p16): Likewise.
1482         (vreinterpret_u16_s8): Likewise.
1483         (vreinterpret_u16_s16): Likewise.
1484         (vreinterpret_u16_s32): Likewise.
1485         (vreinterpret_u16_s64): Likewise.
1486         (vreinterpret_u16_f32): Likewise.
1487         (vreinterpret_u16_u8): Likewise.
1488         (vreinterpret_u16_u32): Likewise.
1489         (vreinterpret_u16_u64): Likewise.
1490         (vreinterpret_u16_p8): Likewise.
1491         (vreinterpret_u16_p16): Likewise.
1492         (vreinterpretq_u16_s8): Likewise.
1493         (vreinterpretq_u16_s16): Likewise.
1494         (vreinterpretq_u16_s32): Likewise.
1495         (vreinterpretq_u16_s64): Likewise.
1496         (vreinterpretq_u16_f32): Likewise.
1497         (vreinterpretq_u16_u8): Likewise.
1498         (vreinterpretq_u16_u32): Likewise.
1499         (vreinterpretq_u16_u64): Likewise.
1500         (vreinterpretq_u16_p8): Likewise.
1501         (vreinterpretq_u16_p16): Likewise.
1502         (vreinterpret_u32_s8): Likewise.
1503         (vreinterpret_u32_s16): Likewise.
1504         (vreinterpret_u32_s32): Likewise.
1505         (vreinterpret_u32_s64): Likewise.
1506         (vreinterpret_u32_f32): Likewise.
1507         (vreinterpret_u32_u8): Likewise.
1508         (vreinterpret_u32_u16): Likewise.
1509         (vreinterpret_u32_u64): Likewise.
1510         (vreinterpret_u32_p8): Likewise.
1511         (vreinterpret_u32_p16): Likewise.
1512         (vreinterpretq_u32_s8): Likewise.
1513         (vreinterpretq_u32_s16): Likewise.
1514         (vreinterpretq_u32_s32): Likewise.
1515         (vreinterpretq_u32_s64): Likewise.
1516         (vreinterpretq_u32_f32): Likewise.
1517         (vreinterpretq_u32_u8): Likewise.
1518         (vreinterpretq_u32_u16): Likewise.
1519         (vreinterpretq_u32_u64): Likewise.
1520         (vreinterpretq_u32_p8): Likewise.
1521         (vreinterpretq_u32_p16): Likewise.
1522
1523 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1524
1525         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
1526         Pattern extended.
1527         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
1528         extended.
1529         (sqabs): Likewise.
1530         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
1531         (vqnegd_s64): Likewise.
1532         (vqabs_s64): Likewise.
1533         (vqabsd_s64): Likewise.
1534
1535 2014-04-22  Richard Henderson  <rth@redhat.com>
1536
1537         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
1538         computation to the top of the loop.
1539
1540 2014-04-22  Renlin  <renlin.li@arm.com>
1541             Jiong Wang  <jiong.wang@arm.com>
1542
1543         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
1544         * config/aarch64/aarch64.c (aarch64_layout_frame)
1545         (aarch64_initial_elimination_offset): Likewise.
1546
1547 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1548
1549         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
1550         Fix indentation.
1551
1552 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
1553
1554         * machmode.h (bitwise_mode_for_mode): Declare.
1555         * stor-layout.h (bitwise_type_for_mode): Likewise.
1556         * stor-layout.c (bitwise_mode_for_mode): New function.
1557         (bitwise_type_for_mode): Likewise.
1558         * builtins.c (fold_builtin_memory_op): Use it instead of
1559         int_mode_for_mode and build_nonstandard_integer_type.
1560
1561 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1562
1563         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
1564         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
1565         (*-*-solaris2*): Simplify.
1566         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
1567         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
1568         *-*-solaris2.9* handling.
1569
1570         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
1571         as bug.
1572         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
1573         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
1574         handling, simplify.
1575         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
1576         * configure: Regenerate.
1577
1578         * config/i386/sol2-9.h: Remove.
1579
1580         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
1581         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
1582         Remove Solaris 9 references.
1583
1584 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
1585
1586         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
1587         (floatuns<GPI:mode><GPF:mode>2): Remove.
1588         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
1589         and floatuns conversions.
1590         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
1591         and floatuns conversions.
1592         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
1593         (w1,w2): New mode attributes for inequal width conversions.
1594
1595 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
1596
1597         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
1598         the output asm format.
1599
1600 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
1601
1602         * config/aarch64/aarch64-simd.md
1603         (aarch64_cm<optab>di): Always split.
1604         (*aarch64_cm<optab>di): New.
1605         (aarch64_cmtstdi): Always split.
1606         (*aarch64_cmtstdi): New.
1607
1608 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1609
1610         PR tree-optimization/60823
1611         * omp-low.c (ipa_simd_modify_function_body): Go through
1612         all SSA_NAMEs and for those refering to vector arguments
1613         which are going to be replaced adjust SSA_NAME_VAR and,
1614         if it is a default definition, change it into a non-default
1615         definition assigned at the beginning of function from new_decl.
1616         (ipa_simd_modify_stmt_ops): Rewritten.
1617         * tree-dfa.c (set_ssa_default_def): When removing default def,
1618         check for NULL loc instead of NULL *loc.
1619
1620 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1621
1622         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
1623         restrictions on core registers for DImode values in Thumb2.
1624
1625 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1626
1627         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
1628         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
1629
1630 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1631
1632         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
1633         (*iordi_notzesidi_di): Likewise.
1634         (*iordi_notsesidi_di): Likewise.
1635
1636 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1637
1638         * config/arm/arm-protos.h (tune_params): New struct members.
1639         * config/arm/arm.c: Initialise tune_params per processor.
1640         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
1641         for speed, based on new tune_params.
1642
1643 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1644
1645         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
1646         added.
1647         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
1648         macro.
1649         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
1650         corrected.
1651         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
1652         * config/aarch64/arm_neon.h (vrnd_f64): Added.
1653         (vrnda_f64): Likewise.
1654         (vrndi_f64): Likewise.
1655         (vrndm_f64): Likewise.
1656         (vrndn_f64): Likewise.
1657         (vrndp_f64): Likewise.
1658         (vrndx_f64): Likewise.
1659
1660 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1661
1662         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
1663         GET_MODE_SIZE argument is enum machine_mode.
1664
1665 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1666
1667         PR target/60910
1668         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
1669         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
1670
1671 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
1672
1673         PR middle-end/60281
1674         * asan.c (asan_emit_stack_protection): Force the base to align to
1675         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
1676         appropriate bits if STRICT_ALIGNMENT.
1677         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
1678         when asan is on.
1679         (expand_used_vars): Leave a space in the stack frame for alignment
1680         if STRICT_ALIGNMENT.
1681
1682 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
1683
1684         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
1685         than a gimple.
1686         (gimple_store_p): Likewise.
1687         (gimple_assign_load_p): Likewise.
1688         (gimple_assign_cast_p): Likewise.
1689         (gimple_clobber_p): Likewise.
1690
1691         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
1692         rather than a gimple.
1693         (gimple_assign_cast_p): Likewise.
1694
1695 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
1696
1697         PR target/60735
1698         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
1699         If mode is DDmode and TARGET_E500_DOUBLE allow move.
1700
1701         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
1702         more debug information for E500 if -mdebug=reg.
1703
1704 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
1705
1706         PR target/60909
1707         * config/i386/i386.c (ix86_expand_builtin)
1708         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
1709         register for target RTX.
1710         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
1711
1712 2014-04-18  Cong Hou  <congh@google.com>
1713
1714         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
1715         the widen-mult pattern by handling two operands with different sizes,
1716         and operands whose size is smaller than half of the result type.
1717
1718 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1719
1720         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
1721         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
1722         (do_estimate_edge_time): Compute it.
1723         * ipa-inline.c (want_inline_small_function_p): Bypass
1724         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
1725
1726 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1727
1728         * ipa-inline.c (spec_rem): New static variable.
1729         (dump_overall_stats): New function.
1730         (dump_inline_stats): New function.
1731
1732 2014-04-18  Richard Henderson  <rth@redhat.com>
1733
1734         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
1735         to GET_MODE_SIZE, not a reg_class_t.
1736
1737 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1738
1739         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
1740         (vsx_xxmrglw_<mode>): Likewise.
1741
1742 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
1743
1744         PR target/60876
1745         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
1746         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
1747         (rs6000_init_hard_regno_mode_ok): Likewise.
1748
1749 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
1750
1751         * ipa-inline.c (inline_small_functions): Account only non-cold
1752         functions.
1753         * doc/invoke.texi (inline-unit-growth): Update documentation.
1754
1755 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
1756
1757         * config/rs6000/rs6000.md (addti3, subti3): New.
1758
1759 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
1760
1761         PR target/60863
1762         * config/i386/i386.c (ix86_expand_clear): Remove outdated
1763         comment.  Check optimize_insn_for_size_p instead of
1764         optimize_insn_for_speed_p.
1765
1766 2014-04-17  Martin Jambor  <mjambor@suse.cz>
1767
1768         * gimple-iterator.c (gsi_start_edge): New function.
1769         * gimple-iterator.h (gsi_start_edge): Declare.
1770         * tree-sra.c (single_non_eh_succ): New function.
1771         (disqualify_ops_if_throwing_stmt): Renamed to
1772         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
1773         having one non-EH successor BB.
1774         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
1775         generate loads into replacements.
1776         (sra_modify_assign): Likewise and and also use the simple path for
1777         such statements.
1778         (sra_modify_function_body): Commit statements on edges.
1779
1780 2014-04-17  Richard Biener  <rguenther@suse.de>
1781
1782         PR middle-end/60849
1783         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
1784         comparison results and add clarifying comment.
1785
1786 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1787
1788         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
1789         (blank_mode): Initialize it.
1790         (emit_mode_size_inline, emit_mode_nunits_inline,
1791         emit_mode_inner_inline): New functions.
1792         (emit_insn_modes_h): Call them and surround their output with
1793         #if GCC_VERSION >= 4001 ... #endif.
1794         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
1795         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
1796         mode_* arrays if the argument is __builtin_constant_p.
1797         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
1798         is enum machine_mode.
1799
1800 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1801
1802         * passes.c (opt_pass::execute): Adjust.
1803         (pass_manager::execute_pass_mode_switching): Likewise.
1804         (early_local_passes::execute): Likewise.
1805         (execute_one_pass): Pass cfun to the pass's execute method.
1806         * tree-pass.h (opt_pass::execute): Add function * argument.
1807         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1808         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1809         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1810         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1811         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1812         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
1813         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
1814         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1815         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1816         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
1817         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
1818         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
1819         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
1820         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
1821         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1822         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1823         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1824         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
1825         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
1826         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1827         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1828         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1829         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1830         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1831         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1832         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1833         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1834         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1835         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1836         Adjust.
1837
1838 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1839
1840         * passes.c (opt_pass::gate): Take function * argument.
1841         (gate_all_early_local_passes): Merge into
1842         (early_local_passes::gate): this.
1843         (gate_all_early_optimizations): Merge into
1844         (all_early_optimizations::gate): this.
1845         (gate_all_optimizations): Mege into
1846         (all_optimizations::gate): this.
1847         (gate_all_optimizations_g): Merge into
1848         (all_optimizations_g::gate): this.
1849         (gate_rest_of_compilation): Mege into
1850         (rest_of_compilation::gate): this.
1851         (gate_postreload): Merge into
1852         (postreload::gate): this.
1853         (dump_one_pass): Pass cfun to the pass's gate method.
1854         (execute_ipa_summary_passes): Likewise.
1855         (execute_one_pass): Likewise.
1856         (ipa_write_summaries_2): Likewise.
1857         (ipa_write_optimization_summaries_1): Likewise.
1858         (ipa_read_summaries_1): Likewise.
1859         (ipa_read_optimization_summaries_1): Likewise.
1860         (execute_ipa_stmt_fixups): Likewise.
1861         * tree-pass.h (opt_pass::gate): Add function * argument.
1862         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
1863         combine-stack-adj.c, combine.c, compare-elim.c,
1864         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1865         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
1866         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
1867         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
1868         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1869         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1870         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1871         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
1872         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
1873         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
1874         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1875         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1876         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
1877         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1878         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1879         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1880         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1881         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1882         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1883         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1884         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1885         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1886         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
1887         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
1888         var-tracking.c, vtable-verify.c, web.c: Adjust.
1889
1890 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1891
1892         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
1893         * configure: Regenerate.
1894
1895 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1896
1897         * passes.c (dump_one_pass): don't check pass->has_gate.
1898         (execute_ipa_summary_passes): Likewise.
1899         (execute_one_pass): Likewise.
1900         (ipa_write_summaries_2): Likewise.
1901         (ipa_write_optimization_summaries_1): Likewise.
1902         (ipa_read_optimization_summaries_1): Likewise.
1903         (execute_ipa_stmt_fixups): Likewise.
1904         * tree-pass.h (pass_data::has_gate): Remove.
1905         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1906         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1907         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1908         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1909         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1910         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
1911         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
1912         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
1913         gimple-low.c, gimple-ssa-isolate-paths.c,
1914         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
1915         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
1916         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1917         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
1918         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
1919         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
1920         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
1921         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1922         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
1923         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1924         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1925         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1926         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1927         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1928         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1929         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1930         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1931         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1932         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1933         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1934         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1935         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1936         Adjust.
1937
1938 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1939
1940         * pass_manager.h (pass_manager::register_dump_files_1): Remove
1941         declaration.
1942         * passes.c (pass_manager::register_dump_files_1): Merge into
1943         (pass_manager::register_dump_files): this, and remove its handling of
1944         properties since the pass always has the properties anyway.
1945         (pass_manager::pass_manager): Adjust.
1946
1947 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1948
1949         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
1950         * passes.c (pass_manager::register_dump_files_1): Remove dead code
1951         dealing with properties.
1952         (pass_manager::register_dump_files): Adjust.
1953
1954 2014-03-20  Mark Wielaard  <mjw@redhat.com>
1955
1956         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
1957         then represent the bound as normal constant value.
1958
1959 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1960
1961         PR target/60847
1962         Forward port from 4.8 branch
1963         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
1964
1965         * config/i386/bmiintrin.h (_blsi_u32): New.
1966         (_blsi_u64): Ditto.
1967         (_blsr_u32): Ditto.
1968         (_blsr_u64): Ditto.
1969         (_blsmsk_u32): Ditto.
1970         (_blsmsk_u64): Ditto.
1971         (_tzcnt_u32): Ditto.
1972         (_tzcnt_u64): Ditto.
1973
1974 2014-04-17  Kito Cheng  <kito@0xlab.org>
1975
1976         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
1977
1978 2014-04-17  Richard Biener  <rguenther@suse.de>
1979
1980         PR middle-end/60849
1981         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
1982         boolean results for comparisons.
1983
1984 2014-04-17  Richard Biener  <rguenther@suse.de>
1985
1986         PR tree-optimization/60836
1987         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
1988         initial PHI args to be gimple values.
1989
1990 2014-04-17   Richard Biener  <rguenther@suse.de>
1991
1992         PR tree-optimization/60841
1993         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
1994         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
1995         of stmts to SLP build.
1996         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
1997         (vect_analyze_slp): Likewise.
1998         (vect_analyze_slp_instance): Likewise.
1999         (vect_build_slp_tree): Limit overall SLP tree growth.
2000         * tree-vectorizer.h (vect_analyze_data_refs,
2001         vect_analyze_slp): Adjust prototypes.
2002
2003 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2004
2005         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
2006         Silvermont.
2007
2008 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2009
2010         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
2011         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
2012         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
2013         for TARGET_SLOW_PSHUFB
2014
2015 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
2016
2017         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
2018         * config/i386/i386.c (intel_cost): Ditto.
2019
2020 2014-04-17  Joey Ye  <joey.ye@arm.com>
2021
2022         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
2023
2024 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2025
2026         * opts.c (common_handle_option): Disable -fipa-reference coorectly
2027         with -fuse-profile.
2028
2029 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2030
2031         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
2032         (type_all_derivations_known_p): New predicate.
2033         (type_all_ctors_visible_p): New predicate.
2034         (type_possibly_instantiated_p): New predicate.
2035         (get_odr_type): Compute all_derivations_known.
2036         (dump_odr_type): Dump the flag.
2037         (maybe_record_type): Cleanup.
2038         (record_target_from_binfo): Add bases_to_consider array;
2039         record bases for types w/o instances and skip CXX destructor.
2040         (possible_polymorphic_call_targets_1): Add bases_to_consider
2041         and consider_construction parameters; check if type may have instance.
2042         (get_polymorphic_call_info): Set maybe_in_construction to true
2043         when we know nothing.
2044         (record_targets_from_bases): Skip CXX destructors; they are
2045         never called for types in construction.
2046         (possible_polymorphic_call_targets): Do not record target when
2047         type may not have instance.
2048
2049 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
2050
2051         PR ipa/60854
2052         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
2053         external aliases alive, too.
2054
2055 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
2056
2057         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
2058         definition.
2059
2060 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
2061
2062         * final.c (compute_alignments): Do not apply loop alignment to a block
2063         falling through to the exit.
2064
2065 2014-04-16  Catherine Moore  <clm@codesourcery.com>
2066
2067         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
2068         Adjust constraints for microMIPS store patterns.
2069
2070 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2071
2072         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
2073
2074 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
2075
2076         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
2077         (append_use): Run at -O0.
2078         (append_vdef): Likewise.
2079         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
2080         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
2081
2082 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
2083
2084         PR tree-optimization/60844
2085         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
2086         (propagate_op_to_single_use, remove_visited_stmt_chain,
2087         linearize_expr, repropagate_negates, reassociate_bb): Use it
2088         instead of gsi_remove.
2089
2090 2014-04-16  Martin Jambor  <mjambor@suse.cz>
2091
2092         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
2093         ipa_transforms_to_apply.
2094         (cgraph_function_versioning): Assert that old_node has empty
2095         ipa_transforms_to_apply.
2096         * trans-mem.c (ipa_tm_create_version): Likewise.
2097         * tree-inline.c (tree_function_versioning): Do not duplicate
2098         ipa_transforms_to_apply.
2099
2100 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2101
2102         PR target/60817
2103         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
2104         x86_64-*-* cases.
2105         Pass necessary as flags on 64-bit Solaris/x86.
2106         Use lowercase relocs for x86_64-*-*.
2107         * configure: Regenerate.
2108
2109 2014-04-15  Jan Hubicka  <jh@suse.cz>
2110
2111         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
2112         (maybe_record_node, likely_target_p): Use it.
2113
2114 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2115
2116         PR target/60839
2117         Revert following patch
2118
2119         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2120
2121         PR target/60735
2122         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2123         software floating point or no floating point registers, do not
2124         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2125         in GPRs that occurs after we tested for GPRs that would never be
2126         true.
2127
2128         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2129         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2130         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2131         specifically allow DDmode, since that does not use the SPE SIMD
2132         instructions.
2133
2134 2014-03-21  Mark Wielaard  <mjw@redhat.com>
2135
2136         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
2137         as unsigned or int depending on type and value used.
2138
2139 2014-04-15  Richard Biener  <rguenther@suse.de>
2140
2141         PR rtl-optimization/56965
2142         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
2143         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
2144         ... here.
2145         * alias.c (true_dependence_1): Do not call
2146         nonoverlapping_component_refs_p.
2147         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
2148         nonoverlapping_component_refs_p.
2149         (indirect_refs_may_alias_p): Likewise.
2150
2151 2014-04-15  Teresa Johnson  <tejohnson@google.com>
2152
2153         * cfg.c (dump_bb_info): Fix flags check.
2154         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
2155
2156 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2157
2158         PR rtl-optimization/60663
2159         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
2160         avoid 0 cost.
2161
2162 2014-04-15  Richard Biener  <rguenther@suse.de>
2163
2164         * lto-streamer.h (LTO_major_version): Bump to 4.
2165
2166 2014-04-15  Richard Biener  <rguenther@suse.de>
2167
2168         * common.opt (lto_partition_model): New enum.
2169         (flto-partition=): Merge separate options with a single with argument,
2170         add -flto-partition=one support.
2171         * flag-types.h (enum lto_partition_model): Declare.
2172         * opts.c (finish_options): Remove duplicate -flto-partition=
2173         option check.
2174         * lto-wrapper.c (run_gcc): Adjust.
2175
2176 2014-04-15  Richard Biener  <rguenther@suse.de>
2177
2178         * alias.c (ncr_compar): New function.
2179         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
2180
2181 2014-04-15  Richard Biener  <rguenther@suse.de>
2182
2183         * alias.c (record_component_aliases): Do not walk BINFOs.
2184
2185 2014-04-15  Richard Biener  <rguenther@suse.de>
2186
2187         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2188         Add struct function argument and adjust.
2189         (find_func_aliases_for_call): Likewise.
2190         (find_func_aliases): Likewise.
2191         (find_func_clobbers): Likewise.
2192         (intra_create_variable_infos): Likewise.
2193         (compute_points_to_sets): Likewise.
2194         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
2195
2196 2014-04-15  Richard Biener  <rguenther@suse.de>
2197
2198         * tree.c (iterative_hash_expr): Use enum tree_code_class
2199         to store TREE_CODE_CLASS.
2200         (tree_block): Likewise.
2201         (tree_set_block): Likewise.
2202         * tree.h (fold_build_pointer_plus_loc): Use
2203         convert_to_ptrofftype_loc.
2204
2205 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
2206
2207         PR plugins/59335
2208         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
2209         added in 4.9.
2210
2211 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
2212
2213         * cfgloop.h (struct loop): Move force_vectorize down.
2214         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
2215         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
2216         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
2217         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
2218         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
2219         * tree-core.h (enum annot_expr_kind): Add new kind values.
2220         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
2221         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
2222         kinds.
2223         * tree.def (ANNOTATE_EXPR): Tweak comment.
2224
2225 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2226
2227         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
2228         cxa_pure_virtual).
2229
2230 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
2231
2232         * tree.h (TYPE_IDENTIFIER): Declare.
2233         * tree.c (subrange_type_for_debug_p): Use it.
2234         * godump.c (go_format_type): Likewise.
2235         * dwarf2out.c (is_cxx_auto, modified_type_die,
2236         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
2237         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
2238
2239 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
2240
2241         PR lto/60820
2242         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
2243
2244 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
2245
2246         * config/i386/i386.c (examine_argument): Return bool.  Return true if
2247         parameter should be passed in memory.
2248         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
2249         (construct_container): Update calls to examine_argument.
2250         (function_arg_advance_64): Ditto.
2251         (return_in_memory_32): Merge with ix86_return_in_memory.
2252         (return_in_memory_64): Ditto.
2253         (return_in_memory_ms_64): Ditto.
2254
2255 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2256
2257         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
2258         * coverage.c (coverage_compute_profile_id): Handle externally visible
2259         symbols.
2260
2261 2014-04-14  Martin Jambor  <mjambor@suse.cz>
2262
2263         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
2264         DECL_DISREGARD_INLINE_LIMITS functions.
2265
2266 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2267
2268         PR target/60827
2269         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
2270
2271 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2272
2273         PR target/60827
2274         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
2275         optimize_insn_for_speed_p instead of
2276         optimize_function_for_speed_p.
2277
2278 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
2279
2280         * doc/invoke.texi (free): Document AArch64.
2281
2282 2014-04-14  Richard Biener  <rguenther@suse.de>
2283
2284         PR tree-optimization/60042
2285         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
2286         (insert_into_preds_of_block): Do not prevent PHI insertion
2287         for REFERENCE exprs here ...
2288         (eliminate_dom_walker::before_dom_children): ... but prevent
2289         their use here under similar conditions when applied to the
2290         IL after PRE optimizations.
2291
2292 2014-04-14  Richard Biener  <rguenther@suse.de>
2293
2294         * passes.def: Move early points-to after early SRA.
2295
2296 2014-04-14  Richard Biener  <rguenther@suse.de>
2297
2298         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
2299         check for which sign-changes we allow when forwarding
2300         a converted value into a switch.
2301
2302 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2303
2304         * stor-layout.c (place_field): Finalize non-constant offset for the
2305         field, if any.
2306
2307 2014-04-14  Richard Biener  <rguenther@suse.de>
2308
2309         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
2310         as argument.
2311         (expand_switch_using_bit_tests_p): Likewise.
2312         (process_switch): Compute and pass on speed_p based on the
2313         switch stmt.
2314         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
2315         optimize_bb_for_speed_p.
2316
2317 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2318
2319         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
2320         * function.h (struct function): Rename has_force_vect_loops into
2321         has_force_vectorize_loops.
2322         * lto-streamer-in.c (input_cfg): Adjust for renaming.
2323         (input_struct_function_base): Likewise.
2324         * lto-streamer-out.c (output_cfg): Likewise.
2325         (output_struct_function_base): Likewise.
2326         * omp-low.c (expand_omp_simd): Likewise.
2327         * tree-cfg.c (move_sese_region_to_fn): Likewise.
2328         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2329         (version_loop_for_if_conversion): Likewise.
2330         (tree_if_conversion): Likewise.
2331         (main_tree_if_conversion): Likewise.
2332         (gate_tree_if_conversion): Likewise.
2333         * tree-inline.c (copy_loops): Likewise.
2334         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
2335         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
2336         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
2337         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
2338         * tree-vectorizer.c (vectorize_loops): Likewise.
2339         * tree-vectorizer.h (unlimited_cost_model): Likewise.
2340
2341 2014-04-14  Richard Biener  <rguenther@suse.de>
2342
2343         PR lto/60720
2344         * lto-streamer-out.c (wrap_refs): New function.
2345         (lto_output): Wrap symbol references in global initializes in
2346         type-preserving MEM_REFs.
2347
2348 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2349
2350         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
2351
2352 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2353
2354         * config/sh/sh.md (setmemqi): New expand pattern.
2355         * config/sh/sh.h (CLEAR_RATIO): Define.
2356         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
2357         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
2358
2359 2014-04-14  Richard Biener  <rguenther@suse.de>
2360
2361         PR middle-end/55022
2362         * fold-const.c (negate_expr_p): Don't negate directional rounding
2363         division.
2364         (fold_negate_expr): Likewise.
2365
2366 2014-04-14  Richard Biener  <rguenther@suse.de>
2367
2368         PR tree-optimization/59817
2369         PR tree-optimization/60453
2370         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
2371         recursion to catch all CHRECs in the scalar evolution and restrict
2372         the predicate for the remains appropriately.
2373
2374 2014-04-12  Catherine Moore  <clm@codesourcery.com>
2375
2376         * config/mips/constraints.md: Add new register constraint "kb".
2377         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
2378         (*movhi_internal): Likewise.
2379         (*movqi_internal): Likewise.
2380         * config/mips/mips.h (M16_STORE_REGS): New register class.
2381         (REG_CLASS_NAMES): Add M16_STORE_REGS.
2382         (REG_CLASS_CONTENTS): Likewise.
2383         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
2384
2385 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2386
2387         PR c/60194
2388         * doc/invoke.texi (-Wformat-signedness): Document it.
2389         (Wformat=2): Mention that this enables -Wformat-signedness.
2390
2391 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2392
2393         * common/config/epiphany/epiphany-common.c
2394         (epiphany_option_optimization_table): Enable section anchors by
2395         default at -O1 or higher.
2396         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
2397         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
2398         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
2399         carries no extra cost.
2400         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
2401         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
2402         * config/epiphany/predicates.md (memclob_operand): New predicate.
2403         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
2404         Use memclob_operand predicate and X constraint for operand 3.
2405
2406 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2407
2408         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
2409         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
2410         its operands.
2411
2412 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2413
2414         PR rtl-optimization/60651
2415         * mode-switching.c (optimize_mode_switching): Make sure to emit
2416         sets of a lower numbered entity before sets of a higher numbered
2417         entity to a mode of the same or lower priority.
2418         When creating a seginfo for a basic block that starts with a code
2419         label, move the insertion point past the code label.
2420         (new_seginfo): Document and enforce requirement that
2421         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
2422         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
2423         * doc/tm.texi: Regenerate.
2424
2425 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2426
2427         PR target/60811
2428         * config/arc/arc.c (arc_save_restore): Fix assert typo.
2429
2430 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
2431
2432         * BASE-VER: Set to 4.10.0.
2433
2434 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2435
2436         PR other/59055
2437         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
2438         * doc/gcc.texi (Service): Update description in the @menu
2439         * doc/invoke.texi (Option Summary): Remove misplaced and
2440         duplicated @menu.
2441
2442 2014-04-11  Steve Ellcey  <sellcey@mips.com>
2443             Jakub Jelinek  <jakub@redhat.com>
2444
2445         PR middle-end/60556
2446         * expr.c (convert_move): Use emit_store_flag_force instead of
2447         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
2448         argument to it.
2449
2450 2014-04-11  Richard Biener  <rguenther@suse.de>
2451
2452         PR middle-end/60797
2453         * varasm.c (assemble_alias): Avoid endless error reporting
2454         recursion by setting TREE_ASM_WRITTEN.
2455
2456 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2457
2458         * config/s390/s390.md: Add a splitter for NOT rtx.
2459
2460 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
2461
2462         PR rtl-optimization/60663
2463         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
2464
2465 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
2466             Jakub Jelinek  <jakub@redhat.com>
2467
2468         PR lto/60567
2469         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
2470         flag from decl_node to node.
2471
2472 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2473
2474         PR debug/60655
2475         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
2476         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
2477         ameliorating the cases where it can be.
2478
2479 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
2480
2481         Revert
2482         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2483
2484         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2485         (loadsync_<mode>): Change mode.
2486         (load_quadpti, store_quadpti): New.
2487         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2488         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2489         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
2490
2491 2014-04-09  Cong Hou  <congh@google.com>
2492
2493         PR testsuite/60773
2494         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
2495         documentation.
2496
2497 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2498
2499         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
2500         instead of vnor to exploit possible fusion opportunity in the
2501         future.
2502         (altivec_expand_vec_perm_const_le): Likewise.
2503
2504 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2505
2506         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2507         (loadsync_<mode>): Change mode.
2508         (load_quadpti, store_quadpti): New.
2509         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2510         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2511
2512 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
2513
2514         PR target/60763
2515         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
2516         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
2517         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
2518
2519 2014-04-08  Richard Biener  <rguenther@suse.de>
2520
2521         PR middle-end/60706
2522         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
2523         a 64bit widest int print double-int similar to on HWI64 hosts.
2524
2525 2014-04-08  Richard Biener  <rguenther@suse.de>
2526
2527         PR tree-optimization/60785
2528         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
2529         default defs properly.
2530
2531 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
2532
2533         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
2534         (Weffc++): Likewise.
2535
2536 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
2537
2538         * ipa-devirt.c (maybe_record_node): When node is not recorded,
2539         set completep to false rather than true.
2540
2541 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
2542
2543         PR target/60504
2544         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
2545         ARM_TARGET2_DWARF_FORMAT.
2546
2547 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
2548
2549         PR target/60609
2550         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
2551         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
2552         ADDR_DIFF_VEC.
2553
2554 2014-04-07  Richard Biener  <rguenther@suse.de>
2555
2556         PR tree-optimization/60766
2557         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
2558         (may_eliminate_iv): Convert cand_value_at result to desired type.
2559
2560 2014-04-07  Jason Merrill  <jason@redhat.com>
2561
2562         PR c++/60731
2563         * common.opt (-fno-gnu-unique): Add.
2564         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
2565
2566 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2567
2568         * haifa-sched.c: Fix outdated function reference and minor
2569         grammar errors in introductory comment.
2570
2571 2014-04-07  Richard Biener  <rguenther@suse.de>
2572
2573         PR middle-end/60750
2574         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
2575         for noreturn calls.
2576         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
2577
2578 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
2579
2580         PR debug/55794
2581         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
2582         size accounting for thunks.
2583         (pa_asm_output_mi_thunk): Use final_start_function() and
2584         final_end_function() to output function start and end directives.
2585
2586 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2587
2588         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
2589         device specific ISA/ feature information. Remove short_sp and
2590         errata_skip ds.  Add avr_device_specific_features enum to have device
2591         specific info.
2592         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
2593         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
2594         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
2595         updated device specific info.
2596         * config/avr/avr-mcus.def: Merge device specific details to
2597         dev_attribute field.
2598         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
2599         errata_skip.
2600         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
2601         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
2602         assembler if RMW isa supported by current device.
2603         * config/avr/genmultilib.awk: Update as device info structure changed.
2604         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
2605
2606 2014-04-04  Cong Hou  <congh@google.com>
2607
2608         PR tree-optimization/60656
2609         * tree-vect-stmts.c (supportable_widening_operation):
2610         Fix a bug that elements in a vector with vect_used_by_reduction
2611         property are incorrectly reordered when the operation on it is not
2612         consistant with the one in reduction operation.
2613
2614 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
2615
2616         PR rtl-optimization/60155
2617         * gcse.c (record_set_data): New function.
2618         (single_set_gcse): New function.
2619         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
2620         (hoist_code): Likewise.
2621         (get_pressure_class_and_nregs): Likewise.
2622
2623 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
2624
2625         * explow.c (probe_stack_range): Emit a final optimization blockage.
2626
2627 2014-04-04  Anthony Green  <green@moxielogic.com>
2628
2629         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
2630         typos.
2631
2632 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
2633
2634         PR ipa/59626
2635         * lto-cgraph.c (input_overwrite_node): Check that partitioning
2636         flags are set only during streaming.
2637         * ipa.c (process_references, walk_polymorphic_call_targets,
2638         symtab_remove_unreachable_nodes): Drop bodies of always inline
2639         after early inlining.
2640         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
2641
2642 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
2643         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2644
2645         PR debug/60655
2646         * dwarf2out.c (const_ok_for_output_1): Reject expressions
2647         containing a NOT.
2648
2649 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2650
2651         PR bootstrap/60743
2652         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
2653         duration.
2654         (cortex_a53_fdivd): Likewise.
2655
2656 2014-04-04  Martin Jambor  <mjambor@suse.cz>
2657
2658         PR ipa/60640
2659         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
2660         Adjust all callers.
2661         * cgraph.c (clone_of_p): Also return true if thunks match.
2662         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
2663         cgraph_function_or_thunk_node and an obsolete comment.
2664         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
2665         file.
2666         (build_function_decl_skip_args): Likewise.
2667         (set_new_clone_decl_and_node_flags): New function.
2668         (duplicate_thunk_for_node): Likewise.
2669         (redirect_edge_duplicating_thunks): Likewise.
2670         (cgraph_clone_node): New parameter args_to_skip, pass it to
2671         redirect_edge_duplicating_thunks which is called instead of
2672         cgraph_redirect_edge_callee.
2673         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
2674         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
2675
2676 2014-04-04  Jeff Law  <law@redhat.com>
2677
2678         PR target/60657
2679         * config/arm/predicates.md (const_int_I_operand): New predicate.
2680         (const_int_M_operand): Similarly.
2681         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
2682         const_int_operand.
2683         (insv_t2, extv_reg, extzv_t2): Likewise.
2684         (load_multiple_with_writeback): Similarly for const_int_I_operand.
2685         (pop_multiple_with_writeback_and_return): Likewise.
2686         (vfp_pop_multiple_with_writeback): Likewise
2687
2688 2014-04-04  Richard Biener  <rguenther@suse.de>
2689
2690         PR ipa/60746
2691         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
2692         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
2693         non-GIMPLE_LABELs.
2694         * gimplify.h (gimple_add_tmp_var_fn): Declare.
2695         * gimplify.c (gimple_add_tmp_var_fn): New function.
2696         * gimple-expr.h (create_tmp_reg_fn): Declare.
2697         * gimple-expr.c (create_tmp_reg_fn): New function.
2698         * gimple-low.c (record_vars_into): Don't change cfun.
2699         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
2700         code generation without cfun.
2701
2702 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
2703
2704         PR bootstrap/60719
2705         * Makefile.in (install-driver): Fix shell scripting.
2706
2707 2014-04-03  Cong Hou  <congh@google.com>
2708
2709         PR tree-optimization/60505
2710         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
2711         threshold of number of iterations below which no vectorization
2712         will be done.
2713         * tree-vect-loop.c (new_loop_vec_info):
2714         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
2715         * tree-vect-loop.c (vect_analyze_loop_operations):
2716         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
2717         * tree-vect-loop.c (vect_transform_loop):
2718         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
2719         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
2720         of iterations of the loop and see if we should build the epilogue.
2721
2722 2014-04-03  Richard Biener  <rguenther@suse.de>
2723
2724         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
2725         (streamer_tree_cache_create): Adjust.
2726         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
2727         to allow optional nodes array.
2728         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
2729         (streamer_tree_cache_append): Likewise.
2730         (streamer_tree_cache_create): Create nodes array optionally
2731         as specified by parameter.
2732         * lto-streamer-out.c (create_output_block): Avoid maintaining
2733         the node array in the writer cache.
2734         (DFS_write_tree): Remove assertion.
2735         (produce_asm_for_decls): Free the out decl state hash table early.
2736         * lto-streamer-in.c (lto_data_in_create): Adjust for
2737         streamer_tree_cache_create prototype change.
2738
2739 2014-04-03  Richard Biener  <rguenther@suse.de>
2740
2741         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
2742         set TREE_CHAIN to NULL_TREE.
2743
2744 2014-04-03  Richard Biener  <rguenther@suse.de>
2745
2746         PR tree-optimization/60740
2747         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
2748         over all GIMPLE_COND operands.
2749
2750 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
2751
2752         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
2753         (Weffc++): Remove Scott's numbering, merge lists and reference
2754         Wnon-virtual-dtor.
2755
2756 2014-04-03  Nick Clifton  <nickc@redhat.com>
2757
2758         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
2759         properly.
2760
2761 2014-04-03  Martin Jambor  <mjambor@suse.cz>
2762
2763         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
2764         mention gcc_unreachable before failing.
2765         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
2766         removed symbols.
2767
2768 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
2769
2770         PR ipa/60659
2771         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
2772         inconsistent code and instead mark the context inconsistent.
2773         (possible_polymorphic_call_targets): For inconsistent contexts
2774         return empty complete list.
2775
2776 2014-04-02  Anthony Green  <green@moxielogic.com>
2777
2778         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
2779         (extendqisi2, extendhisi2): Define.
2780         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
2781         (WCHAR_TYPE): Change to unsigned int.
2782
2783 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2784
2785         PR tree-optimization/60733
2786         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
2787         insertion point for PHI candidates to be the end of the feeding
2788         block for the PHI argument.
2789
2790 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
2791
2792         PR rtl-optimization/60650
2793         * lra-constraints.c (process_alt_operands): Decrease reject for
2794         earlyclobber matching.
2795
2796 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2797
2798         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
2799
2800 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2801
2802         * config/spu/spu.c (pad_bb): Do not crash when the last
2803         insn is CODE_FOR_blockage.
2804
2805 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2806
2807         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
2808         lies outside the target mode.
2809
2810 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2811
2812         PR target/60735
2813         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2814         software floating point or no floating point registers, do not
2815         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2816         in GPRs that occurs after we tested for GPRs that would never be
2817         true.
2818
2819         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2820         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2821         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2822         specifically allow DDmode, since that does not use the SPE SIMD
2823         instructions.
2824
2825 2014-04-02  Richard Biener  <rguenther@suse.de>
2826
2827         PR middle-end/60729
2828         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
2829         MODE_INTs.  Properly use negv_optab.
2830         (expand_abs): Likewise.
2831
2832 2014-04-02  Richard Biener  <rguenther@suse.de>
2833
2834         PR bootstrap/60719
2835         * Makefile.in (install-driver): Guard extra installs with special
2836         names properly.
2837
2838 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
2839
2840         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2841         Document vec_vgbbd.
2842
2843 2014-04-01  Richard Henderson  <rth@redhat.com>
2844
2845         PR target/60704
2846         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
2847         alternative enabled before register allocation.
2848
2849 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
2850
2851         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
2852         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
2853         typo.
2854         (nios2_large_got_address): Remove unneeded 'sym' parameter.
2855         (nios2_got_address): Update nios2_large_got_address call site.
2856         (nios2_delegitimize_address): New function.
2857         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
2858         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
2859         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
2860
2861 2014-04-01  Martin Husemann  <martin@duskware.de>
2862
2863         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
2864         for -mabi=32.
2865
2866 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
2867
2868         PR rtl-optimization/60604
2869         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
2870         check from register_operand.
2871         (register_operand): Redefine in terms of general_operand.
2872         (nonmemory_operand): Use register_operand for the non-constant cases.
2873
2874 2014-04-01  Richard Biener  <rguenther@suse.de>
2875
2876         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
2877
2878 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2879
2880         * doc/invoke.texi (mapp-regs): Clarify.
2881
2882 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
2883
2884         * config/i386/avx512fintrin.h (__v32hi): Define type.
2885         (__v64qi): Likewise.
2886         (_mm512_set1_epi8): Define.
2887         (_mm512_set1_epi16): Define.
2888         (_mm512_set4_epi32): Define.
2889         (_mm512_set4_epi64): Define.
2890         (_mm512_set4_pd): Define.
2891         (_mm512_set4_ps): Define.
2892         (_mm512_setr4_epi64): Define.
2893         (_mm512_setr4_epi32): Define.
2894         (_mm512_setr4_pd): Define.
2895         (_mm512_setr4_ps): Define.
2896         (_mm512_setzero_epi32): Define.
2897
2898 2014-03-31  Martin Jambor  <mjambor@suse.cz>
2899
2900         PR middle-end/60647
2901         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
2902         callsite_arguments_match_p.  Updated all callers.  Also check types of
2903         corresponding formal parameters and actual arguments.
2904         (not_all_callers_have_enough_arguments_p) Renamed to
2905         some_callers_have_mismatched_arguments_p.
2906
2907 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
2908
2909         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
2910
2911 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
2912
2913         PR target/60034
2914         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
2915         section anchor.
2916
2917 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
2918
2919         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
2920         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
2921         Split out
2922         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
2923         Use FMAMODE_NOVF512 mode iterator.
2924         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
2925         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
2926         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
2927         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
2928         Split out
2929         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
2930         Use VF_128_256 mode iterator.
2931         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
2932         Ditto.
2933
2934 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2935
2936         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
2937         static chain if needed.
2938
2939 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2940
2941         PR target/60697
2942         * lra-constraints.c (index_part_to_reg): New.
2943         (process_address): Use it.
2944
2945 2014-03-27  Jeff Law  <law@redhat.com>
2946             Jakub Jelinek  <jakub@redhat.com>
2947
2948         PR target/60648
2949         * expr.c (do_tablejump): Use simplify_gen_binary rather than
2950         gen_rtx_{PLUS,MULT} to build up the address expression.
2951
2952         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
2953         creating non-canonical RTL.
2954
2955 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2956
2957         PR ipa/60243
2958         * ipa-inline.c (want_inline_small_function_p): Short circuit large
2959         functions; reorganize to make cheap checks first.
2960         (inline_small_functions): Do not estimate growth when dumping;
2961         it is expensive.
2962         * ipa-inline.h (inline_summary): Add min_size.
2963         (growth_likely_positive): New function.
2964         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
2965         (set_cond_stmt_execution_predicate): Cleanup.
2966         (estimate_edge_size_and_time): Compute min_size.
2967         (estimate_calls_size_and_time): Likewise.
2968         (estimate_node_size_and_time): Likewise.
2969         (inline_update_overall_summary): Update min_size.
2970         (do_estimate_edge_time): Likewise.
2971         (do_estimate_edge_size): Update.
2972         (do_estimate_edge_hints): Update.
2973         (growth_likely_positive): New function.
2974
2975 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
2976
2977         PR target/60693
2978         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
2979         also if addr has VOIDmode.
2980
2981 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2982
2983         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
2984         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
2985         Declare extern.
2986         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
2987         instructions as well as AdvancedSIMD loads.
2988
2989 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2990
2991         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2992         Use crypto_aese type.
2993         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
2994         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
2995         crypto_aese, crypto_aesmc.  Move to types.md.
2996         * config/arm/types.md (crypto_aes): Split into crypto_aese,
2997         crypto_aesmc.
2998         * config/arm/iterators.md (crypto_type): Likewise.
2999
3000 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
3001
3002         * cgraph.c: Include expr.h and tree-dfa.h.
3003         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
3004         remove LHS.
3005
3006 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
3007
3008         PR target/60675
3009         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
3010         regs from checking multi-reg pseudos.
3011
3012 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3013
3014         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
3015
3016 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
3017
3018         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
3019         if it would clobber the stack pointer, even temporarily.
3020
3021 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
3022
3023         * mode-switching.c: Make small adjustments to the top comment.
3024
3025 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3026
3027         * config/rs6000/constraints.md (wD constraint): New constraint to
3028         match the constant integer to get the top DImode/DFmode out of a
3029         vector in a VSX register.
3030
3031         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
3032         match the constant integer to get the top DImode/DFmode out of a
3033         vector in a VSX register.
3034
3035         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
3036         for ISA 2.07.
3037
3038         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3039         vbpermq builtins.
3040
3041         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
3042         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
3043
3044         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
3045         Optimize vec_extract of 64-bit values, where the value being
3046         extracted is in the top word, where we can use scalar
3047         instructions.  Add direct move and store support.  Combine the big
3048         endian/little endian vector select load support into a single insn.
3049         (vsx_extract_<mode>_internal1): Likewise.
3050         (vsx_extract_<mode>_internal2): Likewise.
3051         (vsx_extract_<mode>_load): Likewise.
3052         (vsx_extract_<mode>_store): Likewise.
3053         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
3054         combined into vsx_extract_<mode>_load.
3055         (vsx_extract_<mode>_one_le): Likewise.
3056
3057         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
3058         define the top 64-bit vector element.
3059
3060         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
3061         constraint.
3062
3063         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3064         Document vec_vbpermq builtin.
3065
3066         PR target/60672
3067         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
3068         enable use of xxsldwi and xxpermdi builtin functions.
3069         (vec_xxpermdi): Likewise.
3070
3071         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3072         Document use of vec_xxsldwi and vec_xxpermdi builtins.
3073
3074 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
3075
3076         PR rtl-optimization/60650
3077         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
3078         first_p.  Use it.
3079         (find_spills_for): New.
3080         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
3081         Spill all pseudos on the second iteration.
3082
3083 2014-03-27  Marek Polacek  <polacek@redhat.com>
3084
3085         PR c/50347
3086         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
3087         types.
3088
3089 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3090
3091         * config/s390/s390.c (s390_can_use_return_insn): Check for
3092         call-saved FPRs on 31 bit.
3093
3094 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
3095
3096         PR middle-end/60682
3097         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
3098         if they need regimplification, just drop them instead of
3099         calling gimple_regimplify_operands on them.
3100
3101 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3102
3103         PR target/60580
3104         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
3105         (aarch64_frame_pointer_required): Adjust logic.
3106         (aarch64_can_eliminate): Adjust logic.
3107         (aarch64_override_options_after_change): Adjust logic.
3108
3109 2014-03-27  Dehao Chen  <dehao@google.com>
3110
3111         * ipa-inline.c (early_inliner): Update node's inline info.
3112
3113 2014-03-26  Dehao Chen  <dehao@google.com>
3114
3115         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
3116         compiler inserted conditional jumps for NAN float check.
3117
3118 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
3119
3120         * ubsan.h (ubsan_create_data): Change second argument's type
3121         to const location_t *.
3122         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
3123         _("<unknown>").
3124         (ubsan_create_data): Change second argument to const location_t *PLOC.
3125         Create Loc field whenever PLOC is non-NULL.
3126         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
3127         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
3128         callers.
3129
3130         PR other/59545
3131         * real.c (real_to_integer2): Change type of low to UHWI.
3132
3133 2014-03-26  Tobias Burnus  <burnus@net-b.de>
3134
3135         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
3136         (CILK_SELF_SPECS): New define.
3137         (driver_self_specs): Use it.
3138
3139 2014-03-26  Richard Biener  <rguenther@suse.de>
3140
3141         * tree-pretty-print.c (percent_K_format): Implement special
3142         case for LTO and its stripped down BLOCK tree.
3143
3144 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
3145
3146         PR sanitizer/60636
3147         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
3148
3149         * tree-vrp.c (simplify_internal_call_using_ranges): If only
3150         one range is range_int_cst_p, but not both, at least optimize
3151         addition/subtraction of 0 and multiplication by 0 or 1.
3152         * gimple-fold.c (gimple_fold_call): Fold
3153         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
3154         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
3155         INTEGER_CSTs, try to fold at least x * 0 and y - y.
3156
3157 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
3158
3159         PR rtl-optimization/60452
3160         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
3161         <case REG>: Return 1 for invalid offsets from the frame pointer.
3162
3163 2014-03-26  Marek Polacek  <polacek@redhat.com>
3164
3165         PR c/37428
3166         * doc/extend.texi (C Extensions): Mention variable-length arrays in
3167         a structure/union.
3168
3169 2014-03-26  Marek Polacek  <polacek@redhat.com>
3170
3171         PR c/39525
3172         * doc/extend.texi (Designated Inits): Describe what happens to omitted
3173         field members.
3174
3175 2014-03-26  Marek Polacek  <polacek@redhat.com>
3176
3177         PR other/59545
3178         * ira-color.c (update_conflict_hard_regno_costs): Perform the
3179         multiplication in unsigned type.
3180
3181 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
3182
3183         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
3184
3185 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
3186
3187         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
3188
3189 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
3190
3191         PR ipa/60315
3192         * cif-code.def (UNREACHABLE) New code.
3193         * ipa-inline.c (inline_small_functions): Skip edges to
3194         __builtlin_unreachable.
3195         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
3196         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
3197         predicate to __bulitin_unreachable.
3198         (set_cond_stmt_execution_predicate): Fix issue when
3199         invert_tree_comparison returns ERROR_MARK.
3200         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
3201         propagate to inline clones.
3202         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
3203         to unreachable.
3204         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
3205         * cgraphclones.c (cgraph_clone_node): If call destination is already
3206         ureachable, do not redirect it back.
3207         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
3208         unreachable.
3209
3210 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
3211
3212         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
3213         Do not modify inline clones.
3214
3215 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3216
3217         * config/i386/i386.md (general_sext_operand): New mode attr.
3218         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
3219         don't generate (sign_extend (const_int)).
3220         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
3221         operands[2].  Use We constraint instead of <i> and
3222         <general_sext_operand> predicate instead of <general_operand>.
3223         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
3224         * config/i386/constraints.md (We): New constraint.
3225         * config/i386/predicates.md (x86_64_sext_operand,
3226         sext_operand): New predicates.
3227
3228 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3229
3230         PR ipa/60600
3231         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
3232         inconsistent devirtualizations to __builtin_unreachable.
3233
3234 2014-03-25  Marek Polacek  <polacek@redhat.com>
3235
3236         PR c/35449
3237         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
3238
3239 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
3240
3241         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
3242         order of elements for big-endian.
3243
3244 2014-03-25  Richard Biener  <rguenther@suse.de>
3245
3246         PR middle-end/60635
3247         * gimplify-me.c (gimple_regimplify_operands): Update the
3248         re-gimplifed stmt.
3249
3250 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3251
3252         PR ipa/59176
3253         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
3254         (lto_output_varpool_node): Likewise.
3255         (input_overwrite_node): Likewise.
3256         (input_varpool_node): Likewise.
3257
3258 2014-03-25  Richard Biener  <rguenther@suse.de>
3259
3260         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
3261         (run_gcc): Likewise.
3262
3263 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3264
3265         * combine.c (simplify_compare_const): Add MODE argument.
3266         Handle mode_width 0 as very large mode_width.
3267         (try_combine, simplify_comparison): Adjust callers.
3268
3269         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
3270         type to avoid signed integer overflow.
3271         * explow.c (plus_constant): Likewise.
3272
3273 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3274
3275         * doc/generic.texi: Correct typos.
3276
3277 2014-03-24  Tobias Burnus  <burnus@net-b.de>
3278
3279         * doc/invoke.texi (-flto): Expand section about
3280         using static libraries with LTO.
3281
3282 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3283
3284         PR rtl-optimization/60501
3285         * optabs.def (addptr3_optab): New optab.
3286         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
3287         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
3288         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
3289
3290         * lra.c (emit_add3_insn): Use the addptr pattern if available.
3291
3292         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
3293
3294 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
3295
3296         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
3297         _mm512_set1_pd.
3298
3299         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
3300         (_mm256_undefined_ps): Define.
3301         (_mm256_undefined_pd): Define.
3302         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
3303         (_mm_undefined_pd): Define.
3304         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
3305         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
3306         (_mm512_undefined_ps): Define.
3307         (_mm512_undefined_pd): Define.
3308         Use _mm*_undefined_*.
3309         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
3310
3311 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
3312
3313         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
3314         (lshr_simd): DI mode added.
3315         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
3316         (aarch64_ushr_simddi): Likewise.
3317         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
3318         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
3319         (vshrd_n_u64): Likewise.
3320
3321 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3322
3323         * Makefile.in (s-macro_list): Depend on cc1.
3324
3325 2014-03-23  Teresa Johnson  <tejohnson@google.com>
3326
3327         * ipa-utils.c (ipa_print_order): Use specified dump file.
3328
3329 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
3330
3331         PR rtl-optimization/60601
3332         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
3333
3334         * gcc.c (eval_spec_function): Initialize save_growing_value.
3335
3336 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
3337
3338         PR sanitizer/60613
3339         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
3340         code == MINUS_EXPR, never swap op0 with op1.
3341
3342         * toplev.c (init_local_tick): Avoid signed integer multiplication
3343         overflow.
3344         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
3345         shift by first operand's bitsize.
3346
3347 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
3348
3349         PR target/60610
3350         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
3351         redefine to 1 or 0.
3352         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
3353         TARGET_ISA_64BIT_P(x).
3354
3355 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3356
3357         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
3358         pattern for vector nor instead of subtract from splat(-1).
3359         (altivec_expand_vec_perm_const_le): Likewise.
3360
3361 2014-03-21  Richard Henderson  <rth@twiddle.net>
3362
3363         PR target/60598
3364         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
3365         related insns after epilogue_completed.
3366
3367 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3368
3369         PR ipa/59176
3370         * cgraph.h (symtab_node): New flag body_removed.
3371         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
3372         when removing bodies.
3373         * symtab.c (dump_symtab_base): Dump body_removed flag.
3374         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
3375         had their bodies removed.
3376
3377 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3378
3379         PR ipa/60419
3380         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
3381         in the border.
3382
3383 2014-03-21  Richard Biener  <rguenther@suse.de>
3384
3385         PR tree-optimization/60577
3386         * tree-core.h (struct tree_base): Document nothrow_flag use
3387         in VAR_DECL_NONALIASED.
3388         * tree.h (VAR_DECL_NONALIASED): New.
3389         (may_be_aliased): Adjust.
3390         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
3391
3392 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3393
3394         * expr.c (expand_expr_real_1): Remove outdated comment.
3395
3396 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
3397
3398         PR middle-end/60597
3399         * ira.c (adjust_cleared_regs): Call copy_rtx on
3400         *reg_equiv[REGNO (loc)].src_p before passing it to
3401         simplify_replace_fn_rtx.
3402
3403         PR target/60568
3404         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
3405         into CONST, put pic register as first operand of PLUS.  Use
3406         gen_const_mem for both 32-bit and 64-bit PIC got loads.
3407
3408 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3409
3410         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
3411
3412 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3413
3414         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
3415         around for store forwarding issue in the FPU on the UT699.
3416         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
3417         loads and operations if -mfix-ut699 is specified.
3418         (divtf3_hq): Tweak attribute.
3419         (sqrttf2_hq): Likewise.
3420
3421 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3422
3423         * calls.c (store_one_arg): Remove incorrect const qualification on the
3424         type of the temporary.
3425         * cfgexpand.c (expand_return): Likewise.
3426         * expr.c (expand_constructor): Likewise.
3427         (expand_expr_real_1): Likewise.
3428
3429 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3430
3431         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
3432         of parts.
3433
3434 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
3435
3436         PR target/60039
3437         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
3438
3439 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
3440
3441         * config/arm/aarch-common-protos.h
3442         (alu_cost_table): Fix spelling of "extend".
3443         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
3444
3445 2014-03-19  Richard Biener  <rguenther@suse.de>
3446
3447         PR middle-end/60553
3448         * tree-core.h (tree_type_common): Re-order pointer members
3449         to reduce recursion depth during GC walks.
3450
3451 2014-03-19  Marek Polacek  <polacek@redhat.com>
3452
3453         PR sanitizer/60569
3454         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
3455         before accessing it.
3456
3457 2014-03-19  Richard Biener  <rguenther@suse.de>
3458
3459         PR lto/59543
3460         * lto-streamer-in.c (input_function): In WPA stage do not drop
3461         debug stmts.
3462
3463 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
3464
3465         PR tree-optimization/60559
3466         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
3467         with build_zero_cst assignment.
3468
3469 2014-03-18  Kai Tietz  <ktietz@redhat.com>
3470
3471         PR rtl-optimization/56356
3472         * sdbout.c (sdbout_parms): Verify that parms'
3473         incoming argument is valid.
3474         (sdbout_reg_parms): Likewise.
3475
3476 2014-03-18  Richard Henderson  <rth@redhat.com>
3477
3478         PR target/60562
3479         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
3480         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
3481         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
3482
3483 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
3484
3485         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
3486         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
3487         Italicize plugin event names in description.  Explain that
3488         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
3489         Remind that no GCC functions should be called after PLUGIN_FINISH.
3490         Explain what pragmas with expansion are.
3491
3492 2014-03-18  Martin Liska  <mliska@suse.cz>
3493
3494         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
3495         gimple call statement is update.
3496         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
3497         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
3498
3499 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3500
3501         PR sanitizer/60557
3502         * ubsan.c (ubsan_instrument_unreachable): Call
3503         initialize_sanitizer_builtins.
3504         (ubsan_pass): Likewise.
3505
3506         PR sanitizer/60535
3507         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
3508         varpool_finalize_decl instead of rest_of_decl_compilation.
3509
3510 2014-03-18  Richard Biener  <rguenther@suse.de>
3511
3512         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
3513         by using bitmap_and_compl instead of bitmap_and_compl_into.
3514         (df_rd_transfer_function): Likewise.
3515
3516 2014-03-18  Richard Biener  <rguenther@suse.de>
3517
3518         * doc/lto.texi (fresolution): Fix typo.
3519
3520 2014-03-18  Richard Biener  <rguenther@suse.de>
3521
3522         * doc/invoke.texi (flto): Update for changes in 4.9.
3523
3524 2014-03-18  Richard Biener  <rguenther@suse.de>
3525
3526         * doc/loop.texi: Remove section on the removed lambda framework.
3527         Update loop docs with recent changes in preserving loop structure.
3528
3529 2014-03-18  Richard Biener  <rguenther@suse.de>
3530
3531         * doc/lto.texi (-fresolution): Document.
3532
3533 2014-03-18  Richard Biener  <rguenther@suse.de>
3534
3535         * doc/contrib.texi: Adjust my name.
3536
3537 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3538
3539         PR ipa/58721
3540         * internal-fn.c: Include diagnostic-core.h.
3541         (expand_BUILTIN_EXPECT): New function.
3542         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
3543         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
3544         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
3545         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
3546         IFN_BUILTIN_EXPECT.
3547         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
3548         Revert 3 argument __builtin_expect code.
3549         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
3550         * gimple-fold.c (gimple_fold_call): Likewise.
3551         * tree.h (fold_builtin_expect): New prototype.
3552         * builtins.c (build_builtin_expect_predicate): Add predictor
3553         argument, if non-NULL, create 3 argument __builtin_expect.
3554         (fold_builtin_expect): No longer static.  Add ARG2 argument,
3555         pass it through to build_builtin_expect_predicate.
3556         (fold_builtin_2): Adjust caller.
3557         (fold_builtin_3): Handle BUILT_IN_EXPECT.
3558         * internal-fn.def (BUILTIN_EXPECT): New.
3559
3560 2014-03-18  Tobias Burnus  <burnus@net-b.de>
3561
3562         PR ipa/58721
3563         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
3564         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
3565         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
3566
3567 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
3568
3569         PR ipa/58721
3570         * predict.c (combine_predictions_for_bb): Fix up formatting.
3571         (expr_expected_value_1, expr_expected_value): Add predictor argument,
3572         fill what it points to if non-NULL.
3573         (tree_predict_by_opcode): Adjust caller, use the predictor.
3574         * predict.def (PRED_COMPARE_AND_SWAP): Add.
3575
3576 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
3577
3578         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
3579         proper constant for the store mode.
3580
3581 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
3582
3583         * symtab.c (change_decl_assembler_name): Fix transparent alias
3584         chain construction.
3585
3586 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
3587
3588         * config/aarch64/aarch64.c: Correct the comments about the
3589         aarch64 stack layout.
3590
3591 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
3592
3593         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
3594         check for GF_OMP_FOR_KIND_FOR.
3595
3596 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
3597
3598         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
3599         ymm and zmm register names.
3600
3601 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
3602
3603         PR target/60516
3604         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
3605         note creation for the 2010-08-31 changes.
3606
3607 2014-03-17  Marek Polacek  <polacek@redhat.com>
3608
3609         PR middle-end/60534
3610         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
3611         as -fno-tree-loop-vectorize.
3612         (expand_omp_simd): Likewise.
3613
3614 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
3615
3616         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
3617         (eligible_for_call_delay): New prototype.
3618         * config/sparc/sparc.c (tls_call_delay): Rename into...
3619         (eligible_for_call_delay): ...this.  Return false if the instruction
3620         cannot be put in the delay slot of a branch.
3621         (eligible_for_restore_insn): Simplify.
3622         (eligible_for_return_delay): Return false if the instruction cannot be
3623         put in the delay slot of a branch and simplify.
3624         (eligible_for_sibcall_delay): Return false if the instruction cannot be
3625         put in the delay slot of a branch.
3626         * config/sparc/sparc.md (fix_ut699): New attribute.
3627         (tls_call_delay): Delete.
3628         (in_call_delay): Reimplement.
3629         (eligible_for_sibcall_delay): Rename into...
3630         (in_sibcall_delay): ...this.
3631         (eligible_for_return_delay): Rename into...
3632         (in_return_delay): ...this.
3633         (in_branch_delay): Reimplement.
3634         (in_uncond_branch_delay): Delete.
3635         (in_annul_branch_delay): Delete.
3636
3637 2014-03-14  Richard Henderson  <rth@redhat.com>
3638
3639         PR target/60525
3640         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
3641         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
3642         (*floathi<X87MODEF>2_i387_with_temp): Remove.
3643         (floathi splitters): Remove.
3644         (float<SWI48x>xf2): New pattern.
3645         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
3646         code that tried to handle DImode for 32-bit, but which was excluded
3647         by the pattern's condition.  Drop allocation of stack temporary.
3648         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
3649         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
3650         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
3651         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
3652         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
3653         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
3654         (*float<SWI48><MODEF>2_sse_interunit): Remove.
3655         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
3656         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
3657         (*float<SWI48x><X87MODEF>2_i387): Remove.
3658         (all float _with_temp splitters): Remove.
3659         (*float<SWI48x><MODEF>2_i387): New pattern.
3660         (*float<SWI48><MODEF>2_sse): New pattern.
3661         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
3662         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
3663
3664 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
3665             Marek Polacek  <polacek@redhat.com>
3666
3667         PR middle-end/60484
3668         * common.opt (dump_base_name_prefixed): New Variable.
3669         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
3670         if x_dump_base_name_prefixed is already set, set it at the end.
3671
3672 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
3673
3674         PR rtl-optimization/60508
3675         * lra-constraints.c (get_reload_reg): Add new parameter
3676         in_subreg_p.
3677         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
3678         Pass the new parameter values.
3679
3680 2014-03-14  Richard Biener  <rguenther@suse.de>
3681
3682         * common.opt: Revert unintented changes from r205065.
3683         * opts.c: Likewise.
3684
3685 2014-03-14  Richard Biener  <rguenther@suse.de>
3686
3687         PR middle-end/60518
3688         * cfghooks.c (split_block): Properly adjust all loops the
3689         block was a latch of.
3690
3691 2014-03-14  Martin Jambor  <mjambor@suse.cz>
3692
3693         PR lto/60461
3694         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
3695         and simplify it.
3696
3697 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
3698
3699         PR target/59396
3700         * config/avr/avr.c (avr_set_current_function): Pass function name
3701         through default_strip_name_encoding before sanity checking instead
3702         of skipping the first char of the assembler name.
3703
3704 2014-03-13  Richard Henderson  <rth@redhat.com>
3705
3706         PR debug/60438
3707         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
3708         (ix86_force_to_memory, ix86_free_from_memory): Remove.
3709         * config/i386/i386-protos.h: Likewise.
3710         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
3711         in the expander instead of a splitter.
3712         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
3713         any possibility of requiring a memory.
3714         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
3715         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
3716         (fp branch splitters): Update for ix86_split_fp_branch.
3717         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
3718         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
3719         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
3720         (*fop_<MODEF>_2_i387): Remove f/r alternative.
3721         (*fop_<MODEF>_3_i387): Likewise.
3722         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
3723         (splitters for the fop_* register patterns): Remove.
3724         (fscalexf4_i387): Rename from *fscalexf4_i387.
3725         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
3726
3727 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3728
3729         PR tree-optimization/59779
3730         * tree-dfa.c (get_ref_base_and_extent): Use double_int
3731         type for bitsize and maxsize instead of HOST_WIDE_INT.
3732
3733 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
3734
3735         PR rtl-optimization/57320
3736         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
3737         the CFG after thread_prologue_and_epilogue_insns.
3738
3739 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
3740
3741         PR rtl-optimization/57189
3742         * lra-constraints.c (process_alt_operands): Disfavor spilling
3743         vector pseudos.
3744
3745 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
3746
3747         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
3748
3749 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3750
3751         PR tree-optimization/59025
3752         PR middle-end/60418
3753         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
3754         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
3755
3756 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
3757
3758         PR target/60486
3759         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
3760         calls of avr_out_plus_1.
3761
3762 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
3763
3764         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
3765         BB's single pred and update the father loop's latch info later.
3766
3767 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
3768
3769         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
3770         (VEC_M): Likewise.
3771         (VEC_N): Likewise.
3772         (VEC_R): Likewise.
3773         (VEC_base): Likewise.
3774         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
3775         registers, we need to swap double words in little endian mode.
3776
3777         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
3778         to be a container mode for 128-bit integer operations added in ISA
3779         2.07.  Unlike TImode and PTImode, the preferred register set is
3780         the Altivec/VMX registers for the 128-bit operations.
3781
3782         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
3783         declarations.
3784         (rs6000_split_128bit_ok_p): Likewise.
3785
3786         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
3787         macros for creating ISA 2.07 normal and overloaded builtin
3788         functions with 3 arguments.
3789         (BU_P8V_OVERLOAD_3): Likewise.
3790         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
3791         for use as overloaded functions.
3792         (VPERM_1TI_UNS): Likewise.
3793         (VSEL_1TI): Likewise.
3794         (VSEL_1TI_UNS): Likewise.
3795         (ST_INTERNAL_1ti): Likewise.
3796         (LD_INTERNAL_1ti): Likewise.
3797         (XXSEL_1TI): Likewise.
3798         (XXSEL_1TI_UNS): Likewise.
3799         (VPERM_1TI): Likewise.
3800         (VPERM_1TI_UNS): Likewise.
3801         (XXPERMDI_1TI): Likewise.
3802         (SET_1TI): Likewise.
3803         (LXVD2X_V1TI): Likewise.
3804         (STXVD2X_V1TI): Likewise.
3805         (VEC_INIT_V1TI): Likewise.
3806         (VEC_SET_V1TI): Likewise.
3807         (VEC_EXT_V1TI): Likewise.
3808         (EQV_V1TI): Likewise.
3809         (NAND_V1TI): Likewise.
3810         (ORC_V1TI): Likewise.
3811         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
3812         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
3813         overloaded builtin.
3814         (VADDUQM): Likewise.
3815         (VSUBCUQ): Likewise.
3816         (VADDEUQM): Likewise.
3817         (VADDECUQ): Likewise.
3818         (VSUBEUQM): Likewise.
3819         (VSUBECUQ): Likewise.
3820
3821         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
3822         __int128_t and __uint128_t types.
3823         (__uint128_type): Likewise.
3824         (altivec_categorize_keyword): Add support for vector __int128_t,
3825         vector __uint128_t, vector __int128, and vector unsigned __int128
3826         as a container type for TImode operations that need to be done in
3827         VSX/Altivec registers.
3828         (rs6000_macro_to_expand): Likewise.
3829         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
3830         to support 128-bit integer instructions vaddcuq, vadduqm,
3831         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
3832         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
3833
3834         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
3835         for V1TImode, and set up preferences to use VSX/Altivec registers.
3836         Setup VSX reload handlers.
3837         (rs6000_debug_reg_global): Likewise.
3838         (rs6000_init_hard_regno_mode_ok): Likewise.
3839         (rs6000_preferred_simd_mode): Likewise.
3840         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
3841         (easy_altivec_constant): Likewise.
3842         (output_vec_const_move): Likewise.
3843         (rs6000_expand_vector_set): Convert V1TImode set and extract to
3844         simple move.
3845         (rs6000_expand_vector_extract): Likewise.
3846         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
3847         addressing.
3848         (rs6000_const_vec): Add support for V1TImode.
3849         (rs6000_emit_le_vsx_load): Swap double words when loading or
3850         storing TImode/V1TImode.
3851         (rs6000_emit_le_vsx_store): Likewise.
3852         (rs6000_emit_le_vsx_move): Likewise.
3853         (rs6000_emit_move): Add support for V1TImode.
3854         (altivec_expand_ld_builtin): Likewise.
3855         (altivec_expand_st_builtin): Likewise.
3856         (altivec_expand_vec_init_builtin): Likewise.
3857         (altivec_expand_builtin): Likewise.
3858         (rs6000_init_builtins): Add support for V1TImode type.  Add
3859         support for ISA 2.07 128-bit integer builtins.  Define type names
3860         for the VSX/Altivec vector types.
3861         (altivec_init_builtins): Add support for overloaded vector
3862         functions with V1TImode type.
3863         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
3864         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
3865         external function.
3866         (rs6000_split_128bit_ok_p): Likewise.
3867         (rs6000_handle_altivec_attribute): Create V1TImode from vector
3868         __int128_t and vector __uint128_t.
3869
3870         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
3871         and mode attributes.
3872         (VSX_M): Likewise.
3873         (VSX_M2): Likewise.
3874         (VSm): Likewise.
3875         (VSs): Likewise.
3876         (VSr): Likewise.
3877         (VSv): Likewise.
3878         (VS_scalar): Likewise.
3879         (VS_double): Likewise.
3880         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
3881
3882         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
3883         we support the ISA 2.07 128-bit integer arithmetic instructions.
3884         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
3885         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
3886         and TImode types for use with the builtin functions.
3887         (V1TI_type_node): Likewise.
3888         (unsigned_V1TI_type_node): Likewise.
3889         (intTI_type_internal_node): Likewise.
3890         (uintTI_type_internal_node): Likewise.
3891
3892         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
3893         128-bit builtin functions.
3894         (UNSPEC_VADDEUQM): Likewise.
3895         (UNSPEC_VADDECUQ): Likewise.
3896         (UNSPEC_VSUBCUQ): Likewise.
3897         (UNSPEC_VSUBEUQM): Likewise.
3898         (UNSPEC_VSUBECUQ): Likewise.
3899         (VM): Add V1TImode to vector mode iterators.
3900         (VM2): Likewise.
3901         (VI_unit): Likewise.
3902         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
3903         (altivec_vaddcuq): Likewise.
3904         (altivec_vsubuqm): Likewise.
3905         (altivec_vsubcuq): Likewise.
3906         (altivec_vaddeuqm): Likewise.
3907         (altivec_vaddecuq): Likewise.
3908         (altivec_vsubeuqm): Likewise.
3909         (altivec_vsubecuq): Likewise.
3910
3911         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
3912         mode iterators.
3913         (BOOL_128): Likewise.
3914         (BOOL_REGS_OUTPUT): Likewise.
3915         (BOOL_REGS_OP1): Likewise.
3916         (BOOL_REGS_OP2): Likewise.
3917         (BOOL_REGS_UNARY): Likewise.
3918         (BOOL_REGS_AND_CR0): Likewise.
3919
3920         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
3921         128-bit integer builtin support.
3922         (vec_vadduqm): Likewise.
3923         (vec_vaddecuq): Likewise.
3924         (vec_vaddeuqm): Likewise.
3925         (vec_vsubecuq): Likewise.
3926         (vec_vsubeuqm): Likewise.
3927         (vec_vsubcuq): Likewise.
3928         (vec_vsubuqm): Likewise.
3929
3930         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3931         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
3932         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
3933         128-bit integer add/subtract to ISA 2.07.
3934
3935 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
3936
3937         * config/arc/arc.c (arc_predicate_delay_insns):
3938         Fix third argument passed to conditionalize_nonjump.
3939
3940 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
3941
3942         * config/aarch64/aarch64-builtins.c
3943         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
3944         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
3945         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
3946         instead of __builtin_lfloor.
3947         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
3948
3949 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3950
3951         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
3952         (tree_ssa_ifcombine_bb_1): New function.
3953         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
3954         is an empty forwarder block to then_bb or vice versa and then_bb
3955         and else_bb are effectively swapped.
3956
3957 2014-03-12  Christian Bruel  <christian.bruel@st.com>
3958
3959         PR target/60264
3960         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
3961         REG_CFA_DEF_CFA note.
3962         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
3963         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
3964
3965 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
3966
3967         PR tree-optimization/60454
3968         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
3969
3970 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3971
3972         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
3973         Do not define target_cpu_default2 to generic.
3974         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
3975         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
3976         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
3977
3978 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3979             Marc Glisse  <marc.glisse@inria.fr>
3980
3981         PR tree-optimization/60502
3982         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
3983         instead of build_low_bits_mask.
3984
3985 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3986
3987         PR middle-end/60482
3988         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
3989         if there are multiple uses, but op doesn't live on E edge.
3990         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
3991         clobber stmts before __builtin_unreachable.
3992
3993 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
3994
3995         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
3996         hard_frame_pointer_rtx.
3997         * cse.c (cse_insn): Remove volatile check.
3998         * cselib.c (cselib_process_insn): Likewise.
3999         * dse.c (scan_insn): Likewise.
4000
4001 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4002
4003         * config/arc/arc.c (conditionalize_nonjump): New function,
4004         broken out of ...
4005         (arc_ifcvt): ... this.
4006         (arc_predicate_delay_insns): Use it.
4007
4008 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
4009
4010         * config/arc/predicates.md (extend_operand): During/after reload,
4011         allow const_int_operand.
4012         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
4013         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
4014         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
4015         to "i".
4016         (umulsi3_highpart_i): Likewise.
4017
4018 2014-03-11  Richard Biener  <rguenther@suse.de>
4019
4020         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
4021         Add asserts to guard possible wrong-code bugs.
4022
4023 2014-03-11  Richard Biener  <rguenther@suse.de>
4024
4025         PR tree-optimization/60429
4026         PR tree-optimization/60485
4027         * tree-ssa-structalias.c (set_union_with_increment): Properly
4028         take into account all fields that overlap the shifted vars.
4029         (do_sd_constraint): Likewise.
4030         (do_ds_constraint): Likewise.
4031         (get_constraint_for_ptr_offset): Likewise.
4032
4033 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
4034
4035         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
4036         (nios2_compute_frame_layout):
4037         Add calculation of cfun->machine->fp_save_offset.
4038         (nios2_expand_prologue): Correct setting of frame pointer register
4039         in prologue.
4040         (nios2_expand_epilogue): Update recovery of stack pointer from
4041         frame pointer accordingly.
4042         (nios2_initial_elimination_offset): Update calculation of offset
4043         for eliminating to HARD_FRAME_POINTER_REGNUM.
4044
4045 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
4046
4047         PR ipa/60457
4048         * ipa.c (symtab_remove_unreachable_nodes): Don't call
4049         cgraph_get_create_node on VAR_DECLs.
4050
4051 2014-03-10  Richard Biener  <rguenther@suse.de>
4052
4053         PR middle-end/60474
4054         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
4055
4056 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
4057
4058         * config/vms/vms.opt (vms_float_format): New variable.
4059
4060 2014-03-08  Tobias Burnus  <burnus@net-b.de>
4061
4062         * doc/invoke.texi (-fcilkplus): Update implementation status.
4063
4064 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
4065             Richard Biener  <rguenther@suse.de>
4066
4067         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
4068         consistently accross all TUs.
4069         (run_gcc): Enable -fshort-double automatically at link at link-time
4070         and disallow override.
4071
4072 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
4073
4074         PR target/58271
4075         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
4076         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
4077         if they can't be used.
4078
4079 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4080
4081         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
4082         for Solaris 11/x86 ld.
4083         * configure: Regenerate.
4084
4085 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4086
4087         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
4088         (LIB_TLS_SPEC): Save as ld_tls_libs.
4089         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
4090         (HAVE_AS_IX86_TLSLDM): New test.
4091         * configure, config.in: Regenerate.
4092         * config/i386/i386.c (legitimize_tls_address): Fall back to
4093         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
4094         cannot support TLS_MODEL_LOCAL_DYNAMIC.
4095         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
4096         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
4097
4098 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
4099
4100         * common.opt (fira-loop-pressure): Mark as optimization.
4101
4102 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
4103
4104         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
4105         an OpenMP mappable type.
4106
4107 2014-03-06  Matthias Klose  <doko@ubuntu.com>
4108
4109         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
4110         MULTILIB_OSDIRNAMES is not defined.
4111
4112 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
4113             Meador Inge  <meadori@codesourcery.com>
4114
4115         PR target/58595
4116         * config/arm/arm.c (arm_tls_symbol_p): Remove.
4117         (arm_legitimize_address): Call legitimize_tls_address for any
4118         arm_tls_referenced_p expression, handle constant addend.  Call it
4119         before testing for !TARGET_ARM.
4120         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
4121
4122 2014-03-06  Richard Biener  <rguenther@suse.de>
4123
4124         PR middle-end/60445
4125         PR lto/60424
4126         PR lto/60427
4127         Revert
4128         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
4129
4130         * tree-streamer.c (record_common_node): Assert we don't record
4131         nodes with type double.
4132         (preload_common_node): Skip type double, complex double and double
4133         pointer since it is now frontend dependent due to fshort-double option.
4134
4135 2014-03-06  Richard Biener  <rguenther@suse.de>
4136
4137         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
4138         or -fno-lto is specified and the linker has full plugin support.
4139         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
4140         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
4141         * lto-wrapper.c (merge_and_complain): Merge compile-time
4142         optimization levels.
4143         (run_gcc): And pass it through to the link options.
4144
4145 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
4146
4147         PR debug/60381
4148         Revert:
4149         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4150         PR debug/59992
4151         * cselib.c (remove_useless_values): Skip to avoid quadratic
4152         behavior if the condition moved from...
4153         (cselib_process_insn): ... here holds.
4154
4155 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
4156
4157         PR plugins/59335
4158         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
4159         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
4160
4161         PR plugins/59335
4162         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
4163         (TM_H): Add x86-tune.def.
4164
4165 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4166
4167         * config/aarch64/aarch64.c (generic_tunings):
4168         Use cortexa57_extra_costs.
4169
4170 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
4171
4172         PR lto/60404
4173         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
4174         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
4175         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
4176         cost for in_lto_p.
4177
4178 2014-03-04  Heiher  <r@hev.cc>
4179
4180         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
4181         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
4182
4183 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
4184
4185         * config/i386/predicates.md (const2356_operand): Change to ...
4186         (const2367_operand): ... this.
4187         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
4188         const2367_operand.
4189         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4190         (*avx512pf_scatterpf<mode>sf): Ditto.
4191         (avx512pf_scatterpf<mode>df): Ditto.
4192         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4193         (*avx512pf_scatterpf<mode>df): Ditto.
4194         * config/i386/i386.c (ix86_expand_builtin): Update
4195         incorrect hint operand error message.
4196
4197 2014-03-04  Richard Biener  <rguenther@suse.de>
4198
4199         * lto-section-in.c (lto_get_section_data): Fix const cast.
4200
4201 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
4202
4203         * tree-streamer.c (record_common_node): Assert we don't record
4204         nodes with type double.
4205         (preload_common_node): Skip type double, complex double and double
4206         pointer since it is now frontend dependent due to fshort-double option.
4207
4208 2014-03-04  Richard Biener  <rguenther@suse.de>
4209
4210         PR lto/60405
4211         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
4212         (lto_input_toplevel_asms): Likewise.
4213         * lto-section-in.c (lto_get_section_data): Instead do it here
4214         for every section.
4215
4216 2014-03-04  Richard Biener  <rguenther@suse.de>
4217
4218         PR tree-optimization/60382
4219         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
4220         dead PHIs a reduction.
4221
4222 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
4223
4224         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
4225         hint value.
4226         (_mm_prefetch): Move out of GCC target("sse") pragma.
4227         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
4228         GCC target("prfchw") pragma.
4229         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
4230         for locality <= 2.
4231         * config/i386/i386.c (ix86_option_override_internal): Enable
4232         -mprfchw with -mprefetchwt1.
4233
4234 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4235
4236         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
4237         Mark as varying.
4238
4239 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4240
4241         * opts.h (CL_PCH_IGNORE): Define.
4242         * targhooks.c (option_affects_pch_p):
4243         Return false for options that have CL_PCH_IGNORE set.
4244         * opt-functions.awk: Process PchIgnore.
4245         * doc/options.texi: Document PchIgnore.
4246
4247         * config/arc/arc.opt (misize): Add PchIgnore property.
4248
4249 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4250
4251         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
4252         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
4253         constraint on constants to permit them being loaded into
4254         GENERAL_REGS or BASE_REGS.
4255
4256 2014-03-03  Nick Clifton  <nickc@redhat.com>
4257
4258         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
4259         anti-cacnonical alternatives.
4260         (negandhi3_real): New pattern.
4261         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
4262
4263 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
4264
4265         * config/avr/avr-mcus.def: Remove atxmega16x1.
4266         * config/avr/avr-tables.opt: Regenerate.
4267         * config/avr/t-multilib: Regenerate.
4268         * doc/avr-mmcu.texi: Regenerate.
4269
4270 2014-03-03  Tobias Grosser  <tobias@grosser.es>
4271             Mircea Namolaru  <mircea.namolaru@inria.fr>
4272
4273         PR tree-optimization/58028
4274         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
4275         scalar dimensions.
4276
4277 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4278
4279         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
4280         not handled by recognizers.
4281
4282 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
4283
4284         PR middle-end/60175
4285         * function.c (expand_function_end): Don't emit
4286         clobber_return_register sequence if clobber_after is a BARRIER.
4287         * cfgexpand.c (construct_exit_block): Append instructions before
4288         return_label to prev_bb.
4289
4290 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4291
4292         * config/rs6000/constraints.md: Document reserved use of "wc".
4293
4294 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4295
4296         PR ipa/60150
4297         * ipa.c (function_and_variable_visibility): When dissolving comdat
4298         group, also set all symbols to local.
4299
4300 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4301
4302         PR ipa/60306
4303
4304         Revert:
4305         2013-12-14   Jan Hubicka  <jh@suse.cz>
4306         PR middle-end/58477
4307         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
4308
4309 2014-03-02  Jon Beniston  <jon@beniston.com>
4310
4311         PR bootstrap/48230
4312         PR bootstrap/50927
4313         PR bootstrap/52466
4314         PR target/46898
4315         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
4316           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
4317         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
4318         (simple_return, *simple_return): New patterns
4319         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
4320         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
4321
4322 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
4323
4324         * dwarf2out.c (gen_subprogram_die): Tidy.
4325
4326 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
4327
4328         PR target/60071
4329         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
4330         (*mov_t_msb_neg_negc): ... this new insn.
4331
4332 2014-02-28  Jason Merrill  <jason@redhat.com>
4333
4334         PR c++/58678
4335         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
4336         function.
4337
4338 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
4339
4340         PR c++/60314
4341         * dwarf2out.c (decltype_auto_die): New static.
4342         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
4343         (gen_type_die_with_usage): Handle 'decltype(auto)'.
4344         (is_cxx_auto): Likewise.
4345
4346 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
4347
4348         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
4349         we are not using general regs only.
4350
4351 2014-02-28  Richard Biener  <rguenther@suse.de>
4352
4353         PR target/60280
4354         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
4355         previous fix and only allow to remove trivial pre-headers
4356         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
4357         (remove_forwarder_block): Properly update the latch of a loop.
4358
4359 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4360
4361         PR debug/59992
4362         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
4363         (cselib_preserved_hash_table): New.
4364         (preserve_constants_and_equivs): Move preserved vals to it.
4365         (cselib_find_slot): Look it up first.
4366         (cselib_init): Initialize it.
4367         (cselib_finish): Release it.
4368         (dump_cselib_table): Dump it.
4369
4370 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4371
4372         PR debug/59992
4373         * cselib.c (remove_useless_values): Skip to avoid quadratic
4374         behavior if the condition moved from...
4375         (cselib_process_insn): ... here holds.
4376
4377 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4378
4379         PR debug/57232
4380         * var-tracking.c (vt_initialize): Apply the same condition to
4381         preserve the CFA base value.
4382
4383 2014-02-28  Joey Ye  <joey.ye@arm.com>
4384
4385         PR target/PR60169
4386         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
4387         if reload in progress or completed.
4388
4389 2014-02-28  Tobias Burnus  <burnus@net-b.de>
4390
4391         PR middle-end/60147
4392         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
4393         NAMELIST_DECL.
4394
4395 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
4396
4397         * doc/tm.texi.in (Condition Code Status): Update documention for
4398         relative locations of cc0-setter and cc0-user.
4399
4400 2014-02-27  Jeff Law  <law@redhat.com>
4401
4402         PR rtl-optimization/52714
4403         * combine.c (try_combine): When splitting an unrecognized PARALLEL
4404         into two independent simple sets, if I3 is a jump, ensure the
4405         pattern we place into I3 is a (set (pc) ...).
4406
4407 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
4408             Jeff Law  <law@redhat.com>
4409
4410         PR rtl-optimization/49847
4411         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
4412         are in different blocks.
4413         * doc/tm.texi (Condition Code Status): Update documention for
4414         relative locations of cc0-setter and cc0-user.
4415
4416 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
4417
4418         PR target/59222
4419         * lra.c (lra_emit_add): Check SUBREG too.
4420
4421 2014-02-27  Andreas Schwab  <schwab@suse.de>
4422
4423         * config/m68k/m68k.c (m68k_option_override): Disable
4424         -flive-range-shrinkage for classic m68k.
4425         (m68k_override_options_after_change): Likewise.
4426
4427 2014-02-27  Marek Polacek  <polacek@redhat.com>
4428
4429         PR middle-end/59223
4430         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
4431         -Wmaybe-uninitialized.
4432
4433 2014-02-27  Alan Modra  <amodra@gmail.com>
4434
4435         PR target/57936
4436         * reload1.c (emit_input_reload_insns): When reload_override_in,
4437         set old to rl->in_reg when rl->in_reg is a subreg.
4438
4439 2014-02-26  Richard Biener  <rguenther@suse.de>
4440
4441         PR bootstrap/60343
4442         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
4443
4444 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4445
4446         * common/config/i386/predicates.md (const1256_operand): Remove.
4447         (const2356_operand): New.
4448         (const_1_to_2_operand): Remove.
4449         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4450         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4451         (*avx512pf_gatherpf<mode>sf): Ditto.
4452         (avx512pf_gatherpf<mode>df): Ditto.
4453         (*avx512pf_gatherpf<mode>df_mask): Ditto.
4454         (*avx512pf_gatherpf<mode>df): Ditto.
4455         (avx512pf_scatterpf<mode>sf): Ditto.
4456         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4457         (*avx512pf_scatterpf<mode>sf): Ditto.
4458         (avx512pf_scatterpf<mode>df): Ditto.
4459         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4460         (*avx512pf_scatterpf<mode>df): Ditto.
4461         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
4462
4463 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
4464
4465         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
4466         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
4467         (_mm512_mask_testn_epi64_mask): Move to ...
4468         * config/i386/avx512cdintrin.h: Here.
4469         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
4470         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
4471         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
4472         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
4473         TARGET_AVX512F from TARGET_AVX512CD.
4474
4475 2014-02-26  Richard Biener  <rguenther@suse.de>
4476
4477         PR ipa/60327
4478         * ipa.c (walk_polymorphic_call_targets): Properly guard
4479         call to inline_update_overall_summary.
4480
4481 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
4482
4483         PR target/60280
4484         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
4485         and latches only if requested.  Fix latch if it is removed.
4486         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
4487         LOOPS_HAVE_PREHEADERS.
4488
4489 2014-02-25  Andrew Pinski  <apinski@cavium.com>
4490
4491         * builtins.c (expand_builtin_thread_pointer): Create a new target
4492         when the target is NULL.
4493
4494 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
4495
4496         PR rtl-optimization/60317
4497         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4498         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4499         * lra-assigns.c: Include params.h.
4500         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
4501         other reload pseudos considerations.
4502
4503 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4504
4505         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
4506         to use canonical form for nor<mode>3.
4507
4508 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4509
4510         PR target/55426
4511         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
4512         conversions.
4513
4514 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4515
4516         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
4517         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
4518         (ix86_handle_option): Handle OPT_mprefetchwt1.
4519         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
4520         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4521         PREFETCHWT1 CPUID.
4522         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4523         OPTION_MASK_ISA_PREFETCHWT1.
4524         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
4525         (PTA_PREFETCHWT1): New.
4526         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
4527         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
4528         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
4529         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
4530         (*prefetch_avx512pf_<mode>_: Change into ...
4531         (*prefetch_prefetchwt1_<mode>: This.
4532         * config/i386/i386.opt (mprefetchwt1): New.
4533         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
4534         (_mm_prefetch): Handle intent to write.
4535         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
4536
4537 2014-02-25  Richard Biener  <rguenther@suse.de>
4538
4539         PR middle-end/60291
4540         * emit-rtl.c (mem_attrs_htab): Remove.
4541         (mem_attrs_htab_hash): Likewise.
4542         (mem_attrs_htab_eq): Likewise.
4543         (set_mem_attrs): Always allocate new mem-attrs when something changed.
4544         (init_emit_once): Do not allocate mem_attrs_htab.
4545
4546 2014-02-25  Richard Biener  <rguenther@suse.de>
4547
4548         PR lto/60319
4549         * lto-opts.c (lto_write_options): Output non-explicit conservative
4550         -fwrapv, -fno-trapv and -fno-strict-overflow.
4551         * lto-wrapper.c (merge_and_complain): Handle merging those options.
4552         (run_gcc): And pass them through.
4553
4554 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4555
4556         * sel-sched.c (calculate_new_fences): New parameter ptime.
4557         Calculate it as a maximum over all fence cycles.
4558         (sel_sched_region_2): Adjust the call to calculate_new_fences.
4559         Print the final schedule timing when sched_verbose.
4560
4561 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4562
4563         PR rtl-optimization/60292
4564         * sel-sched.c (fill_vec_av_set): Do not reset target availability
4565         bit fot the fence instruction.
4566
4567 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
4568
4569         * calls.h: Fix typo in comment.
4570
4571 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
4572
4573         * config/pa/pa.c (pa_output_move_double): Don't valididate when
4574         adjusting offsetable addresses.
4575
4576 2014-02-24  Guozhi Wei  <carrot@google.com>
4577
4578         * sparseset.h (sparseset_pop): Fix the wrong index.
4579
4580 2014-02-24  Walter Lee  <walt@tilera.com>
4581
4582         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
4583         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
4584         triplet.
4585         * common/config/tilegx/tilegx-common.c
4586         (TARGET_DEFAULT_TARGET_FLAGS): Define.
4587         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
4588         (LINK_SPEC): Ditto.
4589         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
4590         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
4591         (tilegx_gimplify_va_arg_expr): Handle big endian.
4592         (tilegx_expand_unaligned_load): Ditto.
4593         (tilegx_expand_unaligned_store): Ditto.
4594         (TARGET_RETURN_IN_MSB): New.
4595         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
4596         (TARGET_ENDIAN_DEFAULT): New.
4597         (TARGET_BIG_ENDIAN): Handle big endian.
4598         (BYTES_BIG_ENDIAN): Ditto.
4599         (WORDS_BIG_ENDIAN): Ditto.
4600         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
4601         (ENDIAN_SPEC): New.
4602         (EXTRA_SPECS): New.
4603         * config/tilegx/tilegx.md (extv): Handle big endian.
4604         (extzv): Ditto.
4605         (insn_st<n>): Ditto.
4606         (insn_st<n>_add<bitsuffix>): Ditto.
4607         (insn_stnt<n>): Ditto.
4608         (insn_stnt<n>_add<bitsuffix>):Ditto.
4609         (vec_interleave_highv8qi): Handle big endian.
4610         (vec_interleave_highv8qi_be): New.
4611         (vec_interleave_highv8qi_le): New.
4612         (insn_v1int_h): Handle big endian.
4613         (vec_interleave_lowv8qi): Handle big endian.
4614         (vec_interleave_lowv8qi_be): New.
4615         (vec_interleave_lowv8qi_le): New.
4616         (insn_v1int_l): Handle big endian.
4617         (vec_interleave_highv4hi): Handle big endian.
4618         (vec_interleave_highv4hi_be): New.
4619         (vec_interleave_highv4hi_le): New.
4620         (insn_v2int_h): Handle big endian.
4621         (vec_interleave_lowv4hi): Handle big endian.
4622         (vec_interleave_lowv4hi_be): New.
4623         (vec_interleave_lowv4hi_le): New.
4624         (insn_v2int_l): Handle big endian.
4625         (vec_interleave_highv2si): Handle big endian.
4626         (vec_interleave_highv2si_be): New.
4627         (vec_interleave_highv2si_le): New.
4628         (insn_v4int_h): Handle big endian.
4629         (vec_interleave_lowv2si): Handle big endian.
4630         (vec_interleave_lowv2si_be): New.
4631         (vec_interleave_lowv2si_le): New.
4632         (insn_v4int_l): Handle big endian.
4633         * config/tilegx/tilegx.opt (mbig-endian): New option.
4634         (mlittle-endian): New option.
4635         * doc/install.texi: Document tilegxbe-linux.
4636         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
4637
4638 2014-02-24  Martin Jambor  <mjambor@suse.cz>
4639
4640         PR ipa/60266
4641         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
4642         there are no parameter descriptors.
4643
4644 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
4645
4646         PR rtl-optimization/60268
4647         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
4648         initialization to ...
4649         (sched_rgn_init): ... here.
4650         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
4651
4652 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4653
4654         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
4655         names.
4656
4657 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
4658
4659         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
4660         definition.
4661
4662 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4663
4664         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
4665         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
4666
4667 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4668
4669         * config/microblaze/predicates.md: Add cmp_op predicate.
4670         * config/microblaze/microblaze.md: Add branch_compare instruction
4671         which uses cmp_op predicate and emits cmp insn before branch.
4672         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
4673         to microblaze_expand_conditional_branch and consolidate logic.
4674         (microblaze_expand_conditional_branch): emit branch_compare
4675         insn instead of handling cmp op separate from branch insn.
4676
4677 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4678
4679         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
4680         to permit subregs.
4681
4682 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4683
4684         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
4685         define_insn with define_expand and new define_insn
4686         *altivec_lve<VI_char>x_internal.
4687         (altivec_stve<VI_char>x): Replace define_insn with define_expand
4688         and new define_insn *altivec_stve<VI_char>x_internal.
4689         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
4690         prototype.
4691         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
4692         lve*x built-ins.
4693         (altivec_expand_stvex_be): New function.
4694
4695 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
4696
4697         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
4698         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
4699         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
4700         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
4701
4702 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
4703
4704         PR target/60298
4705         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
4706         instead of emit_move_insn.
4707
4708 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4709
4710         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
4711         vspltw with vsldoi.
4712         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
4713         gen_altivec_vsumsws.
4714
4715 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4716
4717         * config/rs6000/altivec.md (altivec_lvxl): Rename as
4718         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
4719         (altivec_lvxl_<mode>): New define_expand incorporating
4720         -maltivec=be semantics where needed.
4721         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
4722         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
4723         semantics where needed.
4724         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
4725         (altivec_stvx_<mode>): New define_expand incorporating
4726         -maltivec=be semantics where needed.
4727         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
4728         VM2 iterator instead of V4SI.
4729         (altivec_stvxl_<mode>): New define_expand incorporating
4730         -maltivec=be semantics where needed.
4731         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
4732         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
4733         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
4734         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
4735         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
4736         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
4737         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
4738         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
4739         ALTIVEC_BUILTIN_STVXL.
4740         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
4741         (altivec_expand_stvx_be): Likewise.
4742         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
4743         (altivec_expand_lvx_be): Likewise.
4744         (altivec_expand_stvx_be): Likewise.
4745         (altivec_expand_builtin): Add cases for
4746         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
4747         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
4748         (altivec_init_builtins): Add definitions for
4749         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
4750         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
4751
4752 2014-02-21  Catherine Moore  <clm@codesourcery.com>
4753
4754         * doc/invoke.texi (mvirt, mno-virt): Document.
4755         * config/mips/mips.opt (mvirt): New option.
4756         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
4757
4758 2014-02-21  Richard Biener  <rguenther@suse.de>
4759
4760         PR tree-optimization/60276
4761         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
4762         (STMT_VINFO_MIN_NEG_DIST): New macro.
4763         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
4764         STMT_VINFO_MIN_NEG_DIST.
4765         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
4766         made for negative dependence distances still hold.
4767
4768 2014-02-21  Richard Biener  <rguenther@suse.de>
4769
4770         PR middle-end/60291
4771         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
4772         DECL_INITIAL for globals not in the current function context.
4773
4774 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4775
4776         PR tree-optimization/56490
4777         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
4778         * tree-ssa-uninit.c: Include params.h.
4779         (compute_control_dep_chain): Add num_calls argument, return false
4780         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
4781         num_calls to recursive call.
4782         (find_predicates): Change dep_chain into normal array,
4783         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
4784         variable and adjust compute_control_dep_chain caller.
4785         (find_def_preds): Likewise.
4786
4787 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
4788
4789         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4790         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
4791
4792 2014-02-21  Nick Clifton  <nickc@redhat.com>
4793
4794         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
4795         (pushhi1): Likewise.
4796         (popqi1): Add mode to pre_dec.
4797         (pophi1): Likewise.
4798
4799 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4800
4801         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
4802         mode for mask of V8SFmode permutation.
4803
4804 2014-02-20  Richard Henderson  <rth@redhat.com>
4805
4806         PR c++/60272
4807         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
4808         a new pseudo for OLDVAL.
4809
4810 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
4811
4812         PR target/57896
4813         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
4814         gen_reg_rtx if d->testing_p.
4815         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
4816         if d->testing_p and we will certainly return true.
4817         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
4818         if d->testing_p.
4819
4820 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
4821
4822         * emit-rtl.c (gen_reg_rtx): Assert that
4823         crtl->emit.regno_pointer_align_length is non-zero.
4824
4825 2014-02-20  Richard Henderson  <rth@redhat.com>
4826
4827         PR c++/60272
4828         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
4829         on failure the store back into EXPECT.
4830
4831 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
4832             Sandra Loosemore  <sandra@codesourcery.com>
4833
4834         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
4835         * config/nios2/nios2.c (nios2_function_profiler): Add
4836         -fPIC (flag_pic == 2) support.
4837         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
4838         (nios2_large_offset_p): New function.
4839         (nios2_unspec_reloc_p): Move up position, update to use
4840         nios2_large_offset_p.
4841         (nios2_unspec_address): Remove function.
4842         (nios2_unspec_offset): New function.
4843         (nios2_large_got_address): New function.
4844         (nios2_got_address): Add large offset support.
4845         (nios2_legitimize_tls_address): Update usage of removed and new
4846         functions.
4847         (nios2_symbol_binds_local_p): New function.
4848         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
4849         (nios2_legitimize_address): Update to use nios2_large_offset_p.
4850         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
4851         (nios2_print_operand): Merge H/L processing, add hiadj/lo
4852         processing for (const (unspec ...)).
4853         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
4854
4855 2014-02-20  Richard Biener  <rguenther@suse.de>
4856
4857         * tree-cfg.c (replace_uses_by): Mark altered BBs before
4858         doing the substitution.
4859         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
4860
4861 2014-02-20  Martin Jambor  <mjambor@suse.cz>
4862
4863         PR ipa/55260
4864         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
4865         info when checking whether lattices are bottom.
4866
4867 2014-02-20  Richard Biener  <rguenther@suse.de>
4868
4869         PR middle-end/60221
4870         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
4871         regions at -O0.
4872
4873 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
4874
4875         PR ipa/58555
4876         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
4877         parameter specifying the scaling.
4878         (inline_call): Update.
4879         (want_inline_recursively): Guard division by zero.
4880         (recursive_inlining): Update.
4881         * ipa-inline.h (clone_inlined_nodes): Update.
4882
4883 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4884
4885         PR target/60204
4886         * config/i386/i386.c (classify_argument): Pass structures of size
4887         64 bytes or less in register.
4888
4889 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4890             Kirill Yukhin  <kirill.yukhin@intel.com>
4891
4892         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
4893         (_mm_rcp28_round_ss): Ditto.
4894         (_mm_rsqrt28_round_sd): Ditto.
4895         (_mm_rsqrt28_round_ss): Ditto.
4896         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
4897         (_mm_rcp14_round_ss): Ditto.
4898         (_mm_rsqrt14_round_sd): Ditto.
4899         (_mm_rsqrt14_round_ss): Ditto.
4900         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
4901         the first input operand, get rid of match_dup.
4902         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
4903         attribute to sse.
4904         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4905         Ditto.
4906         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
4907         operand as the first input operand, set type attribute.
4908         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4909         Set type attribute.
4910         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
4911         operand as the first input operand, set type attribute.
4912
4913 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4914
4915         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
4916         bit of zero.
4917
4918 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
4919
4920         PR target/60207
4921         * config/i386/i386.c (construct_container): Remove TFmode check
4922         for X86_64_INTEGER_CLASS.
4923
4924 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
4925
4926         PR target/59794
4927         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
4928         only when -Wpsabi is enabled.
4929
4930 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
4931
4932          PR target/59799
4933         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
4934         passing arrays in registers are the same as for structs, so remove the
4935         special case for them.
4936
4937 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
4938
4939         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
4940         destination type, extract only the valid bits if the source type is not
4941         integral and has a different mode.
4942
4943 2014-02-19  Richard Biener  <rguenther@suse.de>
4944
4945         PR ipa/60243
4946         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
4947         for all calls.
4948
4949 2014-02-19  Richard Biener  <rguenther@suse.de>
4950
4951         PR ipa/60243
4952         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
4953         (ipa_modify_call_arguments): Emit an argument load explicitely and
4954         preserve virtual SSA form there and for the replacement call.
4955         Do not update SSA form nor free dominance info.
4956
4957 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4958
4959         * ipa.c (function_and_variable_visibility): Also clear WEAK
4960         flag when disolving COMDAT_GROUP.
4961
4962 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4963
4964         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
4965         * ipa-prop.c (ipa_set_jf_known_type): Return early when
4966         not devirtualizing.
4967         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
4968         do more sanity checks.
4969         (detect_type_change): Return true when giving up early.
4970         (compute_complex_assign_jump_func): Fix type parameter of
4971         ipa_set_ancestor_jf.
4972         (compute_complex_ancestor_jump_func): Likewise.
4973         (update_jump_functions_after_inlining): Fix updating of
4974         ancestor function.
4975         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
4976
4977 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4978
4979         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
4980         inline clones when edge disappears.
4981
4982 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4983
4984         PR target/60203
4985         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
4986         Split 64-bit moves into 2 patterns.  Do not allow the use of
4987         direct move for TDmode in little endian, since the decimal value
4988         has little endian bytes within a word, but the 64-bit pieces are
4989         ordered in a big endian fashion, and normal subreg's of TDmode are
4990         not allowed.
4991         (mov<mode>_64bit_dm): Likewise.
4992         (movtd_64bit_nodm): Likewise.
4993
4994 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4995
4996         PR tree-optimization/60174
4997         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
4998         statement of an SSA_NAME that occurs in an abnormal PHI node.
4999
5000 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
5001
5002         PR sanitizer/60142
5003         * final.c (SEEN_BB): Remove.
5004         (SEEN_NOTE, SEEN_EMITTED): Renumber.
5005         (final_scan_insn): Don't force_source_line on second
5006         NOTE_INSN_BASIC_BLOCK.
5007
5008 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
5009
5010         PR target/60205
5011         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
5012         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
5013         (type_natural_mode): Warn ABI change when %zmm register is not
5014         available for AVX512F vector value passing.
5015
5016 2014-02-18  Kai Tietz  <ktietz@redhat.com>
5017
5018         PR target/60193
5019         * config/i386/i386.c (ix86_expand_prologue): Use value in
5020         rax register as displacement when restoring %r10 or %rax.
5021         Fix wrong offset when restoring both registers.
5022
5023 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
5024
5025         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
5026         assertion with conditional return.
5027
5028 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
5029             Uros Bizjak  <ubizjak@gmail.com>
5030
5031         PR driver/60233
5032         * config/i386/driver-i386.c (host_detect_local_cpu): If
5033         YMM state is not saved by the OS, also clear has_f16c.  Move
5034         CPUID 0x80000001 handling before YMM state saving checking.
5035
5036 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
5037
5038         PR rtl-optimization/58960
5039         * haifa-sched.c (alloc_global_sched_pressure_data): New,
5040         factored out from ...
5041         (sched_init): ... here.
5042         (free_global_sched_pressure_data): New, factored out from ...
5043         (sched_finish): ... here.
5044         * sched-int.h (free_global_sched_pressure_data): Declare.
5045         * sched-rgn.c (nr_regions_initial): New static global.
5046         (haifa_find_rgns): Initialize it.
5047         (schedule_region): Disable sched-pressure for the newly
5048         generated regions.
5049
5050 2014-02-17  Richard Biener  <rguenther@suse.de>
5051
5052         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
5053         release SSA defs of pattern stmts.
5054
5055 2014-02-17  Richard Biener  <rguenther@suse.de>
5056
5057         * tree-inline.c (expand_call_inline): Release the virtual
5058         operand defined by the call we are about to inline.
5059
5060 2014-02-17  Richard Biener  <rguenther@suse.de>
5061
5062         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
5063
5064 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
5065             Ilya Tocar  <ilya.tocar@intel.com>
5066
5067         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
5068         arguments order in builtin.
5069         (_mm512_permutexvar_epi64): Ditto.
5070         (_mm512_mask_permutexvar_epi64): Ditto
5071         (_mm512_maskz_permutexvar_epi32): Ditto
5072         (_mm512_permutexvar_epi32): Ditto
5073         (_mm512_mask_permutexvar_epi32): Ditto
5074
5075 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5076
5077         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
5078         (p8_vmrgow): Likewise.
5079
5080 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5081
5082         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
5083         endian targets.
5084
5085 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
5086
5087         PR target/60203
5088         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
5089         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
5090         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
5091         using direct move instructions on ISA 2.07.  Also adjust
5092         instruction length for 64-bit.
5093         (mov<mode>_64bit, TFmode/TDmode): Likewise.
5094         (mov<mode>_32bit, TFmode/TDmode): Likewise.
5095
5096 2014-02-15  Alan Modra  <amodra@gmail.com>
5097
5098         PR target/58675
5099         PR target/57935
5100         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
5101         find_replacement on parts of insn rtl that might be reloaded.
5102
5103 2014-02-15  Richard Biener  <rguenther@suse.de>
5104
5105         PR tree-optimization/60183
5106         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
5107         (tree_ssa_phiprop): Calculate and free post-dominators.
5108
5109 2014-02-14  Jeff Law  <law@redhat.com>
5110
5111         PR rtl-optimization/60131
5112         * ree.c (get_extended_src_reg): New function.
5113         (combine_reaching_defs): Use it rather than assuming location of REG.
5114         (find_and_remove_re): Verify first operand of extension is
5115         a REG before adding the insns to the copy list.
5116
5117 2014-02-14  Roland McGrath  <mcgrathr@google.com>
5118
5119         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
5120         * configure: Regenerated.
5121         * config.in: Regenerated.
5122         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
5123         instead of ASM_SHORT.
5124
5125 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
5126             Richard Earnshaw  <rearnsha@arm.com>
5127
5128         PR rtl-optimization/59535
5129         * lra-constraints.c (process_alt_operands): Encourage alternative
5130         when unassigned pseudo class is superset of the alternative class.
5131         (inherit_reload_reg): Don't inherit when optimizing for code size.
5132         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
5133         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
5134         modes not less than 4 for Thumb1.
5135
5136 2014-02-14  Kyle McMartin  <kyle@redhat.com>
5137
5138         PR pch/60010
5139         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
5140
5141 2014-02-14  Richard Biener  <rguenther@suse.de>
5142
5143         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
5144         (get_frame_arg): Drop the assert with langhook types_compatible_p.
5145         Do not strip INDIRECT_REFs.
5146
5147 2014-02-14  Richard Biener  <rguenther@suse.de>
5148
5149         PR lto/60179
5150         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
5151         DECL_FUNCTION_SPECIFIC_TARGET.
5152         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
5153         * tree-streamer-out.c (pack_ts_target_option): Remove.
5154         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
5155         (write_ts_function_decl_tree_pointers): Do not stream
5156         DECL_FUNCTION_SPECIFIC_TARGET.
5157         * tree-streamer-in.c (unpack_ts_target_option): Remove.
5158         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
5159         (lto_input_ts_function_decl_tree_pointers): Do not stream
5160         DECL_FUNCTION_SPECIFIC_TARGET.
5161
5162 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
5163
5164         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
5165         (get_initial_def_for_induction, vectorizable_induction): Ignore
5166         debug stmts when looking for exit_phi.
5167         (vectorizable_live_operation): Fix up condition.
5168
5169 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5170
5171         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
5172         nreverse() because it changes the content of original tree list.
5173
5174 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5175
5176         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
5177         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
5178
5179 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
5180
5181         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
5182         GNU coding standards.
5183
5184 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
5185
5186         PR debug/60152
5187         * dwarf2out.c (gen_subprogram_die): Don't call
5188         add_calling_convention_attribute if subr_die is old_die.
5189
5190 2014-02-13  Sharad Singhai  <singhai@google.com>
5191
5192         * doc/optinfo.texi: Fix order of nodes.
5193
5194 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
5195
5196         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
5197         operands[2], not operands[3].
5198
5199 2014-02-13  Richard Biener  <rguenther@suse.de>
5200
5201         PR bootstrap/59878
5202         * doc/install.texi (ISL): Update recommended version to 0.12.2,
5203         mention the possibility of an in-tree build.
5204         (CLooG): Update recommended version to 0.18.1, mention the
5205         possibility of an in-tree build and clarify that the ISL
5206         bundled with CLooG does not work.
5207
5208 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
5209
5210         PR target/43546
5211         * expr.c (compress_float_constant): If x is a hard register,
5212         extend into a pseudo and then move to x.
5213
5214 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
5215
5216         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
5217         caused by bad second argument to warning_at() with -mhotpatch and
5218         nested functions (e.g. with gfortran).
5219
5220 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
5221
5222         * opts.c (option_name): Remove "enabled by default" rider.
5223
5224 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
5225
5226         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
5227
5228 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
5229             Uros Bizjak  <ubizjak@gmail.com>
5230
5231         PR target/60151
5232         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
5233         * configure: Regenerated.
5234
5235 2014-02-12  Richard Biener  <rguenther@suse.de>
5236
5237         * vec.c (vec_prefix::calculate_allocation): Move as
5238         inline variant to vec.h.
5239         (vec_prefix::calculate_allocation_1): New out-of-line version.
5240         * vec.h (vec_prefix::calculate_allocation_1): Declare.
5241         (vec_prefix::m_has_auto_buf): Rename to ...
5242         (vec_prefix::m_using_auto_storage): ... this.
5243         (vec_prefix::calculate_allocation): Inline the easy cases
5244         and dispatch to calculate_allocation_1 which doesn't need the
5245         prefix address.
5246         (va_heap::reserve): Use gcc_checking_assert.
5247         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
5248         m_using_auto_storage.
5249         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
5250         member and adjust.
5251         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
5252         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
5253         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
5254
5255 2014-02-12  Richard Biener  <rguenther@suse.de>
5256
5257         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
5258         when we found a dependence.
5259
5260 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
5261
5262         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
5263         common code...
5264         (maybe_fold_stmt): ... into this new function.
5265         * omp-low.c (lower_omp): Update comment.
5266
5267         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
5268         last use.
5269
5270         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
5271         dereference.
5272
5273 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
5274
5275         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
5276         identifiers in comments.
5277         (cortexa53_extra_costs): Likewise.
5278         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
5279         (cortexa7_extra_costs): Likewise.
5280         (cortexa12_extra_costs): Likewise.
5281         (cortexa15_extra_costs): Likewise.
5282         (v7m_extra_costs): Likewise.
5283
5284 2014-02-12  Richard Biener  <rguenther@suse.de>
5285
5286         PR middle-end/60092
5287         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
5288         of posix_memalign being successful.
5289         (lower_stmt): Restrict lowering of posix_memalign to when
5290         -ftree-bit-ccp is enabled.
5291
5292 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5293
5294         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
5295         arg_loc.
5296         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
5297
5298 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
5299
5300         PR rtl-optimization/60116
5301         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
5302         other_insn once the combination has been validated.
5303
5304 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
5305
5306         PR lto/59468
5307         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
5308         and wrapper.
5309         * ipa-devirt.c: Include demangle.h
5310         (odr_violation_reported): New static variable.
5311         (add_type_duplicate): Update odr_violations.
5312         (maybe_record_node): Add completep parameter; update it.
5313         (record_target_from_binfo): Add COMPLETEP parameter;
5314         update it as needed.
5315         (possible_polymorphic_call_targets_1): Likewise.
5316         (struct polymorphic_call_target_d): Add nonconstruction_targets;
5317         rename FINAL to COMPLETE.
5318         (record_targets_from_bases): Sanity check we found the binfo;
5319         fix COMPLETEP updating.
5320         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
5321         parameter, fix computing of COMPLETEP.
5322         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
5323         at LTO time do demangling.
5324         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
5325         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
5326         parameter.
5327         (gimple_get_virt_method_for_binfo): Likewise.
5328         * gimple-fold.h (gimple_get_virt_method_for_binfo,
5329         gimple_get_virt_method_for_vtable): Update prototypes.
5330
5331 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
5332
5333         PR target/49008
5334         * genautomata.c (add_presence_absence): Fix typo with
5335         {final_}presence_list.
5336
5337 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5338
5339         PR target/60137
5340         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
5341         for VSX/Altivec vectors that land in GPR registers.
5342
5343 2014-02-11  Richard Henderson  <rth@redhat.com>
5344             Jakub Jelinek  <jakub@redhat.com>
5345
5346         PR debug/59776
5347         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
5348         around drhs if type conversion to lacc->type is not useless.
5349
5350 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5351
5352         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
5353         tuning struct.
5354         (cortex-a57.cortex-a53): Likewise.
5355         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
5356
5357 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5358
5359         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
5360         arm_restrict_it.
5361
5362 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
5363
5364         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
5365         add_options_for_arm_vfp3.
5366
5367 2014-02-11  Jeff Law  <law@redhat.com>
5368
5369         PR middle-end/54041
5370         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
5371         object with an undesirable mode.
5372
5373 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5374
5375         PR libgomp/60107
5376         * config/i386/sol2-9.h: New file.
5377         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
5378         *-*-solaris2.9*): Use it.
5379
5380 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5381
5382         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
5383         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
5384
5385 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5386
5387         * config/microblaze/microblaze.c: Extend mcpu version format
5388
5389 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
5390
5391         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
5392
5393 2014-02-10  Richard Henderson  <rth@redhat.com>
5394
5395         PR target/59927
5396         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
5397         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
5398         ms-abi vs -mno-accumulate-outgoing-args.
5399         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
5400         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
5401         respect to ms-abi.
5402
5403 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5404
5405         PR middle-end/60080
5406         * cfgexpand.c (expand_asm_operands): Attach source location to
5407         ASM_INPUT rtx objects.
5408         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
5409
5410 2014-02-10  Nick Clifton  <nickc@redhat.com>
5411
5412         * config/mn10300/mn10300.c (popcount): New function.
5413         (mn10300_expand_prologue): Include saved registers in stack usage
5414         count.
5415
5416 2014-02-10  Jeff Law  <law@redhat.com>
5417
5418         PR middle-end/52306
5419         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
5420         when changing the SET_DEST of a prior insn to avoid an input reload.
5421
5422 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5423
5424         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
5425         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
5426         -mcall-openbsd, or -mcall-linux.
5427         (CC1_ENDIAN_BIG_SPEC): Remove.
5428         (CC1_ENDIAN_LITTLE_SPEC): Remove.
5429         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5430         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
5431         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
5432         and %cc1_endian_default.
5433         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5434
5435 2014-02-10  Richard Biener  <rguenther@suse.de>
5436
5437         PR tree-optimization/60115
5438         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
5439         MEM_REF handling.  Properly verify that the accesses are not
5440         out of the objects bound.
5441
5442 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5443
5444         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
5445         coretex to cortex.
5446
5447 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
5448
5449         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
5450         proper constants and fix formatting.
5451         (possible_polymorphic_call_targets): Fix formatting.
5452
5453 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
5454             Ilya Tocar  <ilya.tocar@intel.com>
5455
5456         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
5457         (_mm512_loadu_epi32): Renamed into...
5458         (_mm512_loadu_si512): This.
5459         (_mm512_storeu_epi32): Renamed into...
5460         (_mm512_storeu_si512): This.
5461         (_mm512_maskz_ceil_ps): Removed.
5462         (_mm512_maskz_ceil_pd): Ditto.
5463         (_mm512_maskz_floor_ps): Ditto.
5464         (_mm512_maskz_floor_pd): Ditto.
5465         (_mm512_floor_round_ps): Ditto.
5466         (_mm512_floor_round_pd): Ditto.
5467         (_mm512_ceil_round_ps): Ditto.
5468         (_mm512_ceil_round_pd): Ditto.
5469         (_mm512_mask_floor_round_ps): Ditto.
5470         (_mm512_mask_floor_round_pd): Ditto.
5471         (_mm512_mask_ceil_round_ps): Ditto.
5472         (_mm512_mask_ceil_round_pd): Ditto.
5473         (_mm512_maskz_floor_round_ps): Ditto.
5474         (_mm512_maskz_floor_round_pd): Ditto.
5475         (_mm512_maskz_ceil_round_ps): Ditto.
5476         (_mm512_maskz_ceil_round_pd): Ditto.
5477         (_mm512_expand_pd): Ditto.
5478         (_mm512_expand_ps): Ditto.
5479         * config/i386/i386.c (ix86_builtins): Remove
5480         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
5481         (bdesc_args): Ditto.
5482         * config/i386/predicates.md (const1256_operand): New.
5483         (const_1_to_2_operand): Ditto.
5484         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
5485         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5486         (*avx512pf_gatherpf<mode>sf): Ditto.
5487         (avx512pf_gatherpf<mode>df): Ditto.
5488         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5489         (*avx512pf_gatherpf<mode>df): Ditto.
5490         (avx512pf_scatterpf<mode>sf): Ditto.
5491         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5492         (*avx512pf_scatterpf<mode>sf): Ditto.
5493         (avx512pf_scatterpf<mode>df): Ditto.
5494         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5495         (*avx512pf_scatterpf<mode>df): Ditto.
5496         (avx512f_expand<mode>): Removed.
5497         (<shift_insn><mode>3<mask_name>): Change predicate type.
5498
5499 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5500
5501         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
5502         not at the end of datarefs vector use ordered_remove to avoid
5503         reordering datarefs vector.
5504
5505         PR c/59984
5506         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
5507         mark local addressable non-static vars as GOVD_PRIVATE
5508         instead of GOVD_LOCAL.
5509         * omp-low.c (lower_omp_for): Move gimple_bind_vars
5510         and BLOCK_VARS of gimple_bind_block to new_stmt rather
5511         than copying them.
5512
5513         PR middle-end/60092
5514         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
5515         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
5516         assume_aligned or alloc_align attributes.
5517         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
5518         arguments.  Handle also assume_aligned and alloc_align attributes.
5519         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
5520         calls to functions with assume_aligned or alloc_align attributes.
5521         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
5522
5523 2014-02-08  Terry Guo  <terry.guo@arm.com>
5524
5525         * doc/invoke.texi: Document ARM -march=armv7e-m.
5526
5527 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5528
5529         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
5530         flag on __cilkrts_rethrow builtin.
5531
5532         PR ipa/60026
5533         * ipa-cp.c (determine_versionability): Fail at -O0
5534         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
5535         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
5536
5537         Revert:
5538         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5539
5540         PR ipa/60026
5541         * tree-inline.c (copy_forbidden): Fail for
5542         __attribute__((optimize (0))) functions.
5543
5544 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5545
5546         * varpool.c: Include pointer-set.h.
5547         (varpool_remove_unreferenced_decls): Variables in other partitions
5548         will not be output; be however careful to not lose information
5549         about partitioning.
5550
5551 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5552
5553         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
5554         lookup in the vtable constructor.
5555
5556 2014-02-07  Jeff Law  <law@redhat.com>
5557
5558         PR target/40977
5559         * config/m68k/m68k.md (ashldi_extsi): Turn into a
5560         define_insn_and_split.
5561
5562         * ipa-inline.c (inline_small_functions): Fix typos.
5563
5564 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5565
5566         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
5567         (s390_can_use_return_insn): Declare.
5568         * config/s390/s390.h (EPILOGUE_USES): Define.
5569         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
5570         instructions.
5571         (s390_chunkify_start): Handle return JUMP_LABELs.
5572         (s390_early_mach): Emit a main_pool instruction on the entry edge.
5573         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
5574         (s390_can_use_return_insn): New functions.
5575         (s390_fix_long_loop_prediction): Handle conditional returns.
5576         (TARGET_SET_UP_BY_PROLOGUE): Define.
5577         * config/s390/s390.md (ANY_RETURN): New code iterator.
5578         (*creturn, *csimple_return, return, simple_return): New patterns.
5579
5580 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5581
5582         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
5583         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
5584         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
5585         REG_CFA_RESTORE list when deciding not to restore a register.
5586
5587 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5588
5589         * config/s390/s390.c: Include tree-pass.h and context.h.
5590         (s390_early_mach): New function, split out from...
5591         (s390_emit_prologue): ...here.
5592         (pass_data_s390_early_mach): New pass structure.
5593         (pass_s390_early_mach): New class.
5594         (s390_option_override): Create and register early_mach pass.
5595         Move to end of file.
5596
5597 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5598
5599         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
5600         to match for the exit block.
5601
5602 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5603
5604         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
5605         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
5606         Reject misaligned operands.
5607
5608 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5609
5610         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
5611
5612 2014-02-07  Richard Biener  <rguenther@suse.de>
5613
5614         PR middle-end/60092
5615         * gimple-low.c (lower_builtin_posix_memalign): New function.
5616         (lower_stmt): Call it to lower posix_memalign in a way
5617         to make alignment info accessible.
5618
5619 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
5620
5621         PR c++/60082
5622         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
5623         __builtin_setjmp_receiver.
5624
5625 2014-02-07  Richard Biener  <rguenther@suse.de>
5626
5627         PR middle-end/60092
5628         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
5629         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
5630         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5631         Handle BUILT_IN_POSIX_MEMALIGN.
5632         (find_func_clobbers): Likewise.
5633         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5634         (call_may_clobber_ref_p_1): Likewise.
5635
5636 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5637
5638         PR ipa/59918
5639         * ipa-devirt.c (record_target_from_binfo): Remove overactive
5640         sanity check.
5641
5642 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5643
5644         PR ipa/59469
5645         * lto-cgraph.c (lto_output_node): Use
5646         symtab_get_symbol_partitioning_class.
5647         (lto_output_varpool_node): likewise.
5648         (symtab_get_symbol_partitioning_class): Move here from
5649         lto/lto-partition.c
5650         * cgraph.h (symbol_partitioning_class): Likewise.
5651         (symtab_get_symbol_partitioning_class): Declare.
5652
5653 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5654
5655         * ggc.h (ggc_internal_cleared_alloc): New macro.
5656         * vec.h (vec_safe_copy): Handle memory stats.
5657         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
5658         * target-globals.c (save_target_globals): Likewise.
5659
5660 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5661
5662         PR target/60077
5663         * expr.c (emit_move_resolve_push): Export; be bit more selective
5664         on when to clear alias set.
5665         * expr.h (emit_move_resolve_push): Declare.
5666         * function.h (struct function): Add tail_call_marked.
5667         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
5668         * config/i386/i386-protos.h (ix86_expand_push): Remove.
5669         * config/i386/i386.md (TImode move expander): De not call
5670         ix86_expand_push.
5671         (FP push expanders): Preserve memory attributes.
5672         * config/i386/sse.md (push<mode>1): Remove.
5673         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
5674         (ix86_expand_push): Remove.
5675         * config/i386/mmx.md (push<mode>1): Remove.
5676
5677 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5678
5679         PR rtl-optimization/60030
5680         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
5681         lopart with paradoxical subreg before shifting it up by hprec.
5682
5683 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5684
5685         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
5686         Remove extra newline at end of file.
5687         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
5688         (arm_issue_rate): Handle cortexa57.
5689         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
5690         (cortex-a57.cortex-a53): Likewise.
5691
5692 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5693
5694         PR target/59575
5695         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
5696         don't record in REG_FRAME_RELATED_EXPR registers not set in that
5697         bitmask.
5698         (arm_expand_prologue): Adjust all callers.
5699         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
5700         info, registers also at the lowest numbered registers side.  Use
5701         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
5702         XEXP.
5703
5704         PR debug/59992
5705         * var-tracking.c (adjust_mems): Before adding a SET to
5706         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
5707
5708 2014-02-06  Alan Modra  <amodra@gmail.com>
5709
5710         PR target/60032
5711         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
5712         change SDmode to DDmode when lra_in_progress.
5713
5714 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5715
5716         PR middle-end/59150
5717         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
5718         free_data_ref on the dr first, and before goto again also set dr
5719         to the next dr.  For simd_lane_access, free old datarefs[i] before
5720         overwriting it.  For get_vectype_for_scalar_type failure, don't
5721         free_data_ref if simd_lane_access.
5722
5723         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
5724
5725         PR target/60062
5726         * tree.h (opts_for_fn): New inline function.
5727         (opt_for_fn): Define.
5728         * config/i386/i386.c (ix86_function_regparm): Use
5729         opt_for_fn (decl, optimize) instead of optimize.
5730
5731 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5732
5733         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
5734         for SYMBOL_REF in large memory model.
5735
5736 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5737
5738         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
5739         and crypto support.
5740         (cortex-a57): Likewise.
5741         (cortex-a57.cortex-a53): Likewise.
5742
5743 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
5744             Kugan Vivekanandarajah  <kuganv@linaro.org>
5745
5746         * config/arm/arm.c (arm_vector_alignment_reachable): Check
5747         unaligned_access.
5748         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
5749
5750 2014-02-06  Richard Biener  <rguenther@suse.de>
5751
5752         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
5753         set_loop_copy and initialize_original_copy_tables.
5754
5755 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
5756
5757         * config/aarch64/aarch64-simd.md
5758         (aarch64_ashr_simddi): Change QI to SI.
5759
5760 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5761             Jakub Jelinek  <jakub@redhat.com>
5762
5763         PR middle-end/60013
5764         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
5765         of the dataflow.
5766
5767 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5768
5769         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
5770         CODE_FOR_altivec_vpku[hw]um to
5771         CODE_FOR_altivec_vpku[hw]um_direct.
5772         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
5773         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
5774         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
5775         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
5776
5777 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5778
5779         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
5780         generation for -maltivec=be.
5781         (altivec_vsumsws): Simplify redundant test.
5782
5783 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5784
5785         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
5786         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
5787         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
5788         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
5789         gen_altivec_vpkuwum.
5790         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
5791         BYTES_BIG_ENDIAN.
5792         (altivec_vpks<VI_char>ss): Likewise.
5793         (altivec_vpks<VI_char>us): Likewise.
5794         (altivec_vpku<VI_char>us): Likewise.
5795         (altivec_vpku<VI_char>um): Likewise.
5796         (altivec_vpku<VI_char>um_direct): New (copy of
5797         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
5798         internal use).
5799         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
5800         target is little endian and -maltivec=be is not specified.
5801         (*altivec_vupkhs<VU_char>_direct): New (copy of
5802         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
5803         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
5804         target is little endian and -maltivec=be is not specified.
5805         (*altivec_vupkls<VU_char>_direct): New (copy of
5806         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
5807         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
5808         little endian and -maltivec=be is not specified.
5809         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
5810         little endian and -maltivec=be is not specified.
5811
5812 2014-02-05  Richard Henderson  <rth@redhat.com>
5813
5814         PR debug/52727
5815         * combine-stack-adj.c: Revert r206943.
5816         * sched-int.h (struct deps_desc): Add last_args_size.
5817         * sched-deps.c (init_deps): Initialize it.
5818         (sched_analyze_insn): Add OUTPUT dependencies between insns that
5819         contain REG_ARGS_SIZE notes.
5820
5821 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5822
5823         * lto-cgraph.c (asm_nodes_output): Make global.
5824         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
5825         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
5826         (driver_handle_option): Handle OPT_fwpa.
5827
5828 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
5829
5830         PR ipa/59947
5831         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
5832         a comment typo and formatting issue.  If odr_hash hasn't been
5833         created, return vNULL and set *completep to false.
5834
5835         PR middle-end/57499
5836         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
5837         bb with no successors.
5838
5839 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
5840
5841         PR target/59718
5842         * doc/invoke.texi (-march): Clarify documentation for ARM.
5843         (-mtune): Likewise.
5844         (-mcpu): Likewise.
5845
5846 2014-02-05  Richard Biener  <rguenther@suse.de>
5847
5848         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
5849         when not vectorizing because of too many alias checks.
5850         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
5851         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
5852
5853 2014-02-05  Nick Clifton  <nickc@redhat.com>
5854
5855         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
5856         accept extended registers in any mode when compiling for the MN10300.
5857
5858 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
5859
5860         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
5861         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
5862         sanitization attributes.
5863         (can_inline_edge_p): Likewise.
5864         (sanitize_attrs_match_for_inline_p): New function.
5865
5866 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5867
5868         * ipa-prop.c (detect_type_change): Shor circuit testing of
5869         type changes on THIS pointer.
5870
5871 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
5872
5873         PR target/59777
5874         * config/pa/pa.c (legitimize_tls_address): Return original address
5875         if not passed a SYMBOL_REF rtx.
5876         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
5877         addresses.
5878         (pa_emit_move_sequence): Simplify TLS source operands.
5879         (pa_legitimate_constant_p): Reject all TLS constants.
5880         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
5881         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
5882
5883 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5884
5885         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
5886         groups when we know they are controlled by LTO.
5887         * varasm.c (default_binds_local_p_1): If object is in other partition,
5888         it will be resolved locally.
5889
5890 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5891
5892         * config/host-linux.c (linux_gt_pch_use_address): Don't
5893         use SSIZE_MAX because it is not always defined.
5894
5895 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
5896
5897         PR bootstrap/59913
5898         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
5899         threshold for pseudo splitting.
5900         (update_ebb_live_info): Process call argument hard registers and
5901         hard registers from insn definition too.
5902         (max_small_class_regs_num): New constant.
5903         (inherit_in_ebb): Update live hard regs through EBBs.  Update
5904         reloads_num only for small register classes.  Don't split for
5905         outputs of jumps.
5906
5907 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
5908
5909         PR ipa/60058
5910         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
5911         is non-null.
5912
5913 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5914
5915         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
5916         visibility is safe.
5917
5918 2014-02-04  Marek Polacek  <polacek@redhat.com>
5919
5920         * gdbinit.in (pel): Define.
5921
5922 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5923
5924         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
5925         behavior.
5926
5927 2014-02-04  Richard Biener  <rguenther@suse.de>
5928
5929         PR lto/59723
5930         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
5931         in function context local.
5932         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
5933         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
5934         similar to LTO_imported_decl_ref.
5935
5936 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5937
5938         PR tree-optimization/60002
5939         * cgraphclones.c (build_function_decl_skip_args): Clear
5940         DECL_LANG_SPECIFIC.
5941
5942         PR tree-optimization/60023
5943         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
5944         false to gsi_replace.
5945         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
5946         has been in some EH region and vec_stmt could throw, add
5947         vec_stmt into the same EH region.
5948         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
5949         has no lhs, ignore it.
5950         * internal-fn.c (expand_MASK_LOAD): Likewise.
5951
5952         PR ipa/60026
5953         * tree-inline.c (copy_forbidden): Fail for
5954         __attribute__((optimize (0))) functions.
5955
5956         PR other/58712
5957         * omp-low.c (simd_clone_struct_copy): If from->inbranch
5958         is set, copy one less argument.
5959         (expand_simd_clones): Don't subtract clone_info->inbranch
5960         from simd_clone_struct_alloc argument.
5961
5962         PR rtl-optimization/57915
5963         * recog.c (simplify_while_replacing): If all unary/binary/relational
5964         operation arguments are constant, attempt to simplify those.
5965
5966         PR middle-end/59261
5967         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
5968         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
5969
5970 2014-02-04  Richard Biener  <rguenther@suse.de>
5971
5972         PR tree-optimization/60012
5973         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
5974         TBAA disambiguation to all DDRs.
5975
5976 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5977
5978         PR target/59788
5979         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
5980         (LINK_SPEC): Use it for -shared, -shared-libgcc.
5981
5982 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5983
5984         PR ipa/59882
5985         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
5986
5987 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5988
5989         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
5990         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
5991
5992 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5993
5994         PR ipa/59831
5995         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
5996         to figure out targets of polymorphic calls with known decl.
5997         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5998         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
5999         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
6000         (get_polymorphic_call_info): ... here.
6001         (get_polymorphic_call_info_from_invariant): New function.
6002
6003 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6004
6005         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
6006         lookup via vtable pointer; check for type consistency
6007         and turn inconsitent facts into UNREACHABLE.
6008         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
6009         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
6010         type inconsistent querries; return UNREACHABLE instead.
6011
6012 2014-02-03  Richard Henderson  <rth@twiddle.net>
6013
6014         PR tree-opt/59924
6015         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
6016         already processed this node.
6017         (normalize_one_pred_1): Pass along mark_set.
6018         (normalize_one_pred): Create and destroy a pointer_set_t.
6019         (normalize_one_pred_chain): Likewise.
6020
6021 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
6022
6023         PR gcov-profile/58602
6024         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
6025
6026 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
6027
6028         PR ipa/59831
6029         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
6030         -fno-devirtualize; try to devirtualize by the knowledge of
6031         virtual table pointer given by aggregate propagation.
6032         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
6033         (ipa_print_node_jump_functions): Dump also offset that
6034         is relevant for polymorphic calls.
6035         (determine_known_aggregate_parts): Add arg_type parameter; use it
6036         instead of determining the type from pointer type.
6037         (ipa_compute_jump_functions_for_edge): Update call of
6038         determine_known_aggregate_parts.
6039         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
6040         (gimple_get_virt_method_for_binfo): ... here; simplify using
6041         vtable_pointer_value_to_vtable.
6042         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
6043         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
6044         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
6045         (vtable_pointer_value_to_vtable): Break out from ...; handle also
6046         POINTER_PLUS_EXPR.
6047         (vtable_pointer_value_to_binfo): ... here.
6048         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
6049
6050 2014-02-03  Teresa Johnson  <tejohnson@google.com>
6051
6052         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
6053         redef of outer loop index variable.
6054
6055 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
6056
6057         PR c++/53017
6058         PR c++/59211
6059         * doc/extend.texi (Function Attributes): Typo.
6060
6061 2014-02-03  Cong Hou  <congh@google.com>
6062
6063         PR tree-optimization/60000
6064         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
6065         if the vectorized statement is a store.  A store statement can only
6066         appear at the end of pattern statements.
6067
6068 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
6069
6070         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
6071         (ix86_option_override_internal): Default long double to 64-bit for
6072         32-bit Bionic and to 128-bit for 64-bit Bionic.
6073
6074         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
6075         TARGET_LONG_DOUBLE_128 is true.
6076         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
6077
6078         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
6079         (mlong-double-64): Negate -mlong-double-128.
6080         (mlong-double-128): New option.
6081
6082         * config/i386/i386-c.c (ix86_target_macros): Define
6083         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
6084
6085         * doc/invoke.texi: Document -mlong-double-128.
6086
6087 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
6088
6089         PR rtl-optimization/60024
6090         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
6091
6092 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
6093
6094         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
6095
6096 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
6097
6098         PR rtl-optimization/57662
6099         * sel-sched.c (code_motion_path_driver): Do not mark already not
6100         existing blocks in the visiting bitmap.
6101
6102 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
6103
6104         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
6105         on the insn being emitted.
6106
6107 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
6108             Will Deacon  <will.deacon@arm.com>
6109
6110         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
6111
6112 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6113
6114         * config/arm/arm-tables.opt: Regenerate.
6115
6116 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6117
6118         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
6119         for vector types other than V16QImode.
6120         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
6121         define_expand, and call altivec_expand_vec_perm_le when producing
6122         code with little endian element order.
6123         (*altivec_vperm_<mode>_internal): New insn having previous
6124         behavior of altivec_vperm_<mode>.
6125         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
6126         altivec_expand_vec_perm_le when producing code with little endian
6127         element order.
6128         (*altivec_vperm_<mode>_uns_internal): New insn having previous
6129         behavior of altivec_vperm_<mode>_uns.
6130
6131 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6132
6133         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
6134         (altivec_vsumsws): Add handling for -maltivec=be with a little
6135         endian target.
6136         (altivec_vsumsws_direct): New.
6137         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
6138         gen_altivec_vsumsws.
6139
6140 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
6141
6142         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
6143         vtable_pointer_value_to_binfo): New functions.
6144         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
6145         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
6146
6147 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
6148
6149         * config/nios2/nios2.md (load_got_register): Initialize GOT
6150         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
6151         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
6152
6153 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
6154
6155         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
6156         preserverd by passthrough, do not propagate the type.
6157
6158 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6159
6160         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
6161         (mips_atomic_assign_expand_fenv): New function.
6162         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
6163
6164 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6165
6166         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
6167         (__builtin_mips_set_fcsr): Likewise.
6168         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
6169         MIPS_USI_FTYPE_VOID.
6170         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
6171         (mips16_expand_set_fcsr): Likewise.
6172         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
6173         (mips16_set_fcsr_stub): Likewise.
6174         (mips16_get_fcsr_one_only_stub): New class.
6175         (mips16_set_fcsr_one_only_stub): Likewise.
6176         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
6177         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
6178         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
6179         (hard_float): New availability predicate.
6180         (mips_builtins): Add get_fcsr and set_fcsr.
6181         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
6182         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
6183         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
6184         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
6185         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
6186         patterns.
6187
6188 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
6189
6190         * config/mips/mips.c (mips_one_only_stub): New class.
6191         (mips_need_mips16_rdhwr_p): Replace with...
6192         (mips16_rdhwr_stub): ...this new variable.
6193         (mips16_stub_call_address): New function.
6194         (mips16_rdhwr_one_only_stub): New class.
6195         (mips_expand_thread_pointer): Use mips16_stub_call_address.
6196         (mips_output_mips16_rdhwr): Delete.
6197         (mips_finish_stub): New function.
6198         (mips_code_end): Use it to handle rdhwr stubs.
6199
6200 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
6201
6202         PR target/60017
6203         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
6204         when calculating size of integer atomic types.
6205
6206 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
6207
6208         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
6209
6210 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
6211
6212         PR tree-optimization/60003
6213         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
6214         * profile.c (branch_prob): Use gimple_call_builtin_p
6215         to check for BUILT_IN_SETJMP_RECEIVER.
6216         * tree-inline.c (copy_bb): Call notice_special_calls.
6217
6218 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
6219
6220         PR bootstrap/59985
6221         * lra-constraints.c (process_alt_operands): Update reload_sum only
6222         on the first pass.
6223
6224 2014-01-31  Richard Henderson  <rth@redhat.com>
6225
6226         PR middle-end/60004
6227         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
6228         until after else_eh is processed.
6229
6230 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6231
6232         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
6233         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
6234         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
6235         in smmintrin.h, remove them.
6236         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
6237         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
6238         * config/i386/i386.md (ROUND_SAE): Fix value.
6239         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
6240         (const48_operand): New.
6241         * config/i386/subst.md (round), (round_expand): Use
6242         const_4_or_8_to_11_operand.
6243         (round_saeonly), (round_saeonly_expand): Use const48_operand.
6244
6245 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6246
6247         * config/i386/constraints.md (Yk): Swap meaning with k.
6248         * config/i386/i386.md (movhi_internal): Change Yk to k.
6249         (movqi_internal): Ditto.
6250         (*k<logic><mode>): Ditto.
6251         (*andhi_1): Ditto.
6252         (*andqi_1): Ditto.
6253         (kandn<mode>): Ditto.
6254         (*<code>hi_1): Ditto.
6255         (*<code>qi_1): Ditto.
6256         (kxnor<mode>): Ditto.
6257         (kortestzhi): Ditto.
6258         (kortestchi): Ditto.
6259         (kunpckhi): Ditto.
6260         (*one_cmplhi2_1): Ditto.
6261         (*one_cmplqi2_1): Ditto.
6262         * config/i386/sse.md (): Change k to Yk.
6263         (avx512f_load<mode>_mask): Ditto.
6264         (avx512f_blendm<mode>): Ditto.
6265         (avx512f_store<mode>_mask): Ditto.
6266         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
6267         (avx512f_storedqu<mode>_mask): Ditto.
6268         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
6269         Ditto.
6270         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
6271         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
6272         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
6273         (avx512f_maskcmp<mode>3): Ditto.
6274         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
6275         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
6276         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
6277         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
6278         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
6279         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
6280         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
6281         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
6282         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
6283         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
6284         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
6285         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
6286         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
6287         (vec_extract_lo_<mode>_maskm): Ditto.
6288         (vec_extract_hi_<mode>_maskm): Ditto.
6289         (avx512f_vternlog<mode>_mask): Ditto.
6290         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
6291         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
6292         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
6293         (avx512f_<code>v8div16qi2_mask): Ditto.
6294         (avx512f_<code>v8div16qi2_mask_store): Ditto.
6295         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
6296         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
6297         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
6298         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
6299         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6300         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6301         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6302         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6303         (avx512cd_maskb_vec_dupv8di): Ditto.
6304         (avx512cd_maskw_vec_dupv16si): Ditto.
6305         (avx512f_vpermi2var<mode>3_maskz): Ditto.
6306         (avx512f_vpermi2var<mode>3_mask): Ditto.
6307         (avx512f_vpermi2var<mode>3_mask): Ditto.
6308         (avx512f_vpermt2var<mode>3_maskz): Ditto.
6309         (*avx512f_gathersi<mode>): Ditto.
6310         (*avx512f_gathersi<mode>_2): Ditto.
6311         (*avx512f_gatherdi<mode>): Ditto.
6312         (*avx512f_gatherdi<mode>_2): Ditto.
6313         (*avx512f_scattersi<mode>): Ditto.
6314         (*avx512f_scatterdi<mode>): Ditto.
6315         (avx512f_compress<mode>_mask): Ditto.
6316         (avx512f_compressstore<mode>_mask): Ditto.
6317         (avx512f_expand<mode>_mask): Ditto.
6318         * config/i386/subst.md (mask): Change k to Yk.
6319         (mask_scalar_merge): Ditto.
6320         (sd): Ditto.
6321
6322 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
6323
6324         * doc/extend.texi (Vector Extensions): Document ?: in C++.
6325
6326 2014-01-31  Richard Biener  <rguenther@suse.de>
6327
6328         PR middle-end/59990
6329         * builtins.c (fold_builtin_memory_op): Make sure to not
6330         use a floating-point mode or a boolean or enumeral type for
6331         the copy operation.
6332
6333 2014-01-30  DJ Delorie  <dj@redhat.com>
6334
6335         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
6336         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
6337         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6338         whenever main() has an epilogue.
6339
6340 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6341
6342         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
6343         unused variable "field".
6344         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
6345         (vsx_mergeh_<mode>): Likewise.
6346         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
6347         (altivec_vmrghh): Likewise.
6348         (altivec_vmrghw): Likewise.
6349         (altivec_vmrglb): Likewise.
6350         (altivec_vmrglh): Likewise.
6351         (altivec_vmrglw): Likewise.
6352         (altivec_vspltb): Add missing uses.
6353         (altivec_vsplth): Likewise.
6354         (altivec_vspltw): Likewise.
6355         (altivec_vspltsf): Likewise.
6356
6357 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6358
6359         PR target/59923
6360         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
6361         frame related instructions.
6362
6363 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
6364
6365         PR rtl-optimization/59959
6366         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
6367         any reload of register whose subreg is invalid.
6368
6369 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6370
6371         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
6372         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
6373         Add missing return type - void.
6374
6375 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6376
6377         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
6378         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
6379         remove element index adjustment for endian (now handled in vsx.md
6380         and altivec.md).
6381         (altivec_expand_vec_perm_const): Use
6382         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
6383         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
6384         (vsx_xxspltw_<mode>): Adjust element index for little endian.
6385         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
6386         define_expand and a new define_insn *altivec_vspltb_internal;
6387         adjust for -maltivec=be on a little endian target.
6388         (altivec_vspltb_direct): New.
6389         (altivec_vsplth): Divide into a define_expand and a new
6390         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
6391         little endian target.
6392         (altivec_vsplth_direct): New.
6393         (altivec_vspltw): Divide into a define_expand and a new
6394         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
6395         little endian target.
6396         (altivec_vspltw_direct): New.
6397         (altivec_vspltsf): Divide into a define_expand and a new
6398         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
6399         a little endian target.
6400
6401 2014-01-30  Richard Biener  <rguenther@suse.de>
6402
6403         PR tree-optimization/59993
6404         * tree-ssa-forwprop.c (associate_pointerplus): Check we
6405         can propagate form the earlier stmt and avoid the transform
6406         when the intermediate result is needed.
6407
6408 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
6409
6410         * README.Portability: Fix typo.
6411
6412 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
6413
6414         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
6415         comparison_operator with ordered_comparison_operator.
6416
6417 2014-01-30  Nick Clifton  <nickc@redhat.com>
6418
6419         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
6420         Rename to mn10300_store_multiple_regs.
6421         * config/mn10300/mn10300.c: Likewise.
6422         * config/mn10300/mn10300.md (store_movm): Fix typo: call
6423         store_multiple_regs.
6424         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
6425         Call mn10300_store_multiple_regs.
6426
6427 2014-01-30  Nick Clifton  <nickc@redhat.com>
6428             DJ Delorie  <dj@redhat.com>
6429
6430         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
6431         %fp 2 to keep registers after it properly word-aligned.
6432         (rl78_alloc_physical_registers_umul): Handle the case where both
6433         input operands are the same.
6434
6435 2014-01-30  Richard Biener  <rguenther@suse.de>
6436
6437         PR tree-optimization/59903
6438         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
6439         check properly.
6440
6441 2014-01-30  Jason Merrill  <jason@redhat.com>
6442
6443         PR c++/59633
6444         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
6445
6446         PR c++/59645
6447         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
6448
6449 2014-01-30  Richard Biener  <rguenther@suse.de>
6450
6451         PR tree-optimization/59951
6452         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
6453
6454 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
6455
6456         PR target/59784
6457         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
6458         SFmode to DFmode case.
6459
6460 2014-01-29  DJ Delorie  <dj@redhat.com>
6461
6462         * config/msp430/msp430.opt (-minrt): New.
6463         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
6464         if -minrt given.
6465         (ENDFILE_SPEC): Likewise.
6466
6467 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
6468
6469         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
6470         (estimate_function_body_sizes): Use it.
6471
6472 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
6473
6474         PR c++/58561
6475         * dwarf2out.c (is_cxx_auto): New.
6476         (is_base_type): Use it.
6477         (gen_type_die_with_usage): Likewise.
6478
6479 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6480
6481         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
6482         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
6483         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
6484         -maltivec=be with LE targets.
6485         (vsx_mergeh_<mode>): Likewise.
6486         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
6487         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
6488         (altivec_vmrghb): Replace with define_expand and new
6489         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
6490         (altivec_vmrghb_direct): New define_insn.
6491         (altivec_vmrghh): Replace with define_expand and new
6492         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
6493         (altivec_vmrghh_direct): New define_insn.
6494         (altivec_vmrghw): Replace with define_expand and new
6495         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
6496         (altivec_vmrghw_direct): New define_insn.
6497         (*altivec_vmrghsf): Adjust for endianness.
6498         (altivec_vmrglb): Replace with define_expand and new
6499         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
6500         (altivec_vmrglb_direct): New define_insn.
6501         (altivec_vmrglh): Replace with define_expand and new
6502         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
6503         (altivec_vmrglh_direct): New define_insn.
6504         (altivec_vmrglw): Replace with define_expand and new
6505         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
6506         (altivec_vmrglw_direct): New define_insn.
6507         (*altivec_vmrglsf): Adjust for endianness.
6508         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6509         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6510         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6511         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6512         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6513         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6514         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6515         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6516
6517 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6518
6519         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
6520         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
6521         whitespace.
6522
6523 2014-01-29  Richard Biener  <rguenther@suse.de>
6524
6525         PR tree-optimization/58742
6526         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
6527         associate_pointerplus_align.
6528         (associate_pointerplus_diff): New function.
6529         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
6530         and associate_pointerplus_diff.
6531
6532 2014-01-29  Richard Biener  <rguenther@suse.de>
6533
6534         * lto-streamer.h (LTO_major_version): Bump to 3.
6535         (LTO_minor_version): Reset to 0.
6536
6537 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
6538
6539         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
6540         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
6541         (arm_file_start): Generate correct asm header for armv7ve.
6542         * config/arm/bpabi.h: Add multilib support for armv7ve.
6543         * config/arm/driver-arm.c: Change the architectures of cortex-a7
6544         and cortex-a15 to armv7ve.
6545         * config/arm/t-aprofile: Add multilib support for armv7ve.
6546         * doc/invoke.texi: Document -march=armv7ve.
6547
6548 2014-01-29  Richard Biener  <rguenther@suse.de>
6549
6550         PR tree-optimization/58742
6551         * tree-ssa-forwprop.c (associate_plusminus): Return true
6552         if we changed sth, defer EH cleanup to ...
6553         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
6554         (simplify_mult): New function.
6555
6556 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6557
6558         PR middle-end/59917
6559         PR tree-optimization/59920
6560         * tree.c (build_common_builtin_nodes): Remove
6561         __builtin_setjmp_dispatcher initialization.
6562         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
6563         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
6564         instead of gsi_after_labels + manually skipping debug stmts.
6565         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
6566         ignore bbs with IFN_ABNORMAL_DISPATCHER.
6567         * tree-inline.c (copy_edges_for_bb): Remove
6568         can_make_abnormal_goto argument, instead add abnormal_goto_dest
6569         argument.  Ignore computed_goto_p stmts.  Don't call
6570         make_abnormal_goto_edges.  If a call might need abnormal edges
6571         for non-local gotos, see if it already has an edge to
6572         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
6573         with true argument, don't do anything then, otherwise add
6574         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
6575         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
6576         caller.
6577         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
6578         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
6579         (lower_stmt): Don't set data->calls_builtin_setjmp.
6580         (lower_builtin_setjmp): Adjust comment.
6581         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
6582         * tree-cfg.c (found_computed_goto): Remove.
6583         (factor_computed_gotos): Remove.
6584         (make_goto_expr_edges): Return bool, true for computed gotos.
6585         Don't call make_abnormal_goto_edges.
6586         (build_gimple_cfg): Don't set found_computed_goto, don't call
6587         factor_computed_gotos.
6588         (computed_goto_p): No longer static.
6589         (make_blocks): Don't set found_computed_goto.
6590         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
6591         (make_edges): If make_goto_expr_edges returns true, push bb
6592         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
6593         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
6594         vector.  Record mapping between bbs and OpenMP regions if there
6595         are any, adjust make_gimple_omp_edges caller.  Call
6596         handle_abnormal_edges.
6597         (make_abnormal_goto_edges): Remove.
6598         * tree-cfg.h (make_abnormal_goto_edges): Remove.
6599         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
6600         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
6601         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
6602         * internal-fn.def (ABNORMAL_DISPATCHER): New.
6603         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
6604         filling *region also set *region_idx to (*region)->entry->index.
6605
6606         PR other/58712
6607         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
6608         For REGs set ORIGINAL_REGNO.
6609
6610 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
6611
6612         * doc/md.texi: Mention that a target shouldn't implement
6613         vec_widen_(s|u)mul_even/odd pair if it is less efficient
6614         than hi/lo pair.
6615
6616 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6617
6618         PR tree-optimization/59594
6619         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
6620         a copy of the datarefs vector rather than the vector itself.
6621
6622 2014-01-28  Jason Merrill  <jason@redhat.com>
6623
6624         PR c++/53756
6625         * dwarf2out.c (auto_die): New static.
6626         (gen_type_die_with_usage): Handle C++1y 'auto'.
6627         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
6628         on definition.
6629
6630 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
6631
6632         PR target/59672
6633         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
6634         (SPEC_X32): Likewise.
6635         (SPEC_64): Likewise.
6636         * config/i386/i386.c (ix86_option_override_internal): Turn off
6637         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
6638         for TARGET_16BIT.
6639         (x86_file_start): Output .code16gcc for TARGET_16BIT.
6640         * config/i386/i386.h (TARGET_16BIT): New macro.
6641         (TARGET_16BIT_P): Likewise.
6642         * config/i386/i386.opt: Add m16.
6643         * doc/invoke.texi: Document -m16.
6644
6645 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
6646
6647         PR preprocessor/59935
6648         * input.c (location_get_source_line): Bail out on when line number
6649         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
6650
6651 2014-01-28  Richard Biener  <rguenther@suse.de>
6652
6653         PR tree-optimization/58742
6654         * tree-ssa-forwprop.c (associate_plusminus): Handle
6655         pointer subtraction of the form (T)(P + A) - (T)P.
6656
6657 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6658
6659         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
6660         at const_int_cost.
6661
6662 2014-01-28  Richard Biener  <rguenther@suse.de>
6663
6664         Revert
6665         2014-01-28  Richard Biener  <rguenther@suse.de>
6666
6667         PR rtl-optimization/45364
6668         PR rtl-optimization/59890
6669         * var-tracking.c (local_get_addr_clear_given_value): Handle
6670         already cleared slot.
6671         (val_reset): Handle not allocated local_get_addr_cache.
6672         (vt_find_locations): Use post-order on the inverted CFG.
6673
6674 2014-01-28  Richard Biener  <rguenther@suse.de>
6675
6676         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
6677
6678 2014-01-28  Richard Biener  <rguenther@suse.de>
6679
6680         PR rtl-optimization/45364
6681         PR rtl-optimization/59890
6682         * var-tracking.c (local_get_addr_clear_given_value): Handle
6683         already cleared slot.
6684         (val_reset): Handle not allocated local_get_addr_cache.
6685         (vt_find_locations): Use post-order on the inverted CFG.
6686
6687 2014-01-28  Alan Modra  <amodra@gmail.com>
6688
6689         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
6690         * configure.ac <recursive call for build != host>: Define
6691         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
6692         and LD_FOR_BUILD too.
6693         * configure: Regenerate.
6694
6695 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
6696
6697         * config/i386/i386.c (get_builtin_code_for_version): Separate
6698         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
6699         Broadwell from Haswell.
6700
6701 2014-01-27  Steve Ellcey  <sellcey@mips.com>
6702
6703         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6704         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
6705         * config/mips/mips.c (mips_option_override): Change setting
6706         of TARGET_DSP.
6707         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
6708         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
6709         Change from Mask to Var.
6710
6711 2014-01-27  Jeff Law  <law@redhat.com>
6712
6713         * ipa-inline.c (inline_small_functions): Fix typo.
6714
6715 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
6716
6717         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
6718         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
6719         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
6720         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
6721         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
6722         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
6723         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
6724         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
6725         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
6726         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
6727         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
6728         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
6729         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
6730         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
6731         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
6732         (_mm512_storeu_epi64): Ditto.
6733         (_mm512_cmpge_epi32_mask): Ditto.
6734         (_mm512_cmpge_epu32_mask): Ditto.
6735         (_mm512_cmpge_epi64_mask): Ditto.
6736         (_mm512_cmpge_epu64_mask): Ditto.
6737         (_mm512_cmple_epi32_mask): Ditto.
6738         (_mm512_cmple_epu32_mask): Ditto.
6739         (_mm512_cmple_epi64_mask): Ditto.
6740         (_mm512_cmple_epu64_mask): Ditto.
6741         (_mm512_cmplt_epi32_mask): Ditto.
6742         (_mm512_cmplt_epu32_mask): Ditto.
6743         (_mm512_cmplt_epi64_mask): Ditto.
6744         (_mm512_cmplt_epu64_mask): Ditto.
6745         (_mm512_cmpneq_epi32_mask): Ditto.
6746         (_mm512_cmpneq_epu32_mask): Ditto.
6747         (_mm512_cmpneq_epi64_mask): Ditto.
6748         (_mm512_cmpneq_epu64_mask): Ditto.
6749         (_mm512_expand_pd): Ditto.
6750         (_mm512_expand_ps): Ditto.
6751         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
6752         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
6753         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
6754         * config/i386/i386.c (ix86_builtins): Add
6755         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
6756         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
6757         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
6758         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
6759         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
6760         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
6761         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
6762         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
6763         IX86_BUILTIN_PMOVUSQW512_MEM.
6764         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
6765         __builtin_ia32_pmovsqd512mem_mask,
6766         __builtin_ia32_pmovqd512mem_mask,
6767         __builtin_ia32_pmovusqw512mem_mask,
6768         __builtin_ia32_pmovsqw512mem_mask,
6769         __builtin_ia32_pmovqw512mem_mask,
6770         __builtin_ia32_pmovusdw512mem_mask,
6771         __builtin_ia32_pmovsdw512mem_mask,
6772         __builtin_ia32_pmovdw512mem_mask,
6773         __builtin_ia32_pmovqb512mem_mask,
6774         __builtin_ia32_pmovusqb512mem_mask,
6775         __builtin_ia32_pmovsqb512mem_mask,
6776         __builtin_ia32_pmovusdb512mem_mask,
6777         __builtin_ia32_pmovsdb512mem_mask,
6778         __builtin_ia32_pmovdb512mem_mask.
6779         (bdesc_args): Add __builtin_ia32_expanddf512,
6780         __builtin_ia32_expandsf512.
6781         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
6782         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
6783         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
6784         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
6785         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
6786         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
6787         (avx512f_<code>v8div16qi2_mask_store): This.
6788         (avx512f_expand<mode>): New.
6789
6790 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
6791
6792         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
6793         New.
6794         (_mm512_mask_prefetch_i64gather_pd): Ditto.
6795         (_mm512_prefetch_i32scatter_pd): Ditto.
6796         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6797         (_mm512_prefetch_i64scatter_pd): Ditto.
6798         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6799         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
6800         (_mm512_mask_prefetch_i64gather_ps): Ditto.
6801         (_mm512_prefetch_i32scatter_ps): Ditto.
6802         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6803         (_mm512_prefetch_i64scatter_ps): Ditto.
6804         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6805         * config/i386/i386-builtin-types.def: Define
6806         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
6807         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
6808         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
6809         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
6810         IX86_BUILTIN_SCATTERPFQPD.
6811         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
6812         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
6813         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
6814         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
6815         __builtin_ia32_scatterpfqps.
6816         (ix86_expand_builtin): Expand new built-ins.
6817         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
6818         fix memory access data type.
6819         (*avx512pf_gatherpf<mode>_mask): Ditto.
6820         (*avx512pf_gatherpf<mode>): Ditto.
6821         (avx512pf_scatterpf<mode>): Ditto.
6822         (*avx512pf_scatterpf<mode>_mask): Ditto.
6823         (*avx512pf_scatterpf<mode>): Ditto.
6824         (GATHER_SCATTER_SF_MEM_MODE): New.
6825         (avx512pf_gatherpf<mode>df): Ditto.
6826         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6827         (*avx512pf_scatterpf<mode>df): Ditto.
6828
6829 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
6830
6831         PR bootstrap/59934
6832         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
6833         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
6834         reached.
6835
6836 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6837
6838         * common/config/arm/arm-common.c
6839         (arm_rewrite_mcpu): Handle multiple names.
6840         * config/arm/arm.h
6841         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6842
6843 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6844
6845         * gimple-builder.h (create_gimple_tmp): Delete.
6846
6847 2014-01-27  Christian Bruel  <christian.bruel@st.com>
6848
6849         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
6850         words comparisons.
6851
6852 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
6853
6854         * config/pa/pa.md (call): Generate indirect long calls to non-local
6855         functions when outputing 32-bit code.
6856         (call_value): Likewise except for special call to buggy powf function.
6857
6858         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
6859         portable runtime and PIC indirect calls.
6860         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
6861         and PIC call sequences.  Use ldo instead of blr to set return register
6862         in PIC call sequence.
6863
6864 2014-01-25  Walter Lee  <walt@tilera.com>
6865
6866         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
6867         avoid clobbering a live register.
6868
6869 2014-01-25  Walter Lee  <walt@tilera.com>
6870
6871         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
6872         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
6873         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
6874         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
6875
6876 2014-01-25  Walter Lee  <walt@tilera.com>
6877
6878         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
6879         arguments on even registers.
6880         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
6881         STACK_BOUNDARY.
6882         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
6883         (BIGGEST_ALIGNMENT): Ditto.
6884         (BIGGEST_FIELD_ALIGNMENT): Ditto.
6885
6886 2014-01-25  Walter Lee  <walt@tilera.com>
6887
6888         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
6889         insns before bundling.
6890         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
6891
6892 2014-01-25  Walter Lee  <walt@tilera.com>
6893
6894         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
6895         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
6896         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
6897
6898 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6899
6900         * config/mips/constraints.md (kl): Delete.
6901         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
6902         define expands, using...
6903         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
6904         instructions for MIPS16.
6905         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
6906         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
6907
6908 2014-01-25  Walter Lee  <walt@tilera.com>
6909
6910         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
6911         (clzdi2): Ditto.
6912         (ffsdi2): Ditto.
6913
6914 2014-01-25  Walter Lee  <walt@tilera.com>
6915
6916         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
6917         (TARGET_EXPAND_TO_RTL_HOOK): Define.
6918
6919 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6920
6921         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
6922         Handle XOR.
6923
6924 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
6925
6926         * print-rtl.c (in_call_function_usage): New var.
6927         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
6928         EXPR_LIST mode as mode and not as reg note name.
6929
6930         PR middle-end/59561
6931         * cfgloopmanip.c (copy_loop_info): If
6932         loop->warned_aggressive_loop_optimizations, make sure
6933         the flag is set in target loop too.
6934
6935 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6936
6937         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
6938         flag_cilkplus.
6939         * builtins.def: Likewise.
6940         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
6941         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
6942         * ira.c (ira_setup_eliminable_regset): Likewise.
6943         * omp-low.c (gate_expand_omp): Likewise.
6944         (execute_lower_omp): Likewise.
6945         (diagnose_sb_0): Likewise.
6946         (gate_diagnose_omp_blocks): Likewise.
6947         (simd_clone_clauses_extract): Likewise.
6948         (gate): Likewise.
6949
6950 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6951
6952         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
6953         correction for little endian...
6954         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
6955         here.
6956
6957 2014-01-24  Jeff Law  <law@redhat.com>
6958
6959         PR tree-optimization/59919
6960         * tree-vrp.c (find_assert_locations_1): Do not register asserts
6961         for non-returning calls.
6962
6963 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
6964
6965         * common/config/aarch64/aarch64-common.c
6966         (aarch64_rewrite_mcpu): Handle multiple names.
6967         * config/aarch64/aarch64.h
6968         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6969
6970 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
6971
6972         * input.c (add_file_to_cache_tab): Handle the case where fopen
6973         returns NULL.
6974
6975 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
6976
6977         PR target/59929
6978         * config/i386/i386.md (pushsf splitter): Get stack adjustment
6979         from push operand if code of push isn't PRE_DEC.
6980
6981 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6982
6983         PR target/59909
6984         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
6985         -mquad-memory-atomic.  Update -mquad-memory documentation to say
6986         it is only used for non-atomic loads/stores.
6987
6988         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
6989         -mquad-memory or -mquad-memory-atomic switches.
6990
6991         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
6992         -mquad-memory-atomic to ISA 2.07 support.
6993
6994         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
6995         to separate support of normal quad word memory operations (ldq, stq)
6996         from the atomic quad word memory operations.
6997
6998         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6999         support to separate non-atomic quad word operations from atomic
7000         quad word operations.  Disable non-atomic quad word operations in
7001         little endian mode so that we don't have to swap words after the
7002         load and before the store.
7003         (quad_load_store_p): Add comment about atomic quad word support.
7004         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
7005         options printed with -mdebug=reg.
7006
7007         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
7008         -mquad-memory-atomic as the test for whether we have quad word
7009         atomic instructions.
7010         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
7011         or -mp8-vector are used, allow byte/half-word atomic operations.
7012
7013         * config/rs6000/sync.md (load_lockedti): Insure that the address
7014         is a proper indexed or indirect address for the lqarx instruction.
7015         On little endian systems, swap the hi/lo registers after the lqarx
7016         instruction.
7017         (load_lockedpti): Use indexed_or_indirect_operand predicate to
7018         insure the address is valid for the lqarx instruction.
7019         (store_conditionalti): Insure that the address is a proper indexed
7020         or indirect address for the stqcrx. instruction.  On little endian
7021         systems, swap the hi/lo registers before doing the stqcrx.
7022         instruction.
7023         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
7024         insure the address is valid for the stqcrx. instruction.
7025
7026         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7027         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
7028         type of quad memory support is available.
7029
7030 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
7031
7032         PR regression/59915
7033         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
7034         there is a danger of looping.
7035
7036 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
7037
7038         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7039         force flag_ira_loop_pressure if set via command line.
7040
7041 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7042
7043         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
7044         (ashr_simd): New builtin handling DI mode.
7045         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
7046         (aarch64_sshr_simddi): New match pattern.
7047         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
7048         (vshrd_n_s64): Likewise.
7049         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
7050
7051 2014-01-23  Nick Clifton  <nickc@redhat.com>
7052
7053         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
7054         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
7055         favour of mcu specific scripts.
7056         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
7057         430x multilibs.
7058
7059 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
7060             Alex Velenko  <Alex.Velenko@arm.com>
7061
7062         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
7063         (vaddv_s16): Likewise.
7064         (vaddv_s32): Likewise.
7065         (vaddv_u8): Likewise.
7066         (vaddv_u16): Likewise.
7067         (vaddv_u32): Likewise.
7068         (vaddvq_s8): Likewise.
7069         (vaddvq_s16): Likewise.
7070         (vaddvq_s32): Likewise.
7071         (vaddvq_s64): Likewise.
7072         (vaddvq_u8): Likewise.
7073         (vaddvq_u16): Likewise.
7074         (vaddvq_u32): Likewise.
7075         (vaddvq_u64): Likewise.
7076         (vaddv_f32): Likewise.
7077         (vaddvq_f32): Likewise.
7078         (vaddvq_f64): Likewise.
7079         (vmaxv_f32): Likewise.
7080         (vmaxv_s8): Likewise.
7081         (vmaxv_s16): Likewise.
7082         (vmaxv_s32): Likewise.
7083         (vmaxv_u8): Likewise.
7084         (vmaxv_u16): Likewise.
7085         (vmaxv_u32): Likewise.
7086         (vmaxvq_f32): Likewise.
7087         (vmaxvq_f64): Likewise.
7088         (vmaxvq_s8): Likewise.
7089         (vmaxvq_s16): Likewise.
7090         (vmaxvq_s32): Likewise.
7091         (vmaxvq_u8): Likewise.
7092         (vmaxvq_u16): Likewise.
7093         (vmaxvq_u32): Likewise.
7094         (vmaxnmv_f32): Likewise.
7095         (vmaxnmvq_f32): Likewise.
7096         (vmaxnmvq_f64): Likewise.
7097         (vminv_f32): Likewise.
7098         (vminv_s8): Likewise.
7099         (vminv_s16): Likewise.
7100         (vminv_s32): Likewise.
7101         (vminv_u8): Likewise.
7102         (vminv_u16): Likewise.
7103         (vminv_u32): Likewise.
7104         (vminvq_f32): Likewise.
7105         (vminvq_f64): Likewise.
7106         (vminvq_s8): Likewise.
7107         (vminvq_s16): Likewise.
7108         (vminvq_s32): Likewise.
7109         (vminvq_u8): Likewise.
7110         (vminvq_u16): Likewise.
7111         (vminvq_u32): Likewise.
7112         (vminnmv_f32): Likewise.
7113         (vminnmvq_f32): Likewise.
7114         (vminnmvq_f64): Likewise.
7115
7116 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
7117
7118         * config/aarch64/aarch64-simd.md
7119         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
7120         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
7121         (*aarch64_mul3_elt<mode>): Likewise.
7122         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
7123         (*aarch64_mul3_elt_to_64v2df): Likewise.
7124         (*aarch64_mla_elt<mode>): Likewise.
7125         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
7126         (*aarch64_mls_elt<mode>): Likewise.
7127         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
7128         (*aarch64_fma4_elt<mode>): Likewise.
7129         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
7130         (*aarch64_fma4_elt_to_64v2df): Likewise.
7131         (*aarch64_fnma4_elt<mode>): Likewise.
7132         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7133         (*aarch64_fnma4_elt_to_64v2df): Likewise.
7134         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
7135         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
7136         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
7137         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
7138         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7139         (aarch64_sqdmull_lane<mode>_internal): Likewise.
7140         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7141
7142 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7143
7144         * config/aarch64/aarch64-simd.md
7145         (aarch64_be_checked_get_lane<mode>): New define_expand.
7146         * config/aarch64/aarch64-simd-builtins.def
7147         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
7148         New builtin definition.
7149         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
7150         Use new safe be builtin.
7151
7152 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
7153
7154         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
7155         New define_insn.
7156         (aarch64_be_st1<mode>): Likewise.
7157         (aarch_ld1<VALL:mode>): Define_expand modified.
7158         (aarch_st1<VALL:mode>): Likewise.
7159         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
7160         (UNSPEC_ST1): Likewise.
7161
7162 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
7163
7164         * config/microblaze/microblaze.md: Add trap insn and attribute
7165
7166 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
7167
7168         PR preprocessor/58580
7169         * input.h (location_get_source_line): Take an additional line_size
7170         parameter.
7171         (void diagnostics_file_cache_fini): Declare new function.
7172         * input.c (struct fcache): New type.
7173         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
7174         New static constants.
7175         (diagnostic_file_cache_init, total_lines_num)
7176         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
7177         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
7178         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
7179         (get_next_line, read_next_line, goto_next_line, read_line_num):
7180         New static function definitions.
7181         (diagnostic_file_cache_fini): New function.
7182         (location_get_source_line): Take an additional output line_len
7183         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
7184         read_line_num.
7185         * diagnostic.c (diagnostic_finish): Call
7186         diagnostic_file_cache_fini.
7187         (adjust_line): Take an additional input parameter for the length
7188         of the line, rather than calculating it with strlen.
7189         (diagnostic_show_locus): Adjust the use of
7190         location_get_source_line and adjust_line with respect to their new
7191         signature.  While displaying a line now, do not stop at the first
7192         null byte.  Rather, display the zero byte as a space and keep
7193         going until we reach the size of the line.
7194         * Makefile.in: Add vec.o to OBJS-libcommon
7195
7196 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
7197             Ilya Tocar     <ilya.tocar@intel.com>
7198
7199         * config/i386/avx512fintrin.h (_mm512_kmov): New.
7200         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
7201         (__builtin_ia32_kmov16): Ditto.
7202         * config/i386/i386.md (UNSPEC_KMOV): New.
7203         (kmovw): Ditto.
7204
7205 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
7206
7207         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
7208         (_mm512_storeu_si512): Ditto.
7209
7210 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
7211
7212         PR target/52125
7213         * rtl.h (get_referenced_operands): Declare.
7214         * recog.c (get_referenced_operands): New function.
7215         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
7216         operands have been referenced when recording LO_SUM references.
7217
7218 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
7219
7220         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
7221
7222 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7223
7224         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
7225         Enable for generic and recent AMD targets.
7226
7227 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7228
7229         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
7230         ARG_SIZE note when adjustment was eliminated.
7231
7232 2014-01-22  Jeff Law  <law@redhat.com>
7233
7234         PR tree-optimization/59597
7235         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
7236         in file.  Accept new argument REGISTERING and use it to modify
7237         dump output appropriately.
7238         (register_jump_thread): Corresponding changes.
7239         (mark_threaded_blocks): Reinstate code to cancel unprofitable
7240         thread paths involving joiner blocks.  Add code to dump cancelled
7241         jump threading paths.
7242
7243 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
7244
7245         PR rtl-optimization/59477
7246         * lra-constraints.c (inherit_in_ebb): Process call for living hard
7247         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
7248
7249 2014-01-22  Tom Tromey  <tromey@redhat.com>
7250
7251         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
7252         PARAMS.
7253         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
7254
7255 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7256
7257         PR rtl-optimization/59896
7258         * lra-constraints.c (process_alt_operands): Check unused note for
7259         matched operands of insn with no output reloads.
7260
7261 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
7262
7263         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
7264         (mips_move_from_gpr_cost): Likewise.
7265
7266 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7267
7268         PR rtl-optimization/59858
7269         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
7270         ira_class_hard_regs_num.
7271         (process_alt_operands): Increase reject for dying matched operand.
7272
7273 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
7274
7275         PR target/59003
7276         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
7277         smaller than size, perform several stores or loads and stores
7278         at dst + count - size to store or copy all of size bytes, rather
7279         than just last modesize bytes.
7280
7281 2014-01-20  DJ Delorie  <dj@redhat.com>
7282
7283         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
7284         that CLOBBERs are REGs before propogating their values.
7285
7286 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
7287
7288         PR middle-end/59789
7289         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
7290         (cgraph_inline_failed_type): New function.
7291         * cgraph.h (DEFCIFCODE): Add type.
7292         (cgraph_inline_failed_type_t): New enum.
7293         (cgraph_inline_failed_type): New prototype.
7294         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
7295         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
7296         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
7297         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
7298         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
7299         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
7300         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
7301         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
7302         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
7303         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
7304         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
7305         OPTIMIZATION_MISMATCH.
7306         * tree-inline.c (expand_call_inline): Emit errors during
7307         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
7308
7309 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7310
7311         PR target/59685
7312         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
7313         mode attribute in insn output.
7314
7315 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
7316
7317         * output.h (output_constant): Delete.
7318         * varasm.c (output_constant): Make private.
7319
7320 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
7321
7322         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
7323
7324 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7325
7326         PR middle-end/59860
7327         * tree.h (fold_builtin_strcat): New prototype.
7328         * builtins.c (fold_builtin_strcat): No longer static.  Add len
7329         argument, if non-NULL, don't call c_strlen.  Optimize
7330         directly into __builtin_memcpy instead of __builtin_strcpy.
7331         (fold_builtin_2): Adjust fold_builtin_strcat caller.
7332         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
7333
7334 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7335
7336         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7337         for SImode_address_operand operands, having only a REG argument.
7338
7339 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7340
7341         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
7342         loader name using mbig-endian.
7343         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
7344
7345 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7346
7347         * doc/invoke.texi (-march): Clarify documentation for AArch64.
7348         (-mtune): Likewise.
7349         (-mcpu): Likewise.
7350
7351 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
7352
7353         * config/aarch64/aarch64-protos.h
7354         (aarch64_cannot_change_mode_class_ptr): Declare.
7355         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
7356         aarch64_cannot_change_mode_class_ptr): New.
7357         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
7358         backend hook aarch64_cannot_change_mode_class.
7359
7360 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7361
7362         * common/config/aarch64/aarch64-common.c
7363         (aarch64_handle_option): Don't handle any option order logic here.
7364         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
7365         selected_cpu, warn on architecture version mismatch.
7366         (aarch64_override_options): Fix parsing order for option strings.
7367
7368 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7369             Iain Sandoe  <iain@codesourcery.com>
7370
7371         PR bootstrap/59496
7372         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
7373         warning.  Amend comment to reflect current functionality.
7374
7375 2014-01-20  Richard Biener  <rguenther@suse.de>
7376
7377         PR middle-end/59860
7378         * builtins.c (fold_builtin_strcat): Remove case better handled
7379         by tree-ssa-strlen.c.
7380
7381 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
7382
7383         * config/aarch64/aarch64.opt
7384         (mcpu, march, mtune): Make case-insensitive.
7385
7386 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7387
7388         PR target/59880
7389         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7390         if operands[1] is a REG or ZERO_EXTEND of a REG.
7391
7392 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
7393
7394         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
7395
7396 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
7397
7398         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
7399         long non-pic millicode calls.
7400
7401 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7402
7403         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
7404
7405 2014-01-19  Kito Cheng  <kito@0xlab.org>
7406
7407         * builtins.c (expand_movstr): Check movstr expand done or fail.
7408
7409 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7410             H.J. Lu  <hongjiu.lu@intel.com>
7411
7412         PR target/59379
7413         * config/i386/i386.md (*lea<mode>): Zero-extend return register
7414         to DImode for zero-extended addresses.
7415
7416 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
7417
7418         PR rtl-optimization/57763
7419         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
7420         on the new indirect jump_insn and increment LABEL_NUSES (label).
7421
7422 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
7423
7424         PR bootstrap/59580
7425         PR bootstrap/59583
7426         * config.gcc (x86_archs): New variable.
7427         (x86_64_archs): Likewise.
7428         (x86_cpus): Likewise.
7429         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
7430         --with-arch/--with-cpu= options.
7431         Support --with-arch=/--with-cpu={nehalem,westmere,
7432         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
7433
7434 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7435
7436         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
7437         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
7438
7439 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7440
7441         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
7442
7443 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
7444
7445         PR target/58944
7446         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
7447         clear cpp_get_options (parse_in)->warn_unused_macros for
7448         ix86_target_macros_internal with cpp_define.
7449
7450 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
7451
7452         * jump.c (delete_related_insns): Keep (use (insn))s.
7453         * reorg.c (redundant_insn): Check for barriers too.
7454
7455 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7456
7457         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
7458
7459 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
7460
7461         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
7462         call to $$dyncall when TARGET_LONG_CALLS is true.
7463
7464 2014-01-17  Jeff Law  <law@redhat.com>
7465
7466         * ree.c (combine_set_extension): Temporarily disable test for
7467         changing number of hard registers.
7468
7469 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7470
7471         PR middle-end/58125
7472         * ipa-inline-analysis.c (inline_free_summary):
7473         Do not free summary of aliases.
7474
7475 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7476
7477         PR middle-end/59706
7478         * gimplify.c (gimplify_expr): Use create_tmp_var
7479         instead of create_tmp_var_raw.  If cond doesn't have
7480         integral type, don't add the IFN_ANNOTATE builtin at all.
7481
7482 2014-01-17  Martin Jambor  <mjambor@suse.cz>
7483
7484         PR ipa/59736
7485         * ipa-cp.c (prev_edge_clone): New variable.
7486         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
7487         Also resize prev_edge_clone vector.
7488         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
7489         (ipcp_edge_removal_hook): New function.
7490         (ipcp_driver): Register ipcp_edge_removal_hook.
7491
7492 2014-01-17  Andrew Pinski  <apinski@cavium.com>
7493             Steve Ellcey  <sellcey@mips.com>
7494
7495         PR target/59462
7496         * config/mips/mips.c (mips_print_operand): Check operand mode instead
7497         of operator mode.
7498
7499 2014-01-17  Jeff Law  <law@redhat.com>
7500
7501         PR middle-end/57904
7502         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
7503         so that pass_ccp runs first.
7504
7505 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7506
7507         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
7508         (ix86_adjust_cost): Use !TARGET_XXX.
7509         (do_reorder_for_imul): Likewise.
7510         (swap_top_of_ready_list): Likewise.
7511         (ix86_sched_reorder): Likewise.
7512
7513 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7514
7515         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7516         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
7517         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
7518         (intel_memset): New.  Duplicate slm_memset.
7519         (intel_cost): New.  Duplicate slm_cost.
7520         (m_INTEL): New macro.
7521         (processor_target_table): Add "intel".
7522         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
7523         with PROCESSOR_INTEL for "intel".
7524         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
7525         PROCESSOR_SILVERMONT.
7526         (ix86_issue_rate): Likewise.
7527         (ix86_adjust_cost): Likewise.
7528         (ia32_multipass_dfa_lookahead): Likewise.
7529         (swap_top_of_ready_list): Likewise.
7530         (ix86_sched_reorder): Likewise.
7531         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
7532         instead of TARGET_OPT_AGU.
7533         * config/i386/i386.h (TARGET_INTEL): New.
7534         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
7535         (processor_type): Add PROCESSOR_INTEL.
7536         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
7537         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
7538
7539 2014-01-17  Marek Polacek  <polacek@redhat.com>
7540
7541         PR c/58346
7542         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
7543         size is zero.
7544
7545 2014-01-17  Richard Biener  <rguenther@suse.de>
7546
7547         PR tree-optimization/46590
7548         * opts.c (default_options_table): Add entries for
7549         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
7550         all enabled at -O1 but not for -Og.
7551         * common.opt (fbranch-count-reg): Remove Init(1).
7552         (fmove-loop-invariants): Likewise.
7553         (ftree-pta): Likewise.
7554
7555 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7556
7557         * config/i386/i386.c (ix86_data_alignment): For compatibility with
7558         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
7559         decls to at least the GCC 4.8 used alignments.
7560
7561         PR fortran/59440
7562         * tree-nested.c (convert_nonlocal_reference_stmt,
7563         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
7564         of GIMPLE_BIND stmts, adjust associated decls.
7565
7566 2014-01-17  Richard Biener  <rguenther@suse.de>
7567
7568         PR tree-optimization/46590
7569         * vec.h (vec<>::bseach): New member function implementing
7570         binary search according to C89 bsearch.
7571         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
7572         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
7573         bitmap pointer again.  Make accesses_in_loop a flat array.
7574         (mem_ref_obstack): New global.
7575         (outermost_indep_loop): Adjust for mem_ref->stored changes.
7576         (mark_ref_stored): Likewise.
7577         (ref_indep_loop_p_2): Likewise.
7578         (set_ref_stored_in_loop): New helper function.
7579         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
7580         (memref_free): Adjust.
7581         (record_mem_ref_loc): Simplify.
7582         (gather_mem_refs_stmt): Adjust.
7583         (sort_locs_in_loop_postorder_cmp): New function.
7584         (analyze_memory_references): Sort accesses_in_loop after
7585         loop postorder number.
7586         (find_ref_loc_in_loop_cmp): New function.
7587         (for_all_locs_in_loop): Find relevant cluster of locs in
7588         accesses_in_loop and iterate without recursion.
7589         (execute_sm): Avoid uninit warning.
7590         (struct ref_always_accessed): Simplify.
7591         (ref_always_accessed::operator ()): Likewise.
7592         (ref_always_accessed_p): Likewise.
7593         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
7594         loop postorder numbers here.
7595         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
7596         numbers.
7597
7598 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7599
7600         PR c++/57945
7601         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
7602         on decls for which assemble_alias has been called.
7603
7604 2014-01-17  Nick Clifton  <nickc@redhat.com>
7605
7606         * config/msp430/msp430.opt: (mcpu): New option.
7607         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
7608         (msp430_option_override): Parse target_cpu.  If the MCU name
7609         matches a generic string, clear target_mcu.
7610         (msp430_attr): Allow numeric interrupt values up to 63.
7611         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
7612         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
7613         option.
7614         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
7615         Add mcpu matches.
7616         * config/msp430/msp430.md (popm): Use %J rather than %I.
7617         (addsi3): Use msp430_nonimmediate_operand for operand 2.
7618         (addhi_cy_i): Use immediate_operand for operand 2.
7619         * doc/invoke.texi: Document -mcpu option.
7620
7621 2014-01-17  Richard Biener  <rguenther@suse.de>
7622
7623         PR rtl-optimization/38518
7624         * df.h (df_analyze_loop): Declare.
7625         * df-core.c: Include cfgloop.h.
7626         (df_analyze_1): Split out main part of df_analyze.
7627         (df_analyze): Adjust.
7628         (loop_inverted_post_order_compute): New function.
7629         (loop_post_order_compute): Likewise.
7630         (df_analyze_loop): New function avoiding whole-function
7631         postorder computes.
7632         * loop-invariant.c (find_defs): Use df_analyze_loop.
7633         (find_invariants): Adjust.
7634         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
7635
7636 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
7637
7638         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
7639         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
7640
7641 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
7642
7643         * ipa-ref.c (ipa_remove_stmt_references): Fix references
7644         traversal when removing references.
7645
7646 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
7647
7648         PR ipa/59775
7649         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
7650
7651 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
7652
7653         PR middle-end/56791
7654         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
7655         pushing a reload for an autoinc when we had previously reloaded an
7656         inner part of the address.
7657
7658 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
7659
7660         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
7661         field.
7662         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
7663         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
7664         when not giving up or versioning for alias only because of
7665         loop->safelen.
7666         (vect_analyze_data_ref_dependences): Set to true.
7667         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
7668         is a GIMPLE_PHI.
7669         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
7670         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
7671         to the condition.
7672
7673         PR middle-end/58344
7674         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
7675
7676         PR target/59839
7677         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
7678         operand 0 predicate for gathers, use a new pseudo as subtarget.
7679
7680 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7681
7682         PR middle-end/59609
7683         * lra-constraints.c (process_alt_operands): Add printing debug info.
7684         Check absence of input/output reloads for matched operands too.
7685
7686 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7687
7688         PR rtl-optimization/59835
7689         * ira.c (ira_init_register_move_cost): Increase cost for
7690         impossible modes.
7691
7692 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
7693
7694         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
7695
7696 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
7697
7698         PR target/59780
7699         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
7700         non-register objects.  Use gen_(high/low)part more consistently.
7701         Fix assertions.
7702
7703 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
7704
7705         PR target/59844
7706         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
7707         endian support, remove tests for WORDS_BIG_ENDIAN.
7708         (p8_mfvsrd_3_<mode>): Likewise.
7709         (reload_gpr_from_vsx<mode>): Likewise.
7710         (reload_gpr_from_vsxsf): Likewise.
7711         (p8_mfvsrd_4_disf): Likewise.
7712
7713 2014-01-16  Richard Biener  <rguenther@suse.de>
7714
7715         PR rtl-optimization/46590
7716         * lcm.c (compute_antinout_edge): Use postorder iteration.
7717         (compute_laterin): Use inverted postorder iteration.
7718
7719 2014-01-16  Nick Clifton  <nickc@redhat.com>
7720
7721         PR middle-end/28865
7722         * varasm.c (output_constant): Return the number of bytes actually
7723         emitted.
7724         (output_constructor_array_range): Update the field size with the
7725         number of bytes emitted by output_constant.
7726         (output_constructor_regular_field): Likewise.  Also do not
7727         complain if the total number of bytes emitted is now greater
7728         than the expected fieldpos.
7729         * output.h (output_constant): Update prototype and descriptive comment.
7730
7731 2014-01-16  Marek Polacek  <polacek@redhat.com>
7732
7733         PR middle-end/59827
7734         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
7735         it is error_mark_node.
7736
7737 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
7738
7739         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
7740         VALID_AVX256_REG_OR_OI_MODE.
7741
7742 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
7743
7744         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
7745         current procedure should be profiled.
7746
7747 2014-01-15  Andrew Pinski  <apinski@cavium.com>
7748
7749         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
7750         of moving from/to the STACK_REG register class.
7751
7752 2014-01-15  Richard Henderson  <rth@redhat.com>
7753
7754         PR debug/54694
7755         * reginfo.c (global_regs_decl): Globalize.
7756         * rtl.h (global_regs_decl): Declare.
7757         * ira.c (do_reload): Diagnose frame_pointer_needed and it
7758         reserved via global_regs.
7759
7760 2014-01-15  Teresa Johnson  <tejohnson@google.com>
7761
7762         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
7763
7764 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7765
7766         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
7767         and vmulosh rather than call gen_vec_widen_smult_*.
7768         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
7769         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
7770         (vec_widen_smult_even_v16qi): Likewise.
7771         (vec_widen_umult_even_v8hi): Likewise.
7772         (vec_widen_smult_even_v8hi): Likewise.
7773         (vec_widen_umult_odd_v16qi): Likewise.
7774         (vec_widen_smult_odd_v16qi): Likewise.
7775         (vec_widen_umult_odd_v8hi): Likewise.
7776         (vec_widen_smult_odd_v8hi): Likewise.
7777         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
7778         vmuloub rather than call gen_vec_widen_umult_*.
7779         (vec_widen_umult_lo_v16qi): Likewise.
7780         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
7781         vmulosb rather than call gen_vec_widen_smult_*.
7782         (vec_widen_smult_lo_v16qi): Likewise.
7783         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
7784         rather than call gen_vec_widen_umult_*.
7785         (vec_widen_umult_lo_v8hi): Likewise.
7786         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
7787         rather than call gen_vec_widen_smult_*.
7788         (vec_widen_smult_lo_v8hi): Likewise.
7789
7790 2014-01-15  Jeff Law  <law@redhat.com>
7791
7792         PR tree-optimization/59747
7793         * ree.c (find_and_remove_re): Properly handle case where a second
7794         eliminated extension requires widening a copy created for elimination
7795         of a prior extension.
7796         (combine_set_extension): Ensure that the number of hard regs needed
7797         for a destination register does not change when we widen it.
7798
7799 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7800
7801         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
7802         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
7803         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
7804         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
7805         (avr-*-rtems*): Likewise.
7806         (bfin*-rtems*): Likewise.
7807         (moxie-*-rtems*): Likewise.
7808         (h8300-*-rtems*): Likewise.
7809         (i[34567]86-*-rtems*): Likewise.
7810         (lm32-*-rtems*): Likewise.
7811         (m32r-*-rtems*): Likewise.
7812         (m68k-*-rtems*): Likewise.
7813         (microblaze*-*-rtems*): Likewise.
7814         (mips*-*-rtems*): Likewise.
7815         (powerpc-*-rtems*): Likewise.
7816         (sh-*-rtems*): Likewise.
7817         (sparc-*-rtems*): Likewise.
7818         (sparc64-*-rtems*): Likewise.
7819         (v850-*-rtems*): Likewise.
7820         (m32c-*-rtems*): Likewise.
7821
7822 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
7823
7824         PR rtl-optimization/59511
7825         * ira.c (ira_init_register_move_cost): Use memory costs for some
7826         cases of register move cost calculations.
7827         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
7828         instead of BB frequency.
7829         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
7830         * lra-assigns.c (find_hard_regno_for): Ditto.
7831
7832 2014-01-15  Richard Biener  <rguenther@suse.de>
7833
7834         PR tree-optimization/59822
7835         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
7836         (vectorizable_load): Use it to hoist defs of uses of invariant
7837         loads out of the loop.
7838
7839 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
7840             Kugan Vivekanandarajah  <kuganv@linaro.org>
7841
7842         PR target/59695
7843         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
7844         truncation.
7845
7846 2014-01-15  Richard Biener  <rguenther@suse.de>
7847
7848         PR rtl-optimization/59802
7849         * lcm.c (compute_available): Use inverted postorder to seed
7850         the initial worklist.
7851
7852 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7853
7854         PR target/59803
7855         * config/s390/s390.c (s390_preferred_reload_class): Don't return
7856         ADDR_REGS for invalid symrefs in non-PIC code.
7857
7858 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
7859
7860         PR other/58712
7861         * builtins.c (determine_block_size): Initialize *probable_max_size
7862         even if len_rtx is CONST_INT.
7863
7864 2014-01-14  Andrew Pinski  <apinski@cavium.com>
7865
7866         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
7867         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
7868         (cortexa53_tunings): Likewise.
7869         (aarch64_sched_issue_rate): New function.
7870         (TARGET_SCHED_ISSUE_RATE): Define.
7871
7872 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7873
7874         * ira-costs.c (find_costs_and_classes): Add missed
7875         ira_init_register_move_cost_if_necessary.
7876
7877 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7878
7879         PR target/59787
7880         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
7881
7882 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
7883
7884         PR target/59794
7885         * config/i386/i386.c (type_natural_mode): Add a bool parameter
7886         to indicate if type is used for function return value.  Warn ABI
7887         change if the vector mode isn't available for function return value.
7888         (ix86_function_arg_advance): Pass false to type_natural_mode.
7889         (ix86_function_arg): Likewise.
7890         (ix86_gimplify_va_arg): Likewise.
7891         (function_arg_32): Don't warn ABI change.
7892         (ix86_function_value): Pass true to type_natural_mode.
7893         (ix86_return_in_memory): Likewise.
7894         (ix86_struct_value_rtx): Removed.
7895         (TARGET_STRUCT_VALUE_RTX): Likewise.
7896
7897 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7898
7899         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
7900         converting a conditional jump into a conditional return.
7901
7902 2014-01-14  Richard Biener  <rguenther@suse.de>
7903
7904         PR tree-optimization/58921
7905         PR tree-optimization/59006
7906         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
7907         hoisting invariant stmts.
7908         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
7909         invariant loads on the preheader edge if possible.
7910
7911 2014-01-14  Joey Ye  <joey.ye@arm.com>
7912
7913         * doc/plugin.texi (Building GCC plugins): Update to C++.
7914
7915 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
7916
7917         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
7918         (_mm_rcp28_round_ss): Ditto.
7919         (_mm_rsqrt28_round_sd): Ditto.
7920         (_mm_rsqrt28_round_ss): Ditto.
7921         (_mm_rcp28_sd): Ditto.
7922         (_mm_rcp28_ss): Ditto.
7923         (_mm_rsqrt28_sd): Ditto.
7924         (_mm_rsqrt28_ss): Ditto.
7925         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
7926         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
7927         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
7928         (IX86_BUILTIN_RCP28SD): Ditto.
7929         (IX86_BUILTIN_RCP28SS): Ditto.
7930         (IX86_BUILTIN_RSQRT28SD): Ditto.
7931         (IX86_BUILTIN_RSQRT28SS): Ditto.
7932         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
7933         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
7934         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
7935         (ix86_expand_special_args_builtin): Expand new FTYPE.
7936         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
7937         (srcp14<mode>): Make insn unary.
7938         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
7939         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
7940         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
7941         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
7942         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
7943         Fix rounding: make it SAE only.
7944         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7945         Ditto.
7946         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7947         Ditto.
7948         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
7949         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
7950         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
7951         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
7952         (round_saeonly_mask_scalar_operand4): Ditto.
7953         (round_saeonly_mask_scalar_op3): Ditto.
7954         (round_saeonly_mask_scalar_op4): Ditto.
7955
7956 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7957
7958         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7959         Implement -maltivec=be for vec_insert and vec_extract.
7960
7961 2014-01-10  DJ Delorie  <dj@redhat.com>
7962
7963         * config/msp430/msp430.md (call_internal): Don't allow memory
7964         references with SP as the base register.
7965         (call_value_internal): Likewise.
7966         * config/msp430/constraints.md (Yc): New.  For memory references
7967         that don't use SP as a base register.
7968
7969         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
7970         "an integer without a # prefix"
7971         * config/msp430/msp430.md (epilogue_helper): Use it.
7972
7973 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7974
7975         PR target/59617
7976         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
7977         AVX512F gather builtins.
7978         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
7979         on gather decls with INTEGER_TYPE masktype.
7980         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
7981         directly into the builtin rather than hoisting it before loop.
7982
7983         PR tree-optimization/59387
7984         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
7985         (scev_const_prop): If folded_casts and type has undefined overflow,
7986         use force_gimple_operand instead of force_gimple_operand_gsi and
7987         for each added stmt if it is assign with
7988         arith_code_with_undefined_signed_overflow, call
7989         rewrite_to_defined_overflow.
7990         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
7991         gimple-fold.h instead.
7992         (arith_code_with_undefined_signed_overflow,
7993         rewrite_to_defined_overflow): Moved to ...
7994         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
7995         rewrite_to_defined_overflow): ... here.  No longer static.
7996         Include gimplify-me.h.
7997         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
7998         rewrite_to_defined_overflow): New prototypes.
7999
8000 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8001
8002         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
8003
8004 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
8005
8006         * builtins.c (get_object_alignment_2): Minor tweak.
8007         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
8008
8009 2014-01-13  Christian Bruel  <christian.bruel@st.com>
8010
8011         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
8012         optimized non constant lengths.
8013
8014 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
8015
8016         PR libgomp/59194
8017         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
8018         load as __atomic_load_N if possible.
8019
8020 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
8021
8022         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
8023         target parameter.
8024         (rs6000_expand_builtin): Adjust call.
8025
8026 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
8027
8028         PR target/58115
8029         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
8030         * config/rs6000/rs6000.c: Include target-globals.h.
8031         (rs6000_set_current_function): Instead of doing target_reinit
8032         unconditionally, use save_target_globals_default_opts and
8033         restore_target_globals.
8034
8035         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
8036         FPSCR.
8037         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
8038         (rs6000_expand_builtin): Handle mffs and mtfsf.
8039         (rs6000_init_builtins): Define mffs and mtfsf.
8040         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
8041         (rs6000_mffs): New pattern.
8042         (rs6000_mtfsf): New pattern.
8043
8044 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
8045
8046         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
8047         Start narrowing with START.  Apply candidate-use pair
8048         and check overall cost in narrowing.
8049         (iv_ca_prune): Pass new argument.
8050
8051 2014-01-10  Jeff Law  <law@redhat.com>
8052
8053         PR middle-end/59743
8054         * ree.c (combine_reaching_defs): Ensure the defining statement
8055         occurs before the extension when optimizing extensions with
8056         different source and destination hard registers.
8057
8058 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
8059
8060         PR ipa/58585
8061         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
8062         vtables into the type inheritance graph.
8063
8064 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8065
8066         PR rtl-optimization/59754
8067         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
8068         modes in the REGNO != REGNO case.
8069
8070 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8071
8072         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
8073
8074 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8075
8076         PR tree-optimization/59745
8077         * tree-predcom.c (tree_predictive_commoning_loop): Call
8078         free_affine_expand_cache if giving up because components is NULL.
8079
8080         * target-globals.c (save_target_globals): Allocate < 4KB structs using
8081         GC in payload of target_globals struct instead of allocating them on
8082         the heap and the larger structs separately using GC.
8083         * target-globals.h (struct target_globals): Make regs, hard_regs,
8084         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
8085         of GTY((skip)) and change type to void *.
8086         (reset_target_globals): Cast loads from those fields to corresponding
8087         types.
8088
8089 2014-01-10  Steve Ellcey  <sellcey@mips.com>
8090
8091         PR plugins/59335
8092         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
8093         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
8094         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
8095
8096 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
8097
8098         PR target/59744
8099         * aarch64-modes.def (CC_Zmode): New flags mode.
8100         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
8101         represents an equality.
8102         (aarch64_get_condition_code): Handle CC_Zmode.
8103         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
8104
8105 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8106
8107         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
8108         extraction in good case.
8109
8110 2014-01-10  Richard Biener  <rguenther@suse.de>
8111
8112         PR tree-optimization/59374
8113         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
8114         checking after SLP discovery.  Mark stmts not participating
8115         in any SLP instance properly.
8116
8117 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8118
8119         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
8120         when handling a SET rtx.
8121
8122 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8123
8124         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
8125         (cortex-a57): Likewise.
8126         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
8127
8128 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8129
8130         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
8131         non-iwmmxt builtins.
8132
8133 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
8134
8135         PR ipa/58252
8136         PR ipa/59226
8137         * ipa-devirt.c record_target_from_binfo): Take as argument
8138         stack of binfos and lookup matching one for virtual inheritance.
8139         (possible_polymorphic_call_targets_1): Update.
8140
8141 2014-01-10  Huacai Chen  <chenhc@lemote.com>
8142
8143         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
8144         kernel strings for Loongson-2E/2F/3A.
8145
8146 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
8147
8148         PR middle-end/59670
8149         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
8150         is_gimple_call before calling gimple_call_internal_p.
8151
8152 2014-01-09  Steve Ellcey  <sellcey@mips.com>
8153
8154         * Makefile.in (TREE_FLOW_H): Remove.
8155         (TREE_SSA_H): Add file names from tree-flow.h.
8156         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
8157         * tree.h: Remove tree-flow.h reference.
8158         * hash-table.h: Remove tree-flow.h reference.
8159         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
8160         reference with tree-ssa-loop.h.
8161
8162 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8163
8164         * doc/invoke.texi: Add -maltivec={be,le} options, and document
8165         default element-order behavior for -maltivec.
8166         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
8167         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
8168         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
8169         when targeting big endian, at least for now.
8170         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
8171
8172 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8173
8174         PR middle-end/47735
8175         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
8176         var satisfies use_register_for_decl, just take into account type
8177         alignment, rather than decl alignment.
8178
8179         PR tree-optimization/59622
8180         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
8181         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
8182         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
8183         Don't devirtualize for inplace at all.  For targets.length () == 1,
8184         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
8185
8186 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
8187
8188         * config/i386/i386.md (cpu): Remove the unused btver1.
8189
8190 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
8191
8192         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
8193
8194 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8195
8196         PR target/58115
8197         * tree-core.h (struct target_globals): New forward declaration.
8198         (struct tree_target_option): Add globals field.
8199         * tree.h (TREE_TARGET_GLOBALS): Define.
8200         (prepare_target_option_nodes_for_pch): New prototype.
8201         * target-globals.h (struct target_globals): Define even if
8202         !SWITCHABLE_TARGET.
8203         * tree.c (prepare_target_option_node_for_pch,
8204         prepare_target_option_nodes_for_pch): New functions.
8205         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
8206         * config/i386/i386.c: Include target-globals.h.
8207         (ix86_set_current_function): Instead of doing target_reinit
8208         unconditionally, use save_target_globals_default_opts and
8209         restore_target_globals.
8210
8211 2014-01-09  Richard Biener  <rguenther@suse.de>
8212
8213         PR tree-optimization/59715
8214         * tree-cfg.h (split_critical_edges): Declare.
8215         * tree-cfg.c (split_critical_edges): Export.
8216         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
8217
8218 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
8219
8220         * cfgexpand.c (expand_stack_vars): Optionally disable
8221         asan stack protection.
8222         (expand_used_vars): Likewise.
8223         (partition_stack_vars): Likewise.
8224         * asan.c (asan_emit_stack_protection): Optionally disable
8225         after return stack usage.
8226         (instrument_derefs): Optionally disable memory access instrumentation.
8227         (instrument_builtin_call): Likewise.
8228         (instrument_strlen_call): Likewise.
8229         (asan_protect_global): Optionally disable global variables protection.
8230         * doc/invoke.texi: Added doc for new options.
8231         * params.def: Added new options.
8232         * params.h: Likewise.
8233
8234 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8235
8236         PR rtl-optimization/59724
8237         * ifcvt.c (cond_exec_process_if_block): Don't call
8238         flow_find_head_matching_sequence with 0 longest_match.
8239         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
8240         non-active insns if !stop_after.
8241         (try_head_merge_bb): Revert 2014-01-07 changes.
8242
8243 2014-01-08  Jeff Law  <law@redhat.com>
8244
8245         * ree.c (get_sub_rtx): New function, extracted from...
8246         (merge_def_and_ext): Here.
8247         (combine_reaching_defs): Use get_sub_rtx.
8248
8249 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
8250
8251         * cgraph.h (varpool_variable_node): Do not choke on null node.
8252
8253 2014-01-08  Catherine Moore  <clm@codesourcery.com>
8254
8255         * config/mips/mips.md (simple_return): Attempt to use JRC
8256         for microMIPS.
8257         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
8258
8259 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8260
8261         PR rtl-optimization/59137
8262         * reorg.c (steal_delay_list_from_target): Call update_block for
8263         elided insns.
8264         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
8265
8266 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8267
8268         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8269         two duplicate entries.
8270
8271 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8272
8273         Revert:
8274         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
8275
8276         * config/mips/mips.c (mips_truncated_op_cost): New function.
8277         (mips_rtx_costs): Adjust test for BADDU.
8278         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
8279
8280         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
8281
8282         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
8283         (*baddu_si): ...this new pattern.
8284
8285 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8286
8287         PR ipa/59722
8288         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
8289
8290 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8291
8292         PR middle-end/57748
8293         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
8294         inner_reference_p.
8295         (expand_expr, expand_normal): Adjust.
8296         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
8297         inner_reference_p. Use inner_reference_p to expand inner references.
8298         (store_expr): Adjust.
8299         * cfgexpand.c (expand_call_stmt): Adjust.
8300
8301 2014-01-08  Rong Xu  <xur@google.com>
8302
8303         * gcov-io.c (gcov_var): Move from gcov-io.h.
8304         (gcov_position): Ditto.
8305         (gcov_is_error): Ditto.
8306         (gcov_rewrite): Ditto.
8307         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
8308         only part to libgcc/libgcov.h.
8309
8310 2014-01-08  Marek Polacek  <polacek@redhat.com>
8311
8312         PR middle-end/59669
8313         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
8314
8315 2014-01-08  Marek Polacek  <polacek@redhat.com>
8316
8317         PR sanitizer/59667
8318         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
8319
8320 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8321
8322         PR rtl-optimization/59649
8323         * stor-layout.c (get_mode_bounds): For BImode return
8324         0 and STORE_FLAG_VALUE.
8325
8326 2014-01-08  Richard Biener  <rguenther@suse.de>
8327
8328         PR middle-end/59630
8329         * gimple.h (is_gimple_builtin_call): Remove.
8330         (gimple_builtin_call_types_compatible_p): New.
8331         (gimple_call_builtin_p): New overload.
8332         * gimple.c (is_gimple_builtin_call): Remove.
8333         (validate_call): Rename to ...
8334         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
8335         check return types.
8336         (validate_type): New static function.
8337         (gimple_call_builtin_p): New overload and adjust.
8338         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
8339         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
8340         (gimple_fold_stmt_to_constant_1): Likewise.
8341         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
8342
8343 2014-01-08  Richard Biener  <rguenther@suse.de>
8344
8345         PR middle-end/59471
8346         * gimplify.c (gimplify_expr): Gimplify register-register type
8347         VIEW_CONVERT_EXPRs to separate stmts.
8348
8349 2014-01-07  Jeff Law  <law@redhat.com>
8350
8351         PR middle-end/53623
8352         * ree.c (combine_set_extension): Handle case where source
8353         and destination registers in an extension insn are different.
8354         (combine_reaching_defs): Allow source and destination registers
8355         in extension to be different under limited circumstances.
8356         (add_removable_extension): Remove restriction that the
8357         source and destination registers in the extension are the same.
8358         (find_and_remove_re): Emit a copy from the extension's
8359         destination to its source after the defining insn if
8360         the source and destination registers are different.
8361
8362         PR middle-end/59285
8363         * ifcvt.c (merge_if_block): If we are merging a block with more than
8364         one successor with a block with no successors, remove any BARRIER
8365         after the second block.
8366
8367 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
8368
8369         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
8370
8371 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
8372
8373         PR target/59652
8374         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
8375         for 14-bit register offsets when INT14_OK_STRICT is false.
8376
8377 2014-01-07  Roland Stigge  <stigge@antcom.de>
8378             Michael Meissner  <meissner@linux.vnet.ibm.com>
8379
8380         PR 57386/target
8381         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
8382         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
8383
8384 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
8385
8386         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
8387         -mcpu.
8388
8389 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
8390
8391         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
8392         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
8393         rtx is const0_rtx or not.
8394
8395 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
8396
8397         PR target/58115
8398         * target-globals.c (save_target_globals): Remove this_fn_optab
8399         handling.
8400         * toplev.c: Include optabs.h.
8401         (target_reinit): Temporarily restore the global options if another
8402         set of options are in force.
8403
8404 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
8405
8406         PR rtl-optimization/58668
8407         * cfgcleanup.c (flow_find_cross_jump): Don't count
8408         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
8409         to determine what is counted.
8410         (flow_find_head_matching_sequence): Use active_insn_p to determine
8411         what is counted.
8412         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
8413         counting change.
8414         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
8415         determine what is counted.
8416
8417         PR tree-optimization/59643
8418         * tree-predcom.c (split_data_refs_to_components): If one dr is
8419         read and one write, determine_offset fails and the write isn't
8420         in the bad component, just put the read into the bad component.
8421
8422 2014-01-07  Mike Stump  <mikestump@comcast.net>
8423             Jakub Jelinek  <jakub@redhat.com>
8424
8425         PR pch/59436
8426         * tree-core.h (struct tree_optimization_option): Change optabs
8427         type from unsigned char * to void *.
8428         * optabs.c (init_tree_optimization_optabs): Adjust
8429         TREE_OPTIMIZATION_OPTABS initialization.
8430
8431 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
8432
8433         PR target/59644
8434         * config/i386/i386.h (struct machine_function): Add
8435         no_drap_save_restore field.
8436         * config/i386/i386.c (ix86_save_reg): Use
8437         !cfun->machine->no_drap_save_restore instead of
8438         crtl->stack_realign_needed.
8439         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
8440         this function clears frame_pointer_needed.  Set
8441         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
8442         and DRAP reg is needed.
8443
8444 2014-01-06  Marek Polacek  <polacek@redhat.com>
8445
8446         PR c/57773
8447         * doc/implement-c.texi: Mention that other integer types are
8448         permitted as bit-field types in strictly conforming mode.
8449
8450 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
8451
8452         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
8453         is newly allocated.
8454
8455 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
8456
8457         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
8458
8459 2014-01-06  Martin Jambor  <mjambor@suse.cz>
8460
8461         PR ipa/59008
8462         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
8463         to int.
8464         * ipa-prop.c (ipa_print_node_params): Fix indentation.
8465
8466 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
8467
8468         PR debug/59350
8469         PR debug/59510
8470         * var-tracking.c (add_stores): Preserve the value of the source even if
8471         we don't record the store.
8472
8473 2014-01-06  Terry Guo  <terry.guo@arm.com>
8474
8475         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
8476
8477 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
8478
8479         PR bootstrap/59541
8480         * config/darwin.c (darwin_function_section): Adjust return values to
8481         correspond to optimisation changes made in r206070.
8482
8483 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
8484
8485         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
8486         from prefetch_block tune setting.
8487         (nocona_cost): Correct size of prefetch block to 64.
8488
8489 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
8490
8491         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
8492         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
8493         used to save the static chain register in the computation of the offset
8494         from which the FP registers need to be restored.
8495
8496 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
8497
8498         PR tree-optimization/59519
8499         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
8500         ICE if get_current_def (current_new_name) is already non-NULL, as long
8501         as it is a phi result of some other phi in *new_exit_bb that has
8502         the same argument.
8503
8504         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
8505         or vmovdqu* for misaligned_operand.
8506         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
8507         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
8508         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
8509         aligned_mem for AVX512F masked aligned load and store builtins and for
8510         non-temporal moves.
8511
8512 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
8513
8514         PR tree-optimization/59651
8515         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
8516         Address range for negative step should be added by TYPE_SIZE_UNIT.
8517
8518 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
8519
8520         * config/m68k/m68k.c (handle_move_double): Handle pushes with
8521         overlapping registers also for registers other than the stack pointer.
8522
8523 2014-01-03  Marek Polacek  <polacek@redhat.com>
8524
8525         PR other/59661
8526         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
8527         __builtin_FILE.
8528
8529 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
8530
8531         PR target/59625
8532         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
8533         asm goto as jump.
8534
8535         * config/i386/i386.md (MODE_SIZE): New mode attribute.
8536         (push splitter): Use <P:MODE_SIZE> instead of
8537         GET_MODE_SIZE (<P:MODE>mode).
8538         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
8539         (mov -1, reg peephole2): Likewise.
8540         * config/i386/sse.md (*mov<mode>_internal,
8541         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
8542         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
8543         *<code><mode>3, *andnot<mode>3<mask_name>,
8544         <mask_codefor><code><mode>3<mask_name>): Likewise.
8545         * config/i386/subst.md (mask_mode512bit_condition,
8546         sd_mask_mode512bit_condition): Likewise.
8547
8548 2014-01-02  Xinliang David Li  <davidxl@google.com>
8549
8550         PR tree-optimization/59303
8551         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
8552         (dump_predicates): Better output format.
8553         (pred_equal_p): New function.
8554         (is_neq_relop_p): Ditto.
8555         (is_neq_zero_form_p): Ditto.
8556         (pred_expr_equal_p): Ditto.
8557         (pred_neg_p): Ditto.
8558         (simplify_pred): Ditto.
8559         (simplify_preds_2): Ditto.
8560         (simplify_preds_3): Ditto.
8561         (simplify_preds_4): Ditto.
8562         (simplify_preds): Ditto.
8563         (push_pred): Ditto.
8564         (push_to_worklist): Ditto.
8565         (get_pred_info_from_cmp): Ditto.
8566         (is_degenerated_phi): Ditto.
8567         (normalize_one_pred_1): Ditto.
8568         (normalize_one_pred): Ditto.
8569         (normalize_one_pred_chain): Ditto.
8570         (normalize_preds): Ditto.
8571         (normalize_cond_1): Remove function.
8572         (normalize_cond): Ditto.
8573         (is_gcond_subset_of): Ditto.
8574         (is_subset_of_any): Ditto.
8575         (is_or_set_subset_of): Ditto.
8576         (is_and_set_subset_of): Ditto.
8577         (is_norm_cond_subset_of): Ditto.
8578         (pred_chain_length_cmp): Ditto.
8579         (convert_control_dep_chain_into_preds): Type change.
8580         (find_predicates): Ditto.
8581         (find_def_preds): Ditto.
8582         (destroy_predicates_vecs): Ditto.
8583         (find_matching_predicates_in_rest_chains): Ditto.
8584         (use_pred_not_overlap_with_undef_path_pred): Ditto.
8585         (is_pred_expr_subset): Ditto.
8586         (is_pred_chain_subset_of): Ditto.
8587         (is_included_in): Ditto.
8588         (is_superset_of): Ditto.
8589
8590 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8591
8592         Update copyright years.
8593
8594 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8595
8596         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
8597         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
8598         config/arc/arc.md, config/arc/arc.opt,
8599         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
8600         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
8601         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
8602         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
8603         config/linux-protos.h, config/linux.c, config/winnt-c.c,
8604         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
8605         vtable-verify.c, vtable-verify.h: Use the standard form for the
8606         copyright notice.
8607
8608 2014-01-02  Tobias Burnus  <burnus@net-b.de>
8609
8610         * gcc.c (process_command): Update copyright notice dates.
8611         * gcov-dump.c: Ditto.
8612         * gcov.c: Ditto.
8613         * doc/cpp.texi: Bump @copying's copyright year.
8614         * doc/cppinternals.texi: Ditto.
8615         * doc/gcc.texi: Ditto.
8616         * doc/gccint.texi: Ditto.
8617         * doc/gcov.texi: Ditto.
8618         * doc/install.texi: Ditto.
8619         * doc/invoke.texi: Ditto.
8620
8621 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8622
8623         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
8624
8625 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
8626
8627         * config/i386/sse.md (*mov<mode>_internal): Guard
8628         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
8629
8630         PR rtl-optimization/59647
8631         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
8632         new_rtx into UNSIGNED_FLOAT rtxes.
8633 \f
8634 Copyright (C) 2014 Free Software Foundation, Inc.
8635
8636 Copying and distribution of this file, with or without modification,
8637 are permitted in any medium without royalty provided the copyright
8638 notice and this notice are preserved.