gcc/ChangeLog

   1 2013-10-18  Richard Biener  <rguenther@suse.de>
   2
   3         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
   4         or TYPE_UNSIGNED of integral types.
   5         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
   6         NULL_TREE for zero-precision integral types.
   7
   8 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
   9
  10         * config/aarch64/arm_neon.h
  11         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
  12
  13 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
  14
  15         * opth-gen.awk: Define target_flags_explicit.
  16
  17 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
  18
  19         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
  20         fields to the reg_addr array that describes the valid addressing
  21         mode for any register, general purpose registers, floating point
  22         registers, and Altivec registers.
  23         (FIRST_RELOAD_REG_CLASS): Likewise.
  24         (LAST_RELOAD_REG_CLASS): Likewise.
  25         (struct reload_reg_map_type): Likewise.
  26         (reload_reg_map_type): Likewise.
  27         (RELOAD_REG_VALID): Likewise.
  28         (RELOAD_REG_MULTIPLE): Likewise.
  29         (RELOAD_REG_INDEXED): Likewise.
  30         (RELOAD_REG_OFFSET): Likewise.
  31         (RELOAD_REG_PRE_INCDEC): Likewise.
  32         (RELOAD_REG_PRE_MODIFY): Likewise.
  33         (reg_addr): Likewise.
  34         (mode_supports_pre_incdec_p): New helper functions to say whether
  35         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
  36         (mode_supports_pre_modify_p): Likewise.
  37         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
  38         print the valid address mode bits for each mode.
  39         (rs6000_debug_print_mode): Likewise.
  40         (rs6000_debug_reg_global): Likewise.
  41         (rs6000_setup_reg_addr_masks): New function to set up the address
  42         mask bits for each type.
  43         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
  44         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
  45         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
  46         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
  47         PRE_MODIFY are supported.
  48         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
  49         registers, instead of {src,dest}_av_p.
  50         (rs6000_print_options_internal): Tweak the debug output slightly.
  51
  52 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
  53
  54         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
  55         isa attribute.
  56
  57 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
  58
  59         * tree-flow.h (struct omp_region): Move to omp-low.c.
  60         Remove omp_ prototypes and variables.
  61         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
  62         (copy_var_decl): Relocate prototype from tree-flow.h.
  63         * gimple.c (copy_var_decl): Relocate from omp-low.c.
  64         * tree.h: Move prototype to omp-low.h.
  65         * omp-low.h: New File.  Relocate prototypes here.
  66         * omp-low.c (struct omp_region): Make local here.
  67         (root_omp_region): Make static.
  68         (copy_var_decl) Move to gimple.c.
  69         (new_omp_region): Make static.
  70         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
  71         * tree-cfg.c: Include omp-low.h.
  72         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
  73         * gimplify.c: Include omp-low.h.
  74         * tree-parloops.c: Likewise.
  75
  76 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
  77
  78         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
  79         operands are in memory, prefer to force non-matched operand 1 to
  80         the register.
  81
  82 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
  83
  84         PR target/58673
  85         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
  86         restrict TImode addresses to single indirect registers if both
  87         -mquad-memory and -mvsx-timode are used.
  88         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
  89         we should emit load/store quad.  Remove using %y for quad memory
  90         addresses.
  91
  92         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
  93         constraints to allow load/store quad on machines where TImode is
  94         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
  95         for TImode to load integer constants.
  96
  97 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
  98
  99         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
 100
 101 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
 102
 103         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
 104         handling of STACK_REG.
 105
 106 2013-10-17  Richard Biener  <rguenther@suse.de>
 107
 108         PR tree-optimization/58143
 109         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
 110         New function.
 111         (rewrite_to_defined_overflow): Likewise.
 112         (move_computations_dom_walker::before_dom): Rewrite stmts
 113         with undefined signed overflow that are not always executed
 114         into unsigned arithmetic.
 115
 116 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
 117
 118         PR target/57756
 119         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
 120         explicit isa flag to be an options variable, instead of using
 121         global_options_set.  Remove define from rs6000.h.
 122         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
 123
 124         * config/rs6000/rs6000.c (rs6000_option_override_internal):
 125         Initialize rs6000_isa_flags_explicit.
 126         (rs6000_function_specific_save): Add gcc_options* parameter, so
 127         that the powerpc builds after the 2013-10-15 changes.
 128         (rs6000_function_specific_restore): Likewise.
 129
 130 2013-10-16  DJ Delorie  <dj@redhat.com>
 131
 132         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
 133         op is a REG before checking REGNO.
 134         (rl78_alloc_physical_registers): Verify pattern is a SET before
 135         checking SET_SRC.
 136
 137 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 138
 139         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
 140         endianness.
 141         (vec_unpacks_lo_v4sf): Likewise.
 142         (vec_unpacks_float_hi_v4si): Likewise.
 143         (vec_unpacks_float_lo_v4si): Likewise.
 144         (vec_unpacku_float_hi_v4si): Likewise.
 145         (vec_unpacku_float_lo_v4si): Likewise.
 146
 147 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
 148
 149         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
 150         (vsx_concat_v2sf): Likewise.
 151
 152 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
 153
 154         * config/aarch64/aarch64.md
 155         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
 156
 157 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
 158
 159         PR tree-optimization/58697
 160         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
 161         estimated_loop_iterations_int.
 162         (max_stmt_executions_int): Call get_max_loop_iterations_int.
 163         (get_max_loop_iterations_int): New.  HWINT version of
 164         get_max_loop_iterations.
 165         * cfgloop.h: Add prototypes.
 166         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
 167         * loop-unroll.c (decide_peel_once_rolling): Call
 168         get_estimated_loop_iterations_int.
 169         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
 170         * tree-ssa-loop-niter.h: Tweak prototypes.
 171
 172 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
 173
 174         * gengtype-parse.c (struct_field_seq): Ignore access-control
 175         keywords ("public:" etc).
 176
 177 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
 178
 179         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
 180         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
 181
 182 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
 183
 184         * config/arm/arm.opt (mlra): New option.
 185         * config/arm/arm.c (arm_lra_p): New function.
 186         (TARGET_LRA_P): Define.
 187
 188 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
 189
 190         * tree-core.h (tree_code_name): Remove.
 191         * tree.h (get_tree_code_name): New prototype.
 192         * tree.c (tree_code_name): Make static.
 193         (get_tree_code_name): New function.
 194         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
 195         tree_class_check_failed, tree_range_check_failed,
 196         tree_not_class_check_failed, omp_clause_check_failed,
 197         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
 198         wrapper get_tree_code_name instead of calling tree_code_name directly.
 199         * tree-vrp.c (dump_asserts_for): Likewise.
 200         * tree-dump.c (dequeue_and_dump): Likewise.
 201         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
 202         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
 203         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
 204         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
 205         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
 206         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
 207         dump_gimple_omp_for): Likewise.
 208         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
 209         * tree-ssa-pre.c (print_pre_expr): Likewise.
 210         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
 211         * print-tree.c (print_node_brief, print_node): Likewise.
 212         * gimple.c (gimple_check_failed): Likewise.
 213         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
 214         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
 215         * config/mep/mep.c (mep_validate_vliw): Likewise.
 216         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
 217         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
 218
 219 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
 220
 221         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
 222         for AMD bdver3.
 223
 224 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
 225
 226         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
 227         (MULTILIB_MATCHES): Add multilib for -march=v8.
 228
 229 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
 230
 231         PR target/57756
 232         * optc-save-gen.awk: Add extra parameter to the save and restore
 233         target calls.
 234         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
 235         * tree.c (build_optimization_node): New parameter.  Add extra parameter
 236         to call to cl_optimization_save.
 237         (build_target_option_node): New parameter. Add extra parameter
 238         to call to cl_target_option_save.
 239         * tree.h (build_optimization_node): New parameter.
 240         (build_target_option_node): New parameter.
 241         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
 242         build_optimization_node and build_target_option_node.
 243         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
 244         (handle_pragma_push_options): Ditto.
 245         * toplev.c (process_options): Ditto.
 246         * opts.c (init_options_struct): Check for opts_set non-null.
 247         * target.def (target_option.save): New parameter.
 248         (target_option.restore): New parameter.
 249         * tm.texi: Generate.
 250         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
 251         (ix86_pragma_target_parse): Ditto.
 252         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
 253         parameters.
 254         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
 255         to  build_optimization_node and build_target_option_node.
 256         (rs6000_valid_attribute_p): Ditto.
 257         (rs6000_pragma_target_parse): Ditto.
 258         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
 259         data.
 260         * config/i386/i386.h:
 261         TARGET_64BIT_P: New Macro
 262         TARGET_MMX_P: New Macro.
 263         TARGET_3DNOW_P: New Macro.
 264         TARGET_3DNOW_A_P: New Macro.
 265         TARGET_SSE_P: New Macro.
 266         TARGET_SSE2_P: New Macro.
 267         TARGET_SSE3_P: New Macro.
 268         TARGET_SSSE3_P: New Macro.
 269         TARGET_SSE4_1_P: New Macro.
 270         TARGET_SSE4_2_P: New Macro.
 271         TARGET_AVX_P: New Macro.
 272         TARGET_AVX2_P: New Macro.
 273         TARGET_AVX512F_P: New Macro.
 274         TARGET_AVX512PF_P: New Macro.
 275         TARGET_AVX512ER_P: New Macro.
 276         TARGET_AVX512CD_P: New Macro.
 277         TARGET_FMA_P: New Macro.
 278         TARGET_SSE4A_P: New Macro.
 279         TARGET_FMA4_P: New Macro.
 280         TARGET_XOP_P: New Macro.
 281         TARGET_LWP_P: New Macro.
 282         TARGET_ABM_P: New Macro.
 283         TARGET_BMI_P: New Macro.
 284         TARGET_BMI2_P: New Macro.
 285         TARGET_LZCNT_P: New Macro.
 286         TARGET_TBM_P: New Macro.
 287         TARGET_POPCNT_P: New Macro.
 288         TARGET_SAHF_P: New Macro.
 289         TARGET_MOVBE_P: New Macro.
 290         TARGET_CRC32_P: New Macro.
 291         TARGET_AES_P: New Macro.
 292         TARGET_PCLMUL_P: New Macro.
 293         TARGET_CMPXCHG16B_P: New Macro.
 294         TARGET_FSGSBASE_P: New Macro.
 295         TARGET_RDRND_P: New Macro.
 296         TARGET_F16C_P: New Macro.
 297         TARGET_RTM_P: New Macro.
 298         TARGET_HLE_P: New Macro.
 299         TARGET_RDSEED_P: New Macro.
 300         TARGET_PRFCHW_P: New Macro.
 301         TARGET_ADX_P: New Macro.
 302         TARGET_FXSR_P: New Macro.
 303         TARGET_XSAVE_P: New Macro.
 304         TARGET_XSAVEOPT_P: New Macro.
 305         TARGET_LP64_P: New Macro.
 306         TARGET_X32_P: New Macro.
 307         TARGET_FPMATH_DEFAULT_P: New Macro.
 308         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
 309         * config/i386/i386.c (ix86_option_override_internal): New parameters.
 310         opts and opts_set.
 311         Change ix86_tune_string to access opts->x_ix86_tune_string.
 312         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
 313         Change ix86_arch_string to access opts->x_ix86_arch_string.
 314         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
 315         Change ix86_pmode to access opts->x_ix86_pmode.
 316         Change ix86_abi to access opts->x_ix86_abi.
 317         Change ix86_cmodel to access opts->x_ix86_cmodel.
 318         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
 319         Change ix86_isa_flags_explicit to access
 320         opts->x_ix86_isa_flags_explicit.
 321         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
 322         Change ix86_regparm to access opts->x_ix86_regparm.
 323         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
 324         Change ix86_preferred_stack_boundary_arg to access
 325         opts->x_ix86_preferred_stack_boundary_arg.
 326         Change ix86_force_align_arg_pointer to access
 327         opts->x_ix86_force_align_arg_pointer.
 328         Change ix86_incoming_stack_boundar_arg to access
 329         opts->x_ix86_incoming_stack_boundar_arg.
 330         Change ix86_fpmath to access opts->x_ix86_fpmath.
 331         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
 332         Change ix86_recip_name to access opts->x_ix86_recip_name.
 333         Change ix86_stack_protector_guard to access
 334         opts->x_ix86_stack_protector_guard.
 335         Change ix86_tune_memcpy_strategy to access
 336         opts->x_ix86_tune_memcpy_strategy.
 337         Change ix86_tune_memset_strategy to access
 338         opts->x_ix86_tune_memset_strategy.
 339         Change global_options to access opts.
 340         Change global_options_set to access opts_set.
 341         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
 342         Change TARGET_MMX to TARGET_MMX_P (opts->...).
 343         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
 344         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
 345         Change TARGET_SSE to TARGET_SSE_P (opts->...).
 346         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
 347         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
 348         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
 349         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
 350         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
 351         Change TARGET_AVX to TARGET_AVX_P (opts->...).
 352         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
 353         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
 354         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
 355         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
 356         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
 357         Change TARGET_FMA to TARGET_FMA_P (opts->...).
 358         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
 359         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
 360         Change TARGET_XOP to TARGET_XOP_P (opts->...).
 361         Change TARGET_LWP to TARGET_LWP_P (opts->...).
 362         Change TARGET_ABM to TARGET_ABM_P (opts->...).
 363         Change TARGET_BMI to TARGET_BMI_P (opts->...).
 364         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
 365         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
 366         Change TARGET_TBM to TARGET_TBM_P (opts->...).
 367         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
 368         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
 369         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
 370         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
 371         Change TARGET_AES to TARGET_AES_P (opts->...).
 372         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
 373         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
 374         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
 375         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
 376         Change TARGET_F16C to TARGET_F16C_P (opts->...).
 377         Change TARGET_RTM to TARGET_RTM_P (opts->...).
 378         Change TARGET_HLE to TARGET_HLE_P (opts->...).
 379         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
 380         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
 381         Change TARGET_ADX to TARGET_ADX_P (opts->...).
 382         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
 383         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
 384         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
 385         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
 386         Change TARGET_X32 to TARGET_X32_P (opts->...).
 387         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
 388         Change TARGET_FLOAT_RETURNS_IN_80387 to
 389         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
 390         (ix86_function_specific_save): New parameter. Use opts-> fields
 391         to replace global fields.
 392         (ix86_function_specific_restore): Ditto.
 393         (ix86_valid_target_attribute_inner_p): New parameters.
 394         Fix recursive call.
 395         Fix call to ix86_handle_option and set_option.
 396         (ix86_valid_target_attribute_tree): New parameters.
 397         Change global_options to access opts.
 398         Change global_options_set to access opts_set.
 399         Fix call to ix86_valid_target_attribute_inner_p.
 400         Change ix86_tune_string to access opts->x_ix86_tune_string.
 401         Change ix86_arch_string to access opts->x_ix86_arch_string.
 402         Change ix86_fpmath to access opts->x_ix86_fpmath
 403         Fix call to ix86_option_override_internal.
 404         Fix call to ix86_add_new_builtins.
 405         Fix calls to build_optimization_node and build_target_option_node.
 406         (ix86_valid_target_attribute_p): Remove access to global_options.
 407         Use new gcc_options structure func_options.
 408         Fix call to ix86_valid_target_attribute_tree.
 409         Fix call to  build_optimization_node.
 410         (get_builtin_code_for_version): Fix call to
 411         ix86_valid_target_attribute_tree.
 412
 413 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
 414
 415         * Makefile.in (PICFLAG): New.
 416         (enable_host_shared): New.
 417         (INTERNAL_CFLAGS): Use PICFLAG.
 418         (LIBIBERTY): Use pic build of libiberty.a if configured with
 419         --enable-host-shared.
 420         * configure.ac: Add --enable-host-shared, setting up new
 421         PICFLAG variable.
 422         * configure: Regenerate.
 423         * doc/install.texi (--enable-shared): Add note contrasting it with ...
 424         (--enable-host-shared): New option.
 425
 426 2013-10-15  Richard Biener  <rguenther@suse.de>
 427
 428         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
 429         for built-in functions.
 430
 431 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
 432
 433         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
 434         (optimize_range_tests_1): New function,
 435         extracted from optimize_range_tests.
 436         (optimize_range_tests_xor): Similarly.
 437         (optimize_range_tests_diff): New function.
 438         (optimize_range_tests): Use optimize_range_tests_1.
 439
 440 2013-10-15  Cong Hou  <congh@google.com>
 441
 442         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
 443         requirement of the reduction pattern so that one operand of the
 444         reduction operation can come from outside of the loop.
 445
 446 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 447
 448         * config/arm/neon-schedgen.ml: Remove.
 449         * config/arm/cortex-a9-neon.md: Remove comment regarding
 450         neon-schedgen.ml.
 451
 452 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 453
 454         * config/arm/types: Remove old neon types.
 455
 456 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 457
 458         * config/arm/cortex-a7.md
 459         (cortex_a7_neon_type): New.
 460         (cortex_a7_neon_mul): Update for new types.
 461         (cortex_a7_neon_mla): Likewise.
 462         (cortex_a7_neon): Likewise.
 463
 464 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 465
 466         * config/arm/cortex-a15-neon.md
 467         (cortex_a15_neon_type): New,
 468
 469         (cortex_a15_neon_int_1): Remove.
 470         (cortex_a15_neon_int_2): Likewise.
 471         (cortex_a15_neon_int_3): Likewise.
 472         (cortex_a15_neon_int_4): Likewise.
 473         (cortex_a15_neon_int_5): Likewise.
 474         (cortex_a15_neon_vqneg_vqabs): Likewise.
 475         (cortex_a15_neon_vmov): Likewise.
 476         (cortex_a15_neon_vaba): Likewise.
 477         (cortex_a15_neon_vaba_qqq): Likewise.
 478         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 479         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 480         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 481         Likewise.
 482         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 483         (cortex_a15_neon_mla_qqq_8_16): Likewise.
 484         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
 485         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 486         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 487         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
 488         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 489         (cortex_a15_neon_shift_1): Likewise.
 490         (cortex_a15_neon_shift_2): Likewise.
 491         (cortex_a15_neon_shift_3): Likewise.
 492         (cortex_a15_neon_vshl_ddd): Likewise.
 493         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 494         (cortex_a15_neon_vsra_vrsra): Likewise.
 495         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
 496         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
 497         (cortex_a15_neon_bp_3cycle): Likewise.
 498         (cortex_a15_neon_ldm_2): Likewise.
 499         (cortex_a15_neon_stm_2): Likewise.
 500         (cortex_a15_neon_mcr): Likewise.
 501         (cortex_a15_neon_mrc): Likewise.
 502         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
 503         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
 504         (cortex_a15_neon_fp_vmul_ddd): Likewise.
 505         (cortex_a15_neon_fp_vmul_qqd): Likewise.
 506         (cortex_a15_neon_fp_vmla_ddd): Likewise.
 507         (cortex_a15_neon_fp_vmla_qqq): Likewise.
 508         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
 509         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
 510         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 511         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 512         (cortex_a15_neon_bp_simple): Likewise.
 513         (cortex_a15_neon_bp_2cycle): Likewise.
 514         (cortex_a15_neon_bp_3cycle): Likewise.
 515         (cortex_a15_neon_vld1_1_2_regs): Likewise.
 516         (cortex_a15_neon_vld1_3_4_regs): Likewise.
 517         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 518         (cortex_a15_neon_vld2_4_regs): Likewise.
 519         (cortex_a15_neon_vld3_vld4): Likewise.
 520         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 521         (cortex_a15_neon_vst1_3_4_regs): Likewise.
 522         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
 523         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
 524         (cortex_a15_neon_vst3_vst4): Rename to...
 525         (cortex_a15_neon_vst4): This, update for new attributes.
 526         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
 527         (cortex_a15_neon_vld3_vld4_lane): Likewise.
 528         (cortex_a15_neon_vst1_vst2_lane): Likewise.
 529         (cortex_a15_neon_vst3_vst4_lane): Likewise.
 530         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
 531         (cortex_a15_neon_ldm_2): Likewise.
 532         (cortex_a15_neon_stm_2): Likewise.
 533         (cortex_a15_neon_mcr): Likewise.
 534         (cortex_a15_neon_mcr_2_mcrr): Likewise.
 535         (cortex_a15_neon_mrc): Likewise.
 536         (cortex_a15_neon_mrrc): Likewise.
 537
 538         (cortex_a15_neon_abd): New.
 539         (cortex_a15_neon_abd_q): Likewise.
 540         (cortex_a15_neon_aba): Likewise.
 541         (cortex_a15_neon_aba_q): Likewise.
 542         (cortex_a15_neon_acc): Likewise.
 543         (cortex_a15_neon_acc_q): Likewise.
 544         (cortex_a15_neon_arith_basic): Likewise.
 545         (cortex_a15_neon_arith_complex): Likewise.
 546         (cortex_a15_neon_multiply): Likewise.
 547         (cortex_a15_neon_multiply_q): Likewise.
 548         (cortex_a15_neon_mla): Likewise.
 549         (cortex_a15_neon_mla_q): Likewise.
 550         (cortex_a15_neon_sat_mla_long): Likewise.
 551         (cortex_a15_neon_shift_acc): Likewise.
 552         (cortex_a15_neon_shift_imm_basic): Likewise.
 553         (cortex_a15_neon_shift_imm_complex): Likewise.
 554         (cortex_a15_neon_shift_reg_basic): Likewise.
 555         (cortex_a15_neon_shift_reg_basic_q): Likewise.
 556         (cortex_a15_neon_shift_reg_complex): Likewise.
 557         (cortex_a15_neon_shift_reg_complex_q): Likewise.
 558         (cortex_a15_neon_fp_negabs): Likewise
 559         (cortex_a15_neon_fp_arith): Likewise
 560         (cortex_a15_neon_fp_arith_q): Likewise
 561         (cortex_a15_neon_fp_cvt_int): Likewise
 562         (cortex_a15_neon_fp_cvt_int_q): Likewise
 563         (cortex_a15_neon_fp_cvt_16): Likewise
 564         (cortex_a15_neon_fp_mul): Likewise
 565         (cortex_a15_neon_fp_mul_q): Likewise
 566         (cortex_a15_neon_fp_mla): Likewise
 567         (cortex_a15_neon_fp_mla_q): Likewise
 568         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
 569         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
 570         (cortex_a15_neon_bitops): Likewise.
 571         (cortex_a15_neon_bitops_q): Likewise.
 572         (cortex_a15_neon_from_gp): Likewise.
 573         (cortex_a15_neon_from_gp_q): Likewise.
 574         (cortex_a15_neon_tbl3_tbl4): Likewise.
 575         (cortex_a15_neon_zip_q): Likewise.
 576         (cortex_a15_neon_to_gp): Likewise.
 577         (cortex_a15_neon_load_a): Likewise.
 578         (cortex_a15_neon_load_b): Likewise.
 579         (cortex_a15_neon_load_c): Likewise.
 580         (cortex_a15_neon_load_d): Likewise.
 581         (cortex_a15_neon_load_e): Likewise.
 582         (cortex_a15_neon_load_f): Likewise.
 583         (cortex_a15_neon_store_a): Likewise.
 584         (cortex_a15_neon_store_b): Likewise.
 585         (cortex_a15_neon_store_c): Likewise.
 586         (cortex_a15_neon_store_d): Likewise.
 587         (cortex_a15_neon_store_e): Likewise.
 588         (cortex_a15_neon_store_f): Likewise.
 589         (cortex_a15_neon_store_g): Likewise.
 590         (cortex_a15_neon_store_h): Likewise.
 591         (cortex_a15_vfp_to_from_gp): Likewise.
 592
 593 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 594
 595         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
 596
 597         (cortex_a9_neon_vshl_ddd): Remove.
 598         (cortex_a9_neon_vst3_vst4): Likewise.
 599         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
 600
 601         (cortex_a9_neon_bit_ops_q): New.
 602
 603         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
 604         (cortex_a9_neon_int_2): Likewise.
 605         (cortex_a9_neon_int_3): Likewise.
 606         (cortex_a9_neon_int_4): Likewise.
 607         (cortex_a9_neon_int_5): Likewise.
 608         (cortex_a9_neon_vqneg_vqabs): Likewise.
 609         (cortex_a9_neon_vmov): Likewise.
 610         (cortex_a9_neon_vaba): Likewise.
 611         (cortex_a9_neon_vaba_qqq): Likewise.
 612         (cortex_a9_neon_shift_1): Likewise.
 613         (cortex_a9_neon_shift_2): Likewise.
 614         (cortex_a9_neon_shift_3): Likewise.
 615         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 616         (cortex_a9_neon_vsra_vrsra): Likewise.
 617         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 618         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 619         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 620         Likewise.
 621         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 622         (cortex_a9_neon_mla_qqq_8_16): Likewise.
 623         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
 624         Likewise.
 625         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 626         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 627         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
 628         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 629         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
 630         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
 631         (cortex_a9_neon_fp_vsum): Likewise.
 632         (cortex_a9_neon_fp_vmul_ddd): Likewise.
 633         (cortex_a9_neon_fp_vmul_qqd): Likewise.
 634         (cortex_a9_neon_fp_vmla_ddd): Likewise.
 635         (cortex_a9_neon_fp_vmla_qqq): Likewise.
 636         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
 637         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
 638         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 639         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 640         (cortex_a9_neon_bp_simple): Likewise.
 641         (cortex_a9_neon_bp_2cycle): Likewise.
 642         (cortex_a9_neon_bp_3cycle): Likewise.
 643         (cortex_a9_neon_ldr): Likewise.
 644         (cortex_a9_neon_str): Likewise.
 645         (cortex_a9_neon_vld1_1_2_regs): Likewise.
 646         (cortex_a9_neon_vld1_3_4_regs): Likewise.
 647         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 648         (cortex_a9_neon_vld2_4_regs): Likewise.
 649         (cortex_a9_neon_vld3_vld4): Likewise.
 650         (cortex_a9_neon_vld1_vld2_lane): Likewise.
 651         (cortex_a9_neon_vld3_vld4_lane): Likewise.
 652         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
 653         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 654         (cortex_a9_neon_vst1_3_4_regs): Likewise.
 655         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
 656         (cortex_a9_neon_vst1_vst2_lane): Likewise.
 657         (cortex_a9_neon_vst3_vst4_lane): Likewise.
 658         (cortex_a9_neon_mcr): Likewise.
 659         (cortex_a9_neon_mcr_2_mcrr): Likewise.
 660         (cortex_a9_neon_mrc): Likewise.
 661         (cortex_a9_neon_mrrc): Likewise.
 662
 663 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 664
 665         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
 666
 667         (cortex_a8_neon_vshl_ddd): Remove.
 668         (cortex_a8_neon_vst3_vst4): Likewise.
 669         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
 670
 671         (cortex_a8_neon_bit_ops_q): New.
 672
 673         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
 674         (cortex_a8_neon_int_2): Likewise..
 675         (cortex_a8_neon_int_3): Likewise.
 676         (cortex_a8_neon_int_5): Likewise.
 677         (cortex_a8_neon_vqneg_vqabs): Likewise.
 678         (cortex_a8_neon_int_4): Likewise.
 679         (cortex_a8_neon_vaba): Likewise.
 680         (cortex_a8_neon_vaba_qqq): Likewise.
 681         (cortex_a8_neon_shift_1): Likewise.
 682         (cortex_a8_neon_shift_2): Likewise.
 683         (cortex_a8_neon_shift_3): Likewise.
 684         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
 685         (cortex_a8_neon_vsra_vrsra): Likewise.
 686         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 687         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
 688         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
 689         Likewise.
 690         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
 691         (cortex_a8_neon_mla_qqq_8_16): Likewise.
 692         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
 693         Likewise.
 694         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
 695         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
 696         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
 697         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
 698         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
 699         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
 700         (cortex_a8_neon_fp_vsum): Likewise.
 701         (cortex_a8_neon_fp_vmul_ddd): Likewise.
 702         (cortex_a8_neon_fp_vmul_qqd): Likewise.
 703         (cortex_a8_neon_fp_vmla_ddd): Likewise.
 704         (cortex_a8_neon_fp_vmla_qqq): Likewise.
 705         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
 706         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
 707         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
 708         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
 709         (cortex_a8_neon_bp_simple): Likewise.
 710         (cortex_a8_neon_bp_2cycle): Likewise.
 711         (cortex_a8_neon_bp_3cycle): Likewise.
 712         (cortex_a8_neon_ldr): Likewise.
 713         (cortex_a8_neon_str): Likewise.
 714         (cortex_a8_neon_vld1_1_2_regs): Likewise.
 715         (cortex_a8_neon_vld1_3_4_regs): Likewise.
 716         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
 717         (cortex_a8_neon_vld2_4_regs): Likewise.
 718         (cortex_a8_neon_vld3_vld4): Likewise.
 719         (cortex_a8_neon_vld1_vld2_lane): Likewise.
 720         (cortex_a8_neon_vld3_vld4_lane): Likewise.
 721         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
 722         (cortex_a8_neon_vst1_3_4_regs): Likewise.
 723         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
 724         (cortex_a8_neon_vst1_vst2_lane): Likewise.
 725         (cortex_a8_neon_vst3_vst4_lane): Likewise.
 726         (cortex_a8_neon_mcr): Likewise.
 727         (cortex_a8_neon_mcr_2_mcrr): Likewise.
 728         (cortex_a8_neon_mrc): Likewise.
 729         (cortex_a8_neon_mrrc): Likewise.
 730
 731 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 732
 733         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
 734         (fp): New.
 735         * config/aarch64/aarch64-simd.md (neon_type): Remove.
 736         (aarch64_simd_dup<mode>): Add "type" attribute.
 737         (aarch64_dup_lane<mode>): Likewise.
 738         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
 739         (*aarch64_simd_mov<mode>): Likewise.
 740         (aarch64_simd_mov_from_<mode>low): Likewise.
 741         (aarch64_simd_mov_from_<mode>high): Likewise.
 742         (orn<mode>3): Likewise.
 743         (bic<mode>3): Likewise.
 744         (add<mode>3): Likewise.
 745         (sub<mode>3): Likewise.
 746         (mul<mode>3): Likewise.
 747         (*aarch64_mul3_elt<mode>): Likewise.
 748         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
 749         (*aarch64_mul3_elt_to_128df): Likewise.
 750         (*aarch64_mul3_elt_to_64v2df): Likewise.
 751         (neg<mode>2): Likewise.
 752         (abs<mode>2): Likewise.
 753         (abd<mode>_3): Likewise.
 754         (aba<mode>_3): Likewise.
 755         (fabd<mode>_3): Likewise.
 756         (*fabd_scalar<mode>3): Likewise.
 757         (and<mode>3): Likewise.
 758         (ior<mode>3): Likewise.
 759         (xor<mode>3): Likewise.
 760         (one_cmpl<mode>2): Likewise.
 761         (aarch64_simd_vec_set<mode>): Likewise.
 762         (aarch64_simd_lshr<mode>): Likewise.
 763         (aarch64_simd_ashr<mode>): Likewise.
 764         (aarch64_simd_imm_shl<mode>): Likewise.
 765         (aarch64_simd_reg_sshl<mode): Likewise.
 766         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
 767         (aarch64_simd_reg_shl<mode>_signed): Likewise.
 768         (aarch64_simd_vec_setv2di): Likewise.
 769         (aarch64_simd_vec_set<mode>): Likewise.
 770         (aarch64_mla<mode>): Likewise.
 771         (*aarch64_mla_elt<mode>): Likewise.
 772         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
 773         (aarch64_mls<mode>): Likewise.
 774         (*aarch64_mls_elt<mode>): Likewise.
 775         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
 776         (<su><maxmin><mode>3): Likewise.
 777         (move_lo_quad_<mode>): Likewise.
 778         (aarch64_simd_move_hi_quad_<mode>): Likewise.
 779         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
 780         (vec_pack_trunc_<mode>): Likewise.
 781         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
 782         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
 783         (*aarch64_<su>mlal_lo<mode>): Likewise.
 784         (*aarch64_<su>mlal_hi<mode>): Likewise.
 785         (*aarch64_<su>mlsl_lo<mode>): Likewise.
 786         (*aarch64_<su>mlsl_hi<mode>): Likewise.
 787         (*aarch64_<su>mlal<mode>): Likewise.
 788         (*aarch64_<su>mlsl<mode>): Likewise.
 789         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
 790         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
 791         (add<mode>3): Likewise.
 792         (sub<mode>3): Likewise.
 793         (mul<mode>3): Likewise.
 794         (div<mode>3): Likewise.
 795         (neg<mode>2): Likewise.
 796         (abs<mode>2): Likewise.
 797         (fma<mode>4): Likewise.
 798         (*aarch64_fma4_elt<mode>): Likewise.
 799         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
 800         (*aarch64_fma4_elt_to_128df): Likewise.
 801         (*aarch64_fma4_elt_to_64v2df): Likewise.
 802         (fnma<mode>4): Likewise.
 803         (*aarch64_fnma4_elt<mode>): Likewise.
 804         (*aarch64_fnma4_elt_<vswap_width_name><mode>
 805         (*aarch64_fnma4_elt_to_128df): Likewise.
 806         (*aarch64_fnma4_elt_to_64v2df): Likewise.
 807         (<frint_pattern><mode>2): Likewise.
 808         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
 809         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
 810         (vec_unpacks_lo_v4sf): Likewise.
 811         (aarch64_float_extend_lo_v2df): Likewise.
 812         (vec_unpacks_hi_v4sf): Likewise.
 813         (aarch64_float_truncate_lo_v2sf): Likewise.
 814         (aarch64_float_truncate_hi_v4sf): Likewise.
 815         (aarch64_vmls<mode>): Likewise.
 816         (<su><maxmin><mode>3): Likewise.
 817         (<maxmin_uns><mode>3): Likewise.
 818         (reduc_<sur>plus_<mode>): Likewise.
 819         (reduc_<sur>plus_v2di): Likewise.
 820         (reduc_<sur>plus_v2si): Likewise.
 821         (reduc_<sur>plus_<mode>): Likewise.
 822         (aarch64_addpv4sf): Likewise.
 823         (clz<mode>2): Likewise.
 824         (reduc_<maxmin_uns>_<mode>): Likewise.
 825         (reduc_<maxmin_uns>_v2di): Likewise.
 826         (reduc_<maxmin_uns>_v2si): Likewise.
 827         (reduc_<maxmin_uns>_<mode>): Likewise.
 828         (reduc_<maxmin_uns>_v4sf): Likewise.
 829         (aarch64_simd_bsl<mode>_internal): Likewise.
 830         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
 831         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
 832         (aarch64_get_lane<mode>): Likewise.
 833         (*aarch64_combinez<mode>): Likewise.
 834         (aarch64_combine<mode>): Likewise.
 835         (aarch64_simd_combine<mode>): Likewise.
 836         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
 837         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
 838         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
 839         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
 840         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
 841         (aarch64_<sur>h<addsub><mode>): Likewise.
 842         (aarch64_<sur><addsub>hn<mode>): Likewise.
 843         (aarch64_<sur><addsub>hn2<mode>): Likewise.
 844         (aarch64_pmul<mode>): Likewise.
 845         (aarch64_<su_optab><optab><mode>): Likewise.
 846         (aarch64_<sur>qadd<mode>): Likewise.
 847         (aarch64_sqmovun<mode>): Likewise.
 848         (aarch64_<sur>qmovn<mode>): Likewise.
 849         (aarch64_s<optab><mode>): Likewise.
 850         (aarch64_sq<r>dmulh<mode>): Likewise.
 851         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
 852         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
 853         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
 854         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
 855         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 856         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
 857         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
 858         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
 859         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
 860         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
 861         (aarch64_sqdmull<mode>): Likewise.
 862         (aarch64_sqdmull_lane<mode>_internal): Likewise.
 863         (aarch64_sqdmull_n<mode>): Likewise.
 864         (aarch64_sqdmull2<mode>_internal): Likewise.
 865         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
 866         (aarch64_sqdmull2_n<mode>_internal): Likewise.
 867         (aarch64_<sur>shl<mode>): Likewise.
 868         (aarch64_<sur>q<r>shl<mode>
 869         (aarch64_<sur>shll_n<mode>): Likewise.
 870         (aarch64_<sur>shll2_n<mode>): Likewise.
 871         (aarch64_<sur>shr_n<mode>): Likewise.
 872         (aarch64_<sur>sra_n<mode>): Likewise.
 873         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
 874         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
 875         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
 876         (aarch64_cm<optab><mode>): Likewise.
 877         (aarch64_cm<optab>di): Likewise.
 878         (aarch64_cm<optab><mode>): Likewise.
 879         (aarch64_cm<optab>di): Likewise.
 880         (aarch64_cmtst<mode>): Likewise.
 881         (aarch64_cmtstdi): Likewise.
 882         (aarch64_cm<optab><mode>): Likewise.
 883         (*aarch64_fac<optab><mode>): Likewise.
 884         (aarch64_addp<mode>): Likewise.
 885         (aarch64_addpdi): Likewise.
 886         (sqrt<mode>2): Likewise.
 887         (vec_load_lanesoi<mode>): Likewise.
 888         (vec_store_lanesoi<mode>): Likewise.
 889         (vec_load_lanesci<mode>): Likewise.
 890         (vec_store_lanesci<mode>): Likewise.
 891         (vec_load_lanesxi<mode>): Likewise.
 892         (vec_store_lanesxi<mode>): Likewise.
 893         (*aarch64_mov<mode>): Likewise.
 894         (aarch64_ld2<mode>_dreg): Likewise.
 895         (aarch64_ld2<mode>_dreg): Likewise.
 896         (aarch64_ld3<mode>_dreg): Likewise.
 897         (aarch64_ld3<mode>_dreg): Likewise.
 898         (aarch64_ld4<mode>_dreg): Likewise.
 899         (aarch64_ld4<mode>_dreg): Likewise.
 900         (aarch64_tbl1<mode>): Likewise.
 901         (aarch64_tbl2v16qi): Likewise.
 902         (aarch64_combinev16qi): Likewise.
 903         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
 904         (aarch64_st2<mode>_dreg): Likewise.
 905         (aarch64_st2<mode>_dreg): Likewise.
 906         (aarch64_st3<mode>_dreg): Likewise.
 907         (aarch64_st3<mode>_dreg): Likewise.
 908         (aarch64_st4<mode>_dreg): Likewise.
 909         (aarch64_st4<mode>_dreg): Likewise.
 910         (*aarch64_simd_ld1r<mode>): Likewise.
 911         (aarch64_frecpe<mode>): Likewise.
 912         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
 913         (aarch64_frecps<mode>): Likewise.
 914
 915 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
 916
 917         * config/arm/iterators.md (V_elem_ch): New.
 918         (q): Likewise.
 919         (VQH_type): Likewise.
 920         * config/arm/arm.md (is_neon_type): New.
 921         (conds): Use is_neon_type.
 922         (anddi3_insn): Update type attribute.
 923         (xordi3_insn): Likewise.
 924         (one_cmpldi2): Likewise.
 925         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
 926         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
 927         (*movmisalign<mode>_neon_store): Likewise.
 928         (*movmisalign<mode>_neon_load): Likewise.
 929         (vec_set<mode>_internal): Likewise.
 930         (vec_set<mode>_internal): Likewise.
 931         (vec_setv2di_internal): Likewise.
 932         (vec_extract<mode>): Likewise.
 933         (vec_extract<mode>): Likewise.
 934         (vec_extractv2di): Likewise.
 935         (*add<mode>3_neon): Likewise.
 936         (adddi3_neon): Likewise.
 937         (*sub<mode>3_neon): Likewise.
 938         (subdi3_neon): Likewise.
 939         (fma<VCVTF:mode>4): Likewise.
 940         (fma<VCVTF:mode>4_intrinsic): Likewise.
 941         (*fmsub<VCVTF:mode>4): Likewise.
 942         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
 943         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
 944         (ior<mode>3): Likewise.
 945         (and<mode>3): Likewise.
 946         (orn<mode>3_neon): Likewise.
 947         (orndi3_neon): Likewise.
 948         (bic<mode>3_neon): Likewise.
 949         (bicdi3_neon): Likewise.
 950         (xor<mode>3): Likewise.
 951         (one_cmpl<mode>2): Likewise.
 952         (abs<mode>2): Likewise.
 953         (neg<mode>2): Likewise.
 954         (negdi2_neon): Likewise.
 955         (*umin<mode>3_neon): Likewise.
 956         (*umax<mode>3_neon): Likewise.
 957         (*smin<mode>3_neon): Likewise.
 958         (*smax<mode>3_neon): Likewise.
 959         (vashl<mode>3): Likewise.
 960         (vashr<mode>3_imm): Likewise.
 961         (vlshr<mode>3_imm): Likewise.
 962         (ashl<mode>3_signed): Likewise.
 963         (ashl<mode>3_unsigned): Likewise.
 964         (neon_load_count): Likewise.
 965         (ashldi3_neon_noclobber): Likewise.
 966         (ashldi3_neon): Likewise.
 967         (signed_shift_di3_neon): Likewise.
 968         (unsigned_shift_di3_neon): Likewise.
 969         (ashrdi3_neon_imm_noclobber): Likewise.
 970         (lshrdi3_neon_imm_noclobber): Likewise.
 971         (<shift>di3_neon): Likewise.
 972         (widen_ssum<mode>3): Likewise.
 973         (widen_usum<mode>3): Likewise.
 974         (quad_halves_<code>v4si): Likewise.
 975         (quad_halves_<code>v4sf): Likewise.
 976         (quad_halves_<code>v8hi): Likewise.
 977         (quad_halves_<code>v16qi): Likewise.
 978         (reduc_splus_v2di): Likewise.
 979         (neon_vpadd_internal<mode>): Likewise.
 980         (neon_vpsmin<mode>): Likewise.
 981         (neon_vpsmax<mode>): Likewise.
 982         (neon_vpumin<mode>): Likewise.
 983         (neon_vpumax<mode>): Likewise.
 984         (*ss_add<mode>_neon): Likewise.
 985         (*us_add<mode>_neon): Likewise.
 986         (*ss_sub<mode>_neon): Likewise.
 987         (*us_sub<mode>_neon): Likewise.
 988         (neon_vadd<mode>_unspec): Likewise.
 989         (neon_vaddl<mode>): Likewise.
 990         (neon_vaddw<mode>): Likewise.
 991         (neon_vhadd<mode>): Likewise.
 992         (neon_vqadd<mode>): Likewise.
 993         (neon_vaddhn<mode>): Likewise.
 994         (neon_vmul<mode>): Likewise.
 995         (neon_vfms<VCVTF:mode>): Likewise.
 996         (neon_vmlal<mode>): Likewise.
 997         (neon_vmls<mode>): Likewise.
 998         (neon_vmlsl<mode>): Likewise.
 999         (neon_vqdmulh<mode>): Likewise.
1000         (neon_vqdmlal<mode>): Likewise.
1001         (neon_vqdmlsl<mode>): Likewise.
1002         (neon_vmull<mode>): Likewise.
1003         (neon_vqdmull<mode>): Likewise.
1004         (neon_vsub<mode>_unspec): Likewise.
1005         (neon_vsubl<mode>): Likewise.
1006         (neon_vsubw<mode>): Likewise.
1007         (neon_vqsub<mode>): Likewise.
1008         (neon_vhsub<mode>): Likewise.
1009         (neon_vsubhn<mode>): Likewise.
1010         (neon_vceq<mode>): Likewise.
1011         (neon_vcge<mode>): Likewise.
1012         (neon_vcgeu<mode>): Likewise.
1013         (neon_vcgt<mode>): Likewise.
1014         (neon_vcgtu<mode>): Likewise.
1015         (neon_vcle<mode>): Likewise.
1016         (neon_vclt<mode>): Likewise.
1017         (neon_vcage<mode>): Likewise.
1018         (neon_vcagt<mode>): Likewise.
1019         (neon_vtst<mode>): Likewise.
1020         (neon_vabd<mode>): Likewise.
1021         (neon_vabdl<mode>): Likewise.
1022         (neon_vaba<mode>): Likewise.
1023         (neon_vabal<mode>): Likewise.
1024         (neon_vmax<mode>): Likewise.
1025         (neon_vmin<mode>): Likewise.
1026         (neon_vpaddl<mode>): Likewise.
1027         (neon_vpadal<mode>): Likewise.
1028         (neon_vpmax<mode>): Likewise.
1029         (neon_vpmin<mode>): Likewise.
1030         (neon_vrecps<mode>): Likewise.
1031         (neon_vrsqrts<mode>): Likewise.
1032         (neon_vqabs<mode>): Likewise.
1033         (neon_vqneg<mode>): Likewise.
1034         (neon_vcls<mode>): Likewise.
1035         (clz<mode>2): Likewise.
1036         (popcount<mode>2): Likewise.
1037         (neon_vrecpe<mode>): Likewise.
1038         (neon_vrsqrte<mode>): Likewise.
1039         (neon_vget_lane<mode>_sext_internal): Likewise.
1040         (neon_vget_lane<mode>_zext_internal): Likewise.
1041         (neon_vdup_n<mode>): Likewise.
1042         (neon_vdup_n<mode>): Likewise.
1043         (neon_vdup_nv2di): Likewise.
1044         (neon_vdup_lane<mode>_interal): Likewise.
1045         (*neon_vswp<mode>): Likewise.
1046         (neon_vcombine<mode>): Likewise.
1047         (float<mode><V_cvtto>2): Likewise.
1048         (floatuns<mode><V_cvtto>2): Likewise.
1049         (fix_trunc<mode><V_cvtto>2): Likewise.
1050         (fixuns_trunc<mode><V_cvtto>2
1051         (neon_vcvt<mode>): Likewise.
1052         (neon_vcvt<mode>): Likewise.
1053         (neon_vcvtv4sfv4hf): Likewise.
1054         (neon_vcvtv4hfv4sf): Likewise.
1055         (neon_vcvt_n<mode>): Likewise.
1056         (neon_vcvt_n<mode>): Likewise.
1057         (neon_vmovn<mode>): Likewise.
1058         (neon_vqmovn<mode>): Likewise.
1059         (neon_vqmovun<mode>): Likewise.
1060         (neon_vmovl<mode>): Likewise.
1061         (neon_vmul_lane<mode>): Likewise.
1062         (neon_vmul_lane<mode>): Likewise.
1063         (neon_vmull_lane<mode>): Likewise.
1064         (neon_vqdmull_lane<mode>): Likewise.
1065         (neon_vqdmulh_lane<mode>): Likewise.
1066         (neon_vqdmulh_lane<mode>): Likewise.
1067         (neon_vmla_lane<mode>): Likewise.
1068         (neon_vmla_lane<mode>): Likewise.
1069         (neon_vmlal_lane<mode>): Likewise.
1070         (neon_vqdmlal_lane<mode>): Likewise.
1071         (neon_vmls_lane<mode>): Likewise.
1072         (neon_vmls_lane<mode>): Likewise.
1073         (neon_vmlsl_lane<mode>): Likewise.
1074         (neon_vqdmlsl_lane<mode>): Likewise.
1075         (neon_vext<mode>): Likewise.
1076         (neon_vrev64<mode>): Likewise.
1077         (neon_vrev32<mode>): Likewise.
1078         (neon_vrev16<mode>): Likewise.
1079         (neon_vbsl<mode>_internal): Likewise.
1080         (neon_vshl<mode>): Likewise.
1081         (neon_vqshl<mode>): Likewise.
1082         (neon_vshr_n<mode>): Likewise.
1083         (neon_vshrn_n<mode>): Likewise.
1084         (neon_vqshrn_n<mode>): Likewise.
1085         (neon_vqshrun_n<mode>): Likewise.
1086         (neon_vshl_n<mode>): Likewise.
1087         (neon_vqshl_n<mode>): Likewise.
1088         (neon_vqshlu_n<mode>): Likewise.
1089         (neon_vshll_n<mode>): Likewise.
1090         (neon_vsra_n<mode>): Likewise.
1091         (neon_vsri_n<mode>): Likewise.
1092         (neon_vsli_n<mode>): Likewise.
1093         (neon_vtbl1v8qi): Likewise.
1094         (neon_vtbl2v8qi): Likewise.
1095         (neon_vtbl3v8qi): Likewise.
1096         (neon_vtbl4v8qi): Likewise.
1097         (neon_vtbl1v16qi): Likewise.
1098         (neon_vtbl2v16qi): Likewise.
1099         (neon_vcombinev16qi): Likewise.
1100         (neon_vtbx1v8qi): Likewise.
1101         (neon_vtbx2v8qi): Likewise.
1102         (neon_vtbx3v8qi): Likewise.
1103         (neon_vtbx4v8qi): Likewise.
1104         (*neon_vtrn<mode>_insn): Likewise.
1105         (*neon_vzip<mode>_insn): Likewise.
1106         (*neon_vuzp<mode>_insn): Likewise.
1107         (neon_vld1<mode>): Likewise.
1108         (neon_vld1_lane<mode>): Likewise.
1109         (neon_vld1_lane<mode>): Likewise.
1110         (neon_vld1_dup<mode>): Likewise.
1111         (neon_vld1_dup<mode>): Likewise.
1112         (neon_vld1_dupv2di): Likewise.
1113         (neon_vst1<mode>): Likewise.
1114         (neon_vst1_lane<mode>): Likewise.
1115         (neon_vst1_lane<mode>): Likewise.
1116         (neon_vld2<mode>): Likewise.
1117         (neon_vld2<mode>): Likewise.
1118         (neon_vld2_lane<mode>): Likewise.
1119         (neon_vld2_lane<mode>): Likewise.
1120         (neon_vld2_dup<mode>): Likewise.
1121         (neon_vst2<mode>): Likewise.
1122         (neon_vst2<mode>): Likewise.
1123         (neon_vst2_lane<mode>): Likewise.
1124         (neon_vst2_lane<mode>): Likewise.
1125         (neon_vld3<mode>): Likewise.
1126         (neon_vld3qa<mode>): Likewise.
1127         (neon_vld3qb<mode>): Likewise.
1128         (neon_vld3_lane<mode>): Likewise.
1129         (neon_vld3_lane<mode>): Likewise.
1130         (neon_vld3_dup<mode>): Likewise.
1131         (neon_vst3<mode>): Likewise.
1132         (neon_vst3qa<mode>): Likewise.
1133         (neon_vst3qb<mode>): Likewise.
1134         (neon_vst3_lane<mode>): Likewise.
1135         (neon_vst3_lane<mode>): Likewise.
1136         (neon_vld4<mode>): Likewise.
1137         (neon_vld4qa<mode>): Likewise.
1138         (neon_vld4qb<mode>): Likewise.
1139         (neon_vld4_lane<mode>): Likewise.
1140         (neon_vld4_lane<mode>): Likewise.
1141         (neon_vld4_dup<mode>): Likewise.
1142         (neon_vst4<mode>): Likewise.
1143         (neon_vst4qa<mode>): Likewise.
1144         (neon_vst4qb<mode>): Likewise.
1145         (neon_vst4_lane<mode>): Likewise.
1146         (neon_vst4_lane<mode>): Likewise.
1147         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1148         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1149         (neon_vec_<US>mult_lo_<mode>): Likewise.
1150         (neon_vec_<US>mult_hi_<mode>): Likewise.
1151         (neon_vec_<US>shiftl_<mode>): Likewise.
1152         (neon_unpack<US>_<mode>): Likewise.
1153         (neon_vec_<US>mult_<mode>): Likewise.
1154         (vec_pack_trunc_<mode>): Likewise.
1155         (neon_vec_pack_trunc_<mode>): Likewise.
1156         (neon_vabd<mode>_2): Likewise.
1157         (neon_vabd<mode>_3): Likewise.
1158
1159 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1160
1161         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1162         (load_pair): Update type attribute.
1163         (store_pair): Update type attribute.
1164         * config/aarch64/iterators.md (q): New.
1165
1166 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1167
1168         * config/arm/types.md: Add new types for Neon insns.
1169
1170 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1171             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1172             Sergey Lega  <sergey.s.lega@intel.com>
1173             Anna Tikhonova  <anna.tikhonova@intel.com>
1174             Ilya Tocar  <ilya.tocar@intel.com>
1175             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1176             Ilya Verbin  <ilya.verbin@intel.com>
1177             Kirill Yukhin  <kirill.yukhin@intel.com>
1178             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1179
1180         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1181         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1182         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1183         (rcp14<mode>): New.
1184         (srcp14<mode>): Ditto.
1185         (rsqrt14<mode>): Ditto.
1186         (rsqrt14<mode>): Ditto.
1187         (avx512f_vmscalef<mode>): Ditto.
1188         (avx512f_scalef<mode>): Ditto.
1189         (avx512f_getexp<mode>): Ditto.
1190         (avx512f_sgetexp<mode>): Ditto.
1191         (avx512f_fixupimm<mode>): Ditto.
1192         (avx512f_sfixupimm<mode>): Ditto.
1193         (avx512f_rndscale<mode>): Ditto.
1194         (*avx512er_exp2<mode>): Ditto.
1195         (*avx512er_rcp28<mode>): Ditto.
1196         (avx512er_rsqrt28<mode>): Ditto.
1197         (avx512f_getmant<mode>): Ditto.
1198         (avx512f_getmant<mode>): Ditto.
1199         (avx512f_rndscale<mode>): Fix formatting.
1200
1201 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1202
1203         * ipa-utils.h (ipa_edge_within_scc): Declare.
1204         * ipa-cp.c (edge_within_scc): Moved...
1205         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1206
1207 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1208             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1209             Sergey Lega  <sergey.s.lega@intel.com>
1210             Anna Tikhonova  <anna.tikhonova@intel.com>
1211             Ilya Tocar  <ilya.tocar@intel.com>
1212             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1213             Ilya Verbin  <ilya.verbin@intel.com>
1214             Kirill Yukhin  <kirill.yukhin@intel.com>
1215             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1216
1217         * config/i386/predicates.md (const_8_to_15_operand): New.
1218         (const_16_to_31_operand): Ditto.
1219         * config/i386/sse.md (V8FI): New.
1220         (V16FI): Ditto.
1221         (reduc_splus_v8df): Ditto.
1222         (reduc_splus_v16sf): Ditto.
1223         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1224         (vec_extract_hi_<mode>): Ditto.
1225         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1226         (vec_set_lo_<mode>): Ditto.
1227         (vec_set_hi_<mode>): Ditto.
1228         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1229         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1230         (avx512f_pshufd_1): Ditto.
1231         (avx512f_broadcast<mode>): Ditto.
1232         (avx512f_broadcast<mode>): Ditto.
1233         (define_split): Split vec_extract_lo into move.
1234         (ssequartermode): Ditto.
1235         (ssedoublemode): Extened with wider modes.
1236         (vec_extract_lo_<mode>): Ditto.
1237
1238 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1239             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1240             Sergey Lega  <sergey.s.lega@intel.com>
1241             Anna Tikhonova  <anna.tikhonova@intel.com>
1242             Ilya Tocar  <ilya.tocar@intel.com>
1243             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1244             Ilya Verbin  <ilya.verbin@intel.com>
1245             Kirill Yukhin  <kirill.yukhin@intel.com>
1246             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1247
1248         * config/i386/predicates.md (register_or_constm1_operand): New.
1249         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1250         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1251         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1252         UNSPEC_SCATTER_PREFETCH
1253         (VI48_512): New.
1254         (avx512f_ucmp<mode>3): Ditto.
1255         (avx512f_vternlog<mode>): Ditto.
1256         (avx512f_align<mode>): Ditto.
1257         (<shift_insn><mode>3): Ditto.
1258         (avx512f_<rotate>v<mode>): Ditto.
1259         (avx512f_<rotate><mode>): Ditto.
1260         (avx512f_eq<mode>3): Ditto.
1261         (avx512f_eq<mode>3_1): Ditto.
1262         (avx512f_gt<mode>3): Ditto.
1263         (avx512f_testm<mode>3): Ditto.
1264         (avx512f_testnm<mode>3): Ditto.
1265         (avx512pf_gatherpf<mode>): Ditto.
1266         (*avx512pf_gatherpf<mode>_mask): Ditto.
1267         (*avx512pf_gatherpf<mode>): Ditto.
1268         (avx512pf_scatterpf<mode>): Ditto.
1269         (*avx512pf_scatterpf<mode>_mask): Ditto.
1270         (*avx512pf_scatterpf<mode>): Ditto.
1271         (avx512f_vec_dup_gpr<mode>): Ditto.
1272         (clz<mode>2): Ditto.
1273         (conflict<mode>): Ditto.
1274         (REDUC_SMINMAX_MODE): Extened with wider modes.
1275         (reduc_<code>_<mode>): Ditto.
1276         (vlshr<mode>3): Ditto.
1277         (vashl<mode>3): Ditto.
1278
1279 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1280             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1281             Sergey Lega  <sergey.s.lega@intel.com>
1282             Anna Tikhonova  <anna.tikhonova@intel.com>
1283             Ilya Tocar  <ilya.tocar@intel.com>
1284             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1285             Ilya Verbin  <ilya.verbin@intel.com>
1286             Kirill Yukhin  <kirill.yukhin@intel.com>
1287             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1288
1289         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1290         UNSPEC_SCATTER.
1291         (VI48F_512): New.
1292         (avx512fmaskmode): Ditto.
1293         (bcstscalarsuff): Ditto.
1294         (avx512f_blendm<mode>): Ditto.
1295         (cmp_imm_predicate): Ditto.
1296         (avx512f_cmp<mode>3): Ditto.
1297         (avx512f_vec_dup<mode>): Ditto.
1298         (avx512f_vec_dup_mem<mode>): Ditto.
1299         (avx512f_vpermi2var<mode>3): Ditto.
1300         (avx512f_vpermt2var<mode>3): Ditto.
1301         (vec_init<mode>): Ditto.
1302         (avx512f_gathersi<mode>): Ditto.
1303         (*avx512f_gathersi<mode>): Ditto.
1304         (*avx512f_gathersi<mode>_2): Ditto.
1305         (avx512f_gatherdi<mode>): Ditto.
1306         (*avx512f_gatherdi<mode>): Ditto.
1307         (*avx512f_gatherdi<mode>_2): Ditto.
1308         (avx512f_scattersi<mode>): Ditto.
1309         (*avx512f_scattersi<mode>): Ditto.
1310         (avx512f_scatterdi<mode>): Ditto.
1311         (*avx512f_scatterdi<mode>): Ditto.
1312         (sseintprefix): Extened with wider modes.
1313         (VEC_GATHER_IDXSI): Ditto.
1314         (VEC_GATHER_IDXDI): Ditto.
1315         (VEC_GATHER_SRCDI): Ditto.
1316
1317 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1318             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1319
1320         * config/arm/t-aprofile: New file.
1321         * config.gcc: Handle --with-multilib-list option.
1322
1323 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1324
1325         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1326         the input for r2.
1327
1328 2013-10-15  Richard Biener  <rguenther@suse.de>
1329
1330         * tree-loop-distribution.c (build_empty_rdg): Inline into
1331         single user.
1332         (rdg_flag_vertex): Inline into single user.
1333         (rdg_flag_vertex_and_dependent): Likewise.
1334         (build_rdg_partition_for_vertex): Remove processed bitmap.
1335         (rdg_build_partitions): Simplify.
1336
1337 2013-10-15  Richard Biener  <rguenther@suse.de>
1338
1339         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1340         Restructure forwarding through conversions and copies to
1341         avoid performing copy-propagation the wrong way.  Adjust
1342         recursion invocations.
1343         (forward_propagate_addr_expr): Add argument stating if we
1344         are recursing from a single-use.
1345         (ssa_forward_propagate_and_combine): Adjust.
1346
1347 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1348
1349         * dumpfile.h (gcc::dump_manager): New class, to hold state
1350         relating to dumpfile management.
1351         (get_dump_file_name): Remove in favor of method of dump_manager.
1352         (dump_initialized_p): Likewise.
1353         (dump_start): Likewise.
1354         (dump_finish): Likewise.
1355         (dump_switch_p): Likewise.
1356         (dump_register): Likewise.
1357         (get_dump_file_info): Likewise.
1358         * context.c (gcc::context::context): Construct the dump_manager
1359         instance.
1360         * context.h (gcc::context::get_dumps): New.
1361         (gcc::context::m_dumps): New.
1362         * coverage.c (coverage_init): Port to dump_manager API.
1363         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1364         (extra_dump_files_in_use): Likewise.
1365         (extra_dump_files_alloced): Likewise.
1366         (gcc::dump_manager::dump_manager): New.
1367         (dump_register): Convert to...
1368         (gcc::dump_manager::dump_register): ...method, replacing
1369         function-static next_dump with m_next_dump field.
1370         (get_dump_file_info): Convert to...
1371         (gcc::dump_manager::get_dump_file_info): ...method.
1372         (get_dump_file_name): Convert to...
1373         (gcc::dump_manager::get_dump_file_name): ...method.
1374         (dump_start): Convert to...
1375         (gcc::dump_manager::dump_start): ...method.
1376         (dump_finish): Convert to...
1377         (gcc::dump_manager::dump_finish): ...method.
1378         (dump_begin): Replace body with...
1379         (gcc::dump_manager::dump_begin): ...new method.
1380         (dump_phase_enabled_p): Convert to...
1381         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1382         (dump_phase_enabled_p): Convert to...
1383         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1384         (dump_initialized_p):  Convert to...
1385         (gcc::dump_manager::dump_initialized_p): ...method.
1386         (dump_flag_name): Replace body with...
1387         (gcc::dump_manager::dump_flag_name): ...new method.
1388         (dump_enable_all): Convert to...
1389         (gcc::dump_manager::dump_enable_all): ...new method.
1390         (opt_info_enable_passes): Convert to...
1391         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1392         (dump_switch_p_1): Convert to...
1393         (gcc::dump_manager::dump_switch_p_1): ...new method.
1394         (dump_switch_p):  Convert to...
1395         (gcc::dump_manager::dump_switch_p): ...new method.
1396         (opt_info_switch_p): Port to dump_manager API.
1397         (enable_rtl_dump_file): Likewise.
1398         * opts-global.c (handle_common_deferred_options): Port to new
1399         dump_manager API.
1400         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1401         (pass_manager::register_one_dump_file): Likewise.
1402         (pass_manager::register_pass): Likewise.
1403         (pass_init_dump_file): Likewise.
1404         (pass_fini_dump_file): Likewise.
1405         * statistics.c (statistics_early_init): Likewise.
1406
1407 2013-10-14  Richard Biener  <rguenther@suse.de>
1408
1409         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1410         iterative_hash_canonical_type, gimple_canonical_type_hash,
1411         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1412         gimple_register_canonical_type, print_gimple_types_stats,
1413         free_gimple_type_tables): Move to lto/lto.c
1414         (gt-gimple.h): Do not include.
1415         * gimple.h (gimple_register_canonical_type,
1416         print_gimple_types_stats, free_gimple_type_tables): Remove.
1417         * Makefile.in (GTFILES): Remove gimple.c.
1418
1419 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1420
1421         PR target/58716
1422         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1423         scanning for msp430x targets.
1424
1425 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1426
1427         PR bootstrap/58509
1428         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1429         (registers_ok_for_ldd_peep): Move around.
1430         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1431         * config/sparc/sparc.md (widening peepholes): Use it.
1432
1433 2013-10-14  Richard Biener  <rguenther@suse.de>
1434
1435         PR middle-end/58712
1436         PR middle-end/55358
1437         * gimple.c (iterative_hash_canonical_type): Make sure to
1438         record the hash into the correct hashtable slot.
1439
1440 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1441
1442         PR rtl-optimization/58662
1443         * combine.c (try_combine): Take into account death nodes on I2 when
1444         splitting a PARALLEL of two independent SETs.  Fix dump message.
1445
1446 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1447
1448         PR target/51244
1449         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1450         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1451         * config/sh/t-sh (sh_treg_combine.o): New entry.
1452         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1453         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1454         (register_sh_passes): New function.  Register sh_treg_combine pass.
1455         (sh_option_override): Invoke it.
1456         (sh_canonicalize_comparison): Handle op0_preserve_value.
1457         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1458         opportunities.  Canonicalize branch condition.
1459         (nott): Allow only if pseudos can be created for non-SH2A.
1460
1461 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1462
1463         PR target/58690
1464         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1465         (ix86_expand_movmem): Replace copy_addr_to_reg with
1466         ix86_copy_addr_to_reg.
1467         (ix86_expand_setmem): Likewise.
1468
1469 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1470
1471         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1472         provided by ix86_fp_compare_mode instead of CCFPUmode.
1473
1474 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1475
1476         * config/aarch64/arm_neon.h
1477         (vtbx<1,3>_<psu>8): Fix register constriants.
1478
1479 2013-10-11  Jeff Law  <law@redhat.com>
1480
1481         PR tree-optimization/58640
1482         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1483         threading paths that cross over two loop entry points.
1484
1485 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1486
1487         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1488         handle vector float as well.
1489         (*vsx_le_perm_load_v4si): Likewise.
1490         (*vsx_le_perm_store_v2di): Likewise.
1491         (*vsx_le_perm_store_v4si): Likewise.
1492
1493 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1494
1495         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1496         directly to circumvent subtract from splat{31} workaround.
1497         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1498         prototype.
1499         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1500         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1501         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1502         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1503         separate big and little endian logic.
1504         (*altivec_vperm_<mode>_internal): New define_insn.
1505         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1506         separate big and little endian logic.
1507         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1508         (vec_permv16qi): Add little endian logic.
1509
1510 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1511
1512         * doc/extend.texi (returns_nonnull): Remove arguments.
1513
1514 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1515             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1516             Sergey Lega  <sergey.s.lega@intel.com>
1517             Anna Tikhonova  <anna.tikhonova@intel.com>
1518             Ilya Tocar  <ilya.tocar@intel.com>
1519             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1520             Ilya Verbin  <ilya.verbin@intel.com>
1521             Kirill Yukhin  <kirill.yukhin@intel.com>
1522             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1523
1524         * config/i386/sse.md (VI48F_256_512): New.
1525         (avx2_permvar<mode>): Change to ...
1526         (<avx2_avx512f>_permvar<mode>): This.
1527
1528 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1529             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1530             Sergey Lega  <sergey.s.lega@intel.com>
1531             Anna Tikhonova  <anna.tikhonova@intel.com>
1532             Ilya Tocar  <ilya.tocar@intel.com>
1533             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1534             Ilya Verbin  <ilya.verbin@intel.com>
1535             Kirill Yukhin  <kirill.yukhin@intel.com>
1536             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1537
1538         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1539         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1540         * config/i386/sse.md (VI4_AVX): New.
1541         (sf2simodelower): Ditto.
1542         (sse2_cvtps2dq): Change to ...
1543         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1544
1545 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1546             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1547             Sergey Lega  <sergey.s.lega@intel.com>
1548             Anna Tikhonova  <anna.tikhonova@intel.com>
1549             Ilya Tocar  <ilya.tocar@intel.com>
1550             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1551             Ilya Verbin  <ilya.verbin@intel.com>
1552             Kirill Yukhin  <kirill.yukhin@intel.com>
1553             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1554
1555         * config/i386/sse.md (V_512): New.
1556         (VI_512): Ditto.
1557         (vcond<V_512:mode><VF_512:mode>): Ditto.
1558         (vcond<V_512:mode><VI_512:mode>): Ditto.
1559         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1560
1561 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1562             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1563             Sergey Lega  <sergey.s.lega@intel.com>
1564             Anna Tikhonova  <anna.tikhonova@intel.com>
1565             Ilya Tocar  <ilya.tocar@intel.com>
1566             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1567             Ilya Verbin  <ilya.verbin@intel.com>
1568             Kirill Yukhin  <kirill.yukhin@intel.com>
1569             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1570
1571         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1572         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1573         (FMAMODE): Ditto.
1574         (fma<mode>4): Removed condition.
1575         (fms<mode>4): Ditto.
1576         (fnma<mode>4): Ditto.
1577         (fnms<mode>4): Ditto.
1578         (fma4i_fmadd_<mode>): Ditto.
1579         (*fma_fmadd_<mode>): Ditto.
1580         (*fma_fmsub_<mode>): Ditto.
1581         (*fma_fnmadd_<mode>): Ditto.
1582         (*fma_fnmsub_<mode>): Ditto.
1583         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1584         (*fma_fmaddsub_<mode>): Ditto.
1585         (*fma_fmsubadd_<mode>): Ditto.
1586         (*fmai_fmadd_<mode>): Ditto.
1587         (*fmai_fmsub_<mode>): Ditto.
1588         (*fmai_fnmadd_<mode>): Ditto.
1589         (*fmai_fnmsub_<mode>): Ditto.
1590
1591 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1592             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1593             Sergey Lega  <sergey.s.lega@intel.com>
1594             Anna Tikhonova  <anna.tikhonova@intel.com>
1595             Ilya Tocar  <ilya.tocar@intel.com>
1596             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1597             Ilya Verbin  <ilya.verbin@intel.com>
1598             Kirill Yukhin  <kirill.yukhin@intel.com>
1599             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1600
1601         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1602         (VI124_256): Changed to ...
1603         (VI124_256_48_512): This.
1604         (ssepackmode): Extended with wider modes.
1605         (<code><mode>3): Changed iterator.
1606         (*avx2_<code><mode>3): Ditto.
1607         (vec_pack_trunc_<mode>): Ditto.
1608
1609 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1610             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1611             Sergey Lega  <sergey.s.lega@intel.com>
1612             Anna Tikhonova  <anna.tikhonova@intel.com>
1613             Ilya Tocar  <ilya.tocar@intel.com>
1614             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1615             Ilya Verbin  <ilya.verbin@intel.com>
1616             Kirill Yukhin  <kirill.yukhin@intel.com>
1617             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1618
1619         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1620         (VI8F_256_512): Ditto.
1621         (abs<mode>2): Changed iterator.
1622         (avx2_perm<mode>): Changed to ...
1623         (<avx2_avx512f>_perm<mode>): This.
1624         (avx2_perm<mode>_1): Changed to ...
1625         (<avx2_avx512f>_perm<mode>_1): This.
1626
1627 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1628             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1629             Sergey Lega  <sergey.s.lega@intel.com>
1630             Anna Tikhonova  <anna.tikhonova@intel.com>
1631             Ilya Tocar  <ilya.tocar@intel.com>
1632             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1633             Ilya Verbin  <ilya.verbin@intel.com>
1634             Kirill Yukhin  <kirill.yukhin@intel.com>
1635             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1636
1637         * config/i386/sse.md (VI48_AVX512F): New.
1638         (VI48_AVX2): Changed to ...
1639         (VI48_AVX2_48_AVX512F): This.
1640         (avx2_ashrv<mode>): Changed to ...
1641         (<avx2_avx512f>_ashrv<mode>): This.
1642         (avx2_<shift_insn>v<mode>): Changed to ...
1643         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1644
1645 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1646             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1647             Sergey Lega  <sergey.s.lega@intel.com>
1648             Anna Tikhonova  <anna.tikhonova@intel.com>
1649             Ilya Tocar  <ilya.tocar@intel.com>
1650             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1651             Ilya Verbin  <ilya.verbin@intel.com>
1652             Kirill Yukhin  <kirill.yukhin@intel.com>
1653             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1654
1655         * config/i386/sse.md (VI4_AVX512F): New.
1656         (VI8_AVX2_AVX512F): Ditto.
1657         (mul<mode>3): Extended with wider modes.
1658         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1659         (mul<mode>3): Ditto.
1660         (vec_widen_<s>mult_odd_<mode>): Ditto.
1661
1662 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1663             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1664             Sergey Lega  <sergey.s.lega@intel.com>
1665             Anna Tikhonova  <anna.tikhonova@intel.com>
1666             Ilya Tocar  <ilya.tocar@intel.com>
1667             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1668             Ilya Verbin  <ilya.verbin@intel.com>
1669             Kirill Yukhin  <kirill.yukhin@intel.com>
1670             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1671
1672         * config/i386/sse.md (VI2_AVX512F): New.
1673         (VI124_AVX512F): Ditto.
1674         (sseunpackmode): Extended with wider modes.
1675         (sseunpackfltmode): Ditto.
1676         (vec_unpacks_float_hi_<mode>): Ditto.
1677         (vec_unpacks_float_lo_<mode>): Ditto.
1678         (vec_unpacku_float_hi_<mode>): Ditto.
1679         (vec_unpacku_float_lo_<mode>): Ditto.
1680         (vec_unpacks_lo_<mode>): Ditto.
1681         (vec_unpacks_hi_<mode>): Ditto.
1682         (vec_unpacku_lo_<mode>): Ditto.
1683         (vec_unpacku_hi_<mode>): Ditto.
1684
1685 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1686             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1687             Sergey Lega  <sergey.s.lega@intel.com>
1688             Anna Tikhonova  <anna.tikhonova@intel.com>
1689             Ilya Tocar  <ilya.tocar@intel.com>
1690             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1691             Ilya Verbin  <ilya.verbin@intel.com>
1692             Kirill Yukhin  <kirill.yukhin@intel.com>
1693             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1694
1695         * config/i386/i386.md (multdiv): New.
1696         (multdiv_mnemonic): Ditto.
1697         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1698         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1699         (<sse>_vmdiv<mode>3): Removed.
1700
1701 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1702             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1703             Sergey Lega  <sergey.s.lega@intel.com>
1704             Anna Tikhonova  <anna.tikhonova@intel.com>
1705             Ilya Tocar  <ilya.tocar@intel.com>
1706             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1707             Ilya Verbin  <ilya.verbin@intel.com>
1708             Kirill Yukhin  <kirill.yukhin@intel.com>
1709             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1710
1711         * config/i386/sse.md (V): Extended with wider modes.
1712         (VF2): Ditto.
1713         (ssehalfvecmode): Ditto.
1714         (i128): Ditto.
1715         (ssepackfltmode): Ditto.
1716         (avx_vec_concat<mode>): Ditto.
1717         (V_256_512): New iterator.
1718         (VF2_512_256): Ditto.
1719         (si2dfmode): New attribute.
1720         (si2dfmodelower): Ditto.
1721         (sf2dfmode): Ditto.
1722         (concat_tg_mode): Ditto.
1723         (floatv4siv4df2): Changed to ...
1724         (float<si2dfmodelower><mode>2): This.
1725         (avx_cvtps2pd256): Changed to ...
1726         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1727         (vec_pack_trunc_v4df): Changed to ...
1728         (vec_pack_trunc_<mode>): This.
1729         (avx_vpermil<mode>): Changed to ...
1730         (<sse2_avx_avx512f>_vpermil<mode>): This.
1731         (<fixsuffix>fix_truncv8dfv8si2): New.
1732         (vec_pack_sfix_trunc_v8df): Ditto.
1733         (avx512f_rndscale<mode>): Ditto.
1734         (avx512f_roundpd512): Ditto.
1735         (vec_pack_ufix_trunc_<mode>): Updated iterator.
1736
1737 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1738             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1739             Sergey Lega  <sergey.s.lega@intel.com>
1740             Anna Tikhonova  <anna.tikhonova@intel.com>
1741             Ilya Tocar  <ilya.tocar@intel.com>
1742             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1743             Ilya Verbin  <ilya.verbin@intel.com>
1744             Kirill Yukhin  <kirill.yukhin@intel.com>
1745             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1746
1747         * config/i386/i386.md (any_fix): New iterator.
1748         (fixsuffix): New attribute.
1749         * config/i386/sse.md (VF1): Extened with wider modes.
1750         (VI): Ditto.
1751         (VI_AVX2): Ditto.
1752         (VI8): Ditto.
1753         (sseintvecmodelower): Ditto.
1754         (ssescalarmode): Ditto.
1755         (ssescalarnum): Ditto.
1756         (VF1_128_256): New.
1757         (ssexmmmode): Ditto.
1758         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
1759         (<sse>_rcp<mode>2): Change iterator.
1760         (rsqrt<mode>2): Ditto.
1761         (<sse>_rsqrt<mode>2): Ditto.
1762         (avx2_vec_dup<mode>): Ditto.
1763         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1764         (round<mode>2_sfix): Ditto.
1765         (avx2_pbroadcast<mode>): Ditto.
1766         (*andnot<mode>3): Handle XI mode.
1767         (*<code><mode>3): Ditto.
1768         (AVXTOSSEMODE): Removed.
1769         (avx_vpermil<mode>): Changed to ...
1770         (<sse2_avx_avx512f>_vpermil<mode>): This.
1771
1772 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1773             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1774             Sergey Lega  <sergey.s.lega@intel.com>
1775             Anna Tikhonova  <anna.tikhonova@intel.com>
1776             Ilya Tocar  <ilya.tocar@intel.com>
1777             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1778             Ilya Verbin  <ilya.verbin@intel.com>
1779             Kirill Yukhin  <kirill.yukhin@intel.com>
1780             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1781
1782         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
1783         (<sse>_comi): Ditto.
1784         (<sse>_ucomi): Ditto.
1785         (sse_cvtss2siq_2): Ditto.
1786         (sse2_cvtsd2si): Ditto.
1787         (sse2_cvtsd2siq): Ditto.
1788         (sse2_cvttsd2si): Ditto.
1789         (sse2_cvttsd2siq): Ditto.
1790         (<shift_insn><mode>3): Ditto.
1791         (sse2_cvtsi2sdq): Update constraint and prefix.
1792         (sse_cvtsi2ss): Update prefix.
1793         (sse_cvtsi2ssq): Ditto.
1794
1795 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
1796
1797         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
1798         ignore internal calls.
1799
1800 2013-10-11  Richard Biener  <rguenther@suse.de>
1801
1802         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
1803         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
1804         TREE_OVERFLOW is set.
1805
1806 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
1807
1808         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
1809
1810         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
1811         * gimple.h: Likewise.
1812
1813         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
1814
1815         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
1816         macros.
1817
1818         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
1819
1820         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
1821         * configure: Regenerate.
1822
1823 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
1824
1825         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
1826         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
1827         formatting fixes, use pp_colon instead of pp_character (..., ':'),
1828         similarly pp_right_paren.
1829         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
1830         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
1831         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
1832         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
1833         clauses.
1834         (omp_declare_simd_clauses_equal,
1835         omp_remove_redundant_declare_simd_attrs): New functions.
1836         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
1837         (walk_tree_1): Handle new OpenMP 4.0 clauses.
1838         * tree.h (OMP_LOOP_CHECK): Define.
1839         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
1840         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
1841         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
1842         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
1843         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
1844         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
1845         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
1846         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
1847         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
1848         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
1849         OMP_CLAUSE_SIMDLEN_EXPR): Define.
1850         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
1851         (omp_remove_redundant_declare_simd_attrs): New prototype.
1852         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
1853         GIMPLE_OMP_TEAMS): New codes.
1854         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
1855         * omp-low.c (struct omp_context): Add cancel_label and cancellable
1856         fields.
1857         (target_nesting_level): New variable.
1858         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
1859         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
1860         for collapse > 1 static schedule unless ordered.
1861         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
1862         (determine_parallel_type): Adjust get_ws_args_for caller.
1863         (install_var_field): Handle mask & 4 for double indirection.
1864         (scan_sharing_clauses): Ignore shared clause on teams construct.
1865         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
1866         (create_omp_child_function): If inside target or declare target
1867         constructs, set "omp declare target" attribute on the child function.
1868         (find_combined_for): New function.
1869         (scan_omp_parallel): Handle combined loops.
1870         (scan_omp_target, scan_omp_teams): New functions.
1871         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
1872         restrictions and set ctx->cancellable for cancellable constructs.
1873         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
1874         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
1875         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
1876         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
1877         (omp_clause_aligned_alignment): New function.
1878         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
1879         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
1880         on teams constructs.  Handle user defined reductions and new
1881         OpenMP 4.0 clauses.
1882         (lower_reduction_clauses): Don't set placeholder to address of ref
1883         if it has already the right type.
1884         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
1885         (expand_parallel_call): Use the new non-_start suffixed builtins,
1886         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
1887         and GOMP_parallel_end after the call.
1888         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
1889         (expand_omp_for_init_counts): Handle combined loops.
1890         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
1891         loops.
1892         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
1893         end of cancellable loops.
1894         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
1895         Likewise.  Handle collapse > 1 loops.
1896         (expand_omp_simd): Handle combined loops.
1897         (expand_omp_for): Add inner_stmt argument, adjust callers of
1898         expand_omp_for* functions, use expand_omp_for_static*chunk even
1899         for collapse > 1 unless ordered.
1900         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
1901         of cancellable sections.
1902         (expand_omp_single): Remove need_barrier variable, just rely on
1903         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
1904         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
1905         (expand_omp_atomic_load, expand_omp_atomic_store,
1906         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
1907         (expand_omp_target): New function.
1908         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
1909         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
1910         (build_omp_regions_1): Immediately close region for
1911         GF_OMP_TARGET_KIND_UPDATE.
1912         (maybe_add_implicit_barrier_cancel): New function.
1913         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
1914         cancellation.
1915         (lower_omp_single): Likewise.  Add clobber after the barrier.
1916         (lower_omp_taskgroup): New function.
1917         (lower_omp_for): Handle combined loops.  Adjust
1918         lower_rec_input_clauses caller.  Handle cancellation.
1919         (lower_depend_clauses): New function.
1920         (lower_omp_taskreg): Lower depend clauses.  Adjust
1921         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
1922         cancellation.
1923         (lower_omp_target, lower_omp_teams): New functions.
1924         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
1925         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
1926         and GOMP_cancellation_point calls.
1927         (lower_omp): Fold stmts inside of target region.
1928         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
1929         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
1930         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
1931         (BT_FN_VOID_OMPFN_PTR_UINT,
1932         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
1933         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
1934         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
1935         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
1936         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
1937         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
1938         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
1939         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
1940         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
1941         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
1942         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
1943         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
1944         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
1945         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
1946         BUILT_IN_GOMP_PARALLEL_END.
1947         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
1948         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
1949         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
1950         GF_OMP_FOR_KIND_DISTRIBUTE.
1951         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
1952         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
1953         (dump_gimple_omp_return): Print lhs if it has any.
1954         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
1955         gimple_omp_atomic_seq_cst_p.
1956         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
1957         and GIMPLE_OMP_TEAMS.
1958         * langhooks.c (lhd_omp_mappable_type): New function.
1959         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
1960         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
1961         hook.
1962         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
1963         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
1964         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
1965         (struct gimplify_omp_ctx): Add combined_loop field.
1966         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
1967         on stmts inside of target region.
1968         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
1969         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
1970         ORT_TARGET and ORT_TARGET_DATA.
1971         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
1972         Handle GOVD_MAP.
1973         (omp_notice_threadprivate_variable): Complain about threadprivate
1974         variables in target region.
1975         (omp_notice_variable): Complain about vars with non-mappable type
1976         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
1977         (omp_check_private): Ignore ORT_TARGET* regions.
1978         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
1979         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
1980         (find_combined_omp_for): New function.
1981         (gimplify_omp_for): Handle gimplification of combined loops.
1982         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
1983         OMP_TEAMS.
1984         (gimplify_omp_target_update): New function.
1985         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
1986         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
1987         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
1988         (gimplify_body): If fndecl has "omp declare target" attribute, add
1989         implicit ORT_TARGET context around it.
1990         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
1991         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
1992         * tree-nested.c (convert_nonlocal_reference_stmt,
1993         convert_local_reference_stmt, convert_gimple_call): Handle
1994         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
1995         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
1996         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
1997         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
1998         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
1999         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2000         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2001         BUILT_IN_GOMP_LOOP_END_CANCEL,
2002         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2003         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2004         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2005         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2006         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2007         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2008         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2009         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2010         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2011         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2012         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2013         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2014         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2015         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2016         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2017         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2018         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2019         gimple_build_omp_teams): New functions.
2020         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2021         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2022         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2023         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2024         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2025         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2026         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2027         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2028         GF_OMP_ATOMIC_SEQ_CST): New.
2029         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2030         gimple_build_omp_teams): New prototypes.
2031         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2032         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2033         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2034         GIMPLE_OMP_SINGLE as end of range.
2035         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2036         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2037         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2038         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2039         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2040         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2041         gimple_omp_target_kind, gimple_omp_target_set_kind,
2042         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2043         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2044         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2045         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2046         gimple_omp_teams_set_clauses): New inlines.
2047         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2048         and GIMPLE_OMP_TASKGROUP.
2049         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2050         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2051         enum omp_clause_proc_bind_kind): New.
2052         (union omp_clause_subcode): Add depend_kind, map_kind and
2053         proc_bind_kind fields.
2054         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2055         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2056         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2057         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2058         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2059
2060 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2061
2062         * predict.c (tree_estimate_probability): Add new parameter
2063         for estimate_bb_frequencies.
2064         (estimate_bb_frequencies): Add new parameter to force estimation.
2065         (rebuild_frequencies): When max frequency in function is small,
2066         recompute counts from frequencies.
2067         * predict.h (estimate_bb_frequencies): New parameter.
2068
2069 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2070
2071         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2072         optimizations are disabled.
2073
2074 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2075
2076         * coverage.c (coverage_finish): Fix leak of da_file_name.
2077
2078 2013-10-10  Jan Hubicka  <jh@suse.cz>
2079
2080         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2081         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2082         Bobcat and generic.
2083
2084 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2085
2086         PR middle-end/58670
2087         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2088         if any labels are in FALLTHRU_BB, use a special label emitted
2089         immediately after the asm goto insn rather than label_rtx
2090         of the LABEL_DECL.
2091         (expand_asm_stmt): Adjust caller.
2092         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2093         edge if the last insn in predecessor is a jump with single successor,
2094         but it isn't simplejump_p.
2095
2096 2013-10-10  Richard Biener  <rguenther@suse.de>
2097
2098         PR tree-optimization/58656
2099         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2100
2101 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2102
2103         * gimplify.c: Include expr.h and tm_p.h for targets with special
2104         va-arg padding requirements.
2105
2106 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2107
2108         * tree-flow.h: Move some prototypes to gimple.h.
2109         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2110         * gimple.h: Relocate some prototypes from tree-flow.h
2111         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2112         Move to gimplify.c.
2113         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2114         (build_va_arg_indirect_ref): Relocate and make static.
2115         (std_gimplify_va_arg_expr): Relocate here.
2116         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2117         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2118
2119 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2120
2121         * doc/md.texi: Document the mnemonic attribute.
2122
2123 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2124
2125         PR target/57377
2126         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2127         (set_attr_alternative x ...) when searching for user defined
2128         mnemonic attribute.
2129
2130 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2131
2132         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2133
2134 2013-10-09  Easwaran Raman <eraman@google.com>
2135
2136         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2137         * cfgexpand.c (defer_stack_allocation): ...use here
2138         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2139
2140 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2141
2142         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2143         (operand_equal_for_value_replacement): New function, extracted from
2144         value_replacement and enhanced to catch more cases.
2145         (value_replacement): Use operand_equal_for_value_replacement.
2146
2147 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2148
2149         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2150         get_max_loop_iterations.
2151
2152 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2153
2154         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2155         Place comment above function.
2156
2157 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2158
2159         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2160         are not related to tree-cfg.c.
2161         * tree-ssa-address.h: New file.  Relocate prototypes.
2162         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2163         (addr_for_mem_ref): New.  Combine call to get_address_description and
2164         return addr_for_mem_ref.
2165         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2166         * tree-ssa-live.h: Adjust prototypes.
2167         * passes.c: Include tree-ssa-live.h.
2168         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2169         * graphite.c (graphite_transform_loops): Make static.
2170         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2171         make_pass_graphite, pass_data_graphite_transforms,
2172         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2173         * ipa-pure-const.c (warn_function_noreturn): Make static.
2174         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2175         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2176         Relocate from tree-cfg.c
2177         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2178         static.
2179         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2180         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2181         Move to ipa-pure-const.c.
2182         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2183         Relocate from tree-optimize.c.
2184         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2185         make_pass_fixup_cfg): Move to tree-cfg.c.
2186         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2187         Relocate some prototypes.
2188         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2189         * tree-dump.c (dump_function_to_file): Remove prototype.
2190         Add tree-flow.h to the include file.
2191         * tree-dump.h: Remove prototype.
2192         * tree-parloops.h: New File.  Add prototypes.
2193         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2194         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2195         from tree-ssa-loop.c.
2196         * tree-predcom.c (run_tree_predictive_commoning,
2197         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2198         Relocate here from tree-ssa-loop.c.
2199         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2200         ssa_name_values.release ().
2201         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2202         (ssa_name_values): Relocate from tree-flow.h.
2203         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2204         * tree-ssa-loop.c (run_tree_predictive_commoning,
2205         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2206         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2207         make_pass_graphite, pass_data_graphite_transforms,
2208         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2209         tree_parallelize_loops, pass_data_parallelize_loops,
2210         make_pass_parallelize_loops): Move to other files.
2211         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2212         moved here.
2213         * tree.h: Remove prototypes from tree-address.c.
2214
2215 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2216
2217         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2218         (struct int_tree_map): Move to tree-hasher.h
2219         (SCALE, LABEL, PERCENT): Move to gimple.h
2220         * tree-flow-inline.h: Delete.  Move functions to other files.
2221         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2222         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2223         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2224         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2225         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2226         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2227         * tree-hasher.h: Don't include tree-flow.h.
2228         (struct int_tree_map): Relocate from tree-flow.h.
2229         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2230         tree-flow-inline.h and make static.
2231         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2232         tree-flow-inline.h.
2233         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2234         tree-flow-inline.h and make static.
2235         * tree.h (is_global_var, may_be_aliased): Relocate from
2236         tree-flow-inline.h.
2237         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2238         * value-prof.c: No longer include tree-flow-inline.h.
2239         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2240
2241 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2242
2243         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2244         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2245         (enum move_pos): Move to tree-ssa-loop-im.h
2246         * cfgloop.h: Move some prototypes.
2247         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2248         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2249         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2250         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2251         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2252         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2253         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2254         (enum move_pos): Relocate here.
2255         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2256         tree-ssa-loop.c.
2257         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2258         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2259         make_pass_lim): Relocate here from tree-ssa-loop.c.
2260         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2261         tree-ssa-loop.c.
2262         (loop_edge_to_cancel, unloop_loops): Make static.
2263         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2264         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2265         (tree_complete_unroll, gate_tree_complete_unroll,
2266         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2267         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2268         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2269         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2270         (stmt_invariant_in_loop_p): Remove unused function.
2271         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2272         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2273         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2274         (gcov_type_to_double_int): Move to cfgloop.h.
2275         (double_int_cmp, bound_index,
2276         estimate_numbers_of_iterations_loop): Make static.
2277         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2278         (max_loop_iterations): Factor out get_max_loop_iterations.
2279         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2280         cfgloop.c.
2281         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2282         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2283         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2284         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2285         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2286         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2287         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2288         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2289         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2290         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2291         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2292         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2293         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2294         pass_data_complete_unroll, make_pass_complete_unroll,
2295         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2296         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2297         tree-ssa-loop-ivcanon.c.
2298         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2299         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2300         tree-ssa-loop-prefetch.c.
2301         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2302         tree-ssa-loop-im.c.
2303         (get_lsm_tmp_name): Relocate and add suffix parameter.
2304         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2305         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2306         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2307         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2308         (get_estimated_loop_iterations): Factor out accessor from
2309         estimated_loop_iterations in tree-ssa-loop-niter.c.
2310         (get_max_loop_iterations): Factor out accessor from
2311         _max_loop_iterations in tree-ssa-niter.c.
2312         * loop-unroll.c (decide_unroll_constant_iterations,
2313         decide_unroll_runtime_iterations, decide_peel_simple,
2314         decide_unroll_stupid): Use new get_* accessors.
2315
2316 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2317
2318         PR tree-optimization/20318
2319         * doc/extend.texi (returns_nonnull): New function attribute.
2320         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2321         attribute.
2322         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2323         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2324
2325 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2326
2327         PR middle-end/58570
2328         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2329         false if both components are bitfields.
2330
2331 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2332
2333         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2334         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2335         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2336         * config/aarch64/aarch64.h
2337         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2338         * config/aarch64/aarch64-simd-builtins.def
2339         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2340
2341 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2342
2343         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2344
2345 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2346
2347         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2348
2349 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2350
2351         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2352         (vneg_f64): New intrinsic.
2353         (vneg_s8): Asm replaced with C.
2354         (vneg_s16): Likewise.
2355         (vneg_s32): Likewise.
2356         (vneg_s64): New intrinsic.
2357         (vnegq_f32): Asm replaced with C.
2358         (vnegq_f64): Likewise.
2359         (vnegq_s8): Likewise.
2360         (vnegq_s16): Likewise.
2361         (vnegq_s32): Likewise.
2362         (vnegq_s64): Likewise.
2363
2364 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2365
2366         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2367
2368 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2369
2370         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2371         packed stack special handling.
2372         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2373         back to fixed stack slots for FPRs saved due to stdarg.
2374
2375 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2376
2377         * config/s390/s390.c (s390_frame_info): Restructure function.
2378
2379 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2380
2381         * config/s390/s390.c (struct s390_frame_layout): New field
2382         gpr_save_slots.
2383         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2384         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2385         regs_ever_clobbered to char*.
2386         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2387         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2388         (s390_register_info): Enable FPR save slots.  Move/Copy some
2389         functionality into ...
2390         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2391         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2392         function.
2393         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2394         not imply a stack frame.
2395         (s390_init_frame_layout): Remove variable clobbered_regs.
2396         (s390_update_register_info): Remove function.
2397         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2398         cannot be used for register renaming.
2399         (s390_hard_regno_scratch_ok): New function.
2400         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2401         (s390_initial_elimination_offset): Change offset calculation of
2402         the return address pointer.
2403         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2404         (restore_gprs): Set frame related flag.
2405         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2406         (s390_emit_prologue): Call s390_register_info instead of
2407         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2408         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2409         (s390_optimize_prologue): Call s390_optimize_register_info.
2410         Try to remove also FPR slot save/restore INSNs.  Remove frame
2411         related flags from restore INSNs.
2412
2413 2013-10-08  DJ Delorie  <dj@redhat.com>
2414
2415         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2416         (movhi): Likewise.
2417
2418         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2419         avoid conflict with the MI use of %c.
2420         * config/rl78/rl78-real.md: change %c to %C throughout.
2421         * config/rl78/rl78-virt.md: Likewise.
2422
2423 2013-10-08  Jan Hubicka  <jh@suse.cz>
2424
2425         * config/i386/i386.c (ix86_option_override_internal): Switch
2426         to SSE math for -ffast-math when target ISA supports SSE2.
2427
2428 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2429
2430         * tree-flow.h: Remove some prototypes.
2431         * tree.h: Remove some protypes, add a couple.
2432         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2433         using_eh_for_cleanups_p): Add interface routines for front ends.
2434         * tree-eh.h: New file.  Add protoptyes.
2435         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2436         (add_stmt_to_eh_lp_fn): Make static.
2437         (lower_try_finally): Use new using_eh_for_cleanups_p.
2438         * emit-rtl.c: Include tree-eh.h.
2439         * gimple.h: Include tree-eh.h.
2440
2441 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2442
2443         PR tree-optimization/58480
2444         * tree-vrp.c (infer_nonnull_range): New function.
2445         (infer_value_range): Call infer_nonnull_range.
2446
2447 2013-10-08  Dehao Chen  <dehao@google.com>
2448
2449         PR tree-optimization/58619
2450         * tree-inline.c (copy_phis_for_bb): Combine location data
2451         only if non-null.
2452
2453 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2454
2455         PR target/58423
2456         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2457         RTX_FRAME_RELATED_P on INSN.
2458
2459 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2460
2461         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2462         (altivec_expand_vec_perm_const): Call it.
2463
2464 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2465
2466         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2467         sequences for LE VSX loads and stores at expand time.
2468         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2469         prototype.
2470         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2471         (rs6000_gen_le_vsx_permute): New.
2472         (rs6000_gen_le_vsx_load): New.
2473         (rs6000_gen_le_vsx_store): New.
2474         (rs6000_gen_le_vsx_move): New.
2475         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2476         (*vsx_le_perm_load_v4si): New.
2477         (*vsx_le_perm_load_v8hi): New.
2478         (*vsx_le_perm_load_v16qi): New.
2479         (*vsx_le_perm_store_v2di): New.
2480         (*vsx_le_perm_store_v4si): New.
2481         (*vsx_le_perm_store_v8hi): New.
2482         (*vsx_le_perm_store_v16qi): New.
2483         (*vsx_xxpermdi2_le_<mode>): New.
2484         (*vsx_xxpermdi4_le_<mode>): New.
2485         (*vsx_xxpermdi8_le_V8HI): New.
2486         (*vsx_xxpermdi16_le_V16QI): New.
2487         (*vsx_lxvd2x2_le_<mode>): New.
2488         (*vsx_lxvd2x4_le_<mode>): New.
2489         (*vsx_lxvd2x8_le_V8HI): New.
2490         (*vsx_lxvd2x16_le_V16QI): New.
2491         (*vsx_stxvd2x2_le_<mode>): New.
2492         (*vsx_stxvd2x4_le_<mode>): New.
2493         (*vsx_stxvd2x8_le_V8HI): New.
2494         (*vsx_stxvd2x16_le_V16QI): New.
2495
2496 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2497
2498         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2499
2500 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2501
2502         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2503         loop to work also for 31bit ABI.
2504         Save the stack pointer for frame_size > 0.
2505
2506 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2507
2508         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2509         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2510         constraint letters from expanders.
2511         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2512         retry count to general_operand.
2513         ("tabort"): Give operand 0 a mode.
2514         ("tabort_1"): Add mode and constraint letter for operand 0.
2515         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2516
2517 2013-10-04  Jeff Law  <law@redhat.com>
2518
2519         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2520
2521         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2522         out of ...
2523         (thread_across_edge): Here.  Call it.
2524
2525 2013-10-04  Cary Coutant  <ccoutant@google.com>
2526
2527         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2528         discarding a location list expression (or a piece of one).
2529
2530 2013-10-03  Jan Hubicka  <jh@suse.cz>
2531
2532         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2533         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2534         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2535
2536 2013-10-03  Jan Hubicka  <jh@suse.cz>
2537
2538         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2539         accumulate-outgoing-args when producing unwind info.
2540
2541 2013-10-03  Wei Mi  <wmi@google.com>
2542
2543         * lra-constraints.c (insert_move_for_subreg): New function
2544         extracted from simplify_operand_subreg.
2545         (simplify_operand_subreg): Add reload for paradoxical subreg.
2546
2547 2013-10-03  Rong Xu  <xur@google.com>
2548
2549         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2550         the candidate of builtin_expect such that we should fix the
2551         size/time estimation.
2552         (estimate_function_body_sizes): Do the acutally size/time fix-up
2553         for builtin_expect.
2554
2555 2013-10-03  Rong Xu  <xur@google.com>
2556
2557         * predict.c (tree_predict_by_opcode): Get the probability
2558         for builtin_expect from param builtin_expect_probability.
2559         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2560         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2561         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2562
2563 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2564
2565         PR c++/19476
2566         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2567         * calls.c (alloca_call_p): Use get_callee_fndecl.
2568         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2569         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2570         Likewise.
2571         (vrp_visit_stmt): Remove duplicated code.
2572
2573 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2574
2575         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2576         ceildf2, btruncdf2, instead of vsx_* name.
2577
2578         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2579         iterators to only do V2DF and V4SF here.  Move the DF code to
2580         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2581         just 'v' since only vector operations are handled with these insns
2582         after moving the DF support to rs6000.md.
2583         (vsx_sub<mode>3): Likewise.
2584         (vsx_mul<mode>3): Likewise.
2585         (vsx_div<mode>3): Likewise.
2586         (vsx_fre<mode>2): Likewise.
2587         (vsx_neg<mode>2): Likewise.
2588         (vsx_abs<mode>2): Likewise.
2589         (vsx_nabs<mode>2): Likewise.
2590         (vsx_smax<mode>3): Likewise.
2591         (vsx_smin<mode>3): Likewise.
2592         (vsx_sqrt<mode>2): Likewise.
2593         (vsx_rsqrte<mode>2): Likewise.
2594         (vsx_fms<mode>4): Likewise.
2595         (vsx_nfma<mode>4): Likewise.
2596         (vsx_copysign<mode>3): Likewise.
2597         (vsx_btrunc<mode>2): Likewise.
2598         (vsx_floor<mode>2): Likewise.
2599         (vsx_ceil<mode>2): Likewise.
2600         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2601         (vsx_sminsf3): Likewise.
2602         (vsx_fmadf4): Likewise.
2603         (vsx_fmsdf4): Likewise.
2604         (vsx_nfmadf4): Likewise.
2605         (vsx_nfmsdf4): Likewise.
2606         (vsx_cmpdf_internal1): Likewise.
2607
2608         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2609         simpler to select whether a target has SPE or traditional floating
2610         point support in iterators.
2611         (TARGET_DF_SPE): Likewise.
2612         (TARGET_SF_FPR): Likewise.
2613         (TARGET_DF_FPR): Likewise.
2614         (TARGET_SF_INSN): Macros to say whether floating point support
2615         exists for a given operation for expanders.
2616         (TARGET_DF_INSN): Likewise.
2617
2618         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2619         combining of SF/DF mode operations, using both traditional and VSX
2620         registers.
2621         (Fvsx): Likewise.
2622         (Ff): Likewise.
2623         (Fv): Likewise.
2624         (Fs): Likewise.
2625         (Ffre): Likewise.
2626         (FFRE): Likewise.
2627         (abs<mode>2): Combine SF/DF modes using traditional floating point
2628         instructions.  Add support for using the upper DF registers with
2629         VSX support, and SF registers with power8-vector support.  Update
2630         expanders for operations supported by both the SPE and traditional
2631         floating point units.
2632         (abs<mode>2_fpr): Likewise.
2633         (nabs<mode>2): Likewise.
2634         (nabs<mode>2_fpr): Likewise.
2635         (neg<mode>2): Likewise.
2636         (neg<mode>2_fpr): Likewise.
2637         (add<mode>3): Likewise.
2638         (add<mode>3_fpr): Likewise.
2639         (sub<mode>3): Likewise.
2640         (sub<mode>3_fpr): Likewise.
2641         (mul<mode>3): Likewise.
2642         (mul<mode>3_fpr): Likewise.
2643         (div<mode>3): Likewise.
2644         (div<mode>3_fpr): Likewise.
2645         (sqrt<mode>3): Likewise.
2646         (sqrt<mode>3_fpr): Likewise.
2647         (fre<Fs>): Likewise.
2648         (rsqrt<mode>2): Likewise.
2649         (cmp<mode>_fpr): Likewise.
2650         (smax<mode>3): Likewise.
2651         (smin<mode>3): Likewise.
2652         (smax<mode>3_vsx): Likewise.
2653         (smin<mode>3_vsx): Likewise.
2654         (negsf2): Delete SF operations that are merged with DF.
2655         (abssf2): Likewise.
2656         (addsf3): Likewise.
2657         (subsf3): Likewise.
2658         (mulsf3): Likewise.
2659         (divsf3): Likewise.
2660         (fres): Likewise.
2661         (fmasf4_fpr): Likewise.
2662         (fmssf4_fpr): Likewise.
2663         (nfmasf4_fpr): Likewise.
2664         (nfmssf4_fpr): Likewise.
2665         (sqrtsf2): Likewise.
2666         (rsqrtsf_internal1): Likewise.
2667         (smaxsf3): Likewise.
2668         (sminsf3): Likewise.
2669         (cmpsf_internal1): Likewise.
2670         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2671         (negdf2): Delete DF operations that are merged with SF.
2672         (absdf2): Likewise.
2673         (nabsdf2): Likewise.
2674         (adddf3): Likewise.
2675         (subdf3): Likewise.
2676         (muldf3): Likewise.
2677         (divdf3): Likewise.
2678         (fred): Likewise.
2679         (rsqrtdf_internal1): Likewise.
2680         (fmadf4_fpr): Likewise.
2681         (fmsdf4_fpr): Likewise.
2682         (nfmadf4_fpr): Likewise.
2683         (nfmsdf4_fpr): Likewise.
2684         (sqrtdf2): Likewise.
2685         (smaxdf3): Likewise.
2686         (smindf3): Likewise.
2687         (cmpdf_internal1): Likewise.
2688         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2689         (btrunc<mode>2): Delete separate expander, and combine with the
2690         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2691         (btrunc<mode>2_fpr): Likewise.
2692         (ceil<mode>2): Likewise.
2693         (ceil<mode>2_fpr): Likewise.
2694         (floor<mode>2): Likewise.
2695         (floor<mode>2_fpr): Likewise.
2696         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2697         Add support for using the upper registers with VSX and
2698         power8-vector.  Move insns to be closer to the define_expands. On
2699         VSX systems, prefer the traditional form of FMA over the VSX
2700         version, since the traditional form allows the target not to
2701         overlap with the inputs.
2702         (fms<mode>4_fpr): Likewise.
2703         (nfma<mode>4_fpr): Likewise.
2704         (nfms<mode>4_fpr): Likewise.
2705
2706 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2707             Richard Earnshaw  <richard.earnshaw@arm.com>
2708
2709         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2710         (struct mult_cost_table): Likewise.
2711         (struct mem_cost_table): Likewise.
2712         (struct fp_cost_table): Likewise.
2713         (struct vector_cost_table): Likewise.
2714         (cpu_cost_table): Likewise.
2715         * config/arm/arm.opt (mold-rts-costs): New option.
2716         (mnew-generic-costs): Likewise.
2717         * config/arm/arm.c (generic_extra_costs): New table.
2718         (cortexa15_extra_costs): Likewise.
2719         (arm_slowmul_tune): Use NULL as new costs.
2720         (arm_fastmul_tune): Likewise.
2721         (arm_strongarm_tune): Likewise.
2722         (arm_xscale_tune): Likewise.
2723         (arm_9e_tune): Likewise.
2724         (arm_v6t2_tune): Likewise.
2725         (arm_cortex_a5_tune): Likewise.
2726         (arm_cortex_a9_tune): Likewise.
2727         (arm_v6m_tune): Likewise.
2728         (arm_fa726te_tune): Likewise.
2729         (arm_cortex_a15_tune): Use cortex15_extra_costs.
2730         (arm_cortex_tune): Use generict_extra_costs.
2731         (shifter_op_p): New function.
2732         (arm_unspec_cost): Likewise.
2733         (LIBCALL_COST): Define.
2734         (arm_new_rtx_costs): New function.
2735         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
2736         table is available. Use old costs otherwise unless mnew-generic-costs
2737         is specified.
2738         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
2739         (cpu_cost_table): Declare.
2740
2741 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2742
2743         PR target/58460
2744         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
2745         (*subs_mul_imm_<mode>)
2746         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
2747         (*sub_<shift>_<mode>)
2748         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
2749         Remove k constraint.
2750
2751 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
2752
2753         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
2754         code.
2755         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
2756
2757 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2758
2759         * predict.c (probably_never_executed): New function.
2760         (probably_never_executed_bb_p): Invoke probably_never_executed.
2761         (probably_never_executed_edge_p): Ditto.
2762         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
2763         Treat profile insanities conservatively.
2764
2765 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
2766
2767         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
2768
2769 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
2770
2771         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
2772         use it.  Use smaller increase for scratch.  Don't increase reject
2773         for early clobber scratch.
2774         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
2775         setting eliminated regs except setting fp from hfp.
2776         (lra_eliminate): Check lra_insn_recog_data on NULL.
2777
2778 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2779
2780         PR target/58587
2781         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
2782         setting -mvsx-timode by default until the underlying problem is fixed.
2783         (RS6000_CPU, power7 defaults): Likewise.
2784
2785 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2786
2787         * config/x-linux (host-linux.o): Remove header dependencies.
2788         Use $(COMPILE) and $(POSTCOMPILE).
2789         * config/t-linux-android (linux-android.o): Ditto.
2790
2791 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2792
2793         * Makefile.in (expmed.o-warn): Remove.
2794
2795 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2796
2797         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
2798         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
2799
2800 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2801
2802         * dojump.c (do_jump_1): Divide probability between
2803         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
2804
2805 2013-10-02  Tom Tromey  <tromey@redhat.com>
2806
2807         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
2808
2809 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2810
2811         * tree-flow.h: Remove some prototypes.
2812         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
2813         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
2814         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
2815         mark_virtual_phi_result_for_renaming): Relocate here.
2816         * tree-into-ssa.h: Add prototypes.
2817         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
2818         single_pred_before_succ_order.
2819         (blocks_in_phiopt_order): Rename and move to cfganal.c.
2820         (nonfreeing_call_p) Move to gimple.c.
2821         * cfganal.c (single_pred_before_succ_order): Move and renamed from
2822         tree-ssa-phiopt.c.
2823         * basic-block.h (single_pred_before_succ_order): Add prototype.
2824         * gimple.c (nonfreeing_call_p): Relocate here.
2825         * gimple.h: Add prototype.
2826         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
2827         * tree-ssa-dom.h: New file.  Relocate prototypes here.
2828         * tree-ssa.h: Include tree-ssa-dom.h.
2829
2830 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
2831
2832         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
2833         Use $(COMPILE) and $(POSTCOMPILE).
2834
2835         * config/alpha/x-alpha (driver-alpha.o): Ditto.
2836
2837 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2838
2839         * tree-flow.h: Remove some prototypes.
2840         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
2841         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
2842         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
2843         propagate_tree_value*): Move from here to...
2844         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
2845         propagate_tree_value*): Relocate here.
2846         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
2847         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
2848         * gimple-fold.c: Remove gimple-fold.h from include list.
2849         * tree-vrp.c: Remove gimple-fold.h from include list.
2850         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
2851         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
2852         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
2853         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
2854         * sese.c: Add tree-ssa-propagate.h to include list.
2855
2856 2013-10-02  Richard Biener  <rguenther@suse.de>
2857
2858         * tree-loop-distribution.c: Include tree-vectorizer.h for
2859         find_loop_location.
2860         (enum partition_kind): Remove PKIND_REDUCTION.
2861         (struct partition_s): Remove has_writes member, add reduction_p member.
2862         (partition_alloc): Adjust.
2863         (partition_builtin_p): Likewise.
2864         (partition_has_writes): Remove.
2865         (partition_reduction_p): New function.
2866         (partition_merge_into): Likewise.
2867         (generate_code_for_partition): Commonize builtin partition
2868         handling tail.
2869         (rdg_cannot_recompute_vertex_p): Remove.
2870         (already_processed_vertex_p): Likewise.
2871         (rdg_flag_vertex): Do not set has_writes.
2872         (classify_partition): Adjust.
2873         (rdg_build_partitions): Do not set has_writes, treat all
2874         partitions as useful.
2875         (distribute_loop): Record number of library calls generated.  Adjust.
2876         (tree_loop_distribution): Report number of loops and library
2877         calls generated as opt-info.
2878
2879 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
2880
2881         * tree-flow.h: Include new .h files.  Move prototypes.
2882         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
2883         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
2884         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
2885         * tree-pretty-print.h: Add prototypes from tree-flow.h.
2886         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
2887         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
2888         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
2889         * tree.h (get_ref_base_and_extent): Move prototype out.
2890         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
2891         tree-dfa.h.
2892         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
2893         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
2894         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
2895         * tree-scalar-evolution.c: Include tree.h.
2896         * sese.c: Include tree.h.
2897         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
2898         * dwarf2out.c: Include tree-dfa.h.
2899         * tree-chrec.c: Include tree.h.
2900         * tree-data-ref.c: Include tree.h.
2901
2902 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
2903
2904         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
2905         Fix whitespace.
2906
2907 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
2908
2909         * config/t-sol2 (sol2-c.o): Remove header dependencies.
2910         Use $(COMPILE) and $(POSTCOMPILE).
2911         (sol2-cxx.o): Likewise.
2912         (sol2-stubs.o): Likewise.
2913         (sol2.o): Likewise.
2914         * config/x-solaris (host-solaris.o): Likewise.
2915
2916         * config/sparc/t-sparc (sparc.o): Remove.
2917         (sparc-c.o): Remove header dependencies.
2918         Use $(COMPILE) and $(POSTCOMPILE).
2919         * config/sparc/x-sparc: Likewise.
2920
2921 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
2922
2923         * config/arc/arc-opts.h: Add 2013 to Copyright years.
2924         * config/arc/arc700.md: Likewise.
2925         * config/arc/arc-modes.def: Likewise.
2926         * config/arc/arc-simd.h: Likewise.
2927         * config/arc/t-arc-uClibc: Likewise.
2928         * config/arc/t-arc-newlib: Likewise.
2929
2930 2013-10-02  Renlin Li  <renlin.li@arm.com>
2931
2932         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2933         plus_constant.
2934         (aarch64_expand_epilogue): Likewise.
2935
2936 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2937             Yufeng Zhang  <yufeng.zhang@arm.com>
2938
2939         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
2940         declaration.
2941         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
2942         'base_in' represent a conversion and legal_cast_p_1 holds; set
2943         'base_in' with the returned value from get_unwidened.
2944
2945 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2946
2947         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
2948         plus_constant is not used.
2949
2950 2013-10-01  Wei Mi  <wmi@google.com>
2951
2952         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
2953         * config/i386/i386.md: Add define_peephole2 to
2954         break partial reg stall for cvtss2sd/cvtsd2ss.
2955
2956 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2957
2958         * config/arc/arc.c (pass_arc_ifcvt::clone):
2959         Update for ctxt_ -> m_ctxt change.
2960
2961 2013-10-01  Jeff Law  <law@redhat.com>
2962
2963         * tree-ssa-threadupdate.c (struct redirection_data): Delete
2964         outgoing_edge and intermediate_edge fields.  Instead store the path.
2965         (redirection_data::hash): Hash on the last edge's destination index.
2966         (redirection_data::equal): Check the entire thread path.
2967         (lookup_redirectio_data): Corresponding changes.
2968         (create_edge_and_update_destination_phis): Likewise.
2969         (thread_single_edge): Likewise.
2970
2971 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2972             Diego Novillo <dnovillo@google.com>
2973
2974         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
2975         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
2976         (vld32wh_insn, vld32wl_insn): Delete commented-out old
2977         versions of these patterns.
2978
2979         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
2980         (__builtin_arc_aligned): Likewise.
2981
2982         * config/arc/arc.md: Expand adc_0 comment stating the intended
2983         purpose and why it isn't ready.
2984         Replace commented out call_value_via_label_mixed with a
2985         plain comment about bl_s.
2986
2987         * config/arc/arc.c (stdio.h): Don't include directly.
2988         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
2989         Remove if (1) condition.
2990         (arc_encode_section_info): Fix comment.
2991
2992 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2993
2994         * config/arc/arc.c (arc_conditional_register_usage):
2995         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
2996         Also set reg_alloc_order for DMA config regs.
2997
2998 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
2999             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3000
3001         * doc/install.texi (--with-cpu): Mention ARC.
3002         (arc-*-elf32): New paragraph.
3003         (arc-linux-uclibc): Likewise.
3004         * doc/md.texi (Machine Constraints): Add ARC part.
3005         * doc/invoke.texi: (menu): Add ARC Options.
3006         (Machine Dependent Options) <ARC Options>: Add synopsis.
3007         (node ARC Options): Add.
3008         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3009         (ARC Built-in Functions): New section defining
3010         generic ARC built-in functions.
3011         (ARC SIMD Built-in Functions): New section defining SIMD specific
3012         built-in functions.
3013         (Declaring Attributes of Functions): Extended
3014         description of short_call and long_call attributes for ARC and
3015         added index entries.
3016
3017 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3018             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3019             Joern Rennecke  <joern.rennecke@embecosm.com>
3020             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3021             Brendan Kehoe  <brendan@zen.org>
3022             Michael Eager  <eager@eagercon.com>
3023             Simon Cook  <simon.cook@embecosm.com>
3024             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3025
3026         * config/arc, common/config/arc: New directories.
3027
3028 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3029             Brendan Kehoe  <brendan@zen.org>
3030             Simon Cook  <simon.cook@embecosm.com>
3031
3032         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3033
3034 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3035
3036         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3037         * tree-ssa-coalesce.h: New. Move prototype to here.
3038         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3039         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3040         (gimple_can_coalesce_p): Move to...
3041         * gimple.c (gimple_can_coalesce_p): Here.
3042
3043 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3044
3045         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3046         (dump_decl_set): Move to gimple.c.
3047         * gimple.h: Don't include tree-ssa-operands.h.
3048         (dump_decl_set): Add prototype.
3049         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3050         Move to gimple-ssa.h.
3051         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3052         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3053         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3054         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3055         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3056         than PHI_ARG_DEF.
3057         (dump_decl_set): Relocate here.
3058         * gimple-ssa.h: New file.
3059         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3060         Relocate from gimple.h.
3061         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3062         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3063         swap_tree_operands and remove non-ssa path.
3064         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3065         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3066         swap_ssa_operands.
3067         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3068         vect_is_simple_reduction_1): Use swap_ssa_operands.
3069         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3070         (enum need_phi_state): Move to tree-into-ssa.c.
3071         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3072         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3073         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3074         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3075         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3076         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3077         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3078         num_imm_uses): Move to ssa-iterators.h.
3079         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3080         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3081         tree-phinodes.h.
3082         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3083         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3084         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3085         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3086         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3087         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3088         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3089         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3090         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3091         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3092         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3093         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3094         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3095         (set_phi_nodes): Move to tree-phinodes.h.
3096         * tree-ssa-operands.h (enum ssa_op_iter_type,
3097         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3098         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3099         (dump_decl_set): Remove prototype.
3100         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3101         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3102         (set_phi_nodes): Relocate from tree-flow-inline.h.
3103         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3104         tree-flow-inline.h
3105         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3106         include list.  Temporarily add gimple.h to include list.
3107         * ssa-iterators.h: New file.
3108         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3109         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3110         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3111         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3112         Relocate from tree-ssa-operands.h.
3113         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3114         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3115         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3116         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3117         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3118         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3119         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3120         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3121         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3122         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3123         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3124         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3125         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3126         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3127         Relocate from tree-flow-inline.h.
3128         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3129
3130 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3131
3132         * aarch64-simd.md
3133         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3134         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3135         Insert '\t' to output template.
3136         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3137         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3138         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3139         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3140
3141 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3142
3143         * doc/install.texi (Host/target specific installation notes for GCC):
3144         Put @anchor before @heading.
3145         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3146         Use @email for email addresses.
3147
3148 2013-10-01  Jeff Law  <law@redhat.com>
3149
3150         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3151         a vec.  Only delete the path if we create one without successfully
3152         registering a jump thread.
3153         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3154         as a pointer.
3155         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3156         (paths): New vector of jump threading paths.
3157         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3158         (THREAD_PATH): New accessor macro for the entire thread path.
3159         (lookup_redirection_data): Get intermediate and final outgoing edge
3160         from the thread path.
3161         (create_edge_and_update_destination_phis): Copy the threading path.
3162         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3163         jump threading path.
3164         (ssa_redirect_edges): Get edges and block types from the jump threading
3165         path.  Free the path vector.
3166         (thread_block): Get edges from the jump threading path.  Look at the
3167         entire path to see if we thread to a loop exit.  If we cancel a jump
3168         thread request, then free the path vector.
3169         (thread_single_edge): Get edges and block types from the jump threading
3170         path.  Free the path vector.
3171         (thread_through_loop_header): Get edges and block types from the jump
3172         threading path.  Free the path vector.
3173         (mark_threaded_blocks): Iterate over the vector of paths and store
3174         the path on the appropriate edge.  Get edges and block types from the
3175         jump threading path.
3176         (mark_threaded_blocks): Get edges and block types from the jump
3177         threading path.  Free the path vector.
3178         (thread_through_all_blocks): Use the vector of paths rather than
3179         a vector of 3-edge sets.
3180         (register_jump_thread): Accept pointer to a path vector rather
3181         than the path vector itself.  Store the path vector for later use.
3182         Simplify.
3183
3184 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3185             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3186
3187         PR target/58574
3188         * config/s390/s390.c (s390_split_branches): Modify check for table
3189         jump insns.
3190         (s390_chunkify_start): Rearrange table jump insn check in order to
3191         deal with compare and branch insns correctly.
3192
3193 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3194
3195         PR target/58578
3196         Revert
3197         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3198         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3199         define_insn_and_split.
3200         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3201         (shiftsi3_compare): New pattern.
3202         (rrx): New pattern.
3203         * config/arm/unspecs.md (UNSPEC_RRX): New.
3204
3205 2013-10-01  Alan Modra  <amodra@gmail.com>
3206
3207         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3208         casing inout operands.
3209
3210 2013-10-01  Richard Biener  <rguenther@suse.de>
3211
3212         PR tree-optimization/58553
3213         * tree-loop-distribution.c (struct partition_s): Add niter member.
3214         (classify_partition): Populate niter member for the partition
3215         and properly identify whether the relevant store happens before
3216         or after the loop exit.
3217         (generate_memset_builtin): Use niter member from the partition.
3218         (generate_memcpy_builtin): Likewise.
3219
3220 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3221
3222         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3223         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3224
3225 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3226
3227         * basic-block.h (edge_list): Prefix member names with "m_".
3228         * context.h (context): Likewise.
3229         * domwalk.h (dom_walker): Likewise.
3230         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3231         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3232         * hash-table.h (hash_table): Likewise.
3233         * machmode.h (bit_field_mode_iterator): Likewise.
3234         * pass_manager.h (pass_list): Likewise.
3235         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3236         * tree-pass.h (pass_data): Likewise.
3237         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3238         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3239         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3240         * asan.c (pass_data_asan): Update accordingly.
3241         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3242         (control_dependences::control_dependences): Likewise.
3243         (control_dependences::~control_dependences): Likewise.
3244         (control_dependences::~control_dependences): Likewise.
3245         (control_dependences::get_edges_dependent_on): Likewise.
3246         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3247         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3248         * context.c (gcc::context::context): Likewise.
3249         * cprop.c (pass_rtl_cprop::clone): Likewise.
3250         * domwalk.c (dom_walker::walk): Likewise.
3251         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3252         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3253         * mode-switching.c (pass_mode_switching::clone): Likewise.
3254         * passes.c (opt_pass::opt_pass): Likewise.
3255         (pass_manager::pass_manager): Likewise.
3256         * predict.c (pass_strip_predict_hints::clone): Likewise.
3257         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3258         (pass_split_all_insns::clone): Likewise.
3259         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3260         Likewise.
3261         (bit_field_mode_iterator::next_mode): Likewise.
3262         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3263         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3264         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3265         * tree-complex.c (pass_lower_complex::clone): Likewise.
3266         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3267         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3268         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3269         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3270         (pass_fold_builtins::clone): Likewise.
3271         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3272         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3273         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3274         (pass_cd_dce::clone): Likewise.
3275         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3276         (pass_phi_only_cprop::clone): Likewise.
3277         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3278         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3279         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3280         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3281         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3282         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3283         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3284         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3285         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3286         * tree-vrp.c (pass_vrp::clone): Likewise.
3287         * tsan.c (pass_tsan::clone): Likewise.
3288
3289 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3290
3291         PR middle-end/58564
3292         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3293         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3294
3295         PR middle-end/58564
3296         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3297         optimization, punt if sign_bit_p looked through any zero extension.
3298
3299 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3300
3301         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3302         Update redirected out edge count in joiner case.
3303         (ssa_redirect_edges): Common the joiner and non-joiner cases
3304         so that joiner case gets profile updates.
3305
3306 2013-09-30  Richard Biener  <rguenther@suse.de>
3307
3308         PR tree-optimization/58554
3309         * tree-loop-distribution.c (classify_partition): Require
3310         unconditionally executed stores for memcpy and memset recognition.
3311         (tree_loop_distribution): Calculate dominance info.
3312
3313 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3314
3315         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3316         (NO_PROFILE_COUNTERS): Likewise.
3317         (PROFILE_HOOK): Likewise.
3318         (FUNCTION_PROFILER): Likewise.
3319         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3320
3321 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3322
3323         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3324         calls and defines in TARGET_MACHO conditional.
3325         (load_macho_picbase_di): Likewise.
3326         (reload_macho_picbase): Likewise.
3327         (reload_macho_picbase_si): Likewise.
3328         (reload_macho_picbase_di): Likewise.
3329         (nonlocal_goto_receiver): Likewise.
3330
3331 2013-09-30  Nick Clifton  <nickc@redhat.com>
3332
3333         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3334         that use the MSP430X ISA.
3335         (msp430_option_override): Scan -mmcu command line option for any
3336         MCU name that supports the MSP430X ISA.
3337         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3338         -mmcu options which enable the MSP430X ISA.
3339
3340 2013-09-30  Richard Biener  <rguenther@suse.de>
3341
3342         PR middle-end/58532
3343         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3344         before looking for setjmp-like calls.
3345
3346 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3347
3348         PR target/10901
3349         * config/darwin-protos.h (machopic_get_function_picbase): New.
3350         * config/darwin.c (machopic_get_function_picbase): New.
3351         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3352         label for a new func.  (load_macho_picbase_di): Likewise.
3353         (reload_macho_picbase): New expand.
3354         (reload_macho_picbase_si): New insn.
3355         (reload_macho_picbase_di): New insn.
3356         (nonlocal_goto_receiver): New define and split.
3357         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3358         (unspecv enum): Add UNSPECV_NLGR.
3359
3360 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3361
3362         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3363         that altivec registers are correctly sized on Darwin.
3364
3365 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3366
3367         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3368         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3369         * config/x-darwin (host-darwin.o): Likewise.
3370         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3371         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3372         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3373
3374 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3375
3376         * doc/invoke.texi: Fix usage of @tie{} command.
3377
3378 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3379
3380         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3381
3382 2013-09-28  Jan Hubicka  <jh@suse.cz>
3383
3384         * config/i386/x86-tune.def: Add documentation for each of the options;
3385         add whitespace.
3386
3387 2013-09-28  Jan Hubicka  <jh@suse.cz>
3388
3389         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3390         generic.
3391         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3392         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3393         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3394         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3395         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3396
3397 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3398
3399         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3400         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3401         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3402         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3403         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3404         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3405         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3406         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3407         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3408         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3409         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3410         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3411         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3412         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3413         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3414         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3415         gimple.h, godump.c, graphite-clast-to-gimple.c,
3416         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3417         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3418         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3419         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3420         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3421         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3422         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3423         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3424         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3425         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3426         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3427         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3428         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3429         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3430         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3431         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3432         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3433         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3434         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3435         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3436         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3437         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3438         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3439         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3440         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3441         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3442         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3443         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3444         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3445         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3446         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3447         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3448         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3449         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3450         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3451         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3452         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3453         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3454         whitespace before "(".
3455
3456 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3457
3458         * expr.h (extract_bit_field): Remove packedp parameter.
3459         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3460         from forward declaration.
3461         (store_split_bit_field): Remove packedp arg from calls to
3462         extract_fixed_bit_field.
3463         (extract_bit_field_1): Remove packedp parameter and packedp
3464         argument from recursive calls and calls to extract_fixed_bit_field.
3465         (extract_bit_field): Remove packedp parameter and corresponding
3466         arg to extract_bit_field_1.
3467         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3468         to issue warnings.
3469         (extract_split_bit_field): Remove packedp arg from call to
3470         extract_fixed_bit_field.
3471         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3472         (copy_blkmode_from_reg): Likewise.
3473         (copy_blkmode_to_reg): Likewise.
3474         (read_complex_part): Likewise.
3475         (store_field): Likewise.
3476         (expand_expr_real_1): Likewise.
3477         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3478         to extract_bit_field.
3479         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3480         call to extract_bit_field.
3481         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3482         call to extract_bit_field.
3483         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3484         and special packedp behavior from -fstrict-volatile-bitfields
3485         documentation.
3486
3487 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3488
3489         * lra-eliminations.c (init_elim_table): Guard value_p.
3490
3491 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3492
3493         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3494         DFmode, DImode, and SFmode in the upper VSX registers based on the
3495         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3496         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3497         of the VSX register handling.
3498
3499         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3500         (f32_sv): Likewise.
3501         (zero_extendsidi2_lfiwzx): Add support for loading into the
3502         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3503         only do VSX memory options on Altivec registers.
3504         (extendsidi2_lfiwax): Likewise.
3505         (extendsfdf2_fpr): Likewise.
3506         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3507         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3508         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3509         (movdi_internal64): Likewise.
3510
3511 2013-09-27  Xinliang David Li  <davidxl@google.com>
3512
3513         * opts.c (finish_options): Adjust parameters
3514         according to vect cost model.
3515         (common_handle_option): Set dynamic vect cost
3516         model for FDO.
3517         targhooks.c (default_add_stmt_cost): Compute stmt cost
3518         unconditionally.
3519         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3520         Use helper function.
3521         * tree-vectorizer.h (unlimited_cost_model): New function.
3522         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3523         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3524         function.
3525         (vect_enhance_data_refs_alignment): Ditto.
3526         * flag-types.h: New enum.
3527         * common/config/i386/i386-common.c (ix86_option_init_struct):
3528         No need to initialize vect_cost_model flag.
3529         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3530         unconditionally.
3531
3532 2013-09-27  Diego Novillo  <dnovillo@google.com>
3533
3534         * gimple.h (enum ssa_mode): Remove.
3535
3536 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3537
3538         * cfgloop.h (number_of_loops): Fix typo in check for null.
3539
3540 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3541
3542         PR middle-end/58551
3543         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3544         are children of outermost saved_cfun's loop, and set it up to
3545         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3546         if loop != loop0 and SESE region contains bbs that belong to loop0.
3547
3548 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3549
3550         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3551         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3552         (set_address_segment, set_address_base, set_address_index)
3553         (set_address_disp): Accept the argument unconditionally.
3554         (baseness): Remove must_be_base_p and must_be_index_p checks.
3555         (decompose_normal_address): Classify as much as possible in the
3556         main loop.
3557
3558 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3559
3560         * cse.c (count_reg_usage): Handle INT_LIST.
3561         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3562         * reginfo.c (reg_scan_mark_refs): Likewise.
3563         * reload1.c (eliminate_regs_1): Likewise.
3564
3565 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3566
3567         PR middle-end/58547
3568         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3569         signed.
3570
3571 2013-09-27  Richard Biener  <rguenther@suse.de>
3572
3573         PR tree-optimization/58459
3574         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3575         restriction not propagating into loops.
3576
3577 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3578
3579         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3580         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3581         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3582
3583 2013-09-26  Richard Biener  <rguenther@suse.de>
3584
3585         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3586
3587 2013-09-26  Richard Biener  <rguenther@suse.de>
3588
3589         * alias.h (component_uses_parent_alias_set): Rename to ...
3590         (component_uses_parent_alias_set_from): ... this.
3591         * alias.c (component_uses_parent_alias_set): Rename to ...
3592         (component_uses_parent_alias_set_from): ... this and return
3593         the desired parent.
3594         (reference_alias_ptr_type_1): Use the result from
3595         component_uses_parent_alias_set_from instead of stripping
3596         components one at a time.
3597         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3598
3599 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3600
3601         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3602         Move prototypes to...
3603         * tree-ssa-ter.h: New File.  Move prototypes here.
3604         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3605         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3606         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3607         from is_replaceable_p.
3608         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3609         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3610         (process_replaceable): Use ter_is_replaceable_p.
3611         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3612         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3613         newly refactored ssa_is_replaceable_p.
3614         * cfgexpand.c: Include tree-outof-ssa.h.
3615         * ssaexpand.h: Delete.
3616
3617 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3618
3619         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3620         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3621         tree-ssa.c
3622         (create_gimple_tmp): Delete.
3623         (get_expr_type, build_assign, build_type_cast): Move to...
3624         * gimple-builder.c: New File.
3625         (get_expr_type): Relocate from gimple.c.
3626         (build_assign, build_type_cast): Change to only create ssanames.
3627         * gimple.h: Move prototypes to...
3628         * gimple-builder.h: New File. Here.
3629         * tree-ssa.h: And here.
3630         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3631         count_uses_and_derefs): Relocate from gimple.c.
3632         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3633         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3634         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3635         gimple_replace_ssa_lhs.
3636         * asan.c: Include gimple-builder.h.
3637         * Makefile.in: Add gimple-builder.o.
3638
3639 2013-09-26  Richard Biener  <rguenther@suse.de>
3640
3641         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3642         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3643         (loe_visit_block): Use gcc_checking_assert.
3644         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3645         gimple_assign_ssa_name_copy_p.
3646         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3647         change.
3648
3649 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3650
3651         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3652         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3653
3654 2013-09-26  Richard Biener  <rguenther@suse.de>
3655
3656         PR tree-optimization/58539
3657         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3658         the fact that debug statements are not taking part in loop-closed
3659         SSA construction.
3660
3661 2013-09-26  Nick Clifton  <nickc@redhat.com>
3662
3663         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3664         time warning message.
3665         (msp430_print_operand_raw): Delete unused letter parameter.
3666         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3667         (msp430_print_operand_address): New function.
3668         (msp430_print_operand): Move address printing code from here to
3669         new function.
3670         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3671         assembler.
3672         (zero_extendpsisi2): Likewise.
3673         (extendpsisi2): New pattern.
3674         (andneghi3): New pattern.
3675
3676 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3677
3678         * config/aarch64/aarch64.opt (mlra): New option.
3679         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3680         (TARGET_LRA_P): Define.
3681
3682 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3683
3684         * expr.c (expand_assignment): Remove obsolete comment.
3685
3686 2013-09-25  Jeff Law  <law@redhat.com>
3687
3688         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3689         tree-ssa-threadupdate.h.
3690         (register_jump_thread): Similarly.
3691         * tree-ssa-threadupdate.h: New header file.
3692         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3693         * tree-vrp.c: Likewise.
3694         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3695         (thread_around_empty_blocks): Change type of path vector argument to
3696         an edge,type pair from just an edge.  Initialize both elements when
3697         appending to a jump threading path.  Tweak references to elements
3698         appropriately.
3699         (thread_across_edge): Similarly.  Release memory for the elements
3700         as needed.
3701         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3702         (dump_jump_thread_path): New function broken out from
3703         register_jump_thread.
3704         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3705         path vector entries.  Search the path for NULL edges and dump
3706         the path if one is found.  Tweak the conversion of path to 3-edge
3707         form to use the block copy type information embedded in the path.
3708
3709 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3710
3711         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3712
3713 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3714             Vladimir Makarov  <vmakarov@redhat.com>
3715
3716         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3717         from the least significant bit.
3718         (strip_address_mutations): Add bitfield operations handling.
3719         (must_be_index_p): Add shifting and rotate operations handling.
3720         (set_address_base): Use must_be_base_p predicate.
3721         (set_address_index): Use must_be_index_p predicate.
3722
3723 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3724             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3725             Sergey Lega  <sergey.s.lega@intel.com>
3726             Anna Tikhonova  <anna.tikhonova@intel.com>
3727             Ilya Tocar  <ilya.tocar@intel.com>
3728             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3729             Ilya Verbin  <ilya.verbin@intel.com>
3730             Kirill Yukhin  <kirill.yukhin@intel.com>
3731             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3732
3733         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
3734         Use new names.
3735         (ix86_expand_vector_move_misalign): Support new unaligned load and
3736         stores and use new names.
3737         (CODE_FOR_sse2_storedqu): Rename to ...
3738         (CODE_FOR_sse2_storedquv16qi): ... this.
3739         (CODE_FOR_sse2_loaddqu): Rename to ...
3740         (CODE_FOR_sse2_loaddquv16qi): ... this.
3741         (CODE_FOR_avx_loaddqu256): Rename to ...
3742         (CODE_FOR_avx_loaddquv32qi): ... this.
3743         (CODE_FOR_avx_storedqu256): Rename to ...
3744         (CODE_FOR_avx_storedquv32qi): ... this.
3745         * config/i386/i386.md (fpint_logic): New.
3746         * config/i386/sse.md (VMOVE): Extend for AVX512.
3747         (VF): Ditto.
3748         (VF_128_256): New.
3749         (VF_512): Ditto.
3750         (VI_UNALIGNED_LOADSTORE): Ditto.
3751         (sse2_avx_avx512f): Ditto.
3752         (sse2_avx2): Extend for AVX512.
3753         (sse4_1_avx2): Ditto.
3754         (avx2_avx512f): New.
3755         (sse): Extend for AVX512.
3756         (sse2): Ditto.
3757         (sse4_1): Ditto.
3758         (avxsizesuffix): Ditto.
3759         (sseintvecmode): Ditto.
3760         (ssePSmode): Ditto.
3761         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
3762         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
3763         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3764         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
3765         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3766         (<sse2_avx_avx512f>_storedqu<mode): ... this.
3767         (<sse>_movnt<mode>): Replace constraint "x" with "v".
3768         (STORENT_MODE): Extend for AVX512.
3769         (*absneg<mode>2): Replace constraint "x" with "v".
3770         (*mul<mode>3): Ditto.
3771         (*ieee_smin<mode>3): Ditto.
3772         (*ieee_smax<mode>3): Ditto.
3773         (avx_cmp<mode>3): Replace VF with VF_128_256.
3774         (*<sse>_maskcmp<mode>3_comm): Ditto.
3775         (<sse>_maskcmp<mode>3): Ditto.
3776         (<sse>_andnot<mode>3): Extend for AVX512.
3777         (<code><mode>3, anylogic): Replace VF with VF_128_256.
3778         (<code><mode>3, fpint_logic): New.
3779         (*<code><mode>3): Extend for AVX512.
3780         (avx512flogicsuff): New.
3781         (avx512f_<logic><mode>): Ditto.
3782         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
3783         VF_128_256.
3784         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
3785         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
3786         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
3787         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
3788         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
3789         (xop_vpermil2<mode>3): Ditto.
3790         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
3791         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
3792         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
3793         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
3794
3795 2013-09-25  Tom Tromey  <tromey@redhat.com>
3796
3797         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
3798         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
3799         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
3800         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
3801         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
3802         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
3803         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
3804         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
3805         (GRAPHITE_HTAB_H): Remove.
3806
3807 2013-09-25  Tom Tromey  <tromey@redhat.com>
3808
3809         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
3810
3811 2013-09-25  Tom Tromey  <tromey@redhat.com>
3812
3813         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
3814
3815 2013-09-25  Tom Tromey  <tromey@redhat.com>
3816
3817         * config/i386/t-i386 (i386.o): Remove.
3818         (i386-c.o): Use COMPILE and POSTCOMPILE.
3819
3820 2013-09-25  Tom Tromey  <tromey@redhat.com>
3821
3822         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
3823
3824 2013-09-25  Tom Tromey  <tromey@redhat.com>
3825
3826         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
3827         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
3828         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
3829         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
3830         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
3831         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
3832         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
3833         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
3834         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
3835         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
3836         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
3837         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
3838         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
3839         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
3840         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
3841         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
3842         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
3843         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
3844         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
3845         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
3846         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
3847         (tree-ssa-pre.o, tree-ssa-sccvn.o)
3848         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
3849         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
3850         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
3851         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
3852         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
3853         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
3854         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
3855         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
3856         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
3857         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
3858         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
3859         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
3860         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
3861         (graphite-clast-to-gimple.o, graphite-dependences.o)
3862         (graphite-interchange.o, graphite-poly.o)
3863         (graphite-scop-detection.o, graphite-sese-to-poly.o)
3864         (graphite-optimize-isl.o, tree-vect-loop.o)
3865         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
3866         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
3867         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
3868         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
3869         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
3870         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
3871         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
3872         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
3873         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
3874         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
3875         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
3876         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
3877         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
3878         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
3879         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
3880         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
3881         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
3882         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
3883         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
3884         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
3885         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
3886         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
3887         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
3888         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
3889         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
3890         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
3891         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
3892         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
3893         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
3894         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
3895         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
3896         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
3897         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
3898         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
3899         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
3900         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
3901         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
3902         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
3903         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
3904         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
3905         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
3906         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
3907         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
3908         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
3909         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
3910         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
3911         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
3912         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
3913         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
3914         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
3915         (gcov-dump.o): Remove.
3916         (default-c.o): Use COMPILE and POSTCOMPILE.
3917         (CFLAGS-gcc.o): New variable.
3918         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
3919
3920 2013-09-25  Tom Tromey  <tromey@redhat.com>
3921
3922         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
3923         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
3924         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
3925         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
3926         (c-family/c-pragma.o, c-family/c-pretty-print.o)
3927         (c-family/c-semantics.o, c-family/c-ada-spec.o)
3928         (c-family/array-notation-common.o, c-family/stub-objc.o)
3929         (c-family/c-ubsan.o): Remove.
3930
3931 2013-09-25  Tom Tromey  <tromey@redhat.com>
3932
3933         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
3934
3935 2013-09-25  Tom Tromey  <tromey@redhat.com>
3936
3937         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
3938         to add -DENABLE_SHARED_LIBGCC.
3939         (gcc.o): Don't use subshell.
3940
3941 2013-09-25  Tom Tromey  <tromey@redhat.com>
3942
3943         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
3944         * configure.ac: Don't invoke AM_PROG_CC_C_O.
3945         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
3946         * configure, config.in: Rebuild.
3947
3948 2013-09-25  Tom Tromey  <tromey@redhat.com>
3949
3950         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
3951         (COMPILE, POSTCOMPILE): New variables.
3952         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
3953         (DEPFILES): New variable.
3954         Include ".Po" files.
3955         * configure.ac: Add checks for dependency checking.
3956         * configure, aclocal.m4: Regenerate.
3957
3958 2013-09-25  Tom Tromey  <tromey@redhat.com>
3959
3960         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
3961         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
3962
3963 2013-09-25  Tom Tromey  <tromey@redhat.com>
3964
3965         * Makefile.in (generated_files): Add options.h,
3966         target-hooks-def.h, insn-opinit.h,
3967         common/common-target-hooks-def.h, pass-instances.def,
3968         c-family/c-target-hooks-def.h.
3969
3970 2013-09-25  Jeff Law  <law@redhat.com>
3971
3972         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
3973         than foo[foo.length () - 1] to access last member in a vec.
3974         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
3975
3976 2013-09-25  Richard Biener  <rguenther@suse.de>
3977
3978         PR middle-end/58521
3979         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
3980
3981 2013-09-25  Jan Hubicka  <jh@suse.cz>
3982
3983         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
3984         test.
3985
3986 2013-09-25  Marek Polacek  <polacek@redhat.com>
3987
3988         PR sanitizer/58420
3989         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
3990         when determining the type name.
3991
3992 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
3993
3994         * config/sh/sh.md: Fix formatting.
3995
3996 2013-09-24  Xinliang David Li  <davidxl@google.com>
3997
3998         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
3999         max peel iterations parameter.
4000         * param.def: New parameter.
4001         * doc/invoke.texi: Document New parameter.
4002
4003 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4004
4005         * gimple-pretty-print.c: Various whitespace tweaks.
4006         * tree-core.h: Likewise.
4007         * tree-pretty-print.c: Likewise.
4008         * tree-ssa-alias.c: Likewise.
4009         * tree-ssa-copy.c: Likewise.
4010         * tree-ssanames.c: Likewise.
4011         * tree-ssanames.h: Likewise.
4012         * tree-vrp.c: Likewise.
4013
4014 2013-09-24  Alan Modra  <amodra@gmail.com>
4015
4016         PR middle-end/57134
4017         PR middle-end/57586
4018         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4019         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4020         inout operands.
4021
4022 2013-09-24  Richard Biener  <rguenther@suse.de>
4023
4024         PR middle-end/58513
4025         * tree.c (reference_alias_ptr_type): Move ...
4026         * alias.c (reference_alias_ptr_type): ... here and implement
4027         in terms of the new reference_alias_ptr_type_1.
4028         (ref_all_alias_ptr_type_p): New helper.
4029         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4030         use ref_all_alias_ptr_type_p.
4031         (get_deref_alias_set): Add flag_strict_aliasing check here.
4032         (reference_alias_ptr_type_1): New function, split out from ...
4033         (get_alias_set): ... here.
4034         (alias_ptr_types_compatible_p): New function.
4035         * alias.h (reference_alias_ptr_type): Declare.
4036         (alias_ptr_types_compatible_p): Likewise.
4037         * tree.h (reference_alias_ptr_type): Remove.
4038         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4039         to compare MEM_REF alias types.
4040
4041 2013-09-24  Richard Biener  <rguenther@suse.de>
4042
4043         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4044
4045 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4046
4047         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4048         reload helper function arrays into a single array reg_addr.
4049         (reload_fpr_gpr): Likewise.
4050         (reload_gpr_vsx): Likewise.
4051         (reload_vsx_gpr): Likewise.
4052         (struct rs6000_reg_addr): Likewise.
4053         (reg_addr): Likewise.
4054         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4055         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4056         (rs6000_init_hard_regno_mode_ok): Likewise.
4057         (rs6000_secondary_reload_direct_move): Likewise.
4058         (rs6000_secondary_reload): Likewise.
4059
4060         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4061         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4062         power8 changes.  Put wg constraint in alphabetical order.
4063
4064         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4065         for future work to add ISA 2.07 VSX single precision support.
4066         (-mvsx-scalar-double): Change default from -1 to 1, update
4067         documentation comment.
4068         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4069         (-mupper-regs-df): New debug switch to control whether DF values
4070         can go in the traditional Altivec registers.
4071         (-mupper-regs-sf): New debug switch to control whether SF values
4072         can go in the traditional Altivec registers.
4073
4074         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4075         and wy constraints.
4076         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4077         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4078         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4079         (rs6000_debug_legitimate_address_p): Print if we are running
4080         before, during, or after reload.
4081         (rs6000_secondary_reload): Add a comment.
4082         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4083
4084         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4085         constraints.  Update documentation string.
4086         (wd constraint): Likewise.
4087         (wf constraint): Likewise.
4088         (wg constraint): Likewise.
4089         (wn constraint): Likewise.
4090         (ws constraint): Likewise.
4091         (wt constraint): Likewise.
4092         (wx constraint): Likewise.
4093         (wz constraint): Likewise.
4094         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4095         instructions.
4096         (ww constraint): Likewise.
4097         (wy constraint): Likewise.
4098         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4099         in the previous submissions.
4100         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4101
4102 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4103
4104         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4105         (REG_BR_PROB): Say that the probability is stored in an int_list.
4106         * reg-notes.def: Update commentary to mention INT_LIST.
4107         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4108         (INT_LIST): New rtx.
4109         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4110         * rtlanal.c (int_reg_note_p): New function.
4111         (alloc_reg_note): Assert that the note does not have an int argument.
4112         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4113         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4114         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4115         rather than an INSN_LIST.  Handle INT_LIST.
4116         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4117         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4118         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4119         Manipulate them as ints rather than rtxes.
4120         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4121         * regmove.c (copy_src_to_dest): Likewise.
4122         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4123
4124         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4125         into the cases that need it.
4126         * config/arm/arm.c (arm_unwind_emit): Likewise.
4127
4128         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4129         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4130         * loop-doloop.c (add_test, doloop_modify): Likewise.
4131         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4132         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4133         * predict.c (combine_predictions_for_insn): Likewise.
4134         * print-rtl.c (print_rtx): Handle INT_LIST.
4135         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4136         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4137         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4138         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4139         (ix86_split_fp_branch, predict_jump): Likewise.
4140         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4141         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4142         * config/spu/spu.c (ea_load_store_inline): Likewise.
4143
4144         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4145         value of a REG_BR_PROB note.
4146         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4147         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4148         * emit-rtl.c (try_split): Likewise.
4149         * predict.c (br_prob_note_reliable_p): Likewise.
4150         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4151         * reorg.c (mostly_true_jump): Likewise.
4152         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4153         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4154         * config/i386/i386.c (ix86_print_operand): Likewise.
4155         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4156         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4157         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4158         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4159         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4160         * config/sparc/sparc.c (output_cbranch): Likewise.
4161         * config/spu/spu.c (get_branch_target): Likewise.
4162         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4163         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4164
4165 2013-09-23  Jan Hubicka  <jh@suse.cz>
4166
4167         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4168         for ipa-devirt.
4169         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4170         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4171         of external calls
4172         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4173         (gimple_fold_call): Dump inconsistent devirtualizations; add
4174         sanity check for type based devirtualizations.
4175         * ipa-prop.c: Include ipa-utils.h
4176         (ipa_intraprocedural_devirtualization): Add sanity check.
4177         (try_make_edge_direct_virtual_call): Likewise.
4178
4179 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4180
4181         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4182         assignment statements.
4183
4184 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4185
4186         * gimple-pretty-print.c (dump_ssaname_info): New function.
4187         (dump_gimple_phi): Call it.
4188         (pp_gimple_stmt_1): Likewise.
4189         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4190         (range_info_def): Declare.
4191         * tree-pretty-print.c (pp_double_int): New function.
4192         (dump_generic_node): Call it.
4193         * tree-pretty-print.h (pp_double_int): Declare.
4194         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4195         * tree-ssanames.h (range_info_def): New structure.
4196         (value_range_type): Move definition here.
4197         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4198         Declare.
4199         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4200         initialization.
4201         (set_range_info): New function.
4202         (get_range_info): Likewise.
4203         (duplicate_ssa_name_range_info): Likewise.
4204         (duplicate_ssa_name_fn): Check pointer type and call
4205         duplicate_ssa_name_range_info.
4206         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4207         * tree-vrp.c (value_range_type): Remove definition, now in
4208         tree-ssanames.h.
4209         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4210         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4211         (SSA_NAME_RANGE_INFO): New macro.
4212
4213 2013-09-23  Richard Biener  <rguenther@suse.de>
4214
4215         PR tree-optimization/58464
4216         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4217         (phi_trans_add): Change to add conditionally on being not present.
4218         (phi_translate_1): Remove recursion detection here.
4219         (phi_translate): Pre-seed the cache with NULL to catch
4220         recursion here in a more generic way.
4221         (bitmap_find_leader): Adjust comment.
4222         (get_representative_for): Dump value-numbers.
4223         (create_expression_by_pieces): Likewise.
4224         (insert_into_preds_of_block): Likewise.
4225
4226 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4227
4228         PR target/58475
4229         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4230         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4231
4232 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4233
4234         Revert r202780:
4235         2013-09-20  Renlin Li  <renlin.li@arm.com>
4236
4237         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4238         plus_constant.
4239         (aarch64_expand_epilogue): Likewise.
4240         (aarch64_legitimize_reload_address): Likewise.
4241
4242 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4243
4244         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4245         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4246
4247 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4248
4249         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4250         comparison operators when -fno-trapping-math is in effect.
4251         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4252         comparison operators in TFmode and assert that unsupported operators
4253         cannot reach here.
4254         (ia64_print_operand): Likewise.
4255
4256 2013-09-21  Jan Hubicka  <jh@suse.cz>
4257
4258         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4259         (sse_typeless_stores): Enable for core
4260         (sse_load0_by_pxor): Likewise.
4261         (four_jump_limit): Disable for core.
4262         (pad_returns): Likewise.
4263         (avoid_vector_decode): Likewise.
4264         (fuse_cmp_and_branch): Enable for cores.
4265         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4266
4267 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4268
4269         PR middle-end/56791
4270         * config/pa/pa.c (pa_option_override): Disable auto increment and
4271         decrement instructions until reload is completed.
4272
4273         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4274         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4275         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4276
4277 2013-09-20  DJ Delorie  <dj@redhat.com>
4278             Nick Clifton  <nickc@redhat.com>
4279
4280         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4281         (need_to_save): Save bank 0 on interrupts.
4282         (characterize_address): Strip far address wrappers.
4283         (rl78_as_legitimate_address): Likewise.
4284         (transcode_memory_rtx): Likewise.
4285         (rl78_peep_movhi_p): Disable this peephole after devirt.
4286         (rl78_propogate_register_origins): Forget all origins when a
4287         CLOBBER is seen.
4288         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4289         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4290         peephole2's.
4291         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4292         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4293         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4294         relocating.
4295         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4296
4297 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4298
4299         * config/pa/pa.md: In "scc" insn patterns, change output template to
4300         handle const0_rtx in reg_or_0_operand operands.
4301
4302 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4303
4304         PR target/56875
4305         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4306         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4307
4308 2013-09-20  Richard Biener  <rguenther@suse.de>
4309
4310         PR middle-end/58484
4311         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4312         remembering SSA name version and block index.
4313         (new_scev_info_str): Adjust.
4314         (hash_scev_info): Likewise.  Also hash the block index.
4315         (eq_scev_info): Adjust.
4316         (find_var_scev_info): Likewise.
4317         (struct instantiate_cache_entry): Remove.
4318         (struct instantiate_cache_type): Use a htab to map name, block
4319         to chrec.
4320         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4321         (get_instantiated_value_entry): Likewise.
4322         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4323         (instantiate_scev_name): Adjust.
4324
4325 2013-09-20  Jeff Law  <law@redhat.com>
4326
4327         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4328
4329 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4330
4331         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4332         Call aarch64_simd_expand_args to update op[argc].
4333
4334 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4335
4336         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4337         plugin argument.
4338
4339 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4340
4341         * gengtype.c (file_rules): Added rule for *.cc files.
4342         (get_output_file_with_visibility): Give fatal message when no
4343         rules found.
4344
4345 2013-09-20  Renlin Li  <renlin.li@arm.com>
4346
4347         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4348         (aarch64_expand_epilogue): Likewise.
4349         (aarch64_legitimize_reload_address): Likewise.
4350
4351 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4352
4353         PR middle-end/57748
4354         * expr.c (expand_assignment): Remove misalignp code path.
4355
4356 2013-09-20  Marek Polacek  <polacek@redhat.com>
4357
4358         PR sanitizer/58413
4359         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4360         TYPE_PRECISION.  Add asserts.
4361
4362 2013-09-20  Richard Biener  <rguenther@suse.de>
4363
4364         PR tree-optimization/58453
4365         * tree-loop-distribution.c (distribute_loop): Apply the cost
4366         model for -ftree-loop-distribute-patterns, too.
4367
4368 2013-09-20  Richard Biener  <rguenther@suse.de>
4369
4370         PR middle-end/58473
4371         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4372         make type comparison less strict.
4373
4374 2013-09-20  Alan Modra  <amodra@gmail.com>
4375
4376         * configure: Regenerate.
4377         * aclocal.m4: Regenerate.
4378
4379 2013-09-20  Marek Polacek  <polacek@redhat.com>
4380
4381         PR other/58467
4382         * doc/extend.texi: Document that attribute used is meant to be used
4383         on variables with static storage duration.
4384
4385 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4386
4387         PR tree-optimization/58472
4388         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4389         simd_lane_access set inv_p = false.
4390         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4391         the simduid magic VAR_DECL.
4392
4393 2013-09-19  Jan Hubicka  <jh@suse.cz>
4394
4395         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4396
4397 2013-09-17  Jeff Law  <law@redhat.com>
4398
4399         * tree-ssa-dom.c (record_temporary_equivalences): New function
4400         split out of dom_opt_dom_walker::after_dom_children.
4401         (dom_opt_dom_walker::thread_across_edge): Move common code
4402         in here from dom_opt_dom_walker::after_dom_children.
4403         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4404
4405 2013-09-19  Jan Hubicka  <jh@suse.cz>
4406
4407         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4408         (TARGET_GENERIC): Use PROCESOR_GENERIC
4409         (enum processor_type): Unify generic32 and 64.
4410         * i386.md (cpu): Likewise.
4411         * x86-tune.def (use_leave): Enable for generic32.
4412         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4413         * athlon.md: Change generic64 to generic in all occurences.
4414         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4415         (ix86_target_macros_internal): Likewise.
4416         * driver-i386.c (host_detect_local_cpu): Likewise.
4417         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4418         to ..
4419         (generic_memcpy, generic_memset, generic_cost): This one.
4420         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4421         (m_GENERIC32, m_GENERIC64): Remove.
4422         (m_GENERIC): Turn into one flag.
4423         (processor_target): Unify generic tunnings.
4424         (ix86_option_override_internal): Replace generic32/64 by generic.
4425         (ix86_issue_rate): Likewise.
4426         (ix86_adjust_cost): Likewise.
4427
4428 2013-09-19  Jan Hubicka  <jh@suse.cz>
4429
4430         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4431         of speculative flag.
4432
4433 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4434
4435         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4436         GOMP_sections_start, even if !exit_reachable.
4437
4438 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4439
4440         * lra-constraints.c (need_for_all_save_p): Use macro
4441         HARD_REGNO_CALL_PART_CLOBBERED.
4442         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4443         set up pseudo conflict hard regs.
4444
4445 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4446
4447         PR target/58452
4448         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4449         operands.
4450
4451 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4452
4453         PR rtl-optimization/58438
4454         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4455         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4456         from previous subpasses.
4457
4458 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4459
4460         * arm.c (arm_get_frame_offsets): Validate architecture supports
4461         LDRD/STRD before accepting the tuning preference.
4462         (arm_expand_prologue): Likewise.
4463         (arm_expand_epilogue): Likewise.
4464
4465 2013-09-18  Richard Biener  <rguenther@suse.de>
4466
4467         PR tree-optimization/58417
4468         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4469         have chrecs with symbols defined in the loop as operands.
4470         (chrec_fold_multiply): Likewise.
4471         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4472         parameters before folding binary operations.
4473         (struct instantiate_cache_entry_hasher): Remove.
4474         (struct instantiate_cache_type): Use a pointer-map.
4475         (instantiate_cache_type::instantiate_cache_type): New function.
4476         (instantiate_cache_type::get): Likewise.
4477         (instantiate_cache_type::set): Likewise.
4478         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4479         (get_instantiated_value_entry): Likewise.
4480         (global_cache): New global.
4481         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4482         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4483         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4484         (instantiate_scev_name): Adjust.
4485         (instantiate_scev): Construct global instead of local cache.
4486         (resolve_mixers): Likewise.
4487
4488 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4489             Paolo Carlini  <paolo.carlini@oracle.com>
4490
4491         PR c++/58458
4492         * doc/implement-cxx.texi: Fix references to the C++ standards.
4493
4494 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4495
4496         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4497         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4498         simd_lane_access drs, update also DR_ALIGNED_TO.
4499
4500 2013-09-18  Marek Polacek  <polacek@redhat.com>
4501
4502         PR sanitizer/58411
4503         * doc/extend.texi: Document no_sanitize_undefined attribute.
4504         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4505         no_sanitize_undefined attribute.
4506
4507 2013-09-18  Nick Clifton  <nickc@redhat.com>
4508
4509         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4510         (ASM_DECLARE_FUNCTION_NAME): Define.
4511
4512 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4513
4514         * compare-elim.c (find_comparison_dom_walker): New class
4515         (find_comparisons_in_bb): Rename to
4516         find_comparison_dom_walker::before_dom_children
4517         (find_comparisons): Adjust
4518         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4519         adjust.
4520         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4521         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4522         (init_walk_dominator_tree): Remove declaration.
4523         (fini_walk_dominator_tree): Remove declaration.
4524         * fwprop.c (single_def_use_dom_walker): New class
4525         (single_def_use_enter_block): Convert to
4526         single_def_use_dom_walker::before_dom_children.
4527         (single_def_use_leave_block): Convert to
4528         single_def_use_dom_walker::after_dom_children.
4529         (build_single_def_use_links): Adjust.
4530         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4531         class.
4532         (find_candidates_in_block): Convert to
4533         find_candidates_dom_walker::before_dom_children.
4534         (execute_strength_reduction): Adjust.
4535         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4536         (sese_dom_walker): New class.
4537         (sese_dom_walker::sese_dom_walker): New constructor.
4538         (sese_dom_walker::~sese_dom_walker): New destructor.
4539         (build_sese_conditions_before): Convert to
4540         sese_dom_walker::before_dom_children.
4541         (build_sese_conditions_after): Convert to
4542         sese_dom_walker::after_dom_children.
4543         (build_poly_scop): Adjust
4544         * tree-into-ssa.c (rewrite_dom_walker): New class
4545         (rewrite_enter_block): Convert to
4546         rewrite_dom_walker::before_dom_children.
4547         (rewrite_leave_block): Convert to
4548         rewrite_dom_walker::after_dom_children.
4549         (rewrite_update_dom_walker): New class.
4550         (rewrite_update_enter_block): Convert to
4551         rewrite_update_dom_walker::before_dom_children.
4552         (rewrite_update_leave_block): Convert to
4553         rewrite_update_dom_walker::after_dom_children.
4554         (rewrite_blocks, rewrite_into_ssa): Adjust.
4555         (mark_def_dom_walker): New class.
4556         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4557         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4558         (mark_def_sites_blocks): Convert to
4559         mark_def_dom_walker::before_dom_children.
4560         (mark_def_site_blocks): Remove.
4561         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4562         (tree_ssa_dominator_optimize): Adjust.
4563         (dom_thread_across_edge): Convert to method
4564         dom_opt_dom_walker::thread_across_edge.
4565         (dom_opt_enter_block): Convert to member function
4566         dom_opt_dom_walker::before_dom_children.
4567         (dom_opt_leave_block): Convert to member function
4568         dom_opt_dom_walker::after_dom_children.
4569         * tree-ssa-dse.c (dse_dom_walker): New class.
4570         (dse_enter_block): Convert to member function
4571         dse_dom_walker::before_dom_children.
4572         (tree_ssa_dse): Adjust.
4573         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4574         (determine_invariantness_stmt): Convert to method
4575         invariantness_dom_walker::before_dom_children.
4576         (determine_invariantness): Remove
4577         (move_computations_dom_walker): New class.
4578         (move_computations_stmt): Convert to method
4579         move_computations_dom_walker::before_dom_children.
4580         (move_computations, tree_ssa_lim): Adjust.
4581         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4582         (nt_init_block): Make method
4583         notrappping_dom_walker::before_dom_children.
4584         (nt_fini_block): Make
4585         method nontrapping_dom_walker::after_dom_children.
4586         (get_non_trapping): Adjust.
4587         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4588         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4589         (eliminate_leave_block): Make method.
4590         eliminate_dom_walker::after_dom_children.
4591         (eliminate): Adjust
4592         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4593         (strlen_enter_block): Make method
4594         strlen_dom_walker::before_dom_children.
4595         (strlen_leave_block): Make
4596         method strlen_dom_walker::after_dom_children.
4597         (tree_ssa_strlen): Adjust.
4598         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4599         (tree_ssa_uncprop): Adjust.
4600         (uncprop_leave_block): Make method
4601         uncprop_dom_walker::after_dom_children.
4602         (uncprop_leave_block): Make method
4603         uncprop_dom_walker::before_dom_children.
4604
4605 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4606
4607         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4608         branch in same basic block.  Check both src and dest of the move insn.
4609
4610 2013-09-17  Nick Clifton  <nickc@redhat.com>
4611
4612         * config/rl78/rl78-real.md (bf): New pattern.
4613         (bt): New pattern.
4614         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4615         (rl78_print_operand): Do not put a # before a %B.
4616         * config/rl78/rl78.opt: Tweak doc strings.
4617
4618 2013-09-17  DJ Delorie  <dj@redhat.com>
4619
4620         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4621         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4622         (rl78_option_override): Likewise, if -mallregs.
4623         (is_virtual_register): Likewise.
4624         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4625         (REGNO_OK_FOR_BASE_P): Likewise.
4626         * config/rl78/rl78.opt (-mallregs): New.
4627
4628 2013-09-17  Nick Clifton  <nickc@redhat.com>
4629
4630         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4631         For interrupt functions: save all call clobbered registers if the
4632         interrupt handler is not a leaf function.
4633         (rl78_expand_prologue): Always recompute the frame information.
4634         For interrupt functions: only select bank 0 if one of the bank 0
4635         registers is going to be psuhed.
4636
4637 2013-09-17  DJ Delorie  <dj@redhat.com>
4638
4639         * config/rl78/constraints.md: For each W* constraint, rename to C*
4640         and create a W* constraint that checks for an optional ES: prefix
4641         pattern also.
4642         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4643         (es_addr): New.  Used to wrap far addresses.
4644         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4645         (rl78_es_base): New.
4646         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4647         wrapped far addresses.
4648         (rl78_print_operand_1): Unwrap far addresses before processing.
4649         (rl78_lo16): Wrap far addresses in unspecs.
4650         (rl78_es_addr): New.
4651         (rl78_es_base): New.
4652         (insn_ok_now): Check for not-yet-wrapped far addresses.
4653         (transcode_memory_rtx): Properly re-wrap far addresses.
4654
4655 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4656
4657         * config/sparc/t-rtems: Add leon3 multilibs.
4658
4659 2013-09-17  Cong Hou  <congh@google.com>
4660
4661         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4662         when checking the dot production pattern. The type of rhs operand
4663         of multiply is now checked correctly.
4664
4665 2013-09-17  Jeff Law  <law@redhat.com>
4666
4667         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4668         edge implied equivalences into successor phis.
4669         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4670         here from tree-ssa-threadedge.c.
4671         (mark_threaded_blocks): When threading through a joiner, if both
4672         successors of the joiner's clone reach the same block, verify the
4673         PHI arguments are equal.  If not, cancel the jump threading request.
4674         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4675         tree-ssa-threadupdate.c
4676         (thread_across_edge): Don't check PHI argument equality when
4677         threading through joiner block here.
4678
4679 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4680
4681         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4682         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4683         (warn_uninit, warn_uninitialized_vars,
4684         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4685         Move to tree-ssa-uninit.c.
4686         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4687         (has_undefined_value_p): New.  Pass dependant parts of
4688         ssa_undefined_value_p.
4689         (uninit_undefined_value_p): Use has_undefined_value_p.
4690         (warn_uninit, warn_uninitialized_vars,
4691         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4692         Move from tree-ssa.c.
4693         * tree-ssa.h: Adjust prototypes.
4694
4695 2013-09-17  Jan Hubicka  <jh@suse.cz>
4696
4697         PR middle-end/58332
4698         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4699         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4700         FUNCTION_NOT_OPTIMIZED.
4701         * ipa-inline-analysis.c (compute_inline_parameters): Function
4702         not optimized is not inlinable unless it is alwaysinline.
4703         (inline_analyze_function): Force calls in not optimized
4704         function not inlinable.
4705
4706 2013-09-17  Jan Hubicka  <jh@suse.cz>
4707
4708         PR middle-end/58329
4709         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4710         to return NULL.
4711         * ipa.c (function_and_variable_visibility): Likewise.
4712         * ipa-profile.c (ipa_profile): Likewise.
4713
4714 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4715
4716         PR ipa/58398
4717         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4718         attribute, and don't inline the resolver in this case.
4719
4720 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4721
4722         * coverage.c (get_coverage_counts): Add missing newline.
4723
4724 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4725
4726         PR tree-optimization/58088
4727         * fold-const.c (mask_with_trailing_zeros): New function.
4728         (fold_binary_loc): Make sure we don't recurse infinitely
4729         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
4730         Use mask_with_trailing_zeros where appropriate.
4731
4732 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
4733
4734         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
4735         of 'prev' var to get better distance estimation.
4736
4737 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
4738
4739         * tree-inline.h (struct copy_body_data): Add transform_parameter.
4740         * tree-inline.c (is_parameter_of): New predicate.
4741         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
4742         a parameter has been remapped.
4743         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
4744         (optimize_inline_calls): Initialize transform_parameter.
4745         (copy_gimple_seq_and_replace_locals): Likewise.
4746         (tree_function_versioning): Likewise.
4747         (maybe_inline_call_in_expr): Likewise.
4748
4749 2013-09-17  Nick Clifton  <nickc@redhat.com>
4750
4751         * config/msp430/msp430-protos.h: Add prototypes for new functions.
4752         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
4753         interrupt handlers.
4754         (is_attr_func): New function.
4755         (msp430_is_interrupt_func): New function.
4756         (is_naked_func): New function.
4757         (is_reentrant_func): New function.
4758         (is_critical_func): New function.
4759         (msp430_start_function): Add annotations for function attributes.
4760         (msp430_attr): New function.
4761         (msp430_attribute_table): New.
4762         (msp430_function_section): New function.
4763         (TARGET_ASM_FUNCTION_SECTION): Define.
4764         (msp430_builtin): New enum.
4765         (msp430_init_builtins): New function.
4766         (msp430_builtin_devl): New function.
4767         (msp430_expand_builtin): New function.
4768         (TARGET_INIT_BUILTINS): Define.
4769         (TARGET_EXPAND_BUILTINS): Define.
4770         (TARGET_BUILTIN_DECL): Define.
4771         (msp430_expand_prologue): Add support for naked, interrupt,
4772         critical and reentrant functions.
4773         (msp430_expand_epilogue): Likewise.
4774         (msp430_print_operand): Handle 'O' character.
4775         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
4776         NO_TRAMPOLINES.
4777         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
4778         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
4779         (pushm): Use a 'n' rather than an 'i' constraint.
4780         (msp_return): Add generation of the interrupt return instruction.
4781         (disable_interrupts): New pattern.
4782         (enable_interrupts): New pattern.
4783         (push_intr_state): New pattern.
4784         (pop_intr_state): New pattern.
4785         (bic_SR): New pattern.
4786         (bis_SR): New pattern.
4787         * doc/extend.texi: Document MSP430 function attributes and builtin
4788         functions.
4789
4790 2013-09-17  Richard Biener  <rguenther@suse.de>
4791
4792         PR tree-optimization/58432
4793         * tree-loop-distribution.c (tree_loop_distribution): Also
4794         scan PHIs for outside loop uses and seed a partition from them.
4795
4796 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
4797
4798         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
4799         (restructure_reference): Call backtrace_base_for_ref.
4800
4801 2013-09-17  Alan Modra  <amodra@gmail.com>
4802
4803         PR target/57589
4804         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
4805         patch.
4806
4807 2013-09-16  DJ Delorie  <dj@redhat.com>
4808
4809         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
4810         vregs location for RL78/G10.
4811         (rl78_expand_prologue): Avoid SEL on G10.
4812         (rl78_expand_epilogue): Likewise.
4813         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
4814         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
4815         __RL78_G10__ when appropriate.
4816         (ASM_SPEC): Pass -mg10 along to the assembler.
4817         * config/rl78/rl78.md (sel_rb): Disable for G10.
4818         * config/rl78/rl78.opt: Add -mg10 option.
4819         * config/rl78/t-rl78: Add -mg10 multilib.
4820
4821 2013-09-16  Xinliang David Li  <davidxl@google.com>
4822
4823         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
4824         * omp-low.c (omp_max_vf): Ditto.
4825         (expand_omp_simd): Ditto.
4826         * tree-vectorizer.c (vectorize_loops): Ditto.
4827         (gate_vect_slp): Ditto.
4828         (gate_increase_alignment): Ditto.
4829         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
4830         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
4831         (gate_tree_vectorize): Name change.
4832         (tree_vectorize): Ditto.
4833         (pass_vectorize::gate): Call new function.
4834         (pass_vectorize::execute): Ditto.
4835         * opts.c: O3 default setting change.
4836         (finish_options): Check new flag.
4837         * doc/invoke.texi: Document new flags.
4838         * common.opt: New flags.
4839
4840 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
4841
4842         * doc/tm.texi.in (Cond Exec Macros): Remove node.
4843         (Condition Code): Don't reference it.
4844         * doc/tm.texi: Regenerate.
4845
4846 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
4847
4848         PR middle-end/58418
4849         * lra-constraints.c (undo_optional_reloads): Consider all optional
4850         reload even if it did not get a hard reg.
4851
4852 2013-09-16  Teresa Johnson  <tejohnson@google.com>
4853
4854         * dumpfile.c (dump_loc): Remove newline emission.
4855         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
4856         emission to dump_printf_loc calls where missing.
4857         (vect_mark_for_runtime_alias_test): Ditto.
4858         (vect_analyze_data_ref_dependence): Ditto.
4859         (vect_analyze_data_ref_dependences): Ditto.
4860         (vect_slp_analyze_data_ref_dependence): Ditto.
4861         (vect_slp_analyze_data_ref_dependences): Ditto.
4862         (vect_compute_data_ref_alignment): Ditto.
4863         (vect_update_misalignment_for_peel): Ditto.
4864         (vect_verify_datarefs_alignment): Ditto.
4865         (vector_alignment_reachable_p): Ditto.
4866         (vect_get_data_access_cost): Ditto.
4867         (vect_enhance_data_refs_alignment): Ditto.
4868         (vect_find_same_alignment_drs): Ditto.
4869         (vect_analyze_data_refs_alignment): Ditto.
4870         (vect_analyze_group_access): Ditto.
4871         (vect_analyze_data_ref_access): Ditto.
4872         (vect_analyze_data_ref_accesses): Ditto.
4873         (vect_prune_runtime_alias_test_list): Ditto.
4874         (vect_analyze_data_refs): Ditto.
4875         (vect_create_addr_base_for_vector_ref): Ditto.
4876         (vect_create_data_ref_ptr): Ditto.
4877         (vect_grouped_store_supported): Ditto.
4878         (vect_grouped_load_supported): Ditto.
4879         * value-prof.c (check_counter): Ditto.
4880         (check_ic_target): Ditto.
4881         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
4882         (vect_recog_widen_mult_pattern): Ditto.
4883         (vect_recog_widen_sum_pattern): Ditto.
4884         (vect_recog_over_widening_pattern): Ditto.
4885         (vect_recog_widen_shift_pattern): Ditto.
4886         (vect_recog_rotate_pattern): Ditto.
4887         (vect_recog_vector_vector_shift_pattern): Ditto.
4888         (vect_recog_divmod_pattern): Ditto.
4889         (vect_recog_mixed_size_cond_pattern): Ditto.
4890         (vect_recog_bool_pattern): Ditto.
4891         (vect_pattern_recog_1): Ditto.
4892         (vect_pattern_recog): Ditto.
4893         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
4894         (vect_is_simple_iv_evolution): Ditto.
4895         (vect_analyze_scalar_cycles_1): Ditto.
4896         (vect_get_loop_niters): Ditto.
4897         (vect_analyze_loop_1): Ditto.
4898         (vect_analyze_loop_form): Ditto.
4899         (vect_analyze_loop_operations): Ditto.
4900         (vect_analyze_loop_2): Ditto.
4901         (vect_analyze_loop): Ditto.
4902         (report_vect_op): Ditto.
4903         (vect_is_slp_reduction): Ditto.
4904         (vect_is_simple_reduction_1): Ditto.
4905         (vect_get_known_peeling_cost): Ditto.
4906         (vect_estimate_min_profitable_iters): Ditto.
4907         (vect_model_reduction_cost): Ditto.
4908         (vect_model_induction_cost): Ditto.
4909         (get_initial_def_for_induction): Ditto.
4910         (vect_create_epilog_for_reduction): Ditto.
4911         (vectorizable_reduction): Ditto.
4912         (vectorizable_induction): Ditto.
4913         (vectorizable_live_operation): Ditto.
4914         (vect_loop_kill_debug_uses): Ditto.
4915         (vect_transform_loop): Ditto.
4916         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
4917         (vect_stmt_relevant_p): Ditto.
4918         (process_use): Ditto.
4919         (vect_mark_stmts_to_be_vectorized): Ditto.
4920         (vect_model_simple_cost): Ditto.
4921         (vect_model_promotion_demotion_cost): Ditto.
4922         (vect_model_store_cost): Ditto.
4923         (vect_get_store_cost): Ditto.
4924         (vect_model_load_cost): Ditto.
4925         (vect_get_load_cost): Ditto.
4926         (vect_init_vector_1): Ditto.
4927         (vect_get_vec_def_for_operand): Ditto.
4928         (vect_finish_stmt_generation): Ditto.
4929         (vectorizable_call): Ditto.
4930         (vectorizable_conversion): Ditto.
4931         (vectorizable_assignment): Ditto.
4932         (vectorizable_shift): Ditto.
4933         (vectorizable_operation): Ditto.
4934         (vectorizable_store): Ditto.
4935         (vectorizable_load): Ditto.
4936         (vectorizable_condition): Ditto.
4937         (vect_analyze_stmt): Ditto.
4938         (vect_transform_stmt): Ditto.
4939         (vect_is_simple_use): Ditto.
4940         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
4941         (vect_can_advance_ivs_p): Ditto.
4942         (vect_update_ivs_after_vectorizer): Ditto.
4943         (vect_do_peeling_for_loop_bound): Ditto.
4944         (vect_gen_niters_for_prolog_loop): Ditto.
4945         (vect_update_inits_of_drs): Ditto.
4946         (vect_create_cond_for_alias_checks): Ditto.
4947         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
4948         (vect_build_slp_tree_1): Ditto.
4949         (vect_supported_load_permutation_p): Ditto.
4950         (vect_analyze_slp_instance): Ditto.
4951         (vect_analyze_slp): Ditto.
4952         (vect_make_slp_decision): Ditto.
4953         (vect_detect_hybrid_slp): Ditto.
4954         (vect_bb_vectorization_profitable_p): Ditto.
4955         (vect_slp_analyze_bb_1): Ditto.
4956         (vect_update_slp_costs_according_to_vf): Ditto.
4957         (vect_get_mask_element): Ditto.
4958         (vect_transform_slp_perm_load): Ditto.
4959         (vect_schedule_slp_instance): Ditto.
4960         (vect_schedule_slp): Ditto.
4961         (vect_slp_transform_bb): Ditto.
4962         * profile.c (read_profile_edge_counts): Ditto.
4963         (compute_branch_probabilities): Ditto.
4964         * coverage.c (get_coverage_counts): Ditto.
4965
4966 2013-09-16  Diego Novillo  <dnovillo@google.com>
4967
4968         * tree-core.h: Add missing comment lines from refactoring of tree.h.
4969
4970 2013-09-16  Jan Hubicka  <jh@suse.cz>
4971
4972         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
4973         abstract functions; for static functions check the presence of body.
4974
4975 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
4976
4977         * config/aarch64/aarch64-simd-builtins.def (fma): New.
4978         * config/aarch64/aarch64-simd.md
4979         (aarch64_mla_elt<mode>): New.
4980         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4981         (aarch64_mls_elt<mode>): Likewise.
4982         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4983         (aarch64_fma4_elt<mode>): Likewise.
4984         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4985         (aarch64_fma4_elt_to_128v2df): Likewise.
4986         (aarch64_fma4_elt_to_64df): Likewise.
4987         (fnma<mode>4): Likewise.
4988         (aarch64_fnma4_elt<mode>): Likewise.
4989         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4990         (aarch64_fnma4_elt_to_128v2df): Likewise.
4991         (aarch64_fnma4_elt_to_64df): Likewise.
4992         * config/aarch64/iterators.md (VDQSF): New.
4993         * config/aarch64/arm_neon.h
4994         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
4995         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
4996
4997 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
4998
4999         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5000         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5001         (aarch64_mul3_elt_to_128df): Likewise.
5002         (aarch64_mul3_elt_to_64v2df): Likewise.
5003         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5004         (VMUL): New.
5005         (VMUL_CHANGE_NLANES) Likewise.
5006         (h_con): Likewise.
5007         (f): Likewise.
5008         * config/aarch64/arm_neon.h
5009         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5010
5011 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5012
5013         * config/aarch64/arm_neon.h
5014         (vcvtx_high_f32_f64): Fix parameters.
5015
5016 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5017             Uros Bizjak  <ubizjak@gmail.com>
5018
5019         * config/alpha.c: Include tree-ssanames.h.
5020
5021 2013-09-16  Richard Biener  <rguenther@suse.de>
5022
5023         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5024         (dot_rdg_1): Handle control_dd.
5025         (create_edge_for_control_dependence): New function.
5026         (create_rdg_edges): Add control dependences if asked for.
5027         (build_rdg): Likewise.
5028         (generate_loops_for_partition): If there are not necessary
5029         control stmts remove all their dependencies.
5030         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5031         (distribute_loop): Pass on control dependences.
5032         (tree_loop_distribution): Compute control dependences and remove
5033         restriction on number of loop nodes.
5034
5035 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5036
5037         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5038         for internal calls.
5039
5040 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5041
5042         * cse.c (try_const_anchors): Punt on CC modes.
5043
5044 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5045
5046         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5047
5048 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5049
5050         PR target/58382
5051         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5052         calls to word_mode.
5053
5054 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5055
5056         PR target/48094
5057         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5058         seen.
5059         (darwin_objc1_section): Likewise.
5060         (darwin_file_end): Emit Image Info section when required.
5061
5062 2013-09-14  Jan Hubicka  <jh@suse.cz>
5063
5064         * tree-into-ssa.c (gate_into_ssa): New.
5065         (pass_data_build_ssa): Use it.
5066         * cgraph.h (expand_thunk): Update prototype.
5067         * cgraphunit.c (analyze_function): Expand thunks early.
5068         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5069         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5070         set lowered flag; do not add new function.
5071         (assemble_thunks_and_aliases): Update.
5072         * tree-ssa.c (gate_init_datastructures): New gate.
5073         (pass_data_init_datastructures): Use it.
5074
5075 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5076
5077         PR target/58269
5078         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5079         xmm register set described in the psABI.
5080
5081 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5082
5083         * dwarf2out.c (should_emit_struct_debug): Add check
5084         for type_decl variable is not NULL.
5085
5086 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5087
5088         * config.gcc: Use new winnt-c.c target hooks
5089         * config/t-winnt: New file
5090         * config/winnt-c.c: New file
5091         * doc/tm.texi.in: Document new hook
5092         * doc/tm.texi: Regenerated
5093
5094 2013-09-13  Jan Hubicka  <jh@suse.cz>
5095
5096         PR middle-end/58094
5097         * ipa-inline.c (check_callers): New function.
5098         (check_caller_edge): Remove.
5099         (want_inline_function_to_all_callers_p): Also permit alises that are
5100         called dirrectly.
5101         (inline_to_all_callers): Terminate the walk when devirtualization
5102         introduce new calls.
5103
5104 2013-09-13  Jan Hubicka  <jh@suse.cz>
5105
5106         * ipa-inline-analysis.c (struct growth_data): Add node.
5107         (do_estimate_growth_1): Fix detection of recursion.
5108
5109 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5110
5111         PR tree-optimization/58392
5112         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5113         to avoid shadowing of outer loop variable.  If
5114         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5115         replace_by_duplicate_decl simduid of loops that have it set and
5116         set dest_cfun->has_simduid_loops and/or
5117         dest_cfun->has_force_vect_loops.
5118         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5119         instead of maybe_lookup_decl.
5120         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5121         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5122         call.  Copy over force_vect and copy and remap simduid.  Set
5123         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5124         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5125         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5126         caller.  Don't set cfun->has_simduid_loops and/or
5127         cfun->has_force_vect_loops here.
5128         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5129         caller.
5130         (expand_call_inline, tree_function_versioning): Adjust copy_body
5131         callers.
5132
5133 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5134
5135         PR bootstrap/58388
5136         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5137         the assert if the edge was a speculative one.
5138
5139 2013-09-13  Richard Biener  <rguenther@suse.de>
5140
5141         * tree-data-ref.h (known_dependences_p): Move here ...
5142         * tree-loop-distribution.c (known_dependences_p): ... from here.
5143         (dump_rdg_component, debug_rdg_component): Remove.
5144         (dump_rdg): Adjust.
5145         (generate_loops_for_partition): Use gimple_uid instead of
5146         relying on matching stmt visit order.
5147         (rdg_build_partitions): Take starting stmt vector.
5148         (ldist_gen): Merge into ...
5149         (distribute_loop): ... this function.  Do not compute starting
5150         vertices vector.
5151         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5152
5153 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5154
5155         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5156         Set type attribute correctly. Set predicable_short_it attribute.
5157         (cmpsi_shiftsi): Remove %? from output template.
5158
5159 2013-09-13  Richard Biener  <rguenther@suse.de>
5160
5161         * tree-loop-distribution.c (struct rdg_component,
5162         rdg_defs_used_in_other_loops_p, free_rdg_components,
5163         rdg_build_components): Remove.
5164         (stmts_from_loop): Do not record virtual PHIs.
5165         (generate_loops_for_partition): Skip virtual PHIs.
5166         (build_rdg_partition_for_component): Rename to ...
5167         (build_rdg_partition_for_vertex): ... this and adjust.
5168         (rdg_build_partitions): Take a vector of starting vertices
5169         instead of components.  Remove unnecessary leftover handling.
5170         (ldist_gen): Do not build components or record other stores.
5171         (distribute_loop): Do not distribute loops containing stmts
5172         with side-effects.
5173
5174 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5175
5176         PR target/58314
5177         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5178
5179 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5180
5181         * config.gcc: Separate cases for mingw and cygwin targets,
5182         and add 64-bit cygwin target case.
5183
5184         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5185         dll-export inline-functions.
5186         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5187
5188
5189 2013-09-13  Jeff Law  <law@redhat.com>
5190
5191         PR middle-end/58387
5192         Revert:
5193         2013-09-06  Jeff Law  <law@redhat.com>
5194
5195         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5196         edge implied equivalences into successor phis.
5197
5198 2013-09-12  DJ Delorie  <dj@redhat.com>
5199
5200         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5201         separators.
5202
5203 2013-09-12  Brooks Moses  <bmoses@google.com>
5204
5205         PR driver/42955
5206         * Makefile.in: Do not install driver binaries in $(target)/bin.
5207
5208 2013-09-12  DJ Delorie  <dj@redhat.com>
5209
5210         * config/rl78/rl78.opt (mrelax): New.
5211         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5212         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5213
5214         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5215         between SP and FP.
5216         (rl78_expand_epilogue): Likewise.
5217
5218 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5219
5220         PR middle-end/58335
5221         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5222         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5223         pointer elimination with using remove_reg_equal_offset_note.
5224
5225 2013-09-12  DJ Delorie  <dj@redhat.com>
5226
5227         * config/msp430/: New port.
5228         * config.gcc (msp430): Added.
5229         * doc/invoke.texi: Document MSP430 options.
5230         * doc/install.texi: Document msp430-elf
5231         * doc/md.texi: Document msp430-elf
5232         * doc/contrib.texi: Document msp430-elf
5233
5234         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5235         PSImode.
5236
5237 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5238
5239         PR ipa/58389
5240         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5241         reference descriptor is NULL.
5242         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5243         edge to NULL.
5244
5245 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5246
5247         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5248         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5249         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5250         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5251         make_temp_ssa_name): move to tree-ssanames.h
5252         * tree-ssa-alias.h: Move prototype.
5253         * tree-ssa.h: Include tree-ssanames.h.
5254         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5255         * tree-ssanames.h: New.  Move items from tree-flow*.h
5256         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5257
5258 2013-09-12  Richard Biener  <rguenther@suse.de>
5259
5260         PR tree-optimization/58404
5261         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5262         propagate non-invariant addresses into dereferences wrapped
5263         in component references.
5264
5265 2013-09-12  Richard Biener  <rguenther@suse.de>
5266
5267         PR tree-optimization/58402
5268         * passes.def: Move pass_late_warn_uninitialized later.
5269
5270 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5271
5272         * tree-ssa.h: New.  Move content from tree-flow.h and
5273         tree-flow-inline.h.
5274         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5275         Move prototypes belonging to tree-ssa.c.
5276         * tree-flow-inline.h (redirect_edge_var_map_def,
5277         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5278         tree-ssa.h.
5279         * gimple.h: Adjust prototypes.
5280         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5281         to...
5282         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5283         * tree.h: Move prototype to tree-ssa.h.
5284         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5285         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5286         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5287         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5288         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5289         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5290         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5291         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5292         graphite-blocking.c, graphite-clast-to-gimple.c,
5293         graphite-dependences.c, graphite-interchange.c,
5294         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5295         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5296         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5297         ipa-reference.c, ipa-split.c, ipa-utils.c,
5298         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5299         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5300         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5301         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5302         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5303         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5304         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5305         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5306         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5307         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5308         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5309         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5310         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5311         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5312         value-prof.c, var-tracking.c,
5313         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5314
5315 2013-09-12  Richard Biener  <rguenther@suse.de>
5316
5317         PR tree-optimization/58396
5318         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5319         (build_rdg): Take a loop-nest parameter, fix memleaks.
5320         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5321
5322 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5323
5324         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5325         for SLM.
5326
5327 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5328
5329         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5330         The FMA instruction names should have a 'v' prefix.
5331
5332 2013-09-12  Richard Biener  <rguenther@suse.de>
5333
5334         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5335         (dot_rdg): Use popen instead of system in optional code.
5336         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5337         (already_processed_vertex_p): Adjust.
5338         (has_anti_or_output_dependence, predecessor_has_mem_write,
5339         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5340         rdg_flag_uses): Remove.
5341         (rdg_flag_vertex): Simplify.
5342         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5343         remove recursion.
5344         (build_rdg_partition_for_component): Process the first vertex
5345         of a component only.
5346         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5347
5348 2013-09-12  Alan Modra  <amodra@gmail.com>
5349
5350         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5351
5352 2013-09-11  DJ Delorie  <dj@redhat.com>
5353             Nick Clifton  <nickc@redhat.com>
5354
5355         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5356         (rl78_stack_based_mem): New.
5357         * config/rl78/constraints.md (Iv08): New.
5358         (Iv16): New.
5359         (Iv24): New.
5360         (Is09): New.
5361         (Is17): New.
5362         (Is25): New.
5363         (ISsi): New.
5364         (IShi): New.
5365         (ISqi): New.
5366         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5367         (movhi): Likewise.
5368         (movsi): Change from expand to insn-and-split.
5369         (ashrsi3): Clobber AX.
5370         (lshrsi3): New.
5371         (ashlsi3): New.
5372         (cbranchsi4): New.
5373         * config/rl78/rl78.md (CC_REG): Fix.
5374         (addsi3): Allow memory and immediate operands.
5375         (addsi3_internal): Split into...
5376         (addsi3_internal_virt): ...new, and ...
5377         (addsi3_internal_real): ...new.
5378         (subsi): New.
5379         (subsi3_internal_virt): New.
5380         (subsi3_internal_real): New.
5381         (mulsi3): Add memory operand.
5382         (mulsi3_rl78): Likewise.
5383         (mulsi3_g13): Likewise.
5384         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5385         (cbranchqi4_real): Add more constraint options.
5386         (cbranchhi4_real): Expand pattern.
5387         (cbranchhi4_real_signed): New.
5388         (cbranchhi4_real_inverted): New.
5389         (cbranchsi4_real_lt): New.
5390         (cbranchsi4_real_ge): New.
5391         (cbranchsi4_real_signed): New.
5392         (cbranchsi4_real): New.
5393         (peephole2): New.
5394         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5395         constant shifts.
5396         (lshrsi3_virt): Likewise.
5397         (ashlsi3_virt): Likewise.
5398         (cbranchqi4_virt_signed): New.
5399         (cbranchhi4_virt_signed): New.
5400         (cbranchsi4_virt): New.
5401         * config/rl78/rl78.c: Whitespace fixes throughout.
5402         (move_elim_pass): New.
5403         (pass_data_rl78_move_elim): New.
5404         (pass_rl78_move_elim): New.
5405         (make_pass_rl78_move_elim): New.
5406         (rl78_devirt_info): Run devirt earlier.
5407         (rl78_move_elim_info): New.
5408         (rl78_asm_file_start): Register it.
5409         (rl78_split_movsi): New.
5410         (rl78_as_legitimate_address): Allow virtual base registers when
5411         appropriate.
5412         (rl78_addr_space_convert): Remove spurious debug stuff.
5413         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5414         (rl78_print_operand): More cases for not printing '#'.
5415         (rl78_expand_compare): Remove most of the logic.
5416         (content_memory): New.
5417         (clear_content_memory): New.
5418         (get_content_index): New.
5419         (get_content_name): New.
5420         (display_content_memory): New.
5421         (update_content): New.
5422         (record_content): New.
5423         (already_contains): New.
5424         (insn_ok_now): Re-recog insns with virtual registers.
5425         (add_postponed_content_update): New.
5426         (process_postponed_content_update): New.
5427         (gen_and_emit_move): New.
5428         (transcode_memory_rtx): Record new location content.
5429         Use gen_and_emit_move.
5430         (force_into_acc): New.
5431         (move_to_acc): Use gen_and_emit_move.
5432         (move_from_acc): Likewise.
5433         (move_acc_to_reg): Likewise.
5434         (move_to_x): Likewise.
5435         (move_to_hl): Likewise.
5436         (move_to_de): Likewise.
5437         (rl78_alloc_physical_registers_op1): Record location content.
5438         (has_constraint): New.
5439         (rl78_alloc_physical_registers_op2): Record location content.
5440         Optimize use of HL.
5441         (rl78_alloc_physical_registers_ro1): Likewise.
5442         (rl78_alloc_physical_registers_cmp): Likewise.
5443         (rl78_alloc_physical_registers_umul): Likewise.
5444         (rl78_alloc_address_registers_macax): New.
5445         (rl78_alloc_physical_registers): Initialize and set location
5446         content memory as needed.
5447         (rl78_reorg): Make sure split2 is called.
5448         (rl78_rtx_costs): New.
5449
5450 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5451
5452         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5453         for (not (neg ...)) and (neg (not ...)) cases.
5454
5455 2013-09-11  Richard Biener  <rguenther@suse.de>
5456
5457         PR middle-end/58377
5458         * passes.def: Split critical edges before late uninit warning passes.
5459         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5460
5461 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5462
5463         PR tree-optimization/58385
5464         * fold-const.c (build_range_check): If both low and high are NULL,
5465         use omit_one_operand_loc to preserve exp side-effects.
5466
5467 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5468
5469         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5470
5471 2013-09-11  Richard Biener  <rguenther@suse.de>
5472
5473         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5474         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5475         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5476         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5477         stmts_from_loop, known_dependences_p, build_empty_rdg,
5478         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5479         * tree-loop-distribution.c: ... here.
5480         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5481         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5482         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5483         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5484         * tree-loop-distribution.c: ... here.
5485         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5486         (struct partition_s): Add loops member.
5487         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5488         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5489         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5490
5491 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5492             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5493             Sergey Lega  <sergey.s.lega@intel.com>
5494             Anna Tikhonova  <anna.tikhonova@intel.com>
5495             Ilya Tocar  <ilya.tocar@intel.com>
5496             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5497             Ilya Verbin  <ilya.verbin@intel.com>
5498             Kirill Yukhin  <kirill.yukhin@intel.com>
5499             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5500
5501         * config/i386/constraints.md (k): New.
5502         (Yk): Ditto.
5503         * config/i386/i386.c (const regclass_map): Add new mask registers.
5504         (dbx_register_map): Ditto.
5505         (dbx64_register_map): Ditto.
5506         (svr4_dbx_register_map): Ditto.
5507         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5508         disabled.
5509         (ix86_preferred_reload_class): Disable constants for mask registers.
5510         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5511         (ix86_hard_regno_mode_ok): Support new mask registers.
5512         (x86_order_regs_for_local_alloc): Ditto.
5513         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5514         (FIXED_REGISTERS): Add new mask registers.
5515         (CALL_USED_REGISTERS): Ditto.
5516         (REG_ALLOC_ORDER): Ditto.
5517         (VALID_MASK_REG_MODE): New.
5518         (FIRST_MASK_REG): Ditto.
5519         (LAST_MASK_REG): Ditto.
5520         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5521         (MAYBE_MASK_CLASS_P): New.
5522         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5523         (REG_CLASS_CONTENTS): Ditto.
5524         (MASK_REGNO_P): New.
5525         (ANY_MASK_REG_P): Ditto.
5526         (HI_REGISTER_NAMES): Add new mask registers.
5527         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5528         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5529         mask registers.
5530         (attribute "type"): Add mskmov, msklog.
5531         (attribute "length_immediate"): Support them.
5532         (attribute "memory"): Ditto.
5533         (attribute "prefix_0f"): Ditto.
5534         (*movhi_internal): Support new mask registers.
5535         (*movqi_internal): Ditto.
5536         (define_split): Split out clobber pattern is a logic
5537         insn on mask registers.
5538         (*k<logic><mode>): New.
5539         (*andhi_1): Extend to support mask regs.
5540         (*andqi_1): Extend to support mask regs.
5541         (kandn<mode>): New.
5542         (define_split): Split and-not to and and not if operands
5543         are not mask regs.
5544         (*<code><mode>_1): Separate HI mode to new pattern...
5545         (*<code>hi_1): This.
5546         (*<code>qi_1): Extend to support mask regs.
5547         (kxnor<mode>): New.
5548         (kortestzhi): Ditto.
5549         (kortestchi): Ditto.
5550         (kunpckhi): Ditto.
5551         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5552         (*one_cmplhi2_1): ...Here, now with mask registers support.
5553         (*one_cmplqi2_1): Support new mask registers.
5554         (HI/QImode arithmetics splitter): Don't split if mask registers
5555         are used.
5556         (HI/QImode not splitter): Ditto.
5557         * config/i386/predicated.md (mask_reg_operand): New.
5558         (general_reg_operand): Ditto.
5559
5560 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5561
5562         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5563         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5564
5565 2013-09-10  Jeff Law  <law@redhat.com>
5566
5567         PR tree-optimization/58380
5568         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5569         of threading through a buried loop header.
5570
5571         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5572         return value for single successor case.
5573
5574 2013-09-10  Jan Hubicka  <jh@suse.cz>
5575
5576         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5577
5578 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5579
5580         PR target/58361
5581         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5582         support conditional execution.
5583         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5584
5585 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5586
5587         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5588         constraint pass.
5589         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5590         Check destination too to check move insn.
5591         (undo_optional_reloads): Add check that the original peudo did not
5592         changed its allocation and the optional reload was inherited on last
5593         inheritance pass.  Break loop after deciding to keep optional reload.
5594         (lra_undo_inheritance): Add check that inherited pseudo still in
5595         memory.
5596
5597 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5598
5599         * config/aarch64/aarch64.md (generic_sched): New.
5600         * config/aarch64/aarch64-generic.md (load): Make conditional
5601         on generic_sched attribute.
5602         (nonload): Likewise.
5603
5604 2013-09-10  Jan Hubicka  <jh@suse.cz>
5605
5606         * lto-cgraph.c: Include ipa-utils.h.
5607         (compute_ltrans_boundary): Also add possible targets into the boundary.
5608
5609 2013-09-10  Jan Hubicka  <jh@suse.cz>
5610
5611         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5612         VAR_DECL of vtable rather than full expression.
5613
5614 2013-09-10  Jan Hubicka  <jh@suse.cz>
5615             Paolo Carlini  <paolo.carlini@oracle.com>
5616
5617         * cgraphunit.c (analyze_functions): Save input_location, set it
5618         to UNKNOWN_LOCATION and restore it at the end.
5619
5620 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5621
5622         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5623         represented by a thunk.
5624
5625 2013-09-10  Jeff Law  <law@redhat.com>
5626
5627         PR tree-optimization/58343
5628         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5629         jump threading requests through loop headers buried in the middle
5630         of a jump threading path.
5631
5632         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5633         in return value/type.
5634
5635 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5636
5637         PR rtl-optimization/58365
5638         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5639         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5640         it differs.
5641
5642 2013-09-10  Richard Biener  <rguenther@suse.de>
5643
5644         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5645         * tree-data-ref.c (create_rdg_vertices): Collect all data
5646         references, signal failure to the caller, use data-ref API.
5647         (build_rdg): Compute data references only once.  Maintain lifetime
5648         of data references and data dependences from within RDG.
5649         (free_rdg): Free dependence relations.
5650         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5651         inventing extra dependences.
5652         (distribute_loop): Update for RDG API changes.
5653
5654 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5655
5656         * doc/invoke.texi (fms-extensions): Document changed
5657         behavior for ms-abi targets.
5658         * config/i386/i386.c (ix86_option_override_internal):
5659         Set default value of option -fms-extension for ms-abi targets.
5660
5661 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5662
5663         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5664
5665 2013-09-10  Alan Modra  <amodra@gmail.com>
5666
5667         PR target/58330
5668         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5669
5670 2013-09-10  Alan Modra  <amodra@gmail.com>
5671
5672         * config/rs6000/predicates.md (add_cint_operand): New.
5673         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5674         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5675         using add_cint_operand.
5676         (largetoc_high_plus_aix): Likewise.
5677
5678 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5679
5680         PR tree-optimization/58364
5681         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5682         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5683         the current range can't be an unconditional true or false.
5684
5685 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5686
5687         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5688
5689 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5690
5691         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5692
5693 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5694
5695         PR c++/43452
5696         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5697
5698 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5699
5700         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5701         NO_REGS for immediate that can't be moved directly into FP_REGS.
5702
5703 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5704
5705         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5706         comparison with negated operand.
5707         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5708         RTL form.
5709
5710 2013-09-09  Richard Biener  <rguenther@suse.de>
5711
5712         PR middle-end/58326
5713         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5714         of a subloop record all its block as affecting loop-closed SSA form.
5715
5716 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5717
5718         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5719         of an rtx/bitpos pair.
5720         (store_fixed_bit_field): Update accordingly.
5721
5722 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5723
5724         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5725         GEN_INT.
5726         * builtins.c (expand_errno_check): Likewise.
5727         * dwarf2cfi.c (init_return_column_size): Likewise.
5728         * except.c (sjlj_mark_call_sites): Likewise.
5729         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
5730         * lra-constraints.c (emit_inc): Likewise.
5731         * ree.c (combine_set_extension): Likewise.
5732         * regmove.c (fixup_match_2): Likewise.
5733         * reload1.c (inc_for_reload): Likewise.
5734
5735 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5736
5737         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
5738         (force_to_mode, simplify_shift_const_1, simplify_comparison):
5739         Use gen_int_mode with the mode of the associated simplify_* call.
5740         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
5741         * expmed.c (expand_shift_1): Likewise.
5742         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5743         * loop-iv.c (iv_number_of_iterations): Likewise.
5744         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5745         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
5746
5747 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5748
5749         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
5750         of the associated expand_* call.
5751         (asan_emit_stack_protection): Likewise.
5752         * builtins.c (round_trampoline_addr): Likewise.
5753         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
5754         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
5755         (emit_store_flag): Likewise.
5756         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
5757         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
5758         Likewise.
5759         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5760         * ifcvt.c (noce_try_store_flag_constants): Likewise.
5761         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5762         * modulo-sched.c (generate_prolog_epilog): Likewise.
5763         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
5764         (expand_ctz, expand_ffs, expand_unop): Likewise.
5765
5766 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5767
5768         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
5769         of the associated gen_rtx_* call.
5770         * caller-save.c (init_caller_save): Likewise.
5771         * combine.c (find_split_point, make_extraction): Likewise.
5772         (make_compound_operation): Likewise.
5773         * dwarf2out.c (mem_loc_descriptor): Likewise.
5774         * explow.c (plus_constant, probe_stack_range): Likewise.
5775         * expmed.c (expand_mult_const): Likewise.
5776         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
5777         * reload1.c (init_reload): Likewise.
5778         * valtrack.c (cleanup_auto_inc_dec): Likewise.
5779         * var-tracking.c (adjust_mems): Likewise.
5780         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
5781         rather than gen_rtx_fmt_ee.
5782
5783 2013-09-09  Jan Hubicka  <jh@suse.cz>
5784
5785         PR middle-end/58294
5786         * value-prof.c (gimple_ic): Copy also abnormal edges.
5787
5788 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5789
5790         * asan.c (asan_shadow_cst): Use gen_int_mode.
5791
5792 2013-09-08  Jan Hubicka  <jh@suse.cz>
5793
5794         * ipa-profile.c: Add toplevel comment.
5795         (ipa_propagate_frequency_1): Be more conservative when profile is read.
5796         (contains_hot_call_p): New function.
5797         (ipa_propagate_frequency): Set frequencies based on counts when
5798         profile is read.
5799         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
5800         profile; do not tamper with profile after inlining if it is read.
5801
5802 2013-09-08  Jan Hubicka  <jh@suse.cz>
5803
5804         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
5805         speculative edges.
5806
5807 2013-09-08  Jan Hubicka  <jh@suse.cz>
5808
5809         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
5810         summary generation.
5811
5812 2013-09-08  Jeff Law  <law@redhat.com>
5813
5814         PR bootstrap/58340
5815         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
5816         of 'found'.
5817
5818 2013-09-08  Andi Kleen  <ak@linux.intel.com>
5819
5820         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
5821
5822 2013-09-08  Jan Hubicka  <jh@suse.cz>
5823
5824         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
5825
5826 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
5827
5828         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
5829         for non-debug insns.
5830         * lra.c (new_insn_reg): Take the containing insn as a parameter.
5831         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
5832         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
5833         accordingly.
5834
5835 2013-09-08  Jan Hubicka  <jh@suse.cz>
5836
5837         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
5838         targets and devirtualize to BUILT_IN_UNREACHABLE.
5839         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
5840         * ipa.c (walk_polymorphic_call_targets): New function.
5841         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
5842         functions; use the new timevar.
5843         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
5844         was removed from the program.
5845         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
5846         not consider it in the walk when its vtable is dead.
5847         (possible_polymorphic_call_targets_1): Pass anonymous flag to
5848         record_binfo.
5849         (devirt_variable_node_removal_hook): New function.
5850         (possible_polymorphic_call_targets): Also register
5851         devirt_variable_node_removal_hook.
5852         (ipa_devirt): Do not do non-speculative devirtualization.
5853         (gate_ipa_devirt): One execute if devirtualizing speculatively.
5854
5855 2013-09-08  Jan Hubicka  <jh@suse.cz>
5856
5857         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
5858         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
5859         varpool_remove_variable_insertion_hook): Declare.
5860         * varpool.c (varpool_node_hook_list): New structure.
5861         (first_varpool_node_removal_hook,
5862         first_varpool_variable_insertion_hook): New variables.
5863         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
5864         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
5865         varpool_remove_variable_insertion_hook,
5866         varpool_call_variable_insertion_hooks): New functions.
5867         (varpool_remove_node): Use it.
5868
5869 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
5870
5871         PR c++/54941
5872         * diagnostic.c (diagnostic_build_prefix): When s.file is
5873         "<built-in>" don't output line and column numbers.
5874
5875 2013-09-06  Jan Hubicka  <jh@suse.cz>
5876
5877         * cgraphunit.c (expand_thunk): Get body before touching arguments.
5878         * lto-streamer-out.c: Stream thunks, too.
5879         * lto-streamer-in.c (input_function): Pop cfun here
5880         (lto_read_body): Instead of here.
5881
5882 2013-09-06  Caroline Tice  <cmtice@google.com>
5883
5884         * doc/install.texi: Add documentation for the --enable-vtable-verify
5885         and the --disable-libvtv configure options.
5886
5887 2013-09-06  Jeff Law  <law@redhat.com>
5888
5889         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5890         edge implied equivalences into successor phis.
5891
5892 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
5893
5894         * resource.c (mark_referenced_resources): Handle COND_EXEC.
5895
5896 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
5897
5898         * resource.c (mark_target_live_regs): Compute resources taking
5899         into account if a call is predicated or not.
5900
5901 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
5902
5903         * toplev.c (output_stack_usage): Be prepared for suffixes created by
5904         the compiler in the function names.
5905
5906 2013-09-06  Jan Hubicka  <jh@suse.cz>
5907
5908         PR middle-end/58094
5909         * ipa-inline.c (has_caller_p): New function.
5910         (want_inline_function_to_all_callers_p): Use it.
5911         (sum_callers, inline_to_all_callers): Break out from ...
5912         (ipa_inline): ... here.
5913
5914 2013-09-06  Jan Hubicka  <jh@suse.cz>
5915
5916         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
5917         only when AVX is enabled.
5918
5919 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5920
5921         * config/aarch64/aarch64.md
5922         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
5923         is fpsimd_<load/store>2.
5924         (load_pair<mode>): Likewise.
5925         (store_pair<mode>): Likewise.
5926
5927 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5928
5929         * config/arm/types.md (type): Add "mrs" type.
5930         * config/aarch64/aarch64.md
5931         (aarch64_load_tp_hard): Make type "mrs".
5932         * config/arm/arm.md
5933         (load_tp_hard): Make type "mrs".
5934         * config/arm/cortex-a15.md: Update with new attributes.
5935         * config/arm/cortex-a5.md: Update with new attributes.
5936         * config/arm/cortex-a53.md: Update with new attributes.
5937         * config/arm/cortex-a7.md: Update with new attributes.
5938         * config/arm/cortex-a8.md: Update with new attributes.
5939         * config/arm/cortex-a9.md: Update with new attributes.
5940         * config/arm/cortex-m4.md: Update with new attributes.
5941         * config/arm/cortex-r4.md: Update with new attributes.
5942         * config/arm/fa526.md: Update with new attributes.
5943         * config/arm/fa606te.md: Update with new attributes.
5944         * config/arm/fa626te.md: Update with new attributes.
5945         * config/arm/fa726te.md: Update with new attributes.
5946
5947 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5948
5949         * config/aarch64/aarch64.md
5950         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
5951         (*movtf_aarch64): Likewise.
5952         * config/arm/arm.md
5953         (thumb1_movdi_insn): Use "multiple" for type where more than one
5954         instruction is used for a move.
5955         (*arm32_movhf): Likewise.
5956         (*thumb_movdf_insn): Likewise.
5957
5958 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5959
5960         * config/arm/types.md (type): Rename fcpys to fmov.
5961         * config/arm/vfp.md
5962         (*arm_movsi_vfp): Rename type fcpys as fmov.
5963         (*thumb2_movsi_vfp): Likewise
5964         (*movhf_vfp_neon): Likewise
5965         (*movhf_vfp): Likewise
5966         (*movsf_vfp): Likewise
5967         (*thumb2_movsf_vfp): Likewise
5968         (*movsfcc_vfp): Likewise
5969         (*thumb2_movsfcc_vfp): Likewise
5970         * config/aarch64/aarch64-simd.md
5971         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
5972         * config/aarch64/aarch64.md
5973         (*movsi_aarch64): Replace type mov_reg with fmovs.
5974         (*movdi_aarch64): Likewise
5975         (*movsf_aarch64): Likewise
5976         (*movdf_aarch64): Likewise
5977         * config/arm/arm.c
5978         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
5979         * config/arm/iwmmxt.md
5980         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
5981         * config/arm/arm1020e.md: Update with new attributes.
5982         * config/arm/cortex-a15-neon.md: Update with new attributes.
5983         * config/arm/cortex-a5.md: Update with new attributes.
5984         * config/arm/cortex-a53.md: Update with new attributes.
5985         * config/arm/cortex-a7.md: Update with new attributes.
5986         * config/arm/cortex-a8-neon.md: Update with new attributes.
5987         * config/arm/cortex-a9.md: Update with new attributes.
5988         * config/arm/cortex-m4-fpu.md: Update with new attributes.
5989         * config/arm/cortex-r4f.md: Update with new attributes.
5990         * config/arm/marvell-pj4.md: Update with new attributes.
5991         * config/arm/vfp11.md: Update with new attributes.
5992
5993 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
5994
5995         * config/aarch64/aarch64.md
5996         (*madd<mode>): Fix type attribute.
5997         (*maddsi_uxtw): Likewise.
5998         (*msub<mode>): Likewise.
5999         (*msubsi_uxtw): Likewise.
6000         (<su_optab>maddsidi4): Likewise.
6001         (<su_optab>msubsidi4): Likewise.
6002
6003 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6004
6005         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6006         * config/arm/arm.md (core_cycles): Remove fdiv.
6007         * config/arm/vfp.md:
6008         (*sqrtsf2_vfp): Update for attribute changes.
6009         (*sqrtdf2_vfp): Likewise.
6010         * config/aarch64/aarch64.md:
6011         (sqrt<mode>2): Update for attribute changes.
6012         * config/arm/arm1020e.md: Update with new attributes.
6013         * config/arm/cortex-a15-neon.md: Update with new attributes.
6014         * config/arm/cortex-a5.md: Update with new attributes.
6015         * config/arm/cortex-a53.md: Update with new attributes.
6016         * config/arm/cortex-a7.md: Update with new attributes.
6017         * config/arm/cortex-a8-neon.md: Update with new attributes.
6018         * config/arm/cortex-a9.md: Update with new attributes.
6019         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6020         * config/arm/cortex-r4f.md: Update with new attributes.
6021         * config/arm/marvell-pj4.md: Update with new attributes.
6022         * config/arm/vfp11.md: Update with new attributes.
6023
6024 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6025
6026         * config/arm/types.md
6027         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6028         * config/aarch64/aarch64.md
6029         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6030         new attributes.
6031         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6032         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6033         (float<GPI:mode><GPF:mode>2): Likewise.
6034         * config/arm/vfp.md
6035         (*truncsisf2_vfp): Update with new attributes.
6036         (*truncsidf2_vfp): Likewise.
6037         (fixuns_truncsfsi2): Likewise.
6038         (fixuns_truncdfsi2): Likewise.
6039         (*floatsisf2_vfp): Likewise.
6040         (*floatsidf2_vfp): Likewise.
6041         (floatunssisf2): Likewise.
6042         (floatunssidf2): Likewise.
6043         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6044         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6045         * config/arm/arm1020e.md: Update with new attributes.
6046         * config/arm/cortex-a15-neon.md: Update with new attributes.
6047         * config/arm/cortex-a5.md: Update with new attributes.
6048         * config/arm/cortex-a53.md: Update with new attributes.
6049         * config/arm/cortex-a7.md: Update with new attributes.
6050         * config/arm/cortex-a8-neon.md: Update with new attributes.
6051         * config/arm/cortex-a9.md: Update with new attributes.
6052         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6053         * config/arm/cortex-r4f.md: Update with new attributes.
6054         * config/arm/marvell-pj4.md: Update with new attributes.
6055         * config/arm/vfp11.md: Update with new attributes.
6056
6057 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6058
6059         * config/aarch64/arm_neon.h
6060         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6061         (vqtbx<1,2,3,4><q>_s8): Likewise.
6062
6063 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6064
6065         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6066         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6067         (add<mode>3): Add type attribute.
6068         (add<mode>3): Likewise.
6069         (usadd<mode>3): Likewise.
6070         (ssadd<mode>3): Likewise.
6071         (sub<mode>3): Likewise.
6072         (sub<mode>3): Likewise.
6073         (ussub<mode>3): Likewise.
6074         (sssub<mode>3): Likewise.
6075         (ssmulsa3): Likewise.
6076         (usmulusa3): Likewise.
6077         (arm_usatsihi): Likewise.
6078         * config/arm/vfp.md
6079         (*movdi_vfp): Add types for all instructions.
6080         (*movdi_vfp_cortexa8): Likewise.
6081         (*movhf_vfp_neon): Likewise.
6082         (*movhf_vfp): Likewise.
6083         (*movdf_vfp): Likewise.
6084         (*thumb2_movdf_vfp): Likewise.
6085         (*thumb2_movdfcc_vfp): Likewise.
6086         * config/arm/arm.md: Add type attribute to all insn patterns.
6087         (*thumb1_adddi3): Add type attribute.
6088         (*arm_adddi3): Likewise.
6089         (*adddi_sesidi_di): Likewise.
6090         (*adddi_zesidi_di): Likewise.
6091         (*thumb1_addsi3): Likewise.
6092         (addsi3_compare0): Likewise.
6093         (*addsi3_compare0_scratch): Likewise.
6094         (*compare_negsi_si): Likewise.
6095         (cmpsi2_addneg): Likewise.
6096         (*addsi3_carryin_<optab>): Likewise.
6097         (*addsi3_carryin_alt2_<optab>): Likewise.
6098         (*addsi3_carryin_clobercc_<optab>): Likewise.
6099         (*subsi3_carryin): Likewise.
6100         (*subsi3_carryin_const): Likewise.
6101         (*subsi3_carryin_compare): Likewise.
6102         (*subsi3_carryin_compare_const): Likewise.
6103         (*arm_subdi3): Likewise.
6104         (*thumb_subdi3): Likewise.
6105         (*subdi_di_zesidi): Likewise.
6106         (*subdi_di_sesidi): Likewise.
6107         (*subdi_zesidi_di): Likewise.
6108         (*subdi_sesidi_di): Likewise.
6109         (*subdi_zesidi_ze): Likewise.
6110         (thumb1_subsi3_insn): Likewise.
6111         (*arm_subsi3_insn): Likewise.
6112         (*anddi3_insn): Likewise.
6113         (*anddi_zesidi_di): Likewise.
6114         (*anddi_sesdi_di): Likewise.
6115         (*ne_zeroextracts): Likewise.
6116         (*ne_zeroextracts): Likewise.
6117         (*ite_ne_zeroextr): Likewise.
6118         (*ite_ne_zeroextr): Likewise.
6119         (*anddi_notdi_di): Likewise.
6120         (*anddi_notzesidi): Likewise.
6121         (*anddi_notsesidi): Likewise.
6122         (andsi_notsi_si): Likewise.
6123         (thumb1_bicsi3): Likewise.
6124         (*iordi3_insn): Likewise.
6125         (*iordi_zesidi_di): Likewise.
6126         (*iordi_sesidi_di): Likewise.
6127         (*thumb1_iorsi3_insn): Likewise.
6128         (*xordi3_insn): Likewise.
6129         (*xordi_zesidi_di): Likewise.
6130         (*xordi_sesidi_di): Likewise.
6131         (*arm_xorsi3): Likewise.
6132         (*andsi_iorsi3_no): Likewise.
6133         (*smax_0): Likewise.
6134         (*smax_m1): Likewise.
6135         (*arm_smax_insn): Likewise.
6136         (*smin_0): Likewise.
6137         (*arm_smin_insn): Likewise.
6138         (*arm_umaxsi3): Likewise.
6139         (*arm_uminsi3): Likewise.
6140         (*minmax_arithsi): Likewise.
6141         (*minmax_arithsi_): Likewise.
6142         (*satsi_<SAT:code>): Likewise.
6143         (arm_ashldi3_1bit): Likewise.
6144         (arm_ashrdi3_1bit): Likewise.
6145         (arm_lshrdi3_1bit): Likewise.
6146         (*arm_negdi2): Likewise.
6147         (*thumb1_negdi2): Likewise.
6148         (*arm_negsi2): Likewise.
6149         (*thumb1_negsi2): Likewise.
6150         (*negdi_extendsid): Likewise.
6151         (*negdi_zero_extend): Likewise.
6152         (*arm_abssi2): Likewise.
6153         (*thumb1_abssi2): Likewise.
6154         (*arm_neg_abssi2): Likewise.
6155         (*thumb1_neg_abss): Likewise.
6156         (one_cmpldi2): Likewise.
6157         (extend<mode>di2): Likewise.
6158         (*compareqi_eq0): Likewise.
6159         (*arm_extendhisi2addsi): Likewise.
6160         (*arm_movdi): Likewise.
6161         (*thumb1_movdi_insn): Likewise.
6162         (*arm_movt): Likewise.
6163         (*thumb1_movsi_insn): Likewise.
6164         (pic_add_dot_plus_four): Likewise.
6165         (pic_add_dot_plus_eight): Likewise.
6166         (tls_load_dot_plus_eight): Likewise.
6167         (*thumb1_movhi_insn): Likewise.
6168         (*thumb1_movsf_insn): Likewise.
6169         (*movdf_soft_insn): Likewise.
6170         (*thumb_movdf_insn): Likewise.
6171         (cbranchsi4_insn): Likewise.
6172         (cbranchsi4_scratch): Likewise.
6173         (*negated_cbranchsi4): Likewise.
6174         (*tbit_cbranch): Likewise.
6175         (*tlobits_cbranch): Likewise.
6176         (*tstsi3_cbranch): Likewise.
6177         (*cbranchne_decr1): Likewise.
6178         (*addsi3_cbranch): Likewise.
6179         (*addsi3_cbranch_scratch): Likewise.
6180         (*arm_cmpdi_insn): Likewise.
6181         (*arm_cmpdi_unsig): Likewise.
6182         (*arm_cmpdi_zero): Likewise.
6183         (*thumb_cmpdi_zero): Likewise.
6184         (*deleted_compare): Likewise.
6185         (*mov_scc): Likewise.
6186         (*mov_negscc): Likewise.
6187         (*mov_notscc): Likewise.
6188         (*cstoresi_eq0_thumb1_insn): Likewise.
6189         (cstoresi_nltu_thumb1): Likewise.
6190         (cstoresi_ltu_thu): Likewise.
6191         (thumb1_addsi3_addgeu): Likewise.
6192         (*arm_jump): Likewise.
6193         (*thumb_jump): Likewise.
6194         (*check_arch2): Likewise.
6195         (arm_casesi_internal): Likewise.
6196         (thumb1_casesi_dispatch): Likewise.
6197         (*arm_indirect_jump): Likewise.
6198         (*thumb1_indirect_jump): Likewise.
6199         (nop): Likewise.
6200         (*and_scc): Likewise.
6201         (*ior_scc): Likewise.
6202         (*compare_scc): Likewise.
6203         (*cond_move): Likewise.
6204         (*cond_arith): Likewise.
6205         (*cond_sub): Likewise.
6206         (*cmp_ite0): Likewise.
6207         (*cmp_ite1): Likewise.
6208         (*cmp_and): Likewise.
6209         (*cmp_ior): Likewise.
6210         (*ior_scc_scc): Likewise.
6211         (*ior_scc_scc_cmp): Likewise.
6212         (*and_scc_scc): Likewise.
6213         (*and_scc_scc_cmp): Likewise.
6214         (*and_scc_scc_nod): Likewise.
6215         (*negscc): Likewise.
6216         (movcond_addsi): Likewise.
6217         (movcond): Likewise.
6218         (*ifcompare_plus_move): Likewise.
6219         (*if_plus_move): Likewise.
6220         (*ifcompare_move_plus): Likewise.
6221         (*if_move_plus): Likewise.
6222         (*ifcompare_arith_arith): Likewise.
6223         (*if_arith_arith): Likewise.
6224         (*ifcompare_arith_move): Likewise.
6225         (*if_arith_move): Likewise.
6226         (*ifcompare_move_arith): Likewise.
6227         (*if_move_arith): Likewise.
6228         (*ifcompare_move_not): Likewise.
6229         (*if_move_not): Likewise.
6230         (*ifcompare_not_move): Likewise.
6231         (*if_not_move): Likewise.
6232         (*ifcompare_shift_move): Likewise.
6233         (*if_shift_move): Likewise.
6234         (*ifcompare_move_shift): Likewise.
6235         (*if_move_shift): Likewise.
6236         (*ifcompare_shift_shift): Likewise.
6237         (*ifcompare_not_arith): Likewise.
6238         (*ifcompare_arith_not): Likewise.
6239         (*if_arith_not): Likewise.
6240         (*ifcompare_neg_move): Likewise.
6241         (*if_neg_move): Likewise.
6242         (*ifcompare_move_neg): Likewise.
6243         (*if_move_neg): Likewise.
6244         (prologue_thumb1_interwork): Likewise.
6245         (*cond_move_not): Likewise.
6246         (*sign_extract_onebit): Likewise.
6247         (*not_signextract_onebit): Likewise.
6248         (stack_tie): Likewise.
6249         (align_4): Likewise.
6250         (align_8): Likewise.
6251         (consttable_end): Likewise.
6252         (consttable_1): Likewise.
6253         (consttable_2): Likewise.
6254         (consttable_4): Likewise.
6255         (consttable_8): Likewise.
6256         (consttable_16): Likewise.
6257         (*thumb1_tablejump): Likewise.
6258         (prefetch): Likewise.
6259         (force_register_use): Likewise.
6260         (thumb_eh_return): Likewise.
6261         (load_tp_hard): Likewise.
6262         (load_tp_soft): Likewise.
6263         (tlscall): Likewise.
6264         (*arm_movtas_ze): Likewise.
6265         (*arm_rev): Likewise.
6266         (*arm_revsh): Likewise.
6267         (*arm_rev16): Likewise.
6268         * config/arm/thumb2.md
6269         (*thumb2_smaxsi3): Likewise.
6270         (*thumb2_sminsi3): Likewise.
6271         (*thumb32_umaxsi3): Likewise.
6272         (*thumb2_uminsi3): Likewise.
6273         (*thumb2_negdi2): Likewise.
6274         (*thumb2_abssi2): Likewise.
6275         (*thumb2_neg_abss): Likewise.
6276         (*thumb2_movsi_insn): Likewise.
6277         (tls_load_dot_plus_four): Likewise.
6278         (*thumb2_movhi_insn): Likewise.
6279         (*thumb2_mov_scc): Likewise.
6280         (*thumb2_mov_negs): Likewise.
6281         (*thumb2_mov_negs): Likewise.
6282         (*thumb2_mov_nots): Likewise.
6283         (*thumb2_mov_nots): Likewise.
6284         (*thumb2_movsicc_): Likewise.
6285         (*thumb2_movsfcc_soft_insn): Likewise.
6286         (*thumb2_indirect_jump): Likewise.
6287         (*thumb2_and_scc): Likewise.
6288         (*thumb2_ior_scc): Likewise.
6289         (*thumb2_ior_scc_strict_it): Likewise.
6290         (*thumb2_cond_move): Likewise.
6291         (*thumb2_cond_arith): Likewise.
6292         (*thumb2_cond_ari): Likewise.
6293         (*thumb2_cond_sub): Likewise.
6294         (*thumb2_negscc): Likewise.
6295         (*thumb2_movcond): Likewise.
6296         (thumb2_casesi_internal): Likewise.
6297         (thumb2_casesi_internal_pic): Likewise.
6298         (*thumb2_alusi3_short): Likewise.
6299         (*thumb2_mov<mode>_shortim): Likewise.
6300         (*thumb2_addsi_short): Likewise.
6301         (*thumb2_subsi_short): Likewise.
6302         (thumb2_addsi3_compare0): Likewise.
6303         (*thumb2_cbz): Likewise.
6304         (*thumb2_cbnz): Likewise.
6305         (*thumb2_one_cmplsi2_short): Likewise.
6306         (*thumb2_negsi2_short): Likewise.
6307         (*orsi_notsi_si): Likewise.
6308         * config/arm/arm1020e.md: Update with new attributes.
6309         * config/arm/arm1026ejs.md: Update with new attributes.
6310         * config/arm/arm1136jfs.md: Update with new attributes.
6311         * config/arm/arm926ejs.md: Update with new attributes.
6312         * config/arm/cortex-a15.md: Update with new attributes.
6313         * config/arm/cortex-a5.md: Update with new attributes.
6314         * config/arm/cortex-a53.md: Update with new attributes.
6315         * config/arm/cortex-a7.md: Update with new attributes.
6316         * config/arm/cortex-a8.md: Update with new attributes.
6317         * config/arm/cortex-a9.md: Update with new attributes.
6318         * config/arm/cortex-m4.md: Update with new attributes.
6319         * config/arm/cortex-r4.md: Update with new attributes.
6320         * config/arm/fa526.md: Update with new attributes.
6321         * config/arm/fa606te.md: Update with new attributes.
6322         * config/arm/fa626te.md: Update with new attributes.
6323         * config/arm/fa726te.md: Update with new attributes.
6324
6325 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6326
6327         * config/aarch64/aarch64-simd.md
6328         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6329         <vwx> iterator to ensure correct register choice.
6330         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6331         (aarch64_sqdmull_n<mode>): Likewise.
6332         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6333         * config/aarch64/arm_neon.h
6334         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6335         (vml<as><q>_n_<su>16): Likewise.
6336         (vml<as>l_high_lane<q>_<su>16): Likewise.
6337         (vml<as>l_high_n_<su>16): Likewise.
6338         (vml<as>l_lane<q>_<su>16): Likewise.
6339         (vml<as>l_n_<su>16): Likewise.
6340         (vmul<q>_lane<q>_<su>16): Likewise.
6341         (vmul<q>_n_<su>16): Likewise.
6342         (vmull_lane<q>_<su>16): Likewise.
6343         (vmull_n_<su>16): Likewise.
6344         (vmull_high_lane<q>_<su>16): Likewise.
6345         (vmull_high_n_<su>16): Likewise.
6346         (vqrdmulh<q>_n_s16): Likewise.
6347
6348 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6349
6350         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6351         have the correct lane parameter.
6352
6353 2013-09-06  Richard Biener <rguenther@suse.de>
6354
6355         * cfganal.c (control_dependences::~control_dependences):
6356         Properly free all of the vector.
6357
6358 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6359
6360         PR target/58269
6361         * config/i386/i386.c (ix86_conditional_register_usage):
6362         Proper initialize extended SSE registers.
6363
6364 2013-09-06  Jan Hubicka  <jh@suse.cz>
6365
6366         PR tree-optimization/58311
6367         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6368
6369 2013-09-06  Jan Hubicka  <jh@suse.cz>
6370
6371         * Makefile.in (tree-sra.o): Update dependencies.
6372         * tree-sra.c: Include ipa-utils.h
6373         (scan_function): Use recursive_call_p.
6374         (has_caller_p): New function.
6375         (cgraph_for_node_and_aliases): Count also callers of aliases.
6376
6377 2013-09-06  Jan Hubicka  <jh@suse.cz>
6378
6379         PR middle-end/58094
6380         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6381         * tree-tailcall.c: Include ipa-utils.h.
6382         (find_tail_calls): Use it.
6383         * ipa-pure-const.c (check_call): Likewise.
6384         * ipa-utils.c (recursive_call_p): New function.
6385         * ipa-utils.h (recursive_call_p): Dclare.
6386         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6387         (symtab_semantically_equivalent_p): New function.
6388         * Makefile.in (tree-tailcall.o): Update dependencies.
6389
6390 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6391
6392         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6393         non-inlinable part.
6394
6395 2013-09-06  Richard Biener  <rguenther@suse.de>
6396
6397         * lto-streamer.h (lto_global_var_decls): Remove.
6398         * Makefile.in (OBJS): Remove lto-symtab.o.
6399         (lto-symtab.o): Remove.
6400         (GTFILES): Remove lto-symtab.c
6401         * lto-symtab.c: Move to lto/
6402
6403 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6404
6405         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6406         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6407         (UNSPEC_FPINT_RINT): New constant definitions.
6408         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6409         definition with 2 attributes.
6410         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6411         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6412         definitions.
6413
6414 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6415
6416         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6417         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6418         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6419         attribute to "z196_alone".
6420         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6421         "zEC12_simple".
6422
6423 2013-09-06  Richard Biener  <rguenther@suse.de>
6424
6425         * basic-block.h (class control_dependences): New.
6426         * tree-ssa-dce.c (control_dependence_map): Remove.
6427         (cd): New global.
6428         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6429         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6430         find_pdom, find_control_dependence, find_all_control_dependences):
6431         Move to cfganal.c.
6432         (mark_control_dependent_edges_necessary,
6433         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6434         tree_dce_done, perform_tree_ssa_dce): Adjust.
6435         * cfganal.c (set_control_dependence_map_bit,
6436         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6437         find_all_control_dependences): Move from tree-ssa-dce.c and
6438         implement as methods of control_dependences class.
6439         (control_dependences::control_dependences): New.
6440         (control_dependences::~control_dependences): Likewise.
6441         (control_dependences::get_edges_dependent_on): Likewise.
6442         (control_dependences::get_edge): Likewise.
6443
6444 2013-09-04  Jan Hubicka  <jh@suse.cz>
6445
6446         * tree.c (types_same_for_odr): Drop overactive check.
6447         * ipa-devirt.c (hash_type_name): Likewise.
6448
6449 2013-09-04  Jan Hubicka  <jh@suse.cz>
6450
6451         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6452         (analyze_functions): ... here.
6453
6454 2013-09-04  Jan Hubicka  <jh@suse.cz>
6455
6456         PR middle-end/58201
6457         * cgraphunit.c (analyze_functions): Clear AUX fields
6458         after processing; initialize assembler name has.
6459
6460 2013-09-05  Jeff Law  <law@redhat.com>
6461
6462         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6463         from thread_around_empty_block.  Record threading path into PATH.
6464         Recurse if threading through the initial block is successful.
6465         (thread_across_edge): Corresponding changes to slightly simplify.
6466
6467 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6468
6469         * config/aarch64/aarch64.md
6470         (type): Remove frecpe, frecps, frecpx.
6471         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6472         fix to be a TARGET_SIMD instruction.
6473         (aarch64_frecps): Remove.
6474         * config/aarch64/aarch64-simd.md
6475         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6476         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6477
6478 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6479             Sofiane Naci  <sofiane.naci@arm.com>
6480
6481         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6482         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6483         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6484         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6485         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6486         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6487         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6488         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6489         into "clz, "rbit".  Rename "shift" to "shift_imm".
6490         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6491         changes.  Update for attribute changes all occurrences of arlo_* and
6492         shift* types.
6493         * config/arm/arm-fixed.md: Update for attribute changes
6494         all occurrences of arlo_* types.
6495         * config/arm/thumb2.md: Update for attribute changes all occurrences
6496         of arlo_* types.
6497         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6498         (cortexa7_older_only): Likewise.
6499         (cortexa7_younger):  Likewise.
6500         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6501         (1020alu_shift_op): Likewise.
6502         (1020alu_shift_reg_op): Likewise.
6503         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6504         (alu_shift_op): Likewise.
6505         (alu_shift_reg_op): Likewise.
6506         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6507         (11_alu_shift_op): Likewise.
6508         (11_alu_shift_reg_op): Likewise.
6509         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6510         (9_alu_shift_reg_op): Likewise.
6511         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6512         attribute changes.
6513         (cortex_a15_alu_shift): Likewise.
6514         (cortex_a15_alu_shift_reg): Likewise.
6515         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6516         attribute changes.
6517         (cortex_a5_alu_shift): Likewise.
6518         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6519         attribute changes.
6520         (cortex_a53_alu_shift): Likewise.
6521         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6522         attribute changes.
6523         (cortex_a7_alu_reg): Likewise.
6524         (cortex_a7_alu_shift): Likewise.
6525         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6526         attribute changes.
6527         (cortex_a8_alu_shift): Likewise.
6528         (cortex_a8_alu_shift_reg): Likewise.
6529         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6530         (cortex_a9_dp_shift): Likewise.
6531         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6532         attribute changes.
6533         * config/arm/cortex-r4.md
6534         (cortex_r4_alu): Update for attribute changes.
6535         (cortex_r4_mov): Likewise.
6536         (cortex_r4_alu_shift_reg): Likewise.
6537         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6538         (526_alu_shift_op): Likewise.
6539         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6540         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6541         (626te_alu_shift_op): Likewise.
6542         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6543         (726te_alu_shift_op): Likewise.
6544         (726te_alu_shift_reg_op): Likewise.
6545         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6546         (mp626_alu_shift_op): Likewise.
6547         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6548         (pj4_alu_conds): Likewise.
6549         (pj4_shift): Likewise.
6550         (pj4_shift_conds): Likewise.
6551         (pj4_alu_shift): Likewise.
6552         (pj4_alu_shift_conds): Likewise.
6553         * config/aarch64/aarch64.md: Update for attribute change
6554         all occurrences of arlo_* and shift* types.
6555
6556 2013-09-05  Mike Stump  <mikestump@comcast.net>
6557
6558         * tree.h: Move documentation for tree_function_decl to tree-core.h
6559         with the declaration.
6560
6561 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6562
6563         PR target/58139
6564         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6565         looking for widest mode.
6566
6567 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6568
6569         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6570
6571 2013-09-05  Richard Biener  <rguenther@suse.de>
6572
6573         PR tree-optimization/58137
6574         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6575         Do not create vectors of pointers.
6576         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6577         types for the components of the vector initializer.
6578         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6579         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6580
6581 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6582
6583         * ipa-prop.c (remove_described_reference): Accept missing references,
6584         return false if that hppens, otherwise return true.
6585         (cgraph_node_for_jfunc): New function.
6586         (try_decrement_rdesc_refcount): Likewise.
6587         (try_make_edge_direct_simple_call): Use them.
6588         (ipa_edge_removal_hook): Remove references from rdescs.
6589         (ipa_edge_duplication_hook): Clone rdescs and their references
6590         when the new edge has the same caller as the old one.
6591         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6592         reference before removing any edges.
6593
6594 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6595
6596         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6597         initial store.
6598         * thumb2.md (thumb2_storewb_parisi): New pattern.
6599
6600 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6601
6602         * config/aarch64/aarch64-option-extensions.def: Add
6603         AARCH64_OPT_EXTENSION of 'crc'.
6604         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6605         (AARCH64_ISA_CRC): Ditto.
6606         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6607         description of the CRC extension.
6608
6609 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6610
6611         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6612         * config/rs6000/linux.h: Ditto.
6613         * alpha/linux.h: Ditto.
6614         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6615         no_c99_libc_has_function.
6616         * config/c6x/uclinux-elf.h: Ditto.
6617         * config/lm32/uclinux-elf.h: Ditto.
6618         * config/m68k/uclinux.h: Ditto.
6619         * config/moxie/uclinux.h: Ditto.
6620         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6621         (crisv32-*-linux*, cris-*-linux*): Ditto.
6622         * config/bfin/bfin.c: Include "tm_p.h".
6623
6624 2013-09-05  Richard Biener  <rguenther@suse.de>
6625
6626         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6627         check for a definition without a basic-block.
6628
6629 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6630             Sofiane Naci <sofiane.naci@arm.com>
6631
6632         * config/aarch64/aarch64.md
6633         (*movti_aarch64): Rename r_2_f and f_2_r.
6634         (*movsf_aarch64): Likewise.
6635         (*movdf_aarch64): Likewise.
6636         (*movtf_aarch64): Likewise.
6637         (aarch64_movdi_<mode>low): Likewise.
6638         (aarch64_movdi_<mode>high): Likewise.
6639         (aarch64_mov<mode>high_di): Likewise.
6640         (aarch64_mov<mode>low_di): Likewise.
6641         (aarch64_movtilow_tilow): Likewise.
6642         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6643         values to config/arm/types.md
6644         (attribute "conds"): Update for attribute change.
6645         (anddi3_insn): Likewise.
6646         (iordi3_insn): Likewise.
6647         (xordi3_insn): Likewise.
6648         (one_cmpldi2): Likewise.
6649         * config/arm/types.md (type): Add Neon types.
6650         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6651         use "type" attribute.
6652         (movmisalign<mode>_neon_store): Likewise.
6653         (movmisalign<mode>_neon_load): Likewise.
6654         (vec_set<mode>_internal): Likewise.
6655         (vec_setv2di_internal): Likewise.
6656         (vec_extract<mode>): Likewise.
6657         (vec_extractv2di): Likewise.
6658         (add<mode>3_neon): Likewise.
6659         (adddi3_neon): Likewise.
6660         (sub<mode>3_neon): Likewise.
6661         (subdi3_neon): Likewise.
6662         (mul<mode>3_neon): Likewise.
6663         (mul<mode>3add<mode>_neon): Likewise.
6664         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6665         (fma<VCVTF:mode>4)): Likewise.
6666         (fma<VCVTF:mode>4_intrinsic): Likewise.
6667         (fmsub<VCVTF:mode>4)): Likewise.
6668         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6669         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6670         (ior<mode>3): Likewise.
6671         (and<mode>3): Likewise.
6672         (anddi3_neon): Likewise.
6673         (orn<mode>3_neon): Likewise.
6674         (orndi3_neon): Likewise.
6675         (bic<mode>3_neon): Likewise.
6676         (bicdi3_neon): Likewise.
6677         (xor<mode>3): Likewise.
6678         (one_cmpl<mode>2): Likewise.
6679         (abs<mode>2): Likewise.
6680         (neg<mode>2): Likewise.
6681         (umin<mode>3_neon): Likewise.
6682         (umax<mode>3_neon): Likewise.
6683         (smin<mode>3_neon): Likewise.
6684         (smax<mode>3_neon): Likewise.
6685         (vashl<mode>3): Likewise.
6686         (vashr<mode>3_imm): Likewise.
6687         (vlshr<mode>3_imm): Likewise.
6688         (ashl<mode>3_signed): Likewise.
6689         (ashl<mode>3_unsigned): Likewise.
6690         (neon_load_count): Likewise.
6691         (ashldi3_neon_noclobber): Likewise.
6692         (signed_shift_di3_neon): Likewise.
6693         (unsigned_shift_di3_neon): Likewise.
6694         (ashrdi3_neon_imm_noclobber): Likewise.
6695         (lshrdi3_neon_imm_noclobber): Likewise.
6696         (widen_ssum<mode>3): Likewise.
6697         (widen_usum<mode>3): Likewise.
6698         (quad_halves_<code>v4si): Likewise.
6699         (quad_halves_<code>v4sf): Likewise.
6700         (quad_halves_<code>v8hi): Likewise.
6701         (quad_halves_<code>v16qi): Likewise.
6702         (reduc_splus_v2di): Likewise.
6703         (neon_vpadd_internal<mode>): Likewise.
6704         (neon_vpsmin<mode>): Likewise.
6705         (neon_vpsmax<mode>): Likewise.
6706         (neon_vpumin<mode>): Likewise.
6707         (neon_vpumax<mode>): Likewise.
6708         (ss_add<mode>_neon): Likewise.
6709         (us_add<mode>_neon): Likewise.
6710         (ss_sub<mode>_neon): Likewise.
6711         (us_sub<mode>_neon): Likewise.
6712         (neon_vadd<mode>_unspec): Likewise.
6713         (neon_vaddl<mode>): Likewise.
6714         (neon_vaddw<mode>): Likewise.
6715         (neon_vhadd<mode>): Likewise.
6716         (neon_vqadd<mode>): Likewise.
6717         (neon_vaddhn<mode>): Likewise.
6718         (neon_vmul<mode>): Likewise.
6719         (neon_vmla<mode>): Likewise.
6720         (neon_vmlal<mode>): Likewise.
6721         (neon_vmls<mode>): Likewise.
6722         (neon_vmlsl<mode>): Likewise.
6723         (neon_vqdmulh<mode>): Likewise.
6724         (neon_vqdmlal<mode>): Likewise.
6725         (neon_vqdmlsl<mode>): Likewise.
6726         (neon_vmull<mode>): Likewise.
6727         (neon_vqdmull<mode>): Likewise.
6728         (neon_vsub<mode>_unspec): Likewise.
6729         (neon_vsubl<mode>): Likewise.
6730         (neon_vsubw<mode>): Likewise.
6731         (neon_vqsub<mode>): Likewise.
6732         (neon_vhsub<mode>): Likewise.
6733         (neon_vsubhn<mode>): Likewise.
6734         (neon_vceq<mode>): Likewise.
6735         (neon_vcge<mode>): Likewise.
6736         (neon_vcgeu<mode>): Likewise.
6737         (neon_vcgt<mode>): Likewise.
6738         (neon_vcgtu<mode>): Likewise.
6739         (neon_vcle<mode>): Likewise.
6740         (neon_vclt<mode>): Likewise.
6741         (neon_vcage<mode>): Likewise.
6742         (neon_vcagt<mode>): Likewise.
6743         (neon_vtst<mode>): Likewise.
6744         (neon_vabd<mode>): Likewise.
6745         (neon_vabdl<mode>): Likewise.
6746         (neon_vaba<mode>): Likewise.
6747         (neon_vabal<mode>): Likewise.
6748         (neon_vmax<mode>): Likewise.
6749         (neon_vmin<mode>): Likewise.
6750         (neon_vpaddl<mode>): Likewise.
6751         (neon_vpadal<mode>): Likewise.
6752         (neon_vpmax<mode>): Likewise.
6753         (neon_vpmin<mode>): Likewise.
6754         (neon_vrecps<mode>): Likewise.
6755         (neon_vrsqrts<mode>): Likewise.
6756         (neon_vqabs<mode>): Likewise.
6757         (neon_vqneg<mode>): Likewise.
6758         (neon_vcls<mode>): Likewise.
6759         (clz<mode>2): Likewise.
6760         (popcount<mode>2): Likewise.
6761         (neon_vrecpe): Likewise.
6762         (neon_vrsqrte): Likewise.
6763         (neon_vget_lane<mode>_sext_internal): Likewise.
6764         (neon_vget_lane<mode>_zext_internal): Likewise.
6765         (neon_vdup_n<mode>): Likewise.
6766         (neon_vdup_nv2di): Likewise.
6767         (neon_vdpu_lane<mode>_internal): Likewise.
6768         (neon_vswp<mode>): Likewise.
6769         (float<mode><V_cvtto>2): Likewise.
6770         (floatuns<mode><V_cvtto>2): Likewise.
6771         (fix_trunc<mode><V_cvtto>)2): Likewise
6772         (fixuns_trunc<mode><V_cvtto)2): Likewise.
6773         (neon_vcvt<mode>): Likewise.
6774         (neon_vcvtv4sfv4hf): Likewise.
6775         (neon_vcvtv4hfv4sf): Likewise.
6776         (neon_vcvt_n<mode>): Likewise.
6777         (neon_vmovn<mode>): Likewise.
6778         (neon_vqmovn<mode>): Likewise.
6779         (neon_vqmovun<mode>): Likewise.
6780         (neon_vmovl<mode>): Likewise.
6781         (neon_vmul_lane<mode>): Likewise.
6782         (neon_vmull_lane<mode>): Likewise.
6783         (neon_vqdmull_lane<mode>): Likewise.
6784         (neon_vqdmulh_lane<mode>): Likewise.
6785         (neon_vmla_lane<mode>): Likewise.
6786         (neon_vmlal_lane<mode>): Likewise.
6787         (neon_vqdmlal_lane<mode>): Likewise.
6788         (neon_vmls_lane<mode>): Likewise.
6789         (neon_vmlsl_lane<mode>): Likewise.
6790         (neon_vqdmlsl_lane<mode>): Likewise.
6791         (neon_vext<mode>): Likewise.
6792         (neon_vrev64<mode>): Likewise.
6793         (neon_vrev32<mode>): Likewise.
6794         (neon_vrev16<mode>): Likewise.
6795         (neon_vbsl<mode>_internal): Likewise.
6796         (neon_vshl<mode>): Likewise.
6797         (neon_vqshl<mode>): Likewise.
6798         (neon_vshr_n<mode>): Likewise.
6799         (neon_vshrn_n<mode>): Likewise.
6800         (neon_vqshrn_n<mode>): Likewise.
6801         (neon_vqshrun_n<mode>): Likewise.
6802         (neon_vshl_n<mode>): Likewise.
6803         (neon_vqshl_n<mode>): Likewise.
6804         (neon_vqshlu_n<mode>): Likewise.
6805         (neon_vshll_n<mode>): Likewise.
6806         (neon_vsra_n<mode>): Likewise.
6807         (neon_vsri_n<mode>): Likewise.
6808         (neon_vsli_n<mode>): Likewise.
6809         (neon_vtbl1v8qi): Likewise.
6810         (neon_vtbl2v8qi): Likewise.
6811         (neon_vtbl3v8qi): Likewise.
6812         (neon_vtbl4v8qi): Likewise.
6813         (neon_vtbx1v8qi): Likewise.
6814         (neon_vtbx2v8qi): Likewise.
6815         (neon_vtbx3v8qi): Likewise.
6816         (neon_vtbx4v8qi): Likewise.
6817         (neon_vtrn<mode>_internal): Likewise.
6818         (neon_vzip<mode>_internal): Likewise.
6819         (neon_vuzp<mode>_internal): Likewise.
6820         (neon_vld1<mode>): Likewise.
6821         (neon_vld1_lane<mode>): Likewise.
6822         (neon_vld1_dup<mode>): Likewise.
6823         (neon_vld1_dupv2di): Likewise.
6824         (neon_vst1<mode>): Likewise.
6825         (neon_vst1_lane<mode>): Likewise.
6826         (neon_vld2<mode>): Likewise.
6827         (neon_vld2_lane<mode>): Likewise.
6828         (neon_vld2_dup<mode>): Likewise.
6829         (neon_vst2<mode>): Likewise.
6830         (neon_vst2_lane<mode>): Likewise.
6831         (neon_vld3<mode>): Likewise.
6832         (neon_vld3qa<mode>): Likewise.
6833         (neon_vld3qb<mode>): Likewise.
6834         (neon_vld3_lane<mode>): Likewise.
6835         (neon_vld3_dup<mode>): Likewise.
6836         (neon_vst3<mode>): Likewise.
6837         (neon_vst3qa<mode>): Likewise.
6838         (neon_vst3qb<mode>): Likewise.
6839         (neon_vst3_lane<mode>): Likewise.
6840         (neon_vld4<mode>): Likewise.
6841         (neon_vld4qa<mode>): Likewise.
6842         (neon_vld4qb<mode>): Likewise.
6843         (neon_vld4_lane<mode>): Likewise.
6844         (neon_vld4_dup<mode>): Likewise.
6845         (neon_vst4<mode>): Likewise.
6846         (neon_vst4qa<mode>): Likewise.
6847         (neon_vst4qb<mode>): Likewise.
6848         (neon_vst4_lane<mode>): Likewise.
6849         (neon_vec_unpack<US>_lo_<mode>): Likewise.
6850         (neon_vec_unpack<US>_hi_<mode>): Likewise.
6851         (neon_vec_<US>mult_lo_<mode>): Likewise.
6852         (neon_vec_<US>mult_hi_<mode>): Likewise.
6853         (neon_vec_<US>shiftl_<mode>): Likewise.
6854         (neon_unpack<US>_<mode>): Likewise.
6855         (neon_vec_<US>mult_<mode>): Likewise.
6856         (vec_pack_trunc_<mode>): Likewise.
6857         (neon_vec_pack_trunk_<mode>): Likewise.
6858         (neon_vabd<mode>_2): Likewise.
6859         (neon_vabd<mode>_3): Likewise.
6860         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
6861         (thumb2_movsi_vfp): Likewise.
6862         (movdi_vfp): Likewise.
6863         (movdi_vfp_cortexa8): Likewise.
6864         (movhf_vfp_neon): Likewise.
6865         (movhf_vfp): Likewiwse.
6866         (movsf_vfp): Likewiwse.
6867         (thumb2_movsf_vfp): Likewiwse.
6868         (movdf_vfp): Likewise.
6869         (thumb2_movdf_vfp): Likewise.
6870         (movsfcc_vfp): Likewise.
6871         (thumb2_movsfcc_vfp): Likewise.
6872         (movdfcc_vfp): Likewise.
6873         (thumb2_movdfcc_vfp): Likewise.
6874         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
6875         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
6876         (v10_v2c): Likewise.
6877         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
6878         attribute change.
6879         (cortex_a15_neon_int_2): Likewise.
6880         (cortex_a15_neon_int_3): Likewise.
6881         (cortex_a15_neon_int_4): Likewise.
6882         (cortex_a15_neon_int_5): Likewise.
6883         (cortex_a15_neon_vqneg_vqabs): Likewise.
6884         (cortex_a15_neon_vmov): Likewise.
6885         (cortex_a15_neon_vaba): Likewise.
6886         (cortex_a15_neon_vaba_qqq): Likewise.
6887         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6888         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
6889         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
6890         scalar_64_32_long_scalar): Likewise.
6891         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
6892         (cortex_a15_neon_mla_qqq_8_16): Likewise.
6893         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
6894         lotype_qdd_64_32_long): Likewise.
6895         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
6896         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
6897         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
6898         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
6899         (cortex_a15_neon_shift_1): Likewise.
6900         (cortex_a15_neon_shift_2): Likewise.
6901         (cortex_a15_neon_shift_3): Likewise.
6902         (cortex_a15_neon_vshl_ddd): Likewise.
6903         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
6904         (cortex_a15_neon_vsra_vrsra): Likewise.
6905         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
6906         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
6907         (cortex_a15_neon_fp_vmul_ddd): Likewise.
6908         (cortex_a15_neon_fp_vmul_qqd): Likewise.
6909         (cortex_a15_neon_fp_vmla_ddd): Likewise.
6910         (cortex_a15_neon_fp_vmla_qqq): Likewise.
6911         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
6912         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
6913         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
6914         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
6915         (cortex_a15_neon_bp_simple): Likewise.
6916         (cortex_a15_neon_bp_2cycle): Likewise.
6917         (cortex_a15_neon_bp_3cycle): Likewise.
6918         (cortex_a15_neon_vld1_1_2_regs): Likewise.
6919         (cortex_a15_neon_vld1_3_4_regs): Likewise.
6920         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
6921         (cortex_a15_neon_vld2_4_regs): Likewise.
6922         (cortex_a15_neon_vld3_vld4): Likewise.
6923         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
6924         (cortex_a15_neon_vst1_3_4_regs): Likewise.
6925         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
6926         (cortex_a15_neon_vst3_vst4): Likewise.
6927         (cortex_a15_neon_vld1_vld2_lane): Likewise.
6928         (cortex_a15_neon_vld3_vld4_lane" 10
6929         (cortex_a15_neon_vst1_vst2_lane): Likewise.
6930         (cortex_a15_neon_vst3_vst4_lane): Likewise.
6931         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
6932         (cortex_a15_neon_ldm_2): Likewise.0
6933         (cortex_a15_neon_stm_2): Likewise.
6934         (cortex_a15_neon_mcr): Likewise.
6935         (cortex_a15_neon_mcr_2_mcrr): Likewise.
6936         (cortex_a15_neon_mrc): Likewise.
6937         (cortex_a15_neon_mrrc): Likewise.
6938         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
6939         change.
6940         (cortex_a15_alu_shift): Likewise.
6941         (cortex_a15_alu_shift_reg): Likewise.
6942         (cortex_a15_mult32): Likewise.
6943         (cortex_a15_mult64): Likewise.
6944         (cortex_a15_block): Likewise.
6945         (cortex_a15_branch): Likewise.
6946         (cortex_a15_load1): Likewise.
6947         (cortex_a15_load3): Likewise.
6948         (cortex_a15_store1): Likewise.
6949         (cortex_a15_store3): Likewise.
6950         (cortex_a15_call): Likewise.
6951         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
6952         (cortex_a5_f2r): Likewise.
6953         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
6954         change.
6955         (cortex_a53_f2r): Likewise.
6956         * config/arm/cortex-a7.md
6957         (cortex_a7_branch): Update for attribute change.
6958         (cortex_a7_call): Likewise.
6959         (cortex_a7_alu_imm): Likewise.
6960         (cortex_a7_alu_reg): Likewise.
6961         (cortex_a7_alu_shift): Likewise.
6962         (cortex_a7_mul): Likewise.
6963         (cortex_a7_load1): Likewise.
6964         (cortex_a7_store1): Likewise.
6965         (cortex_a7_load2): Likewise.
6966         (cortex_a7_store2): Likewise.
6967         (cortex_a7_load3): Likewise.
6968         (cortex_a7_store3): Likewise.
6969         (cortex_a7_load4): Likewise.
6970         (cortex_a7_store4): Likewise.
6971         (cortex_a7_fpalu): Likewise.
6972         (cortex_a7_fconst): Likewise.
6973         (cortex_a7_fpmuls): Likewise.
6974         (cortex_a7_neon_mul): Likewise.
6975         (cortex_a7_fpmacs): Likewise.
6976         (cortex_a7_neon_mla: Likewise.
6977         (cortex_a7_fpmuld: Likewise.
6978         (cortex_a7_fpmacd: Likewise.
6979         (cortex_a7_fpfmad: Likewise.
6980         (cortex_a7_fdivs: Likewise.
6981         (cortex_a7_fdivd: Likewise.
6982         (cortex_a7_r2f: Likewise.
6983         (cortex_a7_f2r: Likewise.
6984         (cortex_a7_f_flags: Likewise.
6985         (cortex_a7_f_loads: Likewise.
6986         (cortex_a7_f_loadd: Likewise.
6987         (cortex_a7_f_stores: Likewise.
6988         (cortex_a7_f_stored: Likewise.
6989         (cortex_a7_neon): Likewise.
6990         * config/arm/cortex-a8-neon.md
6991         (cortex_a8_neon_mrc): Update for attribute change.
6992         (cortex_a8_neon_mrrc): Likewise.
6993         (cortex_a8_neon_int_1): Likewise.
6994         (cortex_a8_neon_int_2): Likewise.
6995         (cortex_a8_neon_int_3): Likewise.
6996         (cortex_a8_neon_int_4): Likewise.
6997         (cortex_a8_neon_int_5): Likewise.
6998         (cortex_a8_neon_vqneg_vqabs): Likewise.
6999         (cortex_a8_neon_vmov): Likewise.
7000         (cortex_a8_neon_vaba): Likewise.
7001         (cortex_a8_neon_vaba_qqq): Likewise.
7002         (cortex_a8_neon_vsma): Likewise.
7003         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7004         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7005         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7006         long_scalar): Likewise.
7007         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7008         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7009         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7010         long_scalar_qdd_64_32_long): Likewise.
7011         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7012         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7013         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7014         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7015         (cortex_a8_neon_shift_1): Likewise.
7016         (cortex_a8_neon_shift_2): Likewise.
7017         (cortex_a8_neon_shift_3): Likewise.
7018         (cortex_a8_neon_vshl_ddd): Likewise.
7019         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7020         (cortex_a8_neon_vsra_vrsra): Likewise.
7021         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7022         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7023         (cortex_a8_neon_fp_vsum): Likewise.
7024         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7025         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7026         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7027         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7028         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7029         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7030         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7031         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7032         (cortex_a8_neon_bp_simple): Likewise.
7033         (cortex_a8_neon_bp_2cycle): Likewise.
7034         (cortex_a8_neon_bp_3cycle): Likewise.
7035         (cortex_a8_neon_ldr): Likewise.
7036         (cortex_a8_neon_str): Likewise.
7037         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7038         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7039         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7040         (cortex_a8_neon_vld2_4_regs): Likewise.
7041         (cortex_a8_neon_vld3_vld4): Likewise.
7042         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7043         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7044         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7045         (cortex_a8_neon_vst3_vst4): Likewise.
7046         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7047         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7048         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7049         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7050         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7051         (cortex_a8_neon_mcr): Likewise.
7052         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7053         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7054         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7055         change.
7056         (ca9_neon_mrrc): Likewise.
7057         (cortex_a9_neon_int_1): Likewise.
7058         (cortex_a9_neon_int_2): Likewise.
7059         (cortex_a9_neon_int_3): Likewise.
7060         (cortex_a9_neon_int_4): Likewise.
7061         (cortex_a9_neon_int_5): Likewise.
7062         (cortex_a9_neon_vqneg_vqabs): Likewise.
7063         (cortex_a9_neon_vmov): Likewise.
7064         (cortex_a9_neon_vaba): Likewise.
7065         (cortex_a9_neon_vaba_qqq): Likewise.
7066         (cortex_a9_neon_vsma): Likewise.
7067         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7068         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7069         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7070         long_scalar): Likewise.
7071         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7072         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7073         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7074         long_scalar_qdd_64_32_long): Likewise.
7075         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7076         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7077         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7078         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7079         (cortex_a9_neon_shift_1): Likewise.
7080         (cortex_a9_neon_shift_2): Likewise.
7081         (cortex_a9_neon_shift_3): Likewise.
7082         (cortex_a9_neon_vshl_ddd): Likewise.
7083         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7084         (cortex_a9_neon_vsra_vrsra): Likewise.
7085         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7086         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7087         (cortex_a9_neon_fp_vsum): Likewise.
7088         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7089         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7090         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7091         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7092         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7093         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7094         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7095         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7096         (cortex_a9_neon_bp_simple): Likewise.
7097         (cortex_a9_neon_bp_2cycle): Likewise.
7098         (cortex_a9_neon_bp_3cycle): Likewise.
7099         (cortex_a9_neon_ldr): Likewise.
7100         (cortex_a9_neon_str): Likewise.
7101         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7102         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7103         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7104         (cortex_a9_neon_vld2_4_regs): Likewise.
7105         (cortex_a9_neon_vld3_vld4): Likewise.
7106         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7107         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7108         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7109         (cortex_a9_neon_vst3_vst4): Likewise.
7110         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7111         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7112         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7113         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7114         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7115         (cortex_a9_neon_mcr): Likewise.
7116         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7117         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7118         (cortex_a9_fps): Likewise.
7119         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7120         change.
7121         (cortex_m4_fmuls): Likewise.
7122         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7123         change.
7124         (cortex_r4_mrc): Likewise.
7125         * config/arm/iterators.md: Update comment referring to neon_type.
7126         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7127         (iwmmxt_movsi_insn): Likewise.
7128         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7129         attribute change.
7130         (pj4_core_to_vfp): Likewise.
7131         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7132         attribute change.
7133         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7134         (vfp_fstore): Likewise.
7135         * doc/md.texi: Change references to neon_type to refer to type.
7136
7137 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7138
7139         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7140
7141 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7142
7143         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7144         lglobl if not weak.
7145
7146 2013-09-04  Easwaran Raman  <eraman@google.com>
7147
7148         PR middle-end/57370
7149         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7150         (build_and_add_sum): Use it.
7151         (appears_later_in_bb): Simplify code.
7152
7153 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7154
7155         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7156
7157 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7158
7159         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7160
7161 2013-09-04  Jan Hubicka  <jh@suse.cz>
7162
7163         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7164         * ipa-devirt.c: Include diganostic.h
7165         (odr_type_d): Add types and types_set.
7166         (hash_type_name): Work for types with vtables during LTO.
7167         (odr_hasher::remove): Fix comment; destroy types_set.
7168         (add_type_duplicate): New function,
7169         (get_odr_type): Use it.
7170         (dump_type_inheritance_graph): Dump type duplicates.
7171         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7172         graph.
7173         * tree.c (types_same_for_odr): Give exact answers on types with
7174         virtual tables.
7175
7176 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7177
7178         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7179         explaining their differences.
7180
7181 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7182
7183         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7184
7185 2013-09-03  Jeff Law  <law@redhat.com>
7186
7187         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7188         when not threading through a joiner block.  Pass joiner/no joiner
7189         state to register_jump_thread.
7190         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7191         state from argument rather than implying on path length.
7192         Dump the entire jump thread path into debugging dump.
7193         * tree-flow.h (register_jump_thread): Update prototype.
7194
7195 2013-08-29  Xinliang David Li  <davidxl@google.com>
7196
7197         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7198         Remove a trivial gcc_assert.
7199
7200 2013-08-29  Xinliang David Li  <davidxl@google.com>
7201
7202         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7203         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7204         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7205         Delay base decl alignment adjustment.
7206         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7207         * tree-vectorizer.h: New data structure.
7208         (set_dr_misalignment): New function.
7209         (dr_misalignment): Ditto.
7210         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7211         (vectorizable_load): Ditto.
7212         (ensure_base_align): New function.
7213         (vectorize_loops): Add dbg_cnt support.
7214         (execute_vect_slp): Ditto.
7215         * dbgcnt.def: New debug counter.
7216         * Makefile: New dependency.
7217
7218 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7219
7220         Revert:
7221
7222         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7223
7224         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7225
7226 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7227
7228         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7229         function descriptor.
7230
7231 2013-09-03  Richard Biener  <rguenther@suse.de>
7232
7233         * tree-affine.c (add_elt_to_tree): Fix association issue,
7234         avoid useless converts and make sure to always return a
7235         properly typed result.
7236
7237 2013-09-03  Richard Biener  <rguenther@suse.de>
7238
7239         PR middle-end/57656
7240         * fold-const.c (negate_expr_p): Fix division case.
7241         (negate_expr): Likewise.
7242
7243 2013-09-03  Richard Biener  <rguenther@suse.de>
7244
7245         PR lto/58285
7246         * tree-streamer-out.c: Include tm.h.
7247         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7248
7249 2013-09-03  Jan Hubicka  <jh@suse.cz>
7250
7251         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7252
7253 2013-09-03  Alan Modra  <amodra@gmail.com>
7254
7255         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7256         multilibs to big-endian target and vice versa.
7257         * config/rs6000/t-linux64: Use := assignment on all vars.
7258         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7259         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7260         * config/rs6000/t-linux64le: New file.
7261         * config/rs6000/t-linux64bele: New file.
7262         * config/rs6000/t-linux64lebe: New file.
7263
7264 2013-09-02  Jan Hubicka  <jh@suse.cz>
7265
7266         * ipa-inline-transform.c (inline_transform): Do not
7267         optimize_inline_calls when not optimizing.
7268
7269 2013-09-02  Jan Hubicka  <jh@suse.cz>
7270
7271         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7272         duplicated nodes for assembler names.
7273         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7274         hard registers from assembler name hash.
7275
7276 2013-09-02  Jan Hubicka  <jh@suse.cz>
7277
7278         * ipa-split.c (execute_split_functions): Split externally visible
7279         functions called once.
7280
7281 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7282
7283         PR ipa/58106
7284         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7285         linked list.  When finding the correct duplicate, also consider also
7286         the caller in additon to its inlined_to node.
7287
7288 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7289
7290         * config/aarch64/aarch64-simd-builtins.def
7291         (dup_lane_scalar): Remove.
7292         * config/aarch64/aarch64-simd.md
7293         (aarch64_simd_dup): Add 'w->w' alternative.
7294         (aarch64_dup_lane<mode>): Allow for VALL.
7295         (aarch64_dup_lane_scalar<mode>): Remove.
7296         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7297         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7298         (aarch64_get_lane_unsigned<mode>): Likewise.
7299         (aarch64_get_lane<mode>): Likewise.
7300         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7301         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7302         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7303         (VCON): Change container of V2SF.
7304         (vswap_width_name): Likewise.
7305         * config/aarch64/arm_neon.h
7306         (__aarch64_vdup_lane_any): New.
7307         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7308         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7309         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7310
7311 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7312
7313         PR middle-end/56382
7314         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7315         the source or the destination is a single hard register.
7316
7317 2013-09-02  Richard Biener  <rguenther@suse.de>
7318
7319         PR middle-end/57511
7320         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7321         non-linear SCEVs.
7322
7323 2013-09-02  Richard Biener  <rguenther@suse.de>
7324
7325         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7326         arithmetic to sizetype.
7327
7328 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7329
7330         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7331         Find auto-increment use both before and after candidate.
7332
7333 2013-09-02  Marek Polacek  <polacek@redhat.com>
7334
7335         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7336
7337 2013-09-01  Jan Hubicka  <jh@suse.cz>
7338
7339         * Makefile.in: Add ipa-profile.o
7340         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7341         * cgraph.c (struct cgraph_propagate_frequency_data,
7342         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7343         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7344         * cgraph.h (cgraph_propagate_frequency): Remove.
7345         * ipa-inline-analysis.c: Include ipa-utils.h;
7346         drop duplicated cfgloop.h.
7347         (inline_update_callee_summaries): Update.
7348         * ipa-profile.c: New file.
7349         * ipa-utils.h (ipa_propagate_frequency): Declare.
7350         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7351         data-streamer.h, value-prof.h.
7352         (symtab_remove_unreachable_nodes): Update profile.
7353         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7354         account_time_size, cmp_counts, dump_histogram,
7355         ipa_profile_generate_summary, ipa_profile_write_summary,
7356         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7357         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7358         Move to ipa-profile.c.
7359
7360 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7361
7362         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7363
7364 2013-09-01  Jan Hubicka  <jh@suse.cz>
7365
7366         * common.opt (fdevirtualize-speculatively): New function.
7367         * invoke.texi (fdevirtualize-speculatively): Document.
7368         * ipa-devirt.c: Include ipa-inline.h
7369         (likely_target_p): New function.
7370         (ipa_devirt): New function.
7371         (gate_ipa_devirt): New function.
7372         (pass_data_ipa_devirt): New static var.
7373         (pass_ipa_devirt): Likewise.
7374         (make_pass_ipa_devirt): New function.
7375         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7376         (common_handle_option): Disable devirtualization when
7377         value range profiling is available.
7378         * passes.def (pass_ipa_devirt): Add.
7379         * timever.def (TV_IPA_DEVIRT): New timevar.
7380         * tree-pass.h (make_pass_ipa_devirt):
7381
7382 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7383
7384         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7385         include sanitize(undefined).
7386
7387 2013-08-31  Diego Novillo  <dnovillo@google.com>
7388
7389         * Makefile.in (TREE_CORE_H): Define.
7390         (TREE_H): Use.
7391         (GTFILES): Add tree-core.h.
7392         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7393         size the array.
7394         * tree-core.h: New file.
7395         Move all data structures, enum, typedefs, global
7396         declarations and constants from ...
7397         * tree.h: ... here.
7398
7399 2013-08-31  Jan Hubicka  <jh@suse.cz>
7400
7401         * bulitins.c (expand_builtin): Do not early exit for gcov
7402         instrumented functions.
7403
7404 2013-08-31  Marek Polacek  <polacek@redhat.com>
7405
7406         * ubsan.c: Include tm_p.h.
7407
7408 2013-08-31  Jan Hubicka  <jh@suse.cz>
7409
7410         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7411         warning.
7412
7413         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7414         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7415         * tree-cfg.c (verify_gimple_label): ... here.
7416         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7417         (ipa_merge_profiles): New function.
7418         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7419         (lto_input_function_body): Likewise.
7420         * ipa-utils.h (ipa_merge_profiles): Declare.
7421         * lto-streamer.h (lto_input_function_body): Update prototype.
7422         (emit_label_in_global_context_p): Remove.
7423         * lto-symtab.c: Include ipa-utils.h
7424         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7425
7426 2013-08-31  Jan Hubicka  <jh@suse.cz>
7427
7428         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7429
7430 2013-08-31  Jan Hubicka  <jh@suse.cz>
7431
7432         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7433
7434 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7435
7436         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7437         "cmp" RTX before signed_comparison_operator check to account
7438         for "code" changes.
7439
7440 2013-08-30  Jan Hubicka  <jh@suse.cz>
7441
7442         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7443         (detect_type_change_1): Rename to ...
7444         (detect_type_change): ... this one; early return on non-polymorphic
7445         types.
7446         (detect_type_change_ssa): Add comp_type parameter; update
7447         use of detect_type_change.
7448         (compute_complex_assign_jump_func): Add param_type parameter;
7449         update use of detect_type_change_ssa.
7450         (compute_complex_ancestor_jump_func): Likewise.
7451         (ipa_get_callee_param_type): New function.
7452         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7453         update calls to the jump function computation functions.
7454
7455 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7456             Steven Bosscher  <steven@gcc.gnu.org>
7457
7458         * cfgrtl.c (fixup_new_cold_bb): New routine.
7459         (commit_edge_insertions): Invoke fixup_partitions.
7460         (find_partition_fixes): New routine.
7461         (fixup_partitions): Ditto.
7462         (verify_hot_cold_block_grouping): Update comments.
7463         (rtl_verify_edges): Invoke find_partition_fixes.
7464         (rtl_verify_bb_pointers): Update comments.
7465         (rtl_verify_bb_layout): Ditto.
7466         * basic-block.h (probably_never_executed_edge_p): Declare.
7467         (fixup_partitions): Ditto.
7468         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7469         * bb-reorder.c (sanitize_hot_paths): New function.
7470         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7471         sanitize_hot_paths.
7472         * predict.c (probably_never_executed_edge_p): New routine.
7473         * cfg.c (check_bb_profile): Add partition insanity warnings.
7474
7475 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7476
7477         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7478
7479 2013-08-30  Marek Polacek  <polacek@redhat.com>
7480
7481         * Makefile.in (ubsan.o): Add.
7482         (c-family/c-ubsan.o): Add.
7483         (builtins.o): Add ubsan.h dependency.
7484         * ubsan.h: New file.
7485         * ubsan.c: New file.
7486         * common.opt: Add -fsanitize=undefined option.
7487         (flag_sanitize): Add variable.
7488         (fsanitize=): Add option.  Add Driver.
7489         (fsanitize=thread): Remove option.
7490         (fsanitize=address): Likewise.
7491         (static-libubsan): New option.
7492         * doc/invoke.texi: Document the new flag and -static-libubsan.
7493         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7494         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7495         * builtin-attrs.def (ATTR_COLD): Define.
7496         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7497         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7498         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7499         * flag-types.h (sanitize_code): New enum.
7500         * opts.c (common_handle_option): Parse command line arguments
7501         of -fsanitize=.  Add -fsanitize=unreachable option.
7502         * varasm.c (get_variable_section): Adjust.
7503         (assemble_noswitch_variable): Likewise.
7504         (assemble_variable): Likewise.
7505         (output_constant_def_contents): Likewise.
7506         (categorize_decl_for_section): Likewise.
7507         (place_block_symbol): Likewise.
7508         (output_object_block): Likewise.
7509         * builtins.def: Likewise.
7510         * toplev.c (compile_file): Likewise.
7511         (process_options): Likewise.
7512         * cppbuiltin.c: Likewise.
7513         * tsan.c (tsan_pass): Likewise.
7514         (tsan_gate): Likewise.
7515         (tsan_gate_O0): Likewise.
7516         * cfgexpand.c (partition_stack_vars): Likewise.
7517         (expand_stack_vars): Likewise.
7518         (defer_stack_allocation): Likewise.
7519         (expand_used_vars): Likewise.
7520         * cfgcleanup.c (old_insns_match_p): Likewise.
7521         * asan.c (asan_finish_file): Likewise.
7522         (asan_instrument): Likewise.
7523         (gate_asan): Likewise.
7524         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7525         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7526         (asan_global_struct): Use pointer_sized_int_node instead
7527         calling build_nonstandard_integer_type.
7528         (initialize_sanitizer_builtins): Likewise.
7529         (asan_finish_file): Likewise.
7530         * gcc.c: Document %{%:function(args):X}.
7531         (static_spec_functions): Add sanitize.
7532         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7533         store funcval != NULL there.
7534         (do_spec_1): Adjust handle_spec_function caller.
7535         (handle_braces): Allow %:function(args) as condition.
7536         (sanitize_spec_function): New function.
7537         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7538         (LIBUBSAN_SPEC): Likewise.
7539         (LIBUBSAN_EARLY_SPEC): Likewise.
7540         (SANITIZER_SPEC): Handle libubsan.
7541         (SANITIZER_EARLY_SPEC): Likewise.
7542         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7543         instead of fsanitize=address.
7544         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7545         instead of fsanitize=address*.
7546         * builtins.c: Include ubsan.h.
7547         (fold_builtin_0): Instrument __builtin_unreachable.
7548         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7549         instead of flag_asan.
7550         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7551         (pointer_sized_int_node): Define.
7552         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7553
7554 2013-08-30  Mike Stump  <mikestump@comcast.net>
7555
7556         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7557         with RE patterns.
7558
7559 2013-08-29  Jan Hubicka  <jh@suse.cz>
7560
7561         * cgraph.c (cgraph_function_body_availability): Handle weakref
7562         correctly.
7563         * passes.def: Remove pass_fixup_cfg.
7564         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7565         track when we need to remove functions.
7566         (gate_ipa_inline): Execute inlining always; add comment why.
7567         (pass_data_ipa_inline): Remove TODO_remove_functions.
7568         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7569         do not produce summaries.
7570         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7571         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7572         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7573         constant pool and vtable.
7574
7575 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7576
7577         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7578         New arm_neon.h's internal macros to specify 64-bit constants.
7579         Avoid using stdint.h's macros.
7580
7581 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7582
7583         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7584         before applying REG_P.
7585
7586 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7587
7588         PR tree-optimization/58277
7589         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7590         after seeing too many stmts with vdef in between dombb and current
7591         bb, invalidate everything.
7592
7593 2013-08-30  Richard Biener  <rguenther@suse.de>
7594
7595         * fold-const.c (fold_single_bit_test): Fix overflow test.
7596
7597 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7598
7599         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7600         and which can live in a register, always retrieve the value on entry.
7601         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7602         passed by invisible reference specially.
7603         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7604         (vt_add_function_parameter): Correctly deal with a parameter passed by
7605         invisible reference.
7606
7607 2013-08-30  Jan Hubicka  <jh@suse.cz>
7608
7609         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7610
7611 2013-08-30  Richard Biener  <rguenther@suse.de>
7612
7613         PR tree-optimization/58228
7614         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7615         allow invariant loads in nested loop vectorization.
7616
7617 2013-08-30  Richard Biener  <rguenther@suse.de>
7618
7619         PR tree-optimization/58223
7620         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7621         (has_anti_or_output_dependence): ... this and adjust to also
7622         look for output dependences.
7623         (mark_nodes_having_upstream_mem_writes): Adjust.
7624         (rdg_flag_uses): Likewise.
7625
7626 2013-08-30  Richard Biener  <rguenther@suse.de>
7627
7628         PR tree-optimization/58010
7629         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7630         assert that we have a loop-closed PHI.
7631
7632 2013-08-29  Jan Hubicka  <jh@suse.cz>
7633
7634         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7635         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7636         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7637         (lto_free_function_in_decl_state_for_node): New function.
7638
7639 2013-08-29  Xinliang David Li  <davidxl@google.com>
7640
7641         * loop-unroll.c (report_unroll_peel): Minor message change.
7642         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7643         Emit alignment peeling message with default -fopt-info.
7644         (vect_loop_versioning): Emit loop version info message.
7645         * tree-vectorizer.c (vectorize_loops): Minor message change.
7646         (execute_vect_slp): Ditto.
7647
7648 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7649
7650         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7651         of the clone from the DECL_NAME of the original function.
7652
7653 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7654
7655         * passes.c (register_pass): Add overload.
7656         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7657
7658 2013-08-29  Jan Hubicka  <jh@suse.cz>
7659
7660         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7661         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7662         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7663         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7664         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7665         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7666         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7667         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7668
7669 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7670
7671         * dumpfile.c (dump_loc): Output column number.
7672         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7673         * doc/invoke.texi: Document optall -fopt-info flag.
7674         * profile.c (read_profile_edge_counts): Use new dump framework.
7675         (compute_branch_probabilities): Ditto.
7676         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7677         when pass not in any opt group.
7678         * pass_manager.h (pass_manager::get_pass_profile): New method.
7679         * value-prof.c (check_counter): Use new dump framework.
7680         (check_ic_target): Ditto.
7681         * coverage.c (get_coverage_counts): Ditto.
7682         (coverage_init): Setup new dump framework.
7683
7684 2013-08-29  Richard Biener  <rguenther@suse.de>
7685
7686         PR tree-optimization/58246
7687         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7688         handle the dominance check inside a basic-block.
7689
7690 2013-08-29  Richard Biener  <rguenther@suse.de>
7691
7692         PR middle-end/57287
7693         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7694         of default defs that appear in abnormal PHI nodes.
7695
7696 2013-08-29  Richard Biener  <rguenther@suse.de>
7697
7698         PR tree-optimization/57685
7699         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7700         single-use operands to avoid exponential complexity.
7701
7702 2013-08-28  Dehao Chen  <dehao@google.com>
7703
7704         * ipa-inline.c (edge_badness): Fix integer underflow.
7705
7706 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7707
7708         * gtm-builtins.def (_ITM_free): Declare leaf.
7709
7710 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7711
7712         PR target/58067
7713         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7714         (*tls_local_dynamic_base_64_largepic): Likewise.
7715         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7716         Remove predicate from call operand.
7717         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7718         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7719
7720 2013-08-28  Jeff Law  <law@redhat.com>
7721
7722         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7723         checks for the number of predecessors and successors allowed.
7724         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7725         which require copying a joiner block if there is a request which
7726         is a subpath that requires no joiner block copying.
7727
7728 2013-08-28  Jan Hubicka  <jh@suse.cz>
7729
7730         * lto-streamer-out.c (DFS_write_tree_body): Drop
7731         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
7732         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
7733         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
7734         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
7735         Do not read DECL_ERROR_ISSUED.
7736         (unpack_ts_decl_with_vis_value_fields): Do not read
7737         DECL_DEFER_OUTPUT.
7738         (lto_input_ts_binfo_tree_pointers): Do not read
7739         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
7740         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
7741         write DECL_ERROR_ISSUED..
7742         (pack_ts_decl_with_vis_value_fields): Do not write
7743         DECL_DEFER_OUTPUT.
7744         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
7745         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
7746         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
7747         * tree.h (tree_decl_common): Update comment.
7748         (DECL_ERROR_ISSUED): Remove.
7749
7750 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7751
7752         PR middle-end/58257
7753         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
7754
7755 2013-08-28  Jan Hubicka  <jh@suse.cz>
7756
7757         * builtins.def (free): Declare leaf.
7758
7759 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
7760
7761         * gdbhooks.py: New.
7762         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
7763         * configure: Regenerate.
7764
7765 2013-08-27  Martin Jambor  <mjambor@suse.cz>
7766
7767         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
7768         (ipa_ancestor_jf_data): Likewise.
7769         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
7770         (ipa_get_jf_pass_through_type_preserved): New function.
7771         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
7772         (ipa_get_jf_ancestor_type_preserved): New function.
7773         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
7774         (ipa_get_jf_ancestor_result): Likewise.
7775         (propagate_vals_accross_pass_through): Use
7776         ipa_get_jf_pass_through_result to do all the value mappings.
7777         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
7778         type_preserved flag.
7779         (ipa_set_jf_cst_copy): New function.
7780         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
7781         (ipa_set_jf_arith_pass_through): Likewise.
7782         (ipa_set_ancestor_jf): Likewise.
7783         (compute_complex_assign_jump_func): Set type_preserved instead of
7784         punting.
7785         (ipa_compute_jump_functions_for_edge): Likewise.
7786         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
7787         (update_jump_functions_after_inlining): Update type_preserved.
7788         Explicitely create jump functions when combining one with pass_through.
7789         (ipa_write_jump_function): Stream the type_preserved flags.
7790         (ipa_read_jump_function): Likewise.
7791
7792 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
7793             Aldy Hernandez  <aldyh@redhat.com>
7794
7795         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
7796         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
7797         * function.h (struct function): Add has_force_vect_loops and
7798         has_simduid_loops.
7799         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
7800         * gimple.c (gimple_build_omp_critical): Add KIND argument and
7801         handle it.
7802         * gimple.def: Update CLAUSES comments.
7803         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
7804         (gimple_build_omp_for): Add argument to prototype.
7805         (gimple_omp_for_kind): New.
7806         (gimple_omp_for_set_kind): New.
7807         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
7808         GOVD_DATA_SHARE_CLASS.
7809         (enum omp_region_type): Add ORT_SIMD.
7810         (gimple_add_tmp_var): Handle ORT_SIMD.
7811         (gimplify_var_or_parm_decl): Same.
7812         (is_gimple_stmt): Same.
7813         (omp_firstprivatize_variable): Same.
7814         (omp_add_variable): Only use splay_tree_insert if lookup failed.
7815         (omp_notice_variable): Handle ORT_SIMD.
7816         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
7817         (omp_check_private): Handle ORT_SIMD.
7818         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
7819         OMP_CLAUSE_SAFELEN.
7820         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
7821         Handle OMP_CLAUSE_LASTPRIVATE.
7822         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
7823         OMP_CLAUSE_SAFELEN.
7824         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
7825         (gimplify_expr): Handle OMP_SIMD.
7826         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
7827         (expand_GOMP_SIMD_VF): New.
7828         (expand_GOMP_SIMD_LAST_LANE): New.
7829         * internal-fn.def (GOMP_SIMD_LANE): New.
7830         (GOMP_SIMD_VF): New.
7831         (GOMP_SIMD_LAST_LANE): New.
7832         * omp-low.c: Include target.h.
7833         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
7834         OMP_CLAUSE_SAFELEN.
7835         (check_omp_nesting_restrictions): Same.
7836         (omp_max_vf): New.
7837         (lower_rec_simd_input_clauses): New.
7838         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
7839         OMP_CLAUSE_LINEAR.
7840         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
7841         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
7842         (expand_omp_build_assign): New.
7843         (expand_omp_for_init_counts): New.
7844         (expand_omp_for_init_vars): New.
7845         (extract_omp_for_update_vars): New.
7846         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
7847         and rewrite accordingly.
7848         (expand_omp_simd): New.
7849         (expand_omp_for): Use expand_omp_simd.
7850         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
7851         (lower_omp_for): Do not lower the body.
7852         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
7853         in their own loops.
7854         * tree-flow.h (find_omp_clause): Remove prototype.
7855         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
7856         forcing vectorization of the loop, or if flag_tree_vectorize.
7857         (gate_tree_if_conversion): Similarly.
7858         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
7859         gimple_build_omp_for.
7860         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
7861         * tree-parloops (create_parallel_loop): Pass kind argument to
7862         gimple_build_omp_for.
7863         * tree-pretty-print.c (dump_omp_clause): Add cases for
7864         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
7865         OMP_CLAUSE__SIMDUID_.
7866         (dump_generic_node): Handle OMP_SIMD.
7867         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
7868         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
7869         unroll OMP_SIMD loops here.
7870         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
7871         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
7872         loop->safelen.
7873         (vect_analyze_data_refs): Handle simd loops.
7874         * tree-vect-loop.c (vectorizable_live_operation): Handle
7875         IFN_GOMP_SIMD*.
7876         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
7877         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
7878         (vectorizable_load): Same.
7879         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
7880         (struct simduid_to_vf): New.
7881         (simduid_to_vf::hash): New.
7882         (simduid_to-vf::equal): New.
7883         (struct simd_array_to_simduid): New.
7884         (simd_array_to_simduid::hash): New.
7885         (simd_array_to_simduid::equal): New.
7886         (adjust_simduid_builtins): New.
7887         (struct note_simd_array_uses_struct): New.
7888         (note_simd_array_uses_cb): New.
7889         (note_simd_array_uses): New.
7890         (vectorize_loops): Handle simd hints and adjust simd builtins
7891         accordingly.
7892         * tree-vectorizer.h (struct _stmt_vec_info): Add
7893         simd_lane_access_p field.
7894         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
7895         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
7896         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
7897         (omp_clause_code_name): Same.
7898         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
7899         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
7900         * tree.def (OMP_SIMD): New entry.
7901         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
7902         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
7903         (OMP_CLAUSE_DECL): Adjust range for new clauses.
7904         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
7905         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
7906         (OMP_CLAUSE_LINEAR_STEP): New.
7907         (OMP_CLAUSE_SAFELEN_EXPR): New.
7908         (OMP_CLAUSE__SIMDUID__DECL): New.
7909         (find_omp_clause): New prototype.
7910
7911 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
7912
7913         * config/i386/driver-i386.c (host_detect_local_cpu): Update
7914         Haswell processor detection.
7915
7916 2013-08-27  Christian Widmer  <shadow@umbrox.de>
7917
7918         PR target/57927
7919         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
7920         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
7921         AVX2 capable processors.
7922
7923 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
7924
7925         * config/aarch64/arm_neon.h: Replace all inline asm implementations
7926         of vget_low_* with implementations in terms of other intrinsics.
7927
7928 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
7929
7930         PR middle-end/57219
7931         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
7932         values to -1, 0 and 1.
7933
7934 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
7935
7936         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
7937         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
7938         (<optab><mode>3_insn): Remove.
7939         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
7940         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
7941         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
7942         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
7943         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
7944         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
7945         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
7946         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
7947         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
7948         (ror<mode>3_insn): Likewise.
7949         * config/aarch64/predicates.md (aarch64_simd_register): New.
7950
7951 2013-08-27  Richard Biener  <rguenther@suse.de>
7952
7953         PR tree-optimization/57521
7954         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
7955         one edge is non-critical.
7956         (find_phi_replacement_condition): Make sure to use a non-critical
7957         edge.  Cleanup and remove old bug workarounds.
7958         (bb_postdominates_preds): Remove.
7959         (if_convertible_loop_p_1): Do not compute post-dominators.
7960         (combine_blocks): Do not free post-dominators.
7961         (main_tree_if_conversion): Likewise.
7962         (pass_data_if_conversion): Add TODO_verify_ssa.
7963
7964 2013-08-27  DJ Delorie  <dj@redhat.com>
7965
7966         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
7967
7968 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
7969
7970         * function.c (assign_parm_find_data_types): Set passed_mode and
7971         nominal_mode to the TYPE_MODE of nominal_type for the built
7972         pointer type in case of the struct-pass-by-reference.
7973
7974 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
7975
7976         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
7977         (UINT16_TYPE): Change default to "unsigned int".
7978
7979         * config/avr/avr.opt (mfract-convert-truncate): New option.
7980         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
7981         is set, round negative fractional integers according to n1169
7982         when converting to integer types.
7983
7984 2013-08-26  Jan Hubicka  <jh@suse.cz>
7985
7986         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
7987         methods can not be called indirectly when their address is not taken.
7988
7989 2013-08-26  Jan Hubicka  <jh@suse.cz>
7990
7991         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
7992         ctor_for_folding.
7993
7994 2013-08-26  Jan Hubicka  <jh@suse.cz>
7995
7996         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
7997         can be unshared.
7998
7999 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8000
8001         * reload.c (find_valid_class): Allow classes that do not include
8002         FIRST_PSEUDO_REGISTER - 1.
8003
8004 2013-08-26  Jan Hubicka  <jh@suse.cz>
8005
8006         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8007         fix edge count/frequency when speculation failed; fix type check
8008         for the direct call.
8009
8010 2013-08-26  Jan Hubicka  <jh@suse.cz>
8011
8012         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8013
8014 2013-08-26  Jan Hubicka  <jh@suse.cz>
8015
8016         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8017         to be changed by edge redirection.
8018
8019 2013-08-26  Jan Hubicka  <jh@suse.cz>
8020
8021         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8022         formating; add sanity check.
8023         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8024         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8025         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8026         (resolve_noninline_speculation): Update callee keys, too.
8027
8028 2013-08-26  Jan Hubicka  <jh@suse.cz>
8029
8030         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8031         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8032
8033 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8034
8035         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8036         into proper place.
8037
8038 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8039
8040         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8041         (x86_64_elf_select_section): Ditto.
8042         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8043         arguments.
8044         (ix86_pass_by_reference): Ditto.
8045         (output_set_got): Ditto.
8046         (ix86_unary_operator_ok): Ditto.
8047         (ix86_expand_builtin): Ditto.
8048
8049 2013-08-23  Jan Hubicka  <jh@suse.cz>
8050
8051         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8052
8053 2013-08-23  Jan Hubicka  <jh@suse.cz>
8054
8055         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8056         (tree_decl_with_vis): Add FINAL field.
8057
8058 2013-08-23  Jeff Law  <law@redhat.com>
8059
8060         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8061         the debugging dump when the expression is fully redundant.
8062
8063 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8064
8065         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8066         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8067         * pretty-print.c (pp_formatted_text_data): Likewise.
8068         (pp_write_text_to_stream): Likewise.
8069         (pp_write_text_as_dot_label_to_stream): Likewise.
8070         (pp_append_r): Likewise.
8071         (pp_format): Likewise.
8072         (pp_flush): Likewise.
8073         (pp_clear_output_area): Likewise.
8074         (pp_append_text): Likewise.
8075         (pp_formatted_text): Likewise.
8076         (pp_remaining_character_count_for_line): Likewise.
8077         (pp_newline): Likewise.
8078         (pp_character): Likewise.
8079         (output_buffer::~output_buffer): Define.
8080         (pretty_printer::~pretty_printer): Destruct output buffer.
8081         * pretty-print.h (output_buffer::~output_buffer): Declare.
8082         (pretty_printer::~pretty_printer): Declare virtual.
8083
8084 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8085
8086         PR other/57324
8087         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8088         HOST_WIDE_INT_M1U): New macros.
8089         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8090         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8091         unsigned -1 for lshift.
8092         * cse.c (cse_insn): Likewise.
8093         * double-int.c (rshift_double, lshift_double): Likewise.
8094         * builtins.c (fold_builtin_bitop): Likewise.
8095         * combine.c (force_to_mode): Likewise.
8096         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8097         * simplify-rtx.c (simplify_const_unary_operation,
8098         simplify_const_binary_operation): Likewise.
8099         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8100         check_va_list_escapes): Likewise.
8101         * rtlanal.c (nonzero_bits1): Likewise.
8102         * expmed.c (expand_smod_pow2): Likewise.
8103         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8104
8105 2013-08-23  Jan Hubicka  <jh@suse.cz>
8106
8107         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8108         as having address taken.
8109
8110 2013-08-23  Jan Hubicka  <jh@suse.cz>
8111
8112         * ipa-utils.h (method_class_type): Declare.
8113         * ipa-devirt.c (method_class_type): Export.
8114
8115         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8116         do not walk base classes of anonymous types.
8117
8118 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8119
8120         PR rtl-optimization/58220
8121         PR regression/58221
8122         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8123         handle SEQUENCE insns properly.
8124
8125 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8126
8127         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8128         definition.
8129         (pp_newline_and_indent): Likewise.
8130         (pp_separate_with): Likewise.
8131         * pretty-print.c (pp_newline_and_flush): Define.
8132         (pp_newline_and_indent): Likewise.
8133         (pp_separate_with): Likewise.
8134
8135 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8136
8137         PR target/58218
8138         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8139         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8140
8141 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8142
8143         * config/i386/predicates.md (ext_sse_reg_operand): New.
8144         * config/i386/i386.md (*movti_internal): Use
8145         predicate to determine if EVEX is needed.
8146         (*movsi_internal): Ditto.
8147         (*movdf_internal): Ditto.
8148         (*movsf_internal): Ditto.
8149         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8150
8151 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8152
8153         PR tree-optimization/58209
8154         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8155         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8156         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8157         emit POINTER_PLUS_EXPR.
8158         (create_tailcall_accumulator): For pointer result type accumulate in
8159         sizetype type.
8160
8161 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8162
8163         * configure.ac: Add backslashes missing from the last change.
8164         * configure: Regenerate.
8165
8166 2013-08-22  Jan Hubicka  <jh@susue.cz>
8167
8168         * ipa.c (function_and_variable_visibility): First remember function
8169         was global and then make it local.
8170
8171 2013-08-22  Julian Brown  <julian@codesourcery.com>
8172
8173         * configure.ac: Add aarch64 to list of arches which use "nop" in
8174         debug_line test.
8175         * configure: Regenerate.
8176
8177 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8178
8179         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8180         gnu_libc_has_function.
8181         * config/s390/tpf.h: Likewise.
8182
8183 2013-08-22  Jan Hubicka  <jh@susue.cz>
8184
8185         * timevar.c (validate_phases): Add cast.
8186
8187 2013-08-22  Jan Hubicka  <jh@susue.cz>
8188
8189         * timevar.c (validate_phases): Use size_t for memory.
8190         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8191
8192 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8193
8194         * pretty-print.h (output_buffer::output_buffer): Declare.
8195         (pretty_printer::pretty_printer): Likewise.
8196         (pp_construct): Remove.
8197         * pretty-print.c (output_buffer::output_buffer): Define.
8198         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8199         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8200         (print_gimple_expr): Likewise.
8201         (print_gimple_seq): Likewise.
8202         (gimple_dump_bb): Likewise.
8203         * sched-vis.c (dump_value_slim): Likewise.
8204         (dump_insn_slim): Likewise.
8205         (dump_rtl_slim): Likewise.
8206         (str_pattern_slim): Likewise.
8207         * tree-mudflap.c (mf_varname_tree): Likewise.
8208         * graph.c (print_graph_cfg): Likewise.
8209         (start_graph_dump): Likewise.
8210         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8211         placement-new.
8212         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8213         pretty printer initialization.
8214         * coretypes.h (diagnostic_context): Remove superflous type alias
8215         declaration.
8216         (pretty_printer): Likewise.  Declare directly as a class.
8217         (pretty_print_info): Remove declaration as class.
8218         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8219         and pp_clear_output_area.
8220         (asan_add_global): Likewise.
8221
8222 2013-08-22  Jan Hubicka  <jh@suse.cz>
8223
8224         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8225         * ipa-utils.h (update_type_inheritance_graph): Declare.
8226         (possible_polymorphic_call_target_p): Declare.
8227         (possible_polymorphic_call_target_p): New.
8228         * ipa-devirt.c: Update toplevel comments.
8229         (cached_polymorphic_call_targets): Move up.
8230         (odr_type_d): Move ID down.
8231         (polymorphic_type_binfo_p): Update comment.
8232         (odr_hasher::remove): Likewise;
8233         (get_odr_type): Set anonymous_namespace.
8234         (dump_odr_type): Dump it.
8235         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8236         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8237         (record_binfo): Add comment.
8238         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8239         built.
8240         (devirt_node_removal_hook): Do not iCE when cache is freed.
8241         (possible_polymorphic_call_target_p): New predicate.
8242         (update_type_inheritance_graph): New function.
8243
8244 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8245             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8246             Sergey Lega  <sergey.s.lega@intel.com>
8247             Anna Tikhonova  <anna.tikhonova@intel.com>
8248             Ilya Tocar  <ilya.tocar@intel.com>
8249             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8250             Ilya Verbin  <ilya.verbin@intel.com>
8251             Kirill Yukhin  <kirill.yukhin@intel.com>
8252             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8253
8254         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8255         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8256         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8257         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8258         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8259         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8260         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8261         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8262         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8263         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8264         OPT_mavx512pf, OPT_mavx512er cases.
8265         * config/i386/constraints.md (v): New constraint.
8266         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8267         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8268         (bit_AVX512CD): New.
8269         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8270         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8271         * config/i386/i386-c.c (ix86_target_macros_internal):
8272         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8273         __AVX512PF__.
8274         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8275         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8276         * config/i386/i386.c (regclass_map, dbx_register_map)
8277         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8278         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8279         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8280         -mavx512pf options.
8281         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8282         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8283         -mavx512pf options.  Fix formatting.
8284         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8285         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8286         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8287         -mavx512cd, -mavx512pf options.
8288         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8289         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8290         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8291         ALL_SSE_REGS.
8292         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8293         (ix86_set_reg_reg_cost): Ditto.
8294         (x86_order_regs_for_local_alloc): Ditto.
8295         (MAX_VECT_LEN): Extend to 64-byte.
8296         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8297         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8298         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8299         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8300         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8301         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8302         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8303         (SSE_REG_MODE_P): Support new modes.
8304         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8305         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8306         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8307         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8308         (REG_CLASS_CONTENTS): Add new registers.
8309         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8310         (EXT_REX_SSE_REGNO_P): New.
8311         (HI_REGISTER_NAMES): Add new registers.
8312         * config/i386/i386.md: Define constants for new registers.
8313         (mode): Add new 512-bit modes.
8314         (prefix): Support evex prefix.
8315         (isa): Support avx512f, noavx512f, fma_avx512f.
8316         (ssemodesuffix): Add new 512-bit modes.
8317         (movxi): New.
8318         (*movxi_internal_avx512f): Ditto.
8319         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8320         Support MODE_XI.
8321         (*movsi_internal): Likewise.
8322         (*movdf_internal): Likewise.
8323         (*movsf_internal): Likewise.
8324         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8325         (<code><mode>3): Likewise.
8326         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8327         New.
8328         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8329         with the new constraint "v".
8330         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8331         modes.
8332         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8333         with the new constraint "v".
8334         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8335         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8336         (*<plusminus_insn><mode>3): Likewise.
8337         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8338         (*mul<mode>3): Likewise.
8339         (<sse>_vmmul<mode>3): Likewise.
8340         (<sse>_div<mode>3): Likewise.
8341         (<sse>_vmdiv<mode>3): Likewise.
8342         (<sse>_sqrt<mode>2): Likewise.
8343         (<sse>_vmsqrt<mode>2): Likewise.
8344         (*<code><mode>3_finite): Likewise.
8345         (*<code><mode>3) <smaxmin>: Likewise.
8346         (<sse>_vm<code><mode>3): Likewise.
8347         (*<code><mode>3) <any_logic>: Likewise.
8348         (*fma_fmadd_<mode>): Likewise.
8349         (*fma_fmsub_<mode>): Likewise.
8350         (*fma_fnmadd_<mode>): Likewise.
8351         (*fma_fnmsub_<mode>): Likewise.
8352         (*fma_fmaddsub_<mode>): Likewise.
8353         (*fma_fmsubadd_<mode>): Likewise.
8354         (*fmai_fmadd_<mode>): Likewise.
8355         (*fmai_fmsub_<mode>): Likewise.
8356         (*fmai_fnmadd_<mode>): Likewise.
8357         (*fmai_fnmsub_<mode>): Likewise.
8358         (sse_cvtsi2ss): Likewise.
8359         (sse_cvtsi2ssq): Likewise.
8360         (sse_cvtss2si): Likewise.
8361         (sse_cvtss2si_2): Likewise.
8362         (sse_cvtss2siq): Likewise.
8363         (sse_cvtss2siq_2): Likewise.
8364         (sse_cvttss2si): Likewise.
8365         (sse_cvtss2siq_2): Likewise.
8366         (float<sseintvecmodelower><mode>2): Likewise.
8367         (sse2_cvtsd2si_2): Likewise.
8368         (sse2_cvtsd2siq_2): Likewise.
8369         (*<plusminus_insn><mode>3): Likewise.
8370         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8371         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8372         (ashr<mode>3): Likewise.
8373         (<shift_insn><mode>3): Likewise.
8374         (avx2_<code><mode>3): Likewise.
8375         (*avx2_<code><mode>3): Likewise.
8376         (*andnot<mode>3): Likewise.
8377         (*<code><mode>3) <any_logic>: Likewise.
8378         (abs<mode>2): Likewise.
8379         (avx2_permvar<mode>): Likewise.
8380         (avx2_perm<mode>_1): Likewise.
8381         (*avx_vpermilp<mode>): Likewise.
8382         (avx_vpermilvar<mode>3): Likewise.
8383         (avx2_ashrv<mode>): Likewise.
8384         (avx2_<shift_insn>v<mode>): Likewise.
8385         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8386         -mavx512cd.
8387         * doc/rtl.texi: Document XImode.
8388
8389 2013-08-21  Jeff Law  <law@redhat.com>
8390
8391         * tree-flow.h (register_jump_thread): Pass vector of edges
8392         instead of each important edge.
8393         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8394         thread path into a vector and pass that to register_jump_thread.
8395         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8396         passed in edge vector to the current 3-edge form.
8397
8398         Revert:
8399         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8400
8401         * dce.c (fini_dce): Call df_analyze again just in case
8402         delete_unmarked_insns removed anything.
8403
8404 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8405
8406         * reload.h (struct reg_equivs): Rename to ..
8407         (struct reg_equivs_s): .. this.
8408
8409 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8410
8411         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8412
8413 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8414
8415         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8416
8417 2013-08-21  Jeff Law  <law@redhat.com>
8418
8419         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8420         simplify assignments too.  If the RHS collapses to a singleton
8421         range, then return the value for the range.
8422
8423 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8424
8425         * config/i386/sse.md (V16): Rename to...
8426         (VMOVE): this.
8427         (mov<mode>): Update iterator name.
8428         (*mov<mode>_internal): Ditto.
8429         (push<mode>1): Ditto.
8430         (movmisalign<mode>): Ditto.
8431
8432 2013-08-20  Jan Hubicka  <jh@suse.cz>
8433
8434         PR bootstrap/58186
8435         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8436         entry for direct edges.
8437         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8438
8439 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8440
8441         Revert my last two changes, r201865 and r201864:
8442
8443         Revert r201865:
8444         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8445
8446         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8447         instances can own GC refs.
8448
8449         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8450         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8451         (gcc::context::gt_pch_nx): Likewise.
8452         (gcc::context::gt_pch_nx):  Likewise.
8453         * ggc.h (gt_ggc_mx <T>): New.
8454         (gt_pch_nx_with_op <T>): New.
8455         (gt_pch_nx <T>): New.
8456         * passes.c (opt_pass::gt_ggc_mx): New.
8457         (opt_pass::gt_pch_nx): New.
8458         (opt_pass::gt_pch_nx_with_op): New.
8459         (pass_manager::gt_ggc_mx): New.
8460         (pass_manager::gt_pch_nx): New.
8461         (pass_manager::gt_pch_nx_with_op): New.
8462         (pass_manager::operator new): Use
8463         ggc_internal_cleared_alloc_stat rather than xcalloc.
8464         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8465         (pass_manager::gt_ggc_mx): New.
8466         (pass_manager::gt_pch_nx): New.
8467         (pass_manager::gt_pch_nx_with_op): New.
8468         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8469         (opt_pass::operator new): New.
8470         (opt_pass::gt_ggc_mx): New.
8471         (opt_pass::gt_pch_nx): New.
8472         (opt_pass::gt_pch_nx_with_op): New.
8473
8474         Revert r201864:
8475         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8476
8477         * Makefile.in (GTFILES): Add context.h.
8478         * context.c (gcc::context::operator new): New.
8479         (gcc::context::gt_ggc_mx): New.
8480         (gcc::context::gt_pch_nx): New.
8481         (gcc::context::gt_pch_nx): New.
8482         * context.h (gcc::context): Add GTY((user)) marking.
8483         (gcc::context::operator new): New.
8484         (gcc::context::gt_ggc_mx): New.
8485         (gcc::context::gt_pch_nx): New.
8486         (gcc::context::gt_pch_nx): New.
8487         (g): Add GTY marking.
8488         (gt_ggc_mx (gcc::context *)): New.
8489         (gt_pch_nx (gcc::context *)): New.
8490         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8491         void *cookie)): New.
8492         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8493
8494 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8495
8496         * dce.c (fini_dce): Call df_analyze again just in case
8497         delete_unmarked_insns removed anything.
8498
8499 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8500
8501         PR rtl-optimizations/57451
8502         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8503         from crossing split section boundaries.
8504
8505 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8506
8507         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8508         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8509         with MULTLIB_DEFAULTS.
8510
8511 2013-08-20  Nick Clifton  <nickc@redhat.com>
8512
8513         * target.def (narrow_volatile_bitfield): Note that the default
8514         value is false, not !TARGET_STRICT_ALIGN.
8515         * doc/tm.texi: Regenerate.
8516
8517 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8518
8519         Fix LIB_SPEC for systems without libpthread.
8520
8521         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8522         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8523         for Android.
8524         * config/i386/linux-common.h: Likewise.
8525         * config/mips/linux-common.h: Likewise.
8526
8527 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8528
8529         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8530         checks.
8531
8532 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8533
8534         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8535         instances can own GC refs.
8536
8537         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8538         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8539         (gcc::context::gt_pch_nx): Likewise.
8540         (gcc::context::gt_pch_nx):  Likewise.
8541         * ggc.h (gt_ggc_mx <T>): New.
8542         (gt_pch_nx_with_op <T>): New.
8543         (gt_pch_nx <T>): New.
8544         * passes.c (opt_pass::gt_ggc_mx): New.
8545         (opt_pass::gt_pch_nx): New.
8546         (opt_pass::gt_pch_nx_with_op): New.
8547         (pass_manager::gt_ggc_mx): New.
8548         (pass_manager::gt_pch_nx): New.
8549         (pass_manager::gt_pch_nx_with_op): New.
8550         (pass_manager::operator new): Use
8551         ggc_internal_cleared_alloc_stat rather than xcalloc.
8552         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8553         (pass_manager::gt_ggc_mx): New.
8554         (pass_manager::gt_pch_nx): New.
8555         (pass_manager::gt_pch_nx_with_op): New.
8556         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8557         (opt_pass::operator new): New.
8558         (opt_pass::gt_ggc_mx): New.
8559         (opt_pass::gt_pch_nx): New.
8560         (opt_pass::gt_pch_nx_with_op): New.
8561
8562 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8563
8564         * Makefile.in (GTFILES): Add context.h.
8565         * context.c (gcc::context::operator new): New.
8566         (gcc::context::gt_ggc_mx): New.
8567         (gcc::context::gt_pch_nx): New.
8568         (gcc::context::gt_pch_nx): New.
8569         * context.h (gcc::context): Add GTY((user)) marking.
8570         (gcc::context::operator new): New.
8571         (gcc::context::gt_ggc_mx): New.
8572         (gcc::context::gt_pch_nx): New.
8573         (gcc::context::gt_pch_nx): New.
8574         (g): Add GTY marking.
8575         (gt_ggc_mx (gcc::context *)): New.
8576         (gt_pch_nx (gcc::context *)): New.
8577         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8578         void *cookie)): New.
8579         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8580
8581 2013-08-20  Alan Modra  <amodra@gmail.com>
8582
8583         PR target/57865
8584         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8585         (rs6000_emit_epilogue): Likewise.
8586
8587 2013-08-19  Dehao Chen  <dehao@google.com>
8588
8589         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8590
8591 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8592             Jakub Jelinek  <jakub@redhat.com>
8593
8594         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8595         (BUILT_IN_FABSD64): Likewise.
8596         (BUILT_IN_FABSD128): Likewise.
8597         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8598         (fold_builtin_1): Likewise.
8599         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8600         destination and source operands.
8601         (*abstd2_fpr): Likewise.
8602         (*nabstd2_fpr): Likewise.
8603
8604 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8605
8606         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8607         JUMP_P and INSN_P.
8608
8609 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8610
8611         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8612         incomplete.
8613
8614 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8615
8616         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8617         * builtins.c (default_libc_has_function): New.
8618         (gnu_libc_has_function): Ditto.
8619         (no_c99_libc_has_function): Ditto.
8620         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8621         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8622         (fold_builtin_sincos): Likewise.
8623         (fold_builtin_cexp): Likewise.
8624         * builtins.def (DEF_C94_BUILTIN): Likewise.
8625         (DEF_C99_BUILTIN): Likewise.
8626         (DEF_C99_C90RES_BUILTIN): Likewise.
8627         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8628         definitions to using this define.
8629         * config/darwin-protos.h (darwin_libc_has_function): New.
8630         * config/darwin.c (darwin_libc_has_function): Ditto.
8631         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8632         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8633         * config/darwin.h: Ditto.
8634         * config/elfos.h: Ditto.
8635         * config/freebsd.h: Ditto.
8636         * config/i386/cygming.h: Ditto.
8637         * config/i386/djgpp.h: Ditto.
8638         * config/i386/i386-interix.h: Ditto.
8639         * config/microblaze/microblaze.h: Ditto.
8640         * config/mmix/mmix.h: Ditto.
8641         * config/gnu-user.h: Ditto.
8642         * config/ia64/hpux.h: Ditto.
8643         * config/pa/pa-hpux.h: Ditto.
8644         * config/pdp11/pdp11.h: Ditto.
8645         * config/picochip/picochip.h: Ditto.
8646         * config/linux.h: Ditto.
8647         * config/netbsd.h: Ditto.
8648         * config/openbsd.h: Ditto.
8649         * config/rs6000/aix43.h: Ditto.
8650         * config/rs6000/aix51.h: Ditto.
8651         * config/rs6000/aix52.h: Ditto.
8652         * config/rs6000/aix53.h: Ditto.
8653         * config/rs6000/aix61.h: Ditto.
8654         * config/rs6000/darwin.h: Ditto.
8655         * config/rs6000/linux.h: Ditto.
8656         * config/rs6000/linux64.h: Ditto.
8657         * config/s390/tpf.h: Ditto.
8658         * config/sol2-10.h: Ditto.
8659         * config/sol2.h: Ditto.
8660         * config/vms/vms.h: Ditto.
8661         * config/vxworks.h: Ditto.
8662         * config/linux-android.c (linux_android_libc_has_function):
8663         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8664         * config/linux-protos.h (linux_android_libc_has_function):
8665         New declaration.
8666         * config/i386/i386.c (ix86_libc_has_function): New.
8667         * config/i386/i386-protos.h
8668         (ix86_libc_has_function): New declaration.
8669         * config/i386/i386.md
8670         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8671         ("isinf<mode>2): Likewise.
8672         * convert.c (convert_to_integer): Using new target hook
8673         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8674         TARGET_C99_FUNCTIONS.
8675         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8676         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8677         * coretypes.h (function_class): New enum for different
8678         classes of functions.
8679         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8680         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8681         (TARGET_HAS_SINCOS): Likewise.
8682         (TARGET_LIBC_HAS_FUNCTION): New.
8683         * doc/tm.texi: Regenerated.
8684         * targhooks.h (default_libc_has_function): New declaration.
8685         (no_c99_libc_has_function): Ditto.
8686         (gnu_libc_has_function): Ditto.
8687         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8688         and TARGET_HAS_SINCOS.
8689
8690 2013-08-18  Jan Hubicka  <jh@suse.cz>
8691
8692         * Makeifle-in (ipa-devirt.o): New.
8693         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8694         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8695         (analyze_functions): Look into possible targets of polymorphic call.
8696         * dumpfile.c (dump_files): Add type-inheritance dump.
8697         * dumpfile.h (TDI_inheritance): New.
8698         * ipa-devirt.c: New file.
8699         * ipa-utils.h (odr_type_d): Forward declare.
8700         (odr_type): New type.
8701         (build_type_inheritance_graph): Declare.
8702         (possible_polymorphic_call_targets): Declare and introduce inline
8703         variant when only edge is pased.
8704         (dump_possible_polymorphic_call_targets): Likewise.
8705         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8706         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8707         (types_same_for_odr): ... here.
8708         * tree.h (type_in_anonymous_namespace_p): Declare.
8709
8710 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8711
8712         PR tree-optimization/58006
8713         * tree-parloops.c (take_address_of): Don't ICE if get_name
8714         returns NULL.
8715         (eliminate_local_variables_stmt): Remove clobber stmts.
8716
8717 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8718
8719         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8720         error message is issued for an alias to undefined symbol.
8721
8722 2013-08-18  Jan Hubicka  <jh@suse.cz>
8723
8724         * cgraph.c (cgraph_create_indirect_edge): Discover
8725         polymorphic calls and record basic info into indirect_info.
8726         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8727         devirtualization, ignore objc function calls.
8728         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
8729         call with no parm index info.
8730         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
8731         * tree.c (virtual_method_call_p): New function.
8732         * tree.h (virtual_method_call_p): Declare.
8733
8734 2013-08-16  Jan Hubicka  <jh@suse.cz>
8735
8736         PR middle-end/58179
8737         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
8738
8739 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
8740
8741         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
8742         attribute.
8743
8744 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8745
8746         * gengtype.c (type_for_name): Add special-case support for
8747         locating types within the "gcc::" namespace.
8748         (open_base_files): Emit a "using namespace gcc" directive.
8749
8750 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8751
8752         PR target/58160
8753         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
8754         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
8755
8756         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
8757         array instead of each individual operand as a separate argument.
8758         (emit_fusion_gpr_load): Likewise.
8759         (expand_fusion_gpr_load): Add new function declaration.
8760
8761         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
8762         signature to have the operands passed as an array, instead of as
8763         separate arguments.  Allow ZERO_EXTEND to be in the memory
8764         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
8765         depend on the register live/dead flags when peepholes are run.
8766         (expand_fusion_gpr_load): New function to be called from the
8767         peephole2 pass, to change the register that addis sets to be the
8768         target register.
8769         (emit_fusion_gpr_load): Change the calling signature to have the
8770         operands passed as an array, instead of as separate arguments.
8771         Allow ZERO_EXTEND to be in the memory address, and also
8772         SIGN_EXTEND if -mpower8-fusion-sign.
8773
8774         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
8775         unspec enumeration.
8776         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
8777         adjust the register addis loads up in the peephole2 pass.  Do not
8778         depend on the register live/dead state when the peephole pass is done.
8779
8780 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8781
8782         * gengtype.c (create_user_defined_type): Ensure that the kind
8783         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
8784         declaration is seen before the GTY((user)) marking.
8785
8786 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8787
8788         PR target/58105
8789         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
8790
8791 2013-08-16  Jan Hubicka  <jh@suse.cz>
8792
8793         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
8794         arugment expected_type.
8795         (gimple_fold_call): Use it.
8796         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
8797         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
8798         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
8799         (try_make_edge_direct_virtual_call): Likewise.
8800         * tree.c (obj_type_ref_class): New.
8801         * tree.h (obj_type_ref_class): Use it.
8802
8803 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8804
8805         * sched-vis.c (rtl_slim_pp_initialized): Remove.
8806         (rtl_slim_pp): Likewise.
8807         (init_rtl_slim_pretty_print): Likewise.
8808         (dump_value_slim):  Don't call it.  Use local pretty printer.
8809         (dump_insn_slim): Likewise.
8810         (dump_rtl_slim): Likewise.
8811         (str_pattern_slim): Likewise.
8812         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
8813         Simplify.
8814
8815 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
8816
8817         PR tree-optimization/58164
8818         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
8819         walk gimple_goto_dest of GIMPLE_GOTO.
8820
8821         PR tree-optimization/58165
8822         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
8823         bi_call must be the last stmt in a bb, don't split_block, instead
8824         use fallthru edge from it and give up if there is none.
8825         Release conds vector when returning early.
8826
8827 2013-08-14  Xinliang David Li  <davidxl@google.com>
8828
8829         * config/i386/i386.c (ix86_option_override_internal):
8830         Remove unused variable and field.
8831
8832 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8833
8834         PR target/57949
8835         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
8836         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
8837         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
8838         and Linux, correct BLKmode alignment when 128-bit alignment is
8839         required and compatibility flag is not set.
8840         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
8841         for zero-size arguments when compatibility flag is not set.
8842
8843 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
8844
8845         PR tree-optimization/58145
8846         * tree-sra.c (build_ref_for_offset): If prev_base has
8847         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
8848
8849 2013-08-14  Xinliang David Li  <davidxl@google.com>
8850
8851         * config/i386/i386.c (ix86_option_override_internal):
8852         Fix uninitialized variable error.
8853
8854 2013-08-14  Xinliang David Li  <davidxl@google.com>
8855
8856         * config/i386/i386.opt: Define two new options.
8857         * config/i386/x86-tune.def: Add arch selector field in macros.
8858         * config/i386/i386.h: Adjust macro definition.
8859         * config/i386/i386.c (ix86_option_override_internal):
8860         Refactor the code.
8861         (parse_mtune_ctrl_str): New function.
8862         (set_ix86_tune_features): New function.
8863         (ix86_function_specific_restore): Call the new helper function.
8864
8865 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
8866
8867         PR rtl-optimization/57662
8868         * sel-sched.c (code_motion_process_successors): When the current insn
8869         is removed after the recursive traversal, break from the loop.
8870         Add comments and debug printouts.
8871
8872 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
8873             Alexandre Oliva  <aoliva@redhat.com>
8874
8875         PR target/58067
8876         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
8877         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
8878         there also UNSPEC_PLTOFF.
8879
8880 2013-08-14  Marek Polacek  <polacek@redhat.com>
8881
8882         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
8883         NUM_CONDITIONS bit positions.
8884
8885 2013-08-13  Cary Coutant  <ccoutant@google.com>
8886
8887         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
8888         (attr_checksum): Hash vector contents instead of pointer.
8889         (attr_checksum_ordered): Likewise.
8890
8891 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
8892
8893         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
8894         when Pmode != word_mode.  Add length_address attribute.
8895         (sse3_monitor_<mode>): Merge from sse3_monitor and
8896         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
8897         Pmode != word_mode.  Update insn length attribute.
8898         * config/i386/i386.c (ix86_option_override_internal): Update
8899         ix86_gen_monitor selection for merged sse3_monitor insn.
8900
8901 2013-08-13  Julian Brown  <julian@codesourcery.com>
8902
8903         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
8904         perform invalid legitimization on greater-than-word-size modes for
8905         TARGET_E500_DOUBLE.
8906
8907 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
8908
8909         * ira.c (setup_class_translate_array): Use aclass instead of cl
8910         for classes not fully covered by allocno classes.
8911
8912 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
8913
8914         PR tree-optimization/57661
8915         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
8916         * tree-inline.c (tree_function_versioning): Initialize it.
8917         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
8918         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
8919         that is not being copied.
8920
8921         PR sanitizer/56417
8922         * asan.c (instrument_strlen_call): Fix typo in comment.
8923         Use char * type even for the lhs of POINTER_PLUS_EXPR.
8924
8925 2013-08-13  Steve Ellcey  <sellcey@mips.com>
8926
8927         * config/mips/mips.md (prefetch): Use lw instead of ld on
8928         loongson in 32bit mode.
8929
8930 2013-08-13  Nick Clifton  <nickc@redhat.com>
8931
8932         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
8933
8934 2013-08-13  Jan Hubicka  <jh@suse.cz>
8935
8936         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
8937         introduced edge; fix typo in sanity check.
8938         (cgraph_resolve_speculation): Export; improve diagnostic.
8939         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
8940         speculation at type mismatch.
8941         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
8942         (cgraph_resolve_speculation): Declare.
8943         (symtab_can_be_discarded): New function.
8944         * value-prof.c (gimple_ic_transform): Remove actual transform code.
8945         * ipa-inline-transform.c (speculation_removed): New global var.
8946         (clone_inlined_nodes): See if speculation can be removed.
8947         (inline_call): If speculations was removed, we growths may not match.
8948         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
8949         (speculation_useful_p): New function.
8950         (resolve_noninline_speculation): New function.
8951         (inline_small_functions): Resolve useless speculations.
8952         * ipa-inline.h (speculation_useful_p): Declare
8953         * ipa.c (can_replace_by_local_alias): Simplify.
8954         (ipa_profile): Produce speculative calls in non-lto, too;
8955         add simple cost model; produce local aliases.
8956
8957 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
8958
8959         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
8960         PASS_MANAGER_H.
8961
8962 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
8963
8964         * config/i386/i386.c (ix86_function_versions): Use error + inform.
8965
8966 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
8967
8968         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
8969         iterator instead of X87MODEF.
8970
8971 2013-08-12  Perez Read  <netfirewall@gmail.com>
8972
8973         PR target/58132
8974         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
8975         operand 0 for intel asm alternative.
8976         (*movabs<mode>_2): Ditto for operand 1.
8977
8978 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
8979
8980         * config/aarch64/arm_none.h
8981         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
8982
8983 2013-08-12  Nick Clifton  <nickc@redhat.com>
8984
8985         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
8986
8987 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8988
8989         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
8990         expand for QI/HImode operand to produce more effictive code for
8991         unsigned char(short) --> float(double) conversion.
8992
8993 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
8994
8995         * doc/invoke.texi: Mention that -ftls-model does not force the final
8996         model.
8997
8998 2013-08-12  Marek Polacek  <polacek@redhat.com>
8999             Marc Glisse  <marc.glisse@inria.fr>
9000
9001         PR tree-optimization/57980
9002         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9003         when creating -1 constant.
9004
9005 2013-08-10  Jan Hubicka  <jh@suse.cz>
9006
9007         Workaround binutils PR14342.
9008         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9009         (gimple_init_edge_profiler): Likewise.
9010         (gimple_gen_ic_func_profiler): Likewise.
9011
9012 2013-08-09  Jan Hubicka  <jh@suse.cz>
9013
9014         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9015
9016 2013-08-09  Xinliang David Li  <davidxl@google.com>
9017
9018         * config/i386/stringop.def: New file.
9019         * config/i386/stringop.opt: New file.
9020         * config/i386/i386-opts.h: Include stringopt.def.
9021         * config/i386/i386.opt: Include stringopt.opt.
9022         * config/i386/i386.c (ix86_option_override_internal):
9023         Override default size based stringop inline strategies with options.
9024         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9025         New function.
9026
9027 2013-08-09  Jan Hubicka  <jh@suse.cz>
9028
9029         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9030
9031 2013-08-09  Jan Hubicka  <jh@suse.cz>
9032
9033         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9034         CGRAPH_FREQ_MAX.
9035         (dump_cgraph_node): Dump profile-id.
9036         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9037         and common_target_probability.
9038         * lto-cgraph.c (lto_output_edge): Stream common targets.
9039         (lto_output_node): Stream profile ids.
9040         (input_node): Stream profile ids.
9041         (input_edge): Stream common targets.
9042         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9043         * ipa.c: Include value-prof.h
9044         (ipa_profile_generate_summary): Turn indirect call statement histograms
9045         into common targets.
9046         (ipa_profile): Turn common targets into speculative edges.
9047
9048 2013-08-09  Jan Hubicka  <jh@suse.cz>
9049
9050         * cgraph.h (cgraph_node): Add profile_id.
9051         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9052         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9053         (del_node_map): Update.
9054         (find_func_by_funcdef_no): Replace by ...
9055         (find_func_by_profile_id): ... this one.
9056         (gimple_ic_transform): Do not remove useful histograms when
9057         speculation is not done; dump info when indirect call removal
9058         can happen at LTO.
9059         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9060         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9061         (__gcov_indirect_call_profiler_v2): .. this one.
9062         * profile.h (init_node_map): Update.
9063         * coverage.c (coverage_compute_profile_id): New function.
9064         * coverage.h (coverage_compute_profile_id): Declare.
9065         * tree-profile.c (init_ic_make_global_vars): Make
9066         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9067         (gimple_init_edge_profiler): Update prototype of
9068         __gcov_indirect_call_profiler.
9069         (gimple_gen_ic_func_profiler): Simplify.
9070         (tree_profiling): Use init_node_map
9071
9072 2013-08-09  Jan Hubicka  <jh@suse.cz>
9073
9074         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9075         non-speculative refs.
9076         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9077         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9078         (cgraph_set_call_stmt): Likewise.
9079         (cgraph_create_edge_1): Fix release checking compilatoin;
9080         clear lto_stmt_uid.
9081         (cgraph_free_edge): Free indirect info.
9082         (cgraph_turn_edge_to_speculative): New function.
9083         (cgraph_speculative_call_info): New function.
9084         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9085         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9086         (dump_cgraph_node): Dump speculation.
9087         (verify_edge_count_and_frequency): Accept speculative edges.
9088         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9089         (verify_cgraph_node): Handle speculation.
9090         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9091         (cgraph_set_call_stmt): Update prototype.
9092         (cgraph_make_edge_direct): Update prototype.
9093         (cgraph_speculative_call_info): Declare.
9094         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9095         to change; update call of ipa_find_references.
9096         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9097         lto_stmt_uid and speculative flags.
9098         (ipa_dump_references): Dump speculation.
9099         (ipa_clone_references): Clone speculative flag.
9100         (ipa_clone_referring): Likewise.
9101         (ipa_clone_ref): New function.
9102         (ipa_find_reference): Look into lto_stmt_uids
9103         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9104         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9105         (ipa_find_reference): Update declaration.
9106         (ipa_clone_ref): Declare.
9107         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9108         stream speculative flag.
9109         (lto_output_ref): Stream statements uids and speculation.
9110         (input_ref): Likewise.
9111         (input_edge): Stream speuclation.
9112         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9113         (cgraph_set_call_stmt_including_clones): Handle speculation.
9114         * ipa-inline.c (heap_edge_removal_hook): New function.
9115         (inline_small_functions): Register it.
9116         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9117         also initialize refs.
9118         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9119         edge to change.
9120         (try_make_edge_direct_simple_call): Likewise.
9121         (try_make_edge_direct_simple_call): Likewise.
9122         (update_indirect_edges_after_inlining): Likewise.
9123         (remove_described_reference): Look proper lto_stmt_uid.
9124         (propagate_controlled_uses): Likewise.
9125         (propagate_controlled_uses): Liekwise.
9126         * tree-inline.c (copy_bb): Copy speculative edges.
9127         (redirect_all_calls): New function.
9128         (copy_cfg_body): Do redirection after loop info is updated.
9129         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9130
9131 2013-08-09  Jan Hubicka  <jh@suse.cz>
9132
9133         * lto-streamer-out.c (output_function): Renumber PHIs.
9134         * lto-streamer-in.c (input_function): Likewise.
9135
9136 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9137
9138         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9139         (get_lane_unsigned): Likewise.
9140         (dup_lane_scalar): Likewise.
9141         (get_lane): enable for VALL.
9142         * config/aarch64/aarch64-simd.md
9143         (aarch64_dup_lane_scalar<mode>): Remove.
9144         (aarch64_get_lane_signed<mode>): Likewise.
9145         (aarch64_get_lane_unsigned<mode>): Likewise.
9146         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9147         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9148         (aarch64_get_lane<mode>): Enable for all vector modes.
9149         (aarch64_get_lanedi): Remove misleading constraints.
9150         * config/aarch64/arm_neon.h
9151         (__aarch64_vget_lane_any): Define.
9152         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9153         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9154         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9155         * config/aarch64/iterators.md (VDQQH): New.
9156         (VDQQHS): Likewise.
9157         (vwcore): Likewise.
9158
9159 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9160
9161         * configure.ac: Add GAS check for LEON instructions on SPARC.
9162         * configure: Regenerate.
9163         * config.in: Likewise.
9164         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9165         sparc*-*-* block.
9166         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9167         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9168         for LEON or LEON3.
9169         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9170         (AS_LEON_FLAG): New macro.
9171         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9172         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9173         Deal with LEON and LEON3 for the memory model.
9174         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9175         (atomic_compare_and_swap<mode>_1): Likewise.
9176         (*atomic_compare_and_swap<mode>_1): Likewise.
9177
9178 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9179
9180         * config/arm/neon.md (vcond): Fix floating-point vector
9181         comparisons against 0.
9182
9183 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9184
9185         * lra-constraints.c (emit_spill_move): Remove assert.
9186         (process_alt_operands): Add more debugging
9187         output.  Increase reject for spilling into memory.  Decrease
9188         reject for reloading scratch.
9189         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9190
9191 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9192
9193         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9194         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9195         micromips incompatible.  Add nan2008.
9196         (MULTILIB_DIRNAMES): Add nan2008.
9197         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9198         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9199         and micromips incompatible.  Add nan2008.
9200         (MULTILIB_DIRNAMES): Add nan2008.
9201         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9202
9203 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9204
9205         PR rtl-optimization/58079
9206         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9207         simplify_comparison has widened a comparison with an integer.
9208
9209 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9210
9211         * config/arm/neon.md (movmisalign<mode>): Disable when we
9212         don't allow unaligned accesses.
9213         (*movmisalign<mode>_neon_store): Likewise.
9214         (*movmisalign<mode>_neon_load): Likewise.
9215         (*movmisalign<mode>_neon_store): Likewise.
9216         (*movmisalign<mode>_neon_load): Likewise.
9217
9218 2013-08-08  Jan Hubicka  <jh@suse.cz>
9219
9220         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9221         (make_pass_rebuild_cgraph_edges): Also clear references.
9222         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9223         * ipa-inline-transform.c (inline_transform): Remove all references
9224         after inlining.
9225         * cgraphunit.c (expand_function): Remove all references after
9226         expansion.
9227         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9228         (ipa_find_reference): Rewrite to iterator.
9229         (remove_stmt_references): Likewise.
9230         (ipa_clear_stmts_in_references): New function.
9231         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9232         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9233         clear references.
9234         * ipa-split.c (split_function): Remove references in split function.
9235
9236 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9237
9238         PR target/57431
9239         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9240         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9241
9242 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9243
9244         PR target/56979
9245         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9246         suggested mode for the assignment isn't compatible with the
9247         registers required.
9248
9249 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9250
9251         PR target/58065
9252         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9253
9254 2013-08-07  Xinliang David Li  <davidxl@google.com>
9255
9256         * config/i386/i386.opt: New option -mtune-ctrl=.
9257         * config/i386/x86-tune.def: New file.
9258         * config/i386/i386.h: include x86-tune.def.
9259         * config/i386/i386.c (ix86_option_override_internal):
9260         Parsing -mtune-ctrl= option and set tune features.
9261
9262 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9263
9264         PR other/12081
9265         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9266         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9267         to gen_2arg_fn_t.
9268
9269 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9270
9271         * rtl.h (update_alignments): Declare.
9272         * final.c (grow_label_align): New function extracted from...
9273         (shorten_branches): ...here.  Call it.
9274         (update_alignments): New function.
9275         * reorg.c (sibling_labels): New variable.
9276         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9277         the new label along with it onto the sibling_labels vector.
9278         (fill_simple_delay_slots): Adjust call to get_label_before.
9279         (fill_slots_from_thread): Likewise.
9280         (relax_delay_slots): Likewise.
9281         (make_return_insns): Likewise.
9282         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9283
9284 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9285
9286         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9287         document its semantics.
9288         (diagnostic_report_diagnostic): Adjust accordingly.
9289
9290 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9291
9292         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9293         (sparc_option_override): ...and port to new C++ pass API.
9294         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9295
9296 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9297
9298         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9299
9300 2013-08-06  Caroline Tice  <cmtice@google.com>
9301
9302         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9303         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9304         * tree-pass.h: Add pass_vtable_verify.
9305         * varasm.c (assemble_variable): Add code to properly set the comdat
9306         section and name for the .vtable_map_vars section.
9307         (assemble_vtyv_preinit_initializer): New function.
9308         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9309         LINK_ONCE flag.
9310         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9311         * vtable-verify.c: New file.
9312         * vtable-verify.h: New file.
9313         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9314         initialiation levels.
9315         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9316         * passes.def: Insert pass_vtable_verify.
9317         * aclocal.m4: Reorder includes.
9318         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9319         -fvtv-counts options.
9320         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9321         as appropriate, if -fvtable-verify=... is used.
9322         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9323         -fvtable-verify=... is used.
9324         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9325         (vtable-verify.o): Add new build rule.
9326         (GTFILES): Add vtable-verify.c to list.
9327         * common.opt (fvtable-verify=): New flag.
9328         (vtv_priority): Values for fvtable-verify= flag.
9329         (fvtv-counts): New flag.
9330         (fvtv-debug): New flag.
9331         * tree.h (save_vtable_map_decl): New extern function decl.
9332
9333 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9334
9335         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9336         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9337         (pass_data_rl78_devirt): ...new pass_data instance and...
9338         (make_pass_rl78_devirt): ...new function.
9339         (rl78_asm_file_start): Port pass registration to new C++ API.
9340
9341 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9342
9343         * coretypes.h (rtl_opt_pass): Add.
9344         (gcc::context): Add.
9345         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9346         (epiphany_init): Port to new C++ pass API.
9347         (epiphany_optimize_mode_switching): Likewise.
9348         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9349         (pass_manager::get_pass_mode_switching): New.
9350         (pass_manager::get_pass_peephole2): New.
9351         * mode-switching.c (pass_mode_switching): Add clone method.
9352         * recog.c (pass_peephole2): Add clone method.
9353         (pass_split_all_insns): Add clone method.
9354
9355 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9356
9357         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9358         (mips_option_override): ...here, porting to new C++ API for passes.
9359
9360 2013-08-06  Jan Hubicka  <jh@suse.cz>
9361
9362         * cgraph.c (cgraph_get_body): New function based on lto.c
9363         implementation.
9364         * cgraph.h (cgraph_get_body): Declare.
9365         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9366         LTO paths.
9367         * cgraphunit.c (expand_function): Get body prior expanding.
9368         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9369         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9370         really need.
9371         * passes.c (do_per_function_toporder): Get body.
9372         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9373         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9374
9375 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9376
9377         PR fortran/57987
9378         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9379         is not re-finalized.  Rename second parameter to no_collect.
9380
9381 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9382
9383         PR middle-end/58041
9384         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9385         MEM_REF has proper alignment information.
9386
9387 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9388
9389         PR other/12081
9390         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9391         class insn_gen_fn.
9392         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9393         rtx (*) (rtx, ...) with insn_gen_fn.
9394         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9395         insn_gen_fn::stored_funcptr.  Add initializer braces.
9396
9397 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9398
9399         Rewrite how instances of passes are cloned to remove assumptions
9400         about their sizes (thus allowing pass subclasses to have
9401         additional data fields, albeit non-GC-managed ones at this point).
9402
9403         * passes.c (make_pass_instance): Now that passes have clone
9404         methods, rewrite this function to eliminate XNEW and memcpy
9405         calls that used hardcoded sizes.  Since this function no longer
9406         creates pass instances, rename it to...
9407         (add_pass_instance): ...this.  Document the old way that passes were
9408         numbered and flagged, and rework this function to continue using it.
9409         (next_pass_1): Add an initial_pass argument for use by
9410         add_pass_instance.
9411         (position_pass): When adding multiple instances of a pass, use
9412         the pass's clone method, rather than relying on the XNEW/memcpy
9413         within the former make_pass_instance (now add_pass_instance).
9414         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9415         the initial instance of the current pass within the pass manager.
9416
9417 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9418
9419         This is the automated part of the conversion of passes from C
9420         structs to C++ classes.
9421
9422         Patch autogenerated by refactor_passes.py from
9423         https://github.com/davidmalcolm/gcc-refactoring-scripts
9424         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9425
9426         * asan.c (pass_asan): Convert from a global struct to a subclass of
9427         gimple_opt_pass along with...
9428         (pass_data_asan): ...new pass_data instance and...
9429         (make_pass_asan): ...new function.
9430         (pass_asan_O0): Convert from a global struct to a subclass of
9431         gimple_opt_pass along with...
9432         (pass_data_asan_O0): ...new pass_data instance and...
9433         (make_pass_asan_O0): ...new function.
9434         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9435         subclass of rtl_opt_pass along with...
9436         (pass_data_inc_dec): ...new pass_data instance and...
9437         (make_pass_inc_dec): ...new function.
9438         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9439         a subclass of rtl_opt_pass along with...
9440         (pass_data_reorder_blocks): ...new pass_data instance and...
9441         (make_pass_reorder_blocks): ...new function.
9442         (pass_duplicate_computed_gotos): Convert from a global struct to a
9443         subclass of rtl_opt_pass along with...
9444         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9445         (make_pass_duplicate_computed_gotos): ...new function.
9446         (pass_partition_blocks): Convert from a global struct to a subclass of
9447         rtl_opt_pass along with...
9448         (pass_data_partition_blocks): ...new pass_data instance and...
9449         (make_pass_partition_blocks): ...new function.
9450         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9451         struct to a subclass of rtl_opt_pass along with...
9452         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9453         and...
9454         (make_pass_branch_target_load_optimize1): ...new function.
9455         (pass_branch_target_load_optimize2): Convert from a global struct to a
9456         subclass of rtl_opt_pass along with...
9457         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9458         and...
9459         (make_pass_branch_target_load_optimize2): ...new function.
9460         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9461         of rtl_opt_pass along with...
9462         (pass_data_jump): ...new pass_data instance and...
9463         (make_pass_jump): ...new function.
9464         (pass_jump2): Convert from a global struct to a subclass of
9465         rtl_opt_pass along with...
9466         (pass_data_jump2): ...new pass_data instance and...
9467         (make_pass_jump2): ...new function.
9468         * cfgexpand.c (pass_expand): Convert from a global struct to a
9469         subclass of rtl_opt_pass along with...
9470         (pass_data_expand): ...new pass_data instance and...
9471         (make_pass_expand): ...new function.
9472         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9473         of rtl_opt_pass along with...
9474         (pass_data_free_cfg): ...new pass_data instance and...
9475         (make_pass_free_cfg): ...new function.
9476         (pass_into_cfg_layout_mode): Convert from a global struct to a
9477         subclass of rtl_opt_pass along with...
9478         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9479         (make_pass_into_cfg_layout_mode): ...new function.
9480         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9481         subclass of rtl_opt_pass along with...
9482         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9483         (make_pass_outof_cfg_layout_mode): ...new function.
9484         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9485         struct to a subclass of gimple_opt_pass along with...
9486         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9487         (make_pass_build_cgraph_edges): ...new function.
9488         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9489         subclass of gimple_opt_pass along with...
9490         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9491         (make_pass_rebuild_cgraph_edges): ...new function.
9492         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9493         subclass of gimple_opt_pass along with...
9494         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9495         and...
9496         (make_pass_remove_cgraph_callee_edges): ...new function.
9497         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9498         struct to a subclass of rtl_opt_pass along with...
9499         (pass_data_stack_adjustments): ...new pass_data instance and...
9500         (make_pass_stack_adjustments): ...new function.
9501         * combine.c (pass_combine): Convert from a global struct to a subclass
9502         of rtl_opt_pass along with...
9503         (pass_data_combine): ...new pass_data instance and...
9504         (make_pass_combine): ...new function.
9505         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9506         global struct to a subclass of rtl_opt_pass along with...
9507         (pass_data_compare_elim_after_reload): ...new pass_data instance
9508         and...
9509         (make_pass_compare_elim_after_reload): ...new function.
9510         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9511         of rtl_opt_pass along with...
9512         (pass_data_rtl_cprop): ...new pass_data instance and...
9513         (make_pass_rtl_cprop): ...new function.
9514         * cse.c (pass_cse): Convert from a global struct to a subclass of
9515         rtl_opt_pass along with...
9516         (pass_data_cse): ...new pass_data instance and...
9517         (make_pass_cse): ...new function.
9518         (pass_cse2): Convert from a global struct to a subclass of
9519         rtl_opt_pass along with...
9520         (pass_data_cse2): ...new pass_data instance and...
9521         (make_pass_cse2): ...new function.
9522         (pass_cse_after_global_opts): Convert from a global struct to a
9523         subclass of rtl_opt_pass along with...
9524         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9525         (make_pass_cse_after_global_opts): ...new function.
9526         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9527         of rtl_opt_pass along with...
9528         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9529         (make_pass_ud_rtl_dce): ...new function.
9530         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9531         rtl_opt_pass along with...
9532         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9533         (make_pass_fast_rtl_dce): ...new function.
9534         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9535         a subclass of rtl_opt_pass along with...
9536         (pass_data_df_initialize_opt): ...new pass_data instance and...
9537         (make_pass_df_initialize_opt): ...new function.
9538         (pass_df_initialize_no_opt): Convert from a global struct to a
9539         subclass of rtl_opt_pass along with...
9540         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9541         (make_pass_df_initialize_no_opt): ...new function.
9542         (pass_df_finish): Convert from a global struct to a subclass of
9543         rtl_opt_pass along with...
9544         (pass_data_df_finish): ...new pass_data instance and...
9545         (make_pass_df_finish): ...new function.
9546         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9547         rtl_opt_pass along with...
9548         (pass_data_rtl_dse1): ...new pass_data instance and...
9549         (make_pass_rtl_dse1): ...new function.
9550         (pass_rtl_dse2): Convert from a global struct to a subclass of
9551         rtl_opt_pass along with...
9552         (pass_data_rtl_dse2): ...new pass_data instance and...
9553         (make_pass_rtl_dse2): ...new function.
9554         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9555         subclass of rtl_opt_pass along with...
9556         (pass_data_dwarf2_frame): ...new pass_data instance and...
9557         (make_pass_dwarf2_frame): ...new function.
9558         * except.c (pass_set_nothrow_function_flags): Convert from a global
9559         struct to a subclass of rtl_opt_pass along with...
9560         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9561         and...
9562         (make_pass_set_nothrow_function_flags): ...new function.
9563         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9564         subclass of rtl_opt_pass along with...
9565         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9566         and...
9567         (make_pass_convert_to_eh_region_ranges): ...new function.
9568         * final.c (pass_compute_alignments): Convert from a global struct to a
9569         subclass of rtl_opt_pass along with...
9570         (pass_data_compute_alignments): ...new pass_data instance and...
9571         (make_pass_compute_alignments): ...new function.
9572         (pass_final): Convert from a global struct to a subclass of
9573         rtl_opt_pass along with...
9574         (pass_data_final): ...new pass_data instance and...
9575         (make_pass_final): ...new function.
9576         (pass_shorten_branches): Convert from a global struct to a subclass of
9577         rtl_opt_pass along with...
9578         (pass_data_shorten_branches): ...new pass_data instance and...
9579         (make_pass_shorten_branches): ...new function.
9580         (pass_clean_state): Convert from a global struct to a subclass of
9581         rtl_opt_pass along with...
9582         (pass_data_clean_state): ...new pass_data instance and...
9583         (make_pass_clean_state): ...new function.
9584         * function.c (pass_instantiate_virtual_regs): Convert from a global
9585         struct to a subclass of rtl_opt_pass along with...
9586         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9587         (make_pass_instantiate_virtual_regs): ...new function.
9588         (pass_leaf_regs): Convert from a global struct to a subclass of
9589         rtl_opt_pass along with...
9590         (pass_data_leaf_regs): ...new pass_data instance and...
9591         (make_pass_leaf_regs): ...new function.
9592         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9593         subclass of rtl_opt_pass along with...
9594         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9595         and...
9596         (make_pass_thread_prologue_and_epilogue): ...new function.
9597         (pass_match_asm_constraints): Convert from a global struct to a
9598         subclass of rtl_opt_pass along with...
9599         (pass_data_match_asm_constraints): ...new pass_data instance and...
9600         (make_pass_match_asm_constraints): ...new function.
9601         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9602         subclass of rtl_opt_pass along with...
9603         (pass_data_rtl_fwprop): ...new pass_data instance and...
9604         (make_pass_rtl_fwprop): ...new function.
9605         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9606         rtl_opt_pass along with...
9607         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9608         (make_pass_rtl_fwprop_addr): ...new function.
9609         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9610         rtl_opt_pass along with...
9611         (pass_data_rtl_pre): ...new pass_data instance and...
9612         (make_pass_rtl_pre): ...new function.
9613         (pass_rtl_hoist): Convert from a global struct to a subclass of
9614         rtl_opt_pass along with...
9615         (pass_data_rtl_hoist): ...new pass_data instance and...
9616         (make_pass_rtl_hoist): ...new function.
9617         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9618         subclass of gimple_opt_pass along with...
9619         (pass_data_lower_cf): ...new pass_data instance and...
9620         (make_pass_lower_cf): ...new function.
9621         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9622         from a global struct to a subclass of gimple_opt_pass along with...
9623         (pass_data_strength_reduction): ...new pass_data instance and...
9624         (make_pass_strength_reduction): ...new function.
9625         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9626         of rtl_opt_pass along with...
9627         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9628         (make_pass_rtl_ifcvt): ...new function.
9629         (pass_if_after_combine): Convert from a global struct to a subclass of
9630         rtl_opt_pass along with...
9631         (pass_data_if_after_combine): ...new pass_data instance and...
9632         (make_pass_if_after_combine): ...new function.
9633         (pass_if_after_reload): Convert from a global struct to a subclass of
9634         rtl_opt_pass along with...
9635         (pass_data_if_after_reload): ...new pass_data instance and...
9636         (make_pass_if_after_reload): ...new function.
9637         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9638         a subclass of rtl_opt_pass along with...
9639         (pass_data_initialize_regs): ...new pass_data instance and...
9640         (make_pass_initialize_regs): ...new function.
9641         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9642         of ipa_opt_pass_d along with...
9643         (pass_data_ipa_cp): ...new pass_data instance and...
9644         (make_pass_ipa_cp): ...new function.
9645         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9646         global struct to a subclass of gimple_opt_pass along with...
9647         (pass_data_inline_parameters): ...new pass_data instance and...
9648         (make_pass_inline_parameters): ...new function.
9649         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9650         subclass of gimple_opt_pass along with...
9651         (pass_data_early_inline): ...new pass_data instance and...
9652         (make_pass_early_inline): ...new function.
9653         (pass_ipa_inline): Convert from a global struct to a subclass of
9654         ipa_opt_pass_d along with...
9655         (pass_data_ipa_inline): ...new pass_data instance and...
9656         (make_pass_ipa_inline): ...new function.
9657         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9658         struct to a subclass of gimple_opt_pass along with...
9659         (pass_data_local_pure_const): ...new pass_data instance and...
9660         (make_pass_local_pure_const): ...new function.
9661         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9662         ipa_opt_pass_d along with...
9663         (pass_data_ipa_pure_const): ...new pass_data instance and...
9664         (make_pass_ipa_pure_const): ...new function.
9665         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9666         to a subclass of ipa_opt_pass_d along with...
9667         (pass_data_ipa_reference): ...new pass_data instance and...
9668         (make_pass_ipa_reference): ...new function.
9669         * ipa-split.c (pass_split_functions): Convert from a global struct to
9670         a subclass of gimple_opt_pass along with...
9671         (pass_data_split_functions): ...new pass_data instance and...
9672         (make_pass_split_functions): ...new function.
9673         (pass_feedback_split_functions): Convert from a global struct to a
9674         subclass of gimple_opt_pass along with...
9675         (pass_data_feedback_split_functions): ...new pass_data instance and...
9676         (make_pass_feedback_split_functions): ...new function.
9677         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9678         global struct to a subclass of simple_ipa_opt_pass along with...
9679         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9680         instance and...
9681         (make_pass_ipa_function_and_variable_visibility): ...new function.
9682         (pass_ipa_free_inline_summary): Convert from a global struct to a
9683         subclass of simple_ipa_opt_pass along with...
9684         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9685         (make_pass_ipa_free_inline_summary): ...new function.
9686         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9687         subclass of ipa_opt_pass_d along with...
9688         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9689         and...
9690         (make_pass_ipa_whole_program_visibility): ...new function.
9691         (pass_ipa_profile): Convert from a global struct to a subclass of
9692         ipa_opt_pass_d along with...
9693         (pass_data_ipa_profile): ...new pass_data instance and...
9694         (make_pass_ipa_profile): ...new function.
9695         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9696         ipa_opt_pass_d along with...
9697         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9698         (make_pass_ipa_cdtor_merge): ...new function.
9699         * ira.c (pass_ira): Convert from a global struct to a subclass of
9700         rtl_opt_pass along with...
9701         (pass_data_ira): ...new pass_data instance and...
9702         (make_pass_ira): ...new function.
9703         (pass_reload): Convert from a global struct to a subclass of
9704         rtl_opt_pass along with...
9705         (pass_data_reload): ...new pass_data instance and...
9706         (make_pass_reload): ...new function.
9707         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9708         subclass of rtl_opt_pass along with...
9709         (pass_data_cleanup_barriers): ...new pass_data instance and...
9710         (make_pass_cleanup_barriers): ...new function.
9711         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9712         of rtl_opt_pass along with...
9713         (pass_data_loop2): ...new pass_data instance and...
9714         (make_pass_loop2): ...new function.
9715         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9716         rtl_opt_pass along with...
9717         (pass_data_rtl_loop_init): ...new pass_data instance and...
9718         (make_pass_rtl_loop_init): ...new function.
9719         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9720         rtl_opt_pass along with...
9721         (pass_data_rtl_loop_done): ...new pass_data instance and...
9722         (make_pass_rtl_loop_done): ...new function.
9723         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9724         subclass of rtl_opt_pass along with...
9725         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9726         (make_pass_rtl_move_loop_invariants): ...new function.
9727         (pass_rtl_unswitch): Convert from a global struct to a subclass of
9728         rtl_opt_pass along with...
9729         (pass_data_rtl_unswitch): ...new pass_data instance and...
9730         (make_pass_rtl_unswitch): ...new function.
9731         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
9732         subclass of rtl_opt_pass along with...
9733         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
9734         and...
9735         (make_pass_rtl_unroll_and_peel_loops): ...new function.
9736         (pass_rtl_doloop): Convert from a global struct to a subclass of
9737         rtl_opt_pass along with...
9738         (pass_data_rtl_doloop): ...new pass_data instance and...
9739         (make_pass_rtl_doloop): ...new function.
9740         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
9741         a subclass of rtl_opt_pass along with...
9742         (pass_data_lower_subreg): ...new pass_data instance and...
9743         (make_pass_lower_subreg): ...new function.
9744         (pass_lower_subreg2): Convert from a global struct to a subclass of
9745         rtl_opt_pass along with...
9746         (pass_data_lower_subreg2): ...new pass_data instance and...
9747         (make_pass_lower_subreg2): ...new function.
9748         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
9749         struct to a subclass of ipa_opt_pass_d along with...
9750         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
9751         (make_pass_ipa_lto_gimple_out): ...new function.
9752         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
9753         of ipa_opt_pass_d along with...
9754         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
9755         (make_pass_ipa_lto_finish_out): ...new function.
9756         * mode-switching.c (pass_mode_switching): Convert from a global struct
9757         to a subclass of rtl_opt_pass along with...
9758         (pass_data_mode_switching): ...new pass_data instance and...
9759         (make_pass_mode_switching): ...new function.
9760         * modulo-sched.c (pass_sms): Convert from a global struct to a
9761         subclass of rtl_opt_pass along with...
9762         (pass_data_sms): ...new pass_data instance and...
9763         (make_pass_sms): ...new function.
9764         * omp-low.c (pass_expand_omp): Convert from a global struct to a
9765         subclass of gimple_opt_pass along with...
9766         (pass_data_expand_omp): ...new pass_data instance and...
9767         (make_pass_expand_omp): ...new function.
9768         (pass_lower_omp): Convert from a global struct to a subclass of
9769         gimple_opt_pass along with...
9770         (pass_data_lower_omp): ...new pass_data instance and...
9771         (make_pass_lower_omp): ...new function.
9772         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
9773         of gimple_opt_pass along with...
9774         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
9775         (make_pass_diagnose_omp_blocks): ...new function.
9776         * passes.c (pass_early_local_passes): Convert from a global struct to
9777         a subclass of simple_ipa_opt_pass along with...
9778         (pass_data_early_local_passes): ...new pass_data instance and...
9779         (make_pass_early_local_passes): ...new function.
9780         (pass_all_early_optimizations): Convert from a global struct to a
9781         subclass of gimple_opt_pass along with...
9782         (pass_data_all_early_optimizations): ...new pass_data instance and...
9783         (make_pass_all_early_optimizations): ...new function.
9784         (pass_all_optimizations): Convert from a global struct to a subclass
9785         of gimple_opt_pass along with...
9786         (pass_data_all_optimizations): ...new pass_data instance and...
9787         (make_pass_all_optimizations): ...new function.
9788         (pass_all_optimizations_g): Convert from a global struct to a subclass
9789         of gimple_opt_pass along with...
9790         (pass_data_all_optimizations_g): ...new pass_data instance and...
9791         (make_pass_all_optimizations_g): ...new function.
9792         (pass_rest_of_compilation): Convert from a global struct to a subclass
9793         of rtl_opt_pass along with...
9794         (pass_data_rest_of_compilation): ...new pass_data instance and...
9795         (make_pass_rest_of_compilation): ...new function.
9796         (pass_postreload): Convert from a global struct to a subclass of
9797         rtl_opt_pass along with...
9798         (pass_data_postreload): ...new pass_data instance and...
9799         (make_pass_postreload): ...new function.
9800         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
9801         subclass of rtl_opt_pass along with...
9802         (pass_data_gcse2): ...new pass_data instance and...
9803         (make_pass_gcse2): ...new function.
9804         * postreload.c (pass_postreload_cse): Convert from a global struct to
9805         a subclass of rtl_opt_pass along with...
9806         (pass_data_postreload_cse): ...new pass_data instance and...
9807         (make_pass_postreload_cse): ...new function.
9808         * predict.c (pass_profile): Convert from a global struct to a subclass
9809         of gimple_opt_pass along with...
9810         (pass_data_profile): ...new pass_data instance and...
9811         (make_pass_profile): ...new function.
9812         (pass_strip_predict_hints): Convert from a global struct to a subclass
9813         of gimple_opt_pass along with...
9814         (pass_data_strip_predict_hints): ...new pass_data instance and...
9815         (make_pass_strip_predict_hints): ...new function.
9816         * recog.c (pass_peephole2): Convert from a global struct to a subclass
9817         of rtl_opt_pass along with...
9818         (pass_data_peephole2): ...new pass_data instance and...
9819         (make_pass_peephole2): ...new function.
9820         (pass_split_all_insns): Convert from a global struct to a subclass of
9821         rtl_opt_pass along with...
9822         (pass_data_split_all_insns): ...new pass_data instance and...
9823         (make_pass_split_all_insns): ...new function.
9824         (pass_split_after_reload): Convert from a global struct to a subclass
9825         of rtl_opt_pass along with...
9826         (pass_data_split_after_reload): ...new pass_data instance and...
9827         (make_pass_split_after_reload): ...new function.
9828         (pass_split_before_regstack): Convert from a global struct to a
9829         subclass of rtl_opt_pass along with...
9830         (pass_data_split_before_regstack): ...new pass_data instance and...
9831         (make_pass_split_before_regstack): ...new function.
9832         (pass_split_before_sched2): Convert from a global struct to a subclass
9833         of rtl_opt_pass along with...
9834         (pass_data_split_before_sched2): ...new pass_data instance and...
9835         (make_pass_split_before_sched2): ...new function.
9836         (pass_split_for_shorten_branches): Convert from a global struct to a
9837         subclass of rtl_opt_pass along with...
9838         (pass_data_split_for_shorten_branches): ...new pass_data instance
9839         and...
9840         (make_pass_split_for_shorten_branches): ...new function.
9841         * ree.c (pass_ree): Convert from a global struct to a subclass of
9842         rtl_opt_pass along with...
9843         (pass_data_ree): ...new pass_data instance and...
9844         (make_pass_ree): ...new function.
9845         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
9846         subclass of rtl_opt_pass along with...
9847         (pass_data_stack_regs): ...new pass_data instance and...
9848         (make_pass_stack_regs): ...new function.
9849         (pass_stack_regs_run): Convert from a global struct to a subclass of
9850         rtl_opt_pass along with...
9851         (pass_data_stack_regs_run): ...new pass_data instance and...
9852         (make_pass_stack_regs_run): ...new function.
9853         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
9854         subclass of rtl_opt_pass along with...
9855         (pass_data_cprop_hardreg): ...new pass_data instance and...
9856         (make_pass_cprop_hardreg): ...new function.
9857         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
9858         subclass of rtl_opt_pass along with...
9859         (pass_data_reginfo_init): ...new pass_data instance and...
9860         (make_pass_reginfo_init): ...new function.
9861         * regmove.c (pass_regmove): Convert from a global struct to a subclass
9862         of rtl_opt_pass along with...
9863         (pass_data_regmove): ...new pass_data instance and...
9864         (make_pass_regmove): ...new function.
9865         * regrename.c (pass_regrename): Convert from a global struct to a
9866         subclass of rtl_opt_pass along with...
9867         (pass_data_regrename): ...new pass_data instance and...
9868         (make_pass_regrename): ...new function.
9869         * reorg.c (pass_delay_slots): Convert from a global struct to a
9870         subclass of rtl_opt_pass along with...
9871         (pass_data_delay_slots): ...new pass_data instance and...
9872         (make_pass_delay_slots): ...new function.
9873         (pass_machine_reorg): Convert from a global struct to a subclass of
9874         rtl_opt_pass along with...
9875         (pass_data_machine_reorg): ...new pass_data instance and...
9876         (make_pass_machine_reorg): ...new function.
9877         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
9878         of rtl_opt_pass along with...
9879         (pass_data_sched): ...new pass_data instance and...
9880         (make_pass_sched): ...new function.
9881         (pass_sched2): Convert from a global struct to a subclass of
9882         rtl_opt_pass along with...
9883         (pass_data_sched2): ...new pass_data instance and...
9884         (make_pass_sched2): ...new function.
9885         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
9886         to a subclass of rtl_opt_pass along with...
9887         (pass_data_stack_ptr_mod): ...new pass_data instance and...
9888         (make_pass_stack_ptr_mod): ...new function.
9889         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
9890         to a subclass of rtl_opt_pass along with...
9891         (pass_data_rtl_store_motion): ...new pass_data instance and...
9892         (make_pass_rtl_store_motion): ...new function.
9893         * tracer.c (pass_tracer): Convert from a global struct to a subclass
9894         of gimple_opt_pass along with...
9895         (pass_data_tracer): ...new pass_data instance and...
9896         (make_pass_tracer): ...new function.
9897         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
9898         to a subclass of gimple_opt_pass along with...
9899         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
9900         (make_pass_diagnose_tm_blocks): ...new function.
9901         (pass_lower_tm): Convert from a global struct to a subclass of
9902         gimple_opt_pass along with...
9903         (pass_data_lower_tm): ...new pass_data instance and...
9904         (make_pass_lower_tm): ...new function.
9905         (pass_tm_init): Convert from a global struct to a subclass of
9906         gimple_opt_pass along with...
9907         (pass_data_tm_init): ...new pass_data instance and...
9908         (make_pass_tm_init): ...new function.
9909         (pass_tm_mark): Convert from a global struct to a subclass of
9910         gimple_opt_pass along with...
9911         (pass_data_tm_mark): ...new pass_data instance and...
9912         (make_pass_tm_mark): ...new function.
9913         (pass_tm_edges): Convert from a global struct to a subclass of
9914         gimple_opt_pass along with...
9915         (pass_data_tm_edges): ...new pass_data instance and...
9916         (make_pass_tm_edges): ...new function.
9917         (pass_tm_memopt): Convert from a global struct to a subclass of
9918         gimple_opt_pass along with...
9919         (pass_data_tm_memopt): ...new pass_data instance and...
9920         (make_pass_tm_memopt): ...new function.
9921         (pass_ipa_tm): Convert from a global struct to a subclass of
9922         simple_ipa_opt_pass along with...
9923         (pass_data_ipa_tm): ...new pass_data instance and...
9924         (make_pass_ipa_tm): ...new function.
9925         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
9926         subclass of gimple_opt_pass along with...
9927         (pass_data_call_cdce): ...new pass_data instance and...
9928         (make_pass_call_cdce): ...new function.
9929         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
9930         subclass of gimple_opt_pass along with...
9931         (pass_data_build_cfg): ...new pass_data instance and...
9932         (make_pass_build_cfg): ...new function.
9933         (pass_split_crit_edges): Convert from a global struct to a subclass of
9934         gimple_opt_pass along with...
9935         (pass_data_split_crit_edges): ...new pass_data instance and...
9936         (make_pass_split_crit_edges): ...new function.
9937         (pass_warn_function_return): Convert from a global struct to a
9938         subclass of gimple_opt_pass along with...
9939         (pass_data_warn_function_return): ...new pass_data instance and...
9940         (make_pass_warn_function_return): ...new function.
9941         (pass_warn_function_noreturn): Convert from a global struct to a
9942         subclass of gimple_opt_pass along with...
9943         (pass_data_warn_function_noreturn): ...new pass_data instance and...
9944         (make_pass_warn_function_noreturn): ...new function.
9945         (pass_warn_unused_result): Convert from a global struct to a subclass
9946         of gimple_opt_pass along with...
9947         (pass_data_warn_unused_result): ...new pass_data instance and...
9948         (make_pass_warn_unused_result): ...new function.
9949         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
9950         a subclass of gimple_opt_pass along with...
9951         (pass_data_merge_phi): ...new pass_data instance and...
9952         (make_pass_merge_phi): ...new function.
9953         * tree-complex.c (pass_lower_complex): Convert from a global struct to
9954         a subclass of gimple_opt_pass along with...
9955         (pass_data_lower_complex): ...new pass_data instance and...
9956         (make_pass_lower_complex): ...new function.
9957         (pass_lower_complex_O0): Convert from a global struct to a subclass of
9958         gimple_opt_pass along with...
9959         (pass_data_lower_complex_O0): ...new pass_data instance and...
9960         (make_pass_lower_complex_O0): ...new function.
9961         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
9962         subclass of gimple_opt_pass along with...
9963         (pass_data_lower_eh): ...new pass_data instance and...
9964         (make_pass_lower_eh): ...new function.
9965         (pass_refactor_eh): Convert from a global struct to a subclass of
9966         gimple_opt_pass along with...
9967         (pass_data_refactor_eh): ...new pass_data instance and...
9968         (make_pass_refactor_eh): ...new function.
9969         (pass_lower_resx): Convert from a global struct to a subclass of
9970         gimple_opt_pass along with...
9971         (pass_data_lower_resx): ...new pass_data instance and...
9972         (make_pass_lower_resx): ...new function.
9973         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
9974         of gimple_opt_pass along with...
9975         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
9976         (make_pass_lower_eh_dispatch): ...new function.
9977         (pass_cleanup_eh): Convert from a global struct to a subclass of
9978         gimple_opt_pass along with...
9979         (pass_data_cleanup_eh): ...new pass_data instance and...
9980         (make_pass_cleanup_eh): ...new function.
9981         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
9982         to a subclass of simple_ipa_opt_pass along with...
9983         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
9984         (make_pass_ipa_lower_emutls): ...new function.
9985         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
9986         a subclass of gimple_opt_pass along with...
9987         (pass_data_if_conversion): ...new pass_data instance and...
9988         (make_pass_if_conversion): ...new function.
9989         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
9990         subclass of gimple_opt_pass along with...
9991         (pass_data_build_ssa): ...new pass_data instance and...
9992         (make_pass_build_ssa): ...new function.
9993         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
9994         global struct to a subclass of gimple_opt_pass along with...
9995         (pass_data_loop_distribution): ...new pass_data instance and...
9996         (make_pass_loop_distribution): ...new function.
9997         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
9998         subclass of gimple_opt_pass along with...
9999         (pass_data_mudflap_1): ...new pass_data instance and...
10000         (make_pass_mudflap_1): ...new function.
10001         (pass_mudflap_2): Convert from a global struct to a subclass of
10002         gimple_opt_pass along with...
10003         (pass_data_mudflap_2): ...new pass_data instance and...
10004         (make_pass_mudflap_2): ...new function.
10005         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10006         subclass of gimple_opt_pass along with...
10007         (pass_data_mudflap_1): ...new pass_data instance and...
10008         (make_pass_mudflap_1): ...new function.
10009         (pass_mudflap_2): Convert from a global struct to a subclass of
10010         gimple_opt_pass along with...
10011         (pass_data_mudflap_2): ...new pass_data instance and...
10012         (make_pass_mudflap_2): ...new function.
10013         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10014         gimple_opt_pass along with...
10015         (pass_data_nrv): ...new pass_data instance and...
10016         (make_pass_nrv): ...new function.
10017         (pass_return_slot): Convert from a global struct to a subclass of
10018         gimple_opt_pass along with...
10019         (pass_data_return_slot): ...new pass_data instance and...
10020         (make_pass_return_slot): ...new function.
10021         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10022         to a subclass of gimple_opt_pass along with...
10023         (pass_data_object_sizes): ...new pass_data instance and...
10024         (make_pass_object_sizes): ...new function.
10025         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10026         global struct to a subclass of gimple_opt_pass along with...
10027         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10028         and...
10029         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10030         (pass_fixup_cfg): Convert from a global struct to a subclass of
10031         gimple_opt_pass along with...
10032         (pass_data_fixup_cfg): ...new pass_data instance and...
10033         (make_pass_fixup_cfg): ...new function.
10034         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10035         (make_pass_mudflap_1): ...new function.
10036         (pass_mudflap_2): Replace declaration with that of...
10037         (make_pass_mudflap_2): ...new function.
10038         (pass_asan): Replace declaration with that of...
10039         (make_pass_asan): ...new function.
10040         (pass_asan_O0): Replace declaration with that of...
10041         (make_pass_asan_O0): ...new function.
10042         (pass_tsan): Replace declaration with that of...
10043         (make_pass_tsan): ...new function.
10044         (pass_tsan_O0): Replace declaration with that of...
10045         (make_pass_tsan_O0): ...new function.
10046         (pass_lower_cf): Replace declaration with that of...
10047         (make_pass_lower_cf): ...new function.
10048         (pass_refactor_eh): Replace declaration with that of...
10049         (make_pass_refactor_eh): ...new function.
10050         (pass_lower_eh): Replace declaration with that of...
10051         (make_pass_lower_eh): ...new function.
10052         (pass_lower_eh_dispatch): Replace declaration with that of...
10053         (make_pass_lower_eh_dispatch): ...new function.
10054         (pass_lower_resx): Replace declaration with that of...
10055         (make_pass_lower_resx): ...new function.
10056         (pass_build_cfg): Replace declaration with that of...
10057         (make_pass_build_cfg): ...new function.
10058         (pass_early_tree_profile): Replace declaration with that of...
10059         (make_pass_early_tree_profile): ...new function.
10060         (pass_cleanup_eh): Replace declaration with that of...
10061         (make_pass_cleanup_eh): ...new function.
10062         (pass_sra): Replace declaration with that of...
10063         (make_pass_sra): ...new function.
10064         (pass_sra_early): Replace declaration with that of...
10065         (make_pass_sra_early): ...new function.
10066         (pass_early_ipa_sra): Replace declaration with that of...
10067         (make_pass_early_ipa_sra): ...new function.
10068         (pass_tail_recursion): Replace declaration with that of...
10069         (make_pass_tail_recursion): ...new function.
10070         (pass_tail_calls): Replace declaration with that of...
10071         (make_pass_tail_calls): ...new function.
10072         (pass_tree_loop): Replace declaration with that of...
10073         (make_pass_tree_loop): ...new function.
10074         (pass_tree_loop_init): Replace declaration with that of...
10075         (make_pass_tree_loop_init): ...new function.
10076         (pass_lim): Replace declaration with that of...
10077         (make_pass_lim): ...new function.
10078         (pass_tree_unswitch): Replace declaration with that of...
10079         (make_pass_tree_unswitch): ...new function.
10080         (pass_predcom): Replace declaration with that of...
10081         (make_pass_predcom): ...new function.
10082         (pass_iv_canon): Replace declaration with that of...
10083         (make_pass_iv_canon): ...new function.
10084         (pass_scev_cprop): Replace declaration with that of...
10085         (make_pass_scev_cprop): ...new function.
10086         (pass_empty_loop): Replace declaration with that of...
10087         (make_pass_empty_loop): ...new function.
10088         (pass_record_bounds): Replace declaration with that of...
10089         (make_pass_record_bounds): ...new function.
10090         (pass_graphite): Replace declaration with that of...
10091         (make_pass_graphite): ...new function.
10092         (pass_graphite_transforms): Replace declaration with that of...
10093         (make_pass_graphite_transforms): ...new function.
10094         (pass_if_conversion): Replace declaration with that of...
10095         (make_pass_if_conversion): ...new function.
10096         (pass_loop_distribution): Replace declaration with that of...
10097         (make_pass_loop_distribution): ...new function.
10098         (pass_vectorize): Replace declaration with that of...
10099         (make_pass_vectorize): ...new function.
10100         (pass_slp_vectorize): Replace declaration with that of...
10101         (make_pass_slp_vectorize): ...new function.
10102         (pass_complete_unroll): Replace declaration with that of...
10103         (make_pass_complete_unroll): ...new function.
10104         (pass_complete_unrolli): Replace declaration with that of...
10105         (make_pass_complete_unrolli): ...new function.
10106         (pass_parallelize_loops): Replace declaration with that of...
10107         (make_pass_parallelize_loops): ...new function.
10108         (pass_loop_prefetch): Replace declaration with that of...
10109         (make_pass_loop_prefetch): ...new function.
10110         (pass_iv_optimize): Replace declaration with that of...
10111         (make_pass_iv_optimize): ...new function.
10112         (pass_tree_loop_done): Replace declaration with that of...
10113         (make_pass_tree_loop_done): ...new function.
10114         (pass_ch): Replace declaration with that of...
10115         (make_pass_ch): ...new function.
10116         (pass_ccp): Replace declaration with that of...
10117         (make_pass_ccp): ...new function.
10118         (pass_phi_only_cprop): Replace declaration with that of...
10119         (make_pass_phi_only_cprop): ...new function.
10120         (pass_build_ssa): Replace declaration with that of...
10121         (make_pass_build_ssa): ...new function.
10122         (pass_build_alias): Replace declaration with that of...
10123         (make_pass_build_alias): ...new function.
10124         (pass_build_ealias): Replace declaration with that of...
10125         (make_pass_build_ealias): ...new function.
10126         (pass_dominator): Replace declaration with that of...
10127         (make_pass_dominator): ...new function.
10128         (pass_dce): Replace declaration with that of...
10129         (make_pass_dce): ...new function.
10130         (pass_dce_loop): Replace declaration with that of...
10131         (make_pass_dce_loop): ...new function.
10132         (pass_cd_dce): Replace declaration with that of...
10133         (make_pass_cd_dce): ...new function.
10134         (pass_call_cdce): Replace declaration with that of...
10135         (make_pass_call_cdce): ...new function.
10136         (pass_merge_phi): Replace declaration with that of...
10137         (make_pass_merge_phi): ...new function.
10138         (pass_split_crit_edges): Replace declaration with that of...
10139         (make_pass_split_crit_edges): ...new function.
10140         (pass_pre): Replace declaration with that of...
10141         (make_pass_pre): ...new function.
10142         (pass_profile): Replace declaration with that of...
10143         (make_pass_profile): ...new function.
10144         (pass_strip_predict_hints): Replace declaration with that of...
10145         (make_pass_strip_predict_hints): ...new function.
10146         (pass_lower_complex_O0): Replace declaration with that of...
10147         (make_pass_lower_complex_O0): ...new function.
10148         (pass_lower_complex): Replace declaration with that of...
10149         (make_pass_lower_complex): ...new function.
10150         (pass_lower_vector): Replace declaration with that of...
10151         (make_pass_lower_vector): ...new function.
10152         (pass_lower_vector_ssa): Replace declaration with that of...
10153         (make_pass_lower_vector_ssa): ...new function.
10154         (pass_lower_omp): Replace declaration with that of...
10155         (make_pass_lower_omp): ...new function.
10156         (pass_diagnose_omp_blocks): Replace declaration with that of...
10157         (make_pass_diagnose_omp_blocks): ...new function.
10158         (pass_expand_omp): Replace declaration with that of...
10159         (make_pass_expand_omp): ...new function.
10160         (pass_expand_omp_ssa): Replace declaration with that of...
10161         (make_pass_expand_omp_ssa): ...new function.
10162         (pass_object_sizes): Replace declaration with that of...
10163         (make_pass_object_sizes): ...new function.
10164         (pass_strlen): Replace declaration with that of...
10165         (make_pass_strlen): ...new function.
10166         (pass_fold_builtins): Replace declaration with that of...
10167         (make_pass_fold_builtins): ...new function.
10168         (pass_stdarg): Replace declaration with that of...
10169         (make_pass_stdarg): ...new function.
10170         (pass_early_warn_uninitialized): Replace declaration with that of...
10171         (make_pass_early_warn_uninitialized): ...new function.
10172         (pass_late_warn_uninitialized): Replace declaration with that of...
10173         (make_pass_late_warn_uninitialized): ...new function.
10174         (pass_cse_reciprocals): Replace declaration with that of...
10175         (make_pass_cse_reciprocals): ...new function.
10176         (pass_cse_sincos): Replace declaration with that of...
10177         (make_pass_cse_sincos): ...new function.
10178         (pass_optimize_bswap): Replace declaration with that of...
10179         (make_pass_optimize_bswap): ...new function.
10180         (pass_optimize_widening_mul): Replace declaration with that of...
10181         (make_pass_optimize_widening_mul): ...new function.
10182         (pass_warn_function_return): Replace declaration with that of...
10183         (make_pass_warn_function_return): ...new function.
10184         (pass_warn_function_noreturn): Replace declaration with that of...
10185         (make_pass_warn_function_noreturn): ...new function.
10186         (pass_cselim): Replace declaration with that of...
10187         (make_pass_cselim): ...new function.
10188         (pass_phiopt): Replace declaration with that of...
10189         (make_pass_phiopt): ...new function.
10190         (pass_forwprop): Replace declaration with that of...
10191         (make_pass_forwprop): ...new function.
10192         (pass_phiprop): Replace declaration with that of...
10193         (make_pass_phiprop): ...new function.
10194         (pass_tree_ifcombine): Replace declaration with that of...
10195         (make_pass_tree_ifcombine): ...new function.
10196         (pass_dse): Replace declaration with that of...
10197         (make_pass_dse): ...new function.
10198         (pass_nrv): Replace declaration with that of...
10199         (make_pass_nrv): ...new function.
10200         (pass_rename_ssa_copies): Replace declaration with that of...
10201         (make_pass_rename_ssa_copies): ...new function.
10202         (pass_sink_code): Replace declaration with that of...
10203         (make_pass_sink_code): ...new function.
10204         (pass_fre): Replace declaration with that of...
10205         (make_pass_fre): ...new function.
10206         (pass_check_data_deps): Replace declaration with that of...
10207         (make_pass_check_data_deps): ...new function.
10208         (pass_copy_prop): Replace declaration with that of...
10209         (make_pass_copy_prop): ...new function.
10210         (pass_vrp): Replace declaration with that of...
10211         (make_pass_vrp): ...new function.
10212         (pass_uncprop): Replace declaration with that of...
10213         (make_pass_uncprop): ...new function.
10214         (pass_return_slot): Replace declaration with that of...
10215         (make_pass_return_slot): ...new function.
10216         (pass_reassoc): Replace declaration with that of...
10217         (make_pass_reassoc): ...new function.
10218         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10219         (make_pass_rebuild_cgraph_edges): ...new function.
10220         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10221         (make_pass_remove_cgraph_callee_edges): ...new function.
10222         (pass_build_cgraph_edges): Replace declaration with that of...
10223         (make_pass_build_cgraph_edges): ...new function.
10224         (pass_local_pure_const): Replace declaration with that of...
10225         (make_pass_local_pure_const): ...new function.
10226         (pass_tracer): Replace declaration with that of...
10227         (make_pass_tracer): ...new function.
10228         (pass_warn_unused_result): Replace declaration with that of...
10229         (make_pass_warn_unused_result): ...new function.
10230         (pass_diagnose_tm_blocks): Replace declaration with that of...
10231         (make_pass_diagnose_tm_blocks): ...new function.
10232         (pass_lower_tm): Replace declaration with that of...
10233         (make_pass_lower_tm): ...new function.
10234         (pass_tm_init): Replace declaration with that of...
10235         (make_pass_tm_init): ...new function.
10236         (pass_tm_mark): Replace declaration with that of...
10237         (make_pass_tm_mark): ...new function.
10238         (pass_tm_memopt): Replace declaration with that of...
10239         (make_pass_tm_memopt): ...new function.
10240         (pass_tm_edges): Replace declaration with that of...
10241         (make_pass_tm_edges): ...new function.
10242         (pass_split_functions): Replace declaration with that of...
10243         (make_pass_split_functions): ...new function.
10244         (pass_feedback_split_functions): Replace declaration with that of...
10245         (make_pass_feedback_split_functions): ...new function.
10246         (pass_strength_reduction): Replace declaration with that of...
10247         (make_pass_strength_reduction): ...new function.
10248         (pass_ipa_lower_emutls): Replace declaration with that of...
10249         (make_pass_ipa_lower_emutls): ...new function.
10250         (pass_ipa_function_and_variable_visibility): Replace declaration with
10251         that of...
10252         (make_pass_ipa_function_and_variable_visibility): ...new function.
10253         (pass_ipa_tree_profile): Replace declaration with that of...
10254         (make_pass_ipa_tree_profile): ...new function.
10255         (pass_early_local_passes): Replace declaration with that of...
10256         (make_pass_early_local_passes): ...new function.
10257         (pass_ipa_whole_program_visibility): Replace declaration with that
10258         of...
10259         (make_pass_ipa_whole_program_visibility): ...new function.
10260         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10261         (make_pass_ipa_lto_gimple_out): ...new function.
10262         (pass_ipa_increase_alignment): Replace declaration with that of...
10263         (make_pass_ipa_increase_alignment): ...new function.
10264         (pass_ipa_inline): Replace declaration with that of...
10265         (make_pass_ipa_inline): ...new function.
10266         (pass_ipa_free_lang_data): Replace declaration with that of...
10267         (make_pass_ipa_free_lang_data): ...new function.
10268         (pass_ipa_free_inline_summary): Replace declaration with that of...
10269         (make_pass_ipa_free_inline_summary): ...new function.
10270         (pass_ipa_cp): Replace declaration with that of...
10271         (make_pass_ipa_cp): ...new function.
10272         (pass_ipa_reference): Replace declaration with that of...
10273         (make_pass_ipa_reference): ...new function.
10274         (pass_ipa_pure_const): Replace declaration with that of...
10275         (make_pass_ipa_pure_const): ...new function.
10276         (pass_ipa_pta): Replace declaration with that of...
10277         (make_pass_ipa_pta): ...new function.
10278         (pass_ipa_lto_finish_out): Replace declaration with that of...
10279         (make_pass_ipa_lto_finish_out): ...new function.
10280         (pass_ipa_tm): Replace declaration with that of...
10281         (make_pass_ipa_tm): ...new function.
10282         (pass_ipa_profile): Replace declaration with that of...
10283         (make_pass_ipa_profile): ...new function.
10284         (pass_ipa_cdtor_merge): Replace declaration with that of...
10285         (make_pass_ipa_cdtor_merge): ...new function.
10286         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10287         of...
10288         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10289         (pass_init_datastructures): Replace declaration with that of...
10290         (make_pass_init_datastructures): ...new function.
10291         (pass_fixup_cfg): Replace declaration with that of...
10292         (make_pass_fixup_cfg): ...new function.
10293         (pass_expand): Replace declaration with that of...
10294         (make_pass_expand): ...new function.
10295         (pass_instantiate_virtual_regs): Replace declaration with that of...
10296         (make_pass_instantiate_virtual_regs): ...new function.
10297         (pass_rtl_fwprop): Replace declaration with that of...
10298         (make_pass_rtl_fwprop): ...new function.
10299         (pass_rtl_fwprop_addr): Replace declaration with that of...
10300         (make_pass_rtl_fwprop_addr): ...new function.
10301         (pass_jump): Replace declaration with that of...
10302         (make_pass_jump): ...new function.
10303         (pass_jump2): Replace declaration with that of...
10304         (make_pass_jump2): ...new function.
10305         (pass_lower_subreg): Replace declaration with that of...
10306         (make_pass_lower_subreg): ...new function.
10307         (pass_cse): Replace declaration with that of...
10308         (make_pass_cse): ...new function.
10309         (pass_fast_rtl_dce): Replace declaration with that of...
10310         (make_pass_fast_rtl_dce): ...new function.
10311         (pass_ud_rtl_dce): Replace declaration with that of...
10312         (make_pass_ud_rtl_dce): ...new function.
10313         (pass_rtl_dce): Replace declaration with that of...
10314         (make_pass_rtl_dce): ...new function.
10315         (pass_rtl_dse1): Replace declaration with that of...
10316         (make_pass_rtl_dse1): ...new function.
10317         (pass_rtl_dse2): Replace declaration with that of...
10318         (make_pass_rtl_dse2): ...new function.
10319         (pass_rtl_dse3): Replace declaration with that of...
10320         (make_pass_rtl_dse3): ...new function.
10321         (pass_rtl_cprop): Replace declaration with that of...
10322         (make_pass_rtl_cprop): ...new function.
10323         (pass_rtl_pre): Replace declaration with that of...
10324         (make_pass_rtl_pre): ...new function.
10325         (pass_rtl_hoist): Replace declaration with that of...
10326         (make_pass_rtl_hoist): ...new function.
10327         (pass_rtl_store_motion): Replace declaration with that of...
10328         (make_pass_rtl_store_motion): ...new function.
10329         (pass_cse_after_global_opts): Replace declaration with that of...
10330         (make_pass_cse_after_global_opts): ...new function.
10331         (pass_rtl_ifcvt): Replace declaration with that of...
10332         (make_pass_rtl_ifcvt): ...new function.
10333         (pass_into_cfg_layout_mode): Replace declaration with that of...
10334         (make_pass_into_cfg_layout_mode): ...new function.
10335         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10336         (make_pass_outof_cfg_layout_mode): ...new function.
10337         (pass_loop2): Replace declaration with that of...
10338         (make_pass_loop2): ...new function.
10339         (pass_rtl_loop_init): Replace declaration with that of...
10340         (make_pass_rtl_loop_init): ...new function.
10341         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10342         (make_pass_rtl_move_loop_invariants): ...new function.
10343         (pass_rtl_unswitch): Replace declaration with that of...
10344         (make_pass_rtl_unswitch): ...new function.
10345         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10346         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10347         (pass_rtl_doloop): Replace declaration with that of...
10348         (make_pass_rtl_doloop): ...new function.
10349         (pass_rtl_loop_done): Replace declaration with that of...
10350         (make_pass_rtl_loop_done): ...new function.
10351         (pass_web): Replace declaration with that of...
10352         (make_pass_web): ...new function.
10353         (pass_cse2): Replace declaration with that of...
10354         (make_pass_cse2): ...new function.
10355         (pass_df_initialize_opt): Replace declaration with that of...
10356         (make_pass_df_initialize_opt): ...new function.
10357         (pass_df_initialize_no_opt): Replace declaration with that of...
10358         (make_pass_df_initialize_no_opt): ...new function.
10359         (pass_reginfo_init): Replace declaration with that of...
10360         (make_pass_reginfo_init): ...new function.
10361         (pass_inc_dec): Replace declaration with that of...
10362         (make_pass_inc_dec): ...new function.
10363         (pass_stack_ptr_mod): Replace declaration with that of...
10364         (make_pass_stack_ptr_mod): ...new function.
10365         (pass_initialize_regs): Replace declaration with that of...
10366         (make_pass_initialize_regs): ...new function.
10367         (pass_combine): Replace declaration with that of...
10368         (make_pass_combine): ...new function.
10369         (pass_if_after_combine): Replace declaration with that of...
10370         (make_pass_if_after_combine): ...new function.
10371         (pass_ree): Replace declaration with that of...
10372         (make_pass_ree): ...new function.
10373         (pass_partition_blocks): Replace declaration with that of...
10374         (make_pass_partition_blocks): ...new function.
10375         (pass_match_asm_constraints): Replace declaration with that of...
10376         (make_pass_match_asm_constraints): ...new function.
10377         (pass_regmove): Replace declaration with that of...
10378         (make_pass_regmove): ...new function.
10379         (pass_split_all_insns): Replace declaration with that of...
10380         (make_pass_split_all_insns): ...new function.
10381         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10382         (make_pass_fast_rtl_byte_dce): ...new function.
10383         (pass_lower_subreg2): Replace declaration with that of...
10384         (make_pass_lower_subreg2): ...new function.
10385         (pass_mode_switching): Replace declaration with that of...
10386         (make_pass_mode_switching): ...new function.
10387         (pass_sms): Replace declaration with that of...
10388         (make_pass_sms): ...new function.
10389         (pass_sched): Replace declaration with that of...
10390         (make_pass_sched): ...new function.
10391         (pass_ira): Replace declaration with that of...
10392         (make_pass_ira): ...new function.
10393         (pass_reload): Replace declaration with that of...
10394         (make_pass_reload): ...new function.
10395         (pass_clean_state): Replace declaration with that of...
10396         (make_pass_clean_state): ...new function.
10397         (pass_branch_prob): Replace declaration with that of...
10398         (make_pass_branch_prob): ...new function.
10399         (pass_value_profile_transformations): Replace declaration with that
10400         of...
10401         (make_pass_value_profile_transformations): ...new function.
10402         (pass_postreload_cse): Replace declaration with that of...
10403         (make_pass_postreload_cse): ...new function.
10404         (pass_gcse2): Replace declaration with that of...
10405         (make_pass_gcse2): ...new function.
10406         (pass_split_after_reload): Replace declaration with that of...
10407         (make_pass_split_after_reload): ...new function.
10408         (pass_branch_target_load_optimize1): Replace declaration with that
10409         of...
10410         (make_pass_branch_target_load_optimize1): ...new function.
10411         (pass_thread_prologue_and_epilogue): Replace declaration with that
10412         of...
10413         (make_pass_thread_prologue_and_epilogue): ...new function.
10414         (pass_stack_adjustments): Replace declaration with that of...
10415         (make_pass_stack_adjustments): ...new function.
10416         (pass_peephole2): Replace declaration with that of...
10417         (make_pass_peephole2): ...new function.
10418         (pass_if_after_reload): Replace declaration with that of...
10419         (make_pass_if_after_reload): ...new function.
10420         (pass_regrename): Replace declaration with that of...
10421         (make_pass_regrename): ...new function.
10422         (pass_cprop_hardreg): Replace declaration with that of...
10423         (make_pass_cprop_hardreg): ...new function.
10424         (pass_reorder_blocks): Replace declaration with that of...
10425         (make_pass_reorder_blocks): ...new function.
10426         (pass_branch_target_load_optimize2): Replace declaration with that
10427         of...
10428         (make_pass_branch_target_load_optimize2): ...new function.
10429         (pass_leaf_regs): Replace declaration with that of...
10430         (make_pass_leaf_regs): ...new function.
10431         (pass_split_before_sched2): Replace declaration with that of...
10432         (make_pass_split_before_sched2): ...new function.
10433         (pass_compare_elim_after_reload): Replace declaration with that of...
10434         (make_pass_compare_elim_after_reload): ...new function.
10435         (pass_sched2): Replace declaration with that of...
10436         (make_pass_sched2): ...new function.
10437         (pass_stack_regs): Replace declaration with that of...
10438         (make_pass_stack_regs): ...new function.
10439         (pass_stack_regs_run): Replace declaration with that of...
10440         (make_pass_stack_regs_run): ...new function.
10441         (pass_df_finish): Replace declaration with that of...
10442         (make_pass_df_finish): ...new function.
10443         (pass_compute_alignments): Replace declaration with that of...
10444         (make_pass_compute_alignments): ...new function.
10445         (pass_duplicate_computed_gotos): Replace declaration with that of...
10446         (make_pass_duplicate_computed_gotos): ...new function.
10447         (pass_variable_tracking): Replace declaration with that of...
10448         (make_pass_variable_tracking): ...new function.
10449         (pass_free_cfg): Replace declaration with that of...
10450         (make_pass_free_cfg): ...new function.
10451         (pass_machine_reorg): Replace declaration with that of...
10452         (make_pass_machine_reorg): ...new function.
10453         (pass_cleanup_barriers): Replace declaration with that of...
10454         (make_pass_cleanup_barriers): ...new function.
10455         (pass_delay_slots): Replace declaration with that of...
10456         (make_pass_delay_slots): ...new function.
10457         (pass_split_for_shorten_branches): Replace declaration with that of...
10458         (make_pass_split_for_shorten_branches): ...new function.
10459         (pass_split_before_regstack): Replace declaration with that of...
10460         (make_pass_split_before_regstack): ...new function.
10461         (pass_convert_to_eh_region_ranges): Replace declaration with that
10462         of...
10463         (make_pass_convert_to_eh_region_ranges): ...new function.
10464         (pass_shorten_branches): Replace declaration with that of...
10465         (make_pass_shorten_branches): ...new function.
10466         (pass_set_nothrow_function_flags): Replace declaration with that of...
10467         (make_pass_set_nothrow_function_flags): ...new function.
10468         (pass_dwarf2_frame): Replace declaration with that of...
10469         (make_pass_dwarf2_frame): ...new function.
10470         (pass_final): Replace declaration with that of...
10471         (make_pass_final): ...new function.
10472         (pass_rtl_seqabstr): Replace declaration with that of...
10473         (make_pass_rtl_seqabstr): ...new function.
10474         (pass_release_ssa_names): Replace declaration with that of...
10475         (make_pass_release_ssa_names): ...new function.
10476         (pass_early_inline): Replace declaration with that of...
10477         (make_pass_early_inline): ...new function.
10478         (pass_inline_parameters): Replace declaration with that of...
10479         (make_pass_inline_parameters): ...new function.
10480         (pass_update_address_taken): Replace declaration with that of...
10481         (make_pass_update_address_taken): ...new function.
10482         (pass_convert_switch): Replace declaration with that of...
10483         (make_pass_convert_switch): ...new function.
10484         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10485         to a subclass of simple_ipa_opt_pass along with...
10486         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10487         (make_pass_ipa_tree_profile): ...new function.
10488         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10489         subclass of gimple_opt_pass along with...
10490         (pass_data_sra_early): ...new pass_data instance and...
10491         (make_pass_sra_early): ...new function.
10492         (pass_sra): Convert from a global struct to a subclass of
10493         gimple_opt_pass along with...
10494         (pass_data_sra): ...new pass_data instance and...
10495         (make_pass_sra): ...new function.
10496         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10497         gimple_opt_pass along with...
10498         (pass_data_early_ipa_sra): ...new pass_data instance and...
10499         (make_pass_early_ipa_sra): ...new function.
10500         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10501         subclass of gimple_opt_pass along with...
10502         (pass_data_ccp): ...new pass_data instance and...
10503         (make_pass_ccp): ...new function.
10504         (pass_fold_builtins): Convert from a global struct to a subclass of
10505         gimple_opt_pass along with...
10506         (pass_data_fold_builtins): ...new pass_data instance and...
10507         (make_pass_fold_builtins): ...new function.
10508         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10509         subclass of gimple_opt_pass along with...
10510         (pass_data_copy_prop): ...new pass_data instance and...
10511         (make_pass_copy_prop): ...new function.
10512         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10513         global struct to a subclass of gimple_opt_pass along with...
10514         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10515         (make_pass_rename_ssa_copies): ...new function.
10516         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10517         subclass of gimple_opt_pass along with...
10518         (pass_data_dce): ...new pass_data instance and...
10519         (make_pass_dce): ...new function.
10520         (pass_dce_loop): Convert from a global struct to a subclass of
10521         gimple_opt_pass along with...
10522         (pass_data_dce_loop): ...new pass_data instance and...
10523         (make_pass_dce_loop): ...new function.
10524         (pass_cd_dce): Convert from a global struct to a subclass of
10525         gimple_opt_pass along with...
10526         (pass_data_cd_dce): ...new pass_data instance and...
10527         (make_pass_cd_dce): ...new function.
10528         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10529         subclass of gimple_opt_pass along with...
10530         (pass_data_dominator): ...new pass_data instance and...
10531         (make_pass_dominator): ...new function.
10532         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10533         gimple_opt_pass along with...
10534         (pass_data_phi_only_cprop): ...new pass_data instance and...
10535         (make_pass_phi_only_cprop): ...new function.
10536         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10537         subclass of gimple_opt_pass along with...
10538         (pass_data_dse): ...new pass_data instance and...
10539         (make_pass_dse): ...new function.
10540         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10541         a subclass of gimple_opt_pass along with...
10542         (pass_data_forwprop): ...new pass_data instance and...
10543         (make_pass_forwprop): ...new function.
10544         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10545         struct to a subclass of gimple_opt_pass along with...
10546         (pass_data_tree_ifcombine): ...new pass_data instance and...
10547         (make_pass_tree_ifcombine): ...new function.
10548         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10549         subclass of gimple_opt_pass along with...
10550         (pass_data_ch): ...new pass_data instance and...
10551         (make_pass_ch): ...new function.
10552         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10553         subclass of gimple_opt_pass along with...
10554         (pass_data_tree_loop): ...new pass_data instance and...
10555         (make_pass_tree_loop): ...new function.
10556         (pass_tree_loop_init): Convert from a global struct to a subclass of
10557         gimple_opt_pass along with...
10558         (pass_data_tree_loop_init): ...new pass_data instance and...
10559         (make_pass_tree_loop_init): ...new function.
10560         (pass_lim): Convert from a global struct to a subclass of
10561         gimple_opt_pass along with...
10562         (pass_data_lim): ...new pass_data instance and...
10563         (make_pass_lim): ...new function.
10564         (pass_tree_unswitch): Convert from a global struct to a subclass of
10565         gimple_opt_pass along with...
10566         (pass_data_tree_unswitch): ...new pass_data instance and...
10567         (make_pass_tree_unswitch): ...new function.
10568         (pass_predcom): Convert from a global struct to a subclass of
10569         gimple_opt_pass along with...
10570         (pass_data_predcom): ...new pass_data instance and...
10571         (make_pass_predcom): ...new function.
10572         (pass_vectorize): Convert from a global struct to a subclass of
10573         gimple_opt_pass along with...
10574         (pass_data_vectorize): ...new pass_data instance and...
10575         (make_pass_vectorize): ...new function.
10576         (pass_graphite): Convert from a global struct to a subclass of
10577         gimple_opt_pass along with...
10578         (pass_data_graphite): ...new pass_data instance and...
10579         (make_pass_graphite): ...new function.
10580         (pass_graphite_transforms): Convert from a global struct to a subclass
10581         of gimple_opt_pass along with...
10582         (pass_data_graphite_transforms): ...new pass_data instance and...
10583         (make_pass_graphite_transforms): ...new function.
10584         (pass_check_data_deps): Convert from a global struct to a subclass of
10585         gimple_opt_pass along with...
10586         (pass_data_check_data_deps): ...new pass_data instance and...
10587         (make_pass_check_data_deps): ...new function.
10588         (pass_iv_canon): Convert from a global struct to a subclass of
10589         gimple_opt_pass along with...
10590         (pass_data_iv_canon): ...new pass_data instance and...
10591         (make_pass_iv_canon): ...new function.
10592         (pass_scev_cprop): Convert from a global struct to a subclass of
10593         gimple_opt_pass along with...
10594         (pass_data_scev_cprop): ...new pass_data instance and...
10595         (make_pass_scev_cprop): ...new function.
10596         (pass_record_bounds): Convert from a global struct to a subclass of
10597         gimple_opt_pass along with...
10598         (pass_data_record_bounds): ...new pass_data instance and...
10599         (make_pass_record_bounds): ...new function.
10600         (pass_complete_unroll): Convert from a global struct to a subclass of
10601         gimple_opt_pass along with...
10602         (pass_data_complete_unroll): ...new pass_data instance and...
10603         (make_pass_complete_unroll): ...new function.
10604         (pass_complete_unrolli): Convert from a global struct to a subclass of
10605         gimple_opt_pass along with...
10606         (pass_data_complete_unrolli): ...new pass_data instance and...
10607         (make_pass_complete_unrolli): ...new function.
10608         (pass_parallelize_loops): Convert from a global struct to a subclass
10609         of gimple_opt_pass along with...
10610         (pass_data_parallelize_loops): ...new pass_data instance and...
10611         (make_pass_parallelize_loops): ...new function.
10612         (pass_loop_prefetch): Convert from a global struct to a subclass of
10613         gimple_opt_pass along with...
10614         (pass_data_loop_prefetch): ...new pass_data instance and...
10615         (make_pass_loop_prefetch): ...new function.
10616         (pass_iv_optimize): Convert from a global struct to a subclass of
10617         gimple_opt_pass along with...
10618         (pass_data_iv_optimize): ...new pass_data instance and...
10619         (make_pass_iv_optimize): ...new function.
10620         (pass_tree_loop_done): Convert from a global struct to a subclass of
10621         gimple_opt_pass along with...
10622         (pass_data_tree_loop_done): ...new pass_data instance and...
10623         (make_pass_tree_loop_done): ...new function.
10624         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10625         struct to a subclass of gimple_opt_pass along with...
10626         (pass_data_cse_reciprocals): ...new pass_data instance and...
10627         (make_pass_cse_reciprocals): ...new function.
10628         (pass_cse_sincos): Convert from a global struct to a subclass of
10629         gimple_opt_pass along with...
10630         (pass_data_cse_sincos): ...new pass_data instance and...
10631         (make_pass_cse_sincos): ...new function.
10632         (pass_optimize_bswap): Convert from a global struct to a subclass of
10633         gimple_opt_pass along with...
10634         (pass_data_optimize_bswap): ...new pass_data instance and...
10635         (make_pass_optimize_bswap): ...new function.
10636         (pass_optimize_widening_mul): Convert from a global struct to a
10637         subclass of gimple_opt_pass along with...
10638         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10639         (make_pass_optimize_widening_mul): ...new function.
10640         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10641         subclass of gimple_opt_pass along with...
10642         (pass_data_phiopt): ...new pass_data instance and...
10643         (make_pass_phiopt): ...new function.
10644         (pass_cselim): Convert from a global struct to a subclass of
10645         gimple_opt_pass along with...
10646         (pass_data_cselim): ...new pass_data instance and...
10647         (make_pass_cselim): ...new function.
10648         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10649         subclass of gimple_opt_pass along with...
10650         (pass_data_phiprop): ...new pass_data instance and...
10651         (make_pass_phiprop): ...new function.
10652         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10653         subclass of gimple_opt_pass along with...
10654         (pass_data_pre): ...new pass_data instance and...
10655         (make_pass_pre): ...new function.
10656         (pass_fre): Convert from a global struct to a subclass of
10657         gimple_opt_pass along with...
10658         (pass_data_fre): ...new pass_data instance and...
10659         (make_pass_fre): ...new function.
10660         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10661         subclass of gimple_opt_pass along with...
10662         (pass_data_reassoc): ...new pass_data instance and...
10663         (make_pass_reassoc): ...new function.
10664         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10665         subclass of gimple_opt_pass along with...
10666         (pass_data_sink_code): ...new pass_data instance and...
10667         (make_pass_sink_code): ...new function.
10668         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10669         subclass of gimple_opt_pass along with...
10670         (pass_data_strlen): ...new pass_data instance and...
10671         (make_pass_strlen): ...new function.
10672         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10673         struct to a subclass of gimple_opt_pass along with...
10674         (pass_data_build_alias): ...new pass_data instance and...
10675         (make_pass_build_alias): ...new function.
10676         (pass_build_ealias): Convert from a global struct to a subclass of
10677         gimple_opt_pass along with...
10678         (pass_data_build_ealias): ...new pass_data instance and...
10679         (make_pass_build_ealias): ...new function.
10680         (pass_ipa_pta): Convert from a global struct to a subclass of
10681         simple_ipa_opt_pass along with...
10682         (pass_data_ipa_pta): ...new pass_data instance and...
10683         (make_pass_ipa_pta): ...new function.
10684         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10685         subclass of gimple_opt_pass along with...
10686         (pass_data_uncprop): ...new pass_data instance and...
10687         (make_pass_uncprop): ...new function.
10688         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10689         global struct to a subclass of gimple_opt_pass along with...
10690         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10691         (make_pass_late_warn_uninitialized): ...new function.
10692         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10693         to a subclass of gimple_opt_pass along with...
10694         (pass_data_init_datastructures): ...new pass_data instance and...
10695         (make_pass_init_datastructures): ...new function.
10696         (pass_early_warn_uninitialized): Convert from a global struct to a
10697         subclass of gimple_opt_pass along with...
10698         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10699         (make_pass_early_warn_uninitialized): ...new function.
10700         (pass_update_address_taken): Convert from a global struct to a
10701         subclass of gimple_opt_pass along with...
10702         (pass_data_update_address_taken): ...new pass_data instance and...
10703         (make_pass_update_address_taken): ...new function.
10704         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10705         struct to a subclass of gimple_opt_pass along with...
10706         (pass_data_release_ssa_names): ...new pass_data instance and...
10707         (make_pass_release_ssa_names): ...new function.
10708         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10709         subclass of gimple_opt_pass along with...
10710         (pass_data_stdarg): ...new pass_data instance and...
10711         (make_pass_stdarg): ...new function.
10712         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10713         global struct to a subclass of gimple_opt_pass along with...
10714         (pass_data_convert_switch): ...new pass_data instance and...
10715         (make_pass_convert_switch): ...new function.
10716         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10717         to a subclass of gimple_opt_pass along with...
10718         (pass_data_tail_recursion): ...new pass_data instance and...
10719         (make_pass_tail_recursion): ...new function.
10720         (pass_tail_calls): Convert from a global struct to a subclass of
10721         gimple_opt_pass along with...
10722         (pass_data_tail_calls): ...new pass_data instance and...
10723         (make_pass_tail_calls): ...new function.
10724         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10725         struct to a subclass of gimple_opt_pass along with...
10726         (pass_data_lower_vector): ...new pass_data instance and...
10727         (make_pass_lower_vector): ...new function.
10728         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
10729         gimple_opt_pass along with...
10730         (pass_data_lower_vector_ssa): ...new pass_data instance and...
10731         (make_pass_lower_vector_ssa): ...new function.
10732         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
10733         to a subclass of gimple_opt_pass along with...
10734         (pass_data_slp_vectorize): ...new pass_data instance and...
10735         (make_pass_slp_vectorize): ...new function.
10736         (pass_ipa_increase_alignment): Convert from a global struct to a
10737         subclass of simple_ipa_opt_pass along with...
10738         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
10739         (make_pass_ipa_increase_alignment): ...new function.
10740         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
10741         gimple_opt_pass along with...
10742         (pass_data_vrp): ...new pass_data instance and...
10743         (make_pass_vrp): ...new function.
10744         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
10745         subclass of simple_ipa_opt_pass along with...
10746         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
10747         (make_pass_ipa_free_lang_data): ...new function.
10748         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
10749         gimple_opt_pass along with...
10750         (pass_data_tsan): ...new pass_data instance and...
10751         (make_pass_tsan): ...new function.
10752         (pass_tsan_O0): Convert from a global struct to a subclass of
10753         gimple_opt_pass along with...
10754         (pass_data_tsan_O0): ...new pass_data instance and...
10755         (make_pass_tsan_O0): ...new function.
10756         * var-tracking.c (pass_variable_tracking): Convert from a global
10757         struct to a subclass of rtl_opt_pass along with...
10758         (pass_data_variable_tracking): ...new pass_data instance and...
10759         (make_pass_variable_tracking): ...new function.
10760         * web.c (pass_web): Convert from a global struct to a subclass of
10761         rtl_opt_pass along with...
10762         (pass_data_web): ...new pass_data instance and...
10763         (make_pass_web): ...new function.
10764         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
10765         declaration with that of...
10766         (make_pass_mode_switch_use): ...new function.
10767         (pass_resolve_sw_modes): Replace declaration with that of...
10768         (make_pass_resolve_sw_modes): ...new function.
10769         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
10770         from a global struct to a subclass of rtl_opt_pass along with...
10771         (pass_data_mode_switch_use): ...new pass_data instance and...
10772         (make_pass_mode_switch_use): ...new function.
10773         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
10774         from a global struct to a subclass of rtl_opt_pass along with...
10775         (pass_data_resolve_sw_modes): ...new pass_data instance and...
10776         (make_pass_resolve_sw_modes): ...new function.
10777         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
10778         struct to a subclass of rtl_opt_pass along with...
10779         (pass_data_insert_vzeroupper): ...new pass_data instance and...
10780         (make_pass_insert_vzeroupper): ...new function.
10781         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
10782         global struct to a subclass of rtl_opt_pass along with...
10783         (pass_data_work_around_errata): ...new pass_data instance and...
10784         (make_pass_work_around_errata): ...new function.
10785         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
10786         struct to a subclass of rtl_opt_pass along with...
10787         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
10788         (make_pass_mips_machine_reorg2): ...new function.
10789
10790 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10791
10792         * passes.c (pass_manager::operator new): New.
10793
10794 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
10795
10796         Handwritten part of conversion of passes to C++ classes.
10797
10798         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
10799         (toplev.o): Add dep on PASS_MANAGER_H.
10800         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
10801         of early local pases to reflect this moving from a global to a
10802         member of gcc::pass_manager.
10803         (cgraph_add_new_function): Likewise.
10804         * lto-cgraph.c (lto_output_node): Update for conversion of
10805         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
10806         * passes.c (opt_pass::clone): New.
10807         (opt_pass::gate): New.
10808         (opt_pass::execute): New.
10809         (opt_pass::opt_pass): New.
10810         (pass_manager::execute_early_local_passes): New.
10811         (pass_manager::execute_pass_mode_switching): new.
10812         (finish_optimization_passes): Convert to...
10813         (pass_manager::finish_optimization_passes): ...this.
10814         (finish_optimization_passes): Update for conversion of passes to
10815         C++ classes.
10816         (register_dump_files_1): Use has_gate since we cannot portably
10817         check a vtable entry against NULL.
10818         (dump_one_pass): Likewise.
10819         (ipa_write_summaries_2): Likewise.
10820         (ipa_write_optimization_summaries_1): Likewise.
10821         (ipa_read_summaries_1): Likewise.
10822         (ipa_read_optimization_summaries_1): Likewise.
10823         (execute_ipa_stmt_fixups): Likewise.
10824         (pass_manager::pass_manager): Rewrite pass-creation, invoking
10825         pass-creation functions rather than wiring up globals, and
10826         storing the results in fields of pass_manager generated using
10827         pass-instances.def.
10828         (pass_manager::dump_profile_report): Update for conversion of
10829         passes to C++ classes.
10830         (pass_manager::execute_ipa_summary_passes): Likewise.
10831         (execute_one_ipa_transform_pass): Likewise.
10832         (execute_one_pass): Use has_gate and has_execute since we cannot
10833         portably check a vtable entry against NULL.
10834         * pass_manager.h (pass_manager::finish_optimization_passes): New.
10835         (pass_manager): Use pass-instances.def to add fields for the
10836         various pass instances.
10837         * toplev.c (finalize): Update for move of
10838         finish_optimization_passes to a method of gcc::pass_manager.
10839         * toplev.h (finish_optimization_passes): Move to method of class
10840         pass_manager.
10841         * tree-pass.h (struct pass_data): New.
10842         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
10843         (opt_pass::gate): Convert to virtual function.
10844         (opt_pass::~opt_pass): New.
10845         (opt_pass::clone): New.
10846         (opt_pass::execute): Convert to virtual function.
10847         (opt_pass::opt_pass): New.
10848         (opt_pass::ctxt_): new.
10849         (gimple_opt_pass): Convert to subclass of opt_pass.
10850         (gimple_opt_pass::gimple_opt_pass): New.
10851         (rtl_opt_pass): Convert to subclass of opt_pass.
10852         (rtl_opt_pass::rtl_opt_pass): New.
10853         (ipa_opt_pass_d): Convert to subclass of opt_pass.
10854         (ipa_opt_pass_d::ipa_opt_pass_d): New.
10855         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
10856         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
10857         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
10858         invocation of pass_mode_switching to reflect this moving from a
10859         global to a member of gcc::pass_manager.
10860         (ix86_option_override): Rework how pass_insert_vzeroupper is
10861         added to the pass_manager to reflect autogenerated changes.
10862         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
10863
10864 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
10865
10866         PR rtl-optimization/57708
10867         * recog.c (peep2_find_free_register): Validate all regs in a
10868         multi-reg mode.
10869
10870 2013-08-05  Jan Hubicka  <jh@suse.cz>
10871
10872         PR lto/57602
10873         * cgraph.c (verify_cgraph_node): Accept local flags from other
10874         partitions.
10875         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
10876         (function_and_variable_visibility): Likewise.
10877         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
10878
10879 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10880
10881         * graph.c (init_graph_slim_pretty_print): Remove.
10882         (print_graph_cfg): Do not call it.  Use local pretty printer.
10883         (start_graph_dump): Likewise.
10884
10885 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10886
10887         * gimple-pretty-print.c (buffer): Remove.
10888         (initialized): Likewise.
10889         (maybe_init_pretty_print): Likewise.
10890         (print_gimple_stmt): Do not call it.  Use non-static local
10891         pretty_printer variable.
10892         (print_gimple_expr): Likewise.
10893         (print_gimple_seq): Likewise.
10894         (gimple_dump_bb): Likewise.
10895
10896 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10897
10898         * asan.c (asan_pp): Remove.
10899         (asan_pp_initialized): Likewise.
10900         (asan_pp_initialize): Likewise.
10901         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
10902         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
10903         (asan_add_global): Likewise.
10904
10905 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10906
10907         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
10908         * diagnostic.h (diagnostic_flush_buffer): Adjust.
10909         * pretty-print.c (pp_formatted_text_data): Likewise.
10910         (pp_indent): Rename from pp_base_indent.
10911         (pp_format): Rename from pp_base_format.
10912         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
10913         (pp_format_verbatim): Rename from pp_base_format_verbatim.
10914         (pp_flush): Rename from pp_base_flush.
10915         (pp_set_line_maximum_length): Rename from
10916         pp_base_set_line_maximum_length.
10917         (pp_clear_output_area): Rename from pp_base_clear_output_area.
10918         (pp_set_prefix): Rename from pp_base_set_prefix.
10919         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
10920         (pp_emit_prefix): Rename from pp_base_emit_prefix.
10921         (pp_append_text): Rename from pp_base_append_text.
10922         (pp_formatted_text): Rename from pp_base_formatted_text.
10923         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
10924         (pp_remaining_character_count_for_line): Rename from
10925         pp_base_remaining_character_count_for_line.
10926         (pp_newline): Rename from pp_base_newline.
10927         (pp_character): Rename from pp_base_character.
10928         (pp_string): Rename from pp_base_string.
10929         (pp_maybe_space): Rename from pp_base_maybe_space.
10930         * asan.c (asan_pp_string): Adjust.
10931         (asan_emit_stack_protection): Likewise.
10932         (asan_add_global): Likewise.
10933         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
10934         * tree-mudflap.c (mf_varname_tree): Likewise.
10935         * tree-pretty-print.c (pp_tree_identifier): Rename from
10936         pp_base_tree_identifier.
10937         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
10938         Declare as function.
10939
10940 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10941
10942         * pretty-print.h (pp_bar_bar): New.
10943         (pp_ampersand_ampersand): Likewise.
10944         (pp_less_equal): Likewise.
10945         (pp_greater_equal): Likewise.
10946         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
10947         printer functions instead of pp_string or operators and punctuators.
10948         (dump_gimple_call): Likewise.
10949         (dump_gimple_omp_for): Likewise.
10950         (dump_gimple_transaction): Likewise.
10951         (dump_gimple_phi): Likewise.
10952         (pp_gimple_stmt_1): Likewise.
10953         * sched-vis.c (print_insn): Likewise.
10954         * tree-mudflap.c (mf_varname_tree): Likewise.
10955         * tree-pretty-print.c (dump_block_node): Likewise.
10956         (dump_generic_node): Likewise.
10957
10958 2013-08-02  Jan Hubicka  <jh@suse.cz>
10959
10960         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
10961         boundaries.
10962         * lto-streamer-out.c (tree_is_indexable): Results decls and
10963         parm decls are not indexable.
10964         (DFS_write_tree_body): Do not follow args and results.
10965         (hash_tree): Likewise.
10966         (output_functions): Rearrange so struct function is needed
10967         only when real body is output; be able to also ouptut abstract
10968         functions; output DECL_ARGUMENTS and DECL_RESULT.
10969         (lto_output): When not in WPA, ale store abstract functions.
10970         (write_symbol): Do not care about RESULT_DECL.
10971         (output_symbol_p): Handle correctly sbtract decls.
10972         * lto-streamer-in.c (input_function): Rearrange so struct
10973         function can be NULL at entry; allow streaming of
10974         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
10975         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
10976         sanity check during LTO.
10977         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
10978         RESULT_DECl and DECL_ARGUMENTS.
10979         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10980         Likewise.
10981
10982 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
10983
10984         * pretty-print.h (pp_underscore): New.
10985         (pp_comma): Tidy.
10986         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
10987         printer functions instead of pp_character.
10988         (dump_binary_rhs): Likewise.
10989         (dump_ternary_rhs): Likewise.
10990         (dump_gimple_call_args): Likewise.
10991         (pp_points_to_solution): Likewise.
10992         (dump_gimple_call): Likewise.
10993         (dump_gimple_switch): Likewise.
10994         (dump_gimple_cond): Likewise.
10995         (dump_gimple_bind): Likewise.
10996         (dump_gimple_try): Likewise.
10997         (dump_gimple_omp_for): Likewise.
10998         (dump_gimple_omp_continue): Likewise.
10999         (dump_gimple_omp_single): Likewise.
11000         (dump_gimple_omp_sections): Likewise.
11001         (dump_gimple_omp_block): Likewise.
11002         (dump_gimple_omp_critical): Likewise.
11003         (dump_gimple_transaction): Likewise.
11004         (dump_gimple_asm): Likewise.
11005         (dump_gimple_phi): Likewise.
11006         (dump_gimple_omp_parallel): Likewise.
11007         (dump_gimple_omp_task): Likewise.
11008         (dump_gimple_omp_atomic_load): Likewise.
11009         (dump_gimple_omp_atomic_store): Likewise.
11010         (dump_gimple_mem_ops): Likewise.
11011         (pp_gimple_stmt_1): Likewise.
11012         (pp_cfg_jump): Likewise.
11013         (dump_implicit_edges): Likewise.
11014         (gimple_dump_bb_for_graph): Likewise.
11015         * graph.c (draw_cfg_node): Likewise.
11016         * langhooks.c (lhd_print_error_function): Likewise.
11017         * sched-vis.c (print_exp): Likewise.
11018         (print_value): Likewise.
11019         (print_pattern): Likewise.
11020         (print_insn): Likewise.
11021         (rtl_dump_bb_for_graph): Likewise.
11022         * tree-pretty-print.c (dump_function_declaration): Likewise.
11023         (dump_array_domain): Likewise.
11024         (dump_omp_clause): Likewise.
11025         (dump_location): Likewise.
11026         (dump_generic_node): Likewise.
11027         (print_struct_decl): Likewise.
11028         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11029
11030 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11031
11032         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11033         candidate table when replacing a candidate statement.
11034         (replace_rhs_if_not_dup): Likewise.
11035         (replace_one_candidate): Likewise.
11036
11037 2013-08-02  Jan Hubicka  <jh@suse.cz>
11038             Martin Liska  <marxin.liska@gmail.com>
11039
11040         * cgraphunit.c (add_new_function): Fix logic when adding from
11041         late IPA pass.
11042         (assemble_thunk): Rename to ...
11043         (expand_thunk); .. this one; export; get it working with
11044         general functions; make produced gimple valid.
11045         * cgraph.h (expand_thunk): Declare.
11046
11047 2013-08-02  Jan Hubicka  <jh@suse.cz>
11048
11049         * ipa-cp.c (gather_context_independent_values): Use
11050         ipa_get_param_move_cost.
11051         (get_replacement_map): Remove PARAM; move parameter folding
11052         into tree-inline.c
11053         (create_specialized_node): Update.
11054         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11055         assert that we have gimple body; update move_cost.
11056         (count_formal_params): Assert that we have gimple body.
11057         (ipa_dump_param): New function.
11058         (ipa_alloc_node_params): Break out from ...
11059         (ipa_initialize_node_params): ... here.
11060         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11061         (ipa_write_node_info): Stream move costs.
11062         (ipa_read_node_info): Read move costs.
11063         (ipa_update_after_lto_read): Do not recompute node params.
11064         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11065         (ipa_get_param): Check we are not in WPA.
11066         (ipa_get_param_move_cost): New.
11067         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11068         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11069         parm numbers to be present.
11070
11071 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11072
11073         PR rtl-optimization/58048
11074         * lra-constraints.c (process_alt_operands): Don't check asm
11075         operand on register.
11076
11077 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11078
11079         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11080         the implied StoreLoad barrier for atomic operations if before.
11081
11082 2013-08-02  Jan Hubicka  <jh@suse.cz>
11083             Martin Liska  <marxin.liska@gmail.com>
11084
11085         * cgraph.c (cgraph_function_body_availability): Do not check
11086         cgraph flags.
11087         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11088         symtab_node_availability): Declare.
11089         * ipa.c (can_replace_by_local_alias): New.
11090         (function_and_variable_visibility): Use it.
11091         * symtab.c (symtab_for_node_and_aliases,
11092         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11093
11094 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11095
11096         PR rtl-optimization/57963
11097         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11098         (lra_constraints): Use them.
11099
11100 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11101
11102         * config/arm/types.md (define_attr "type"): Add "load_acq"
11103         and "store_rel".
11104         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11105         changes.
11106         (cortex_a53_store1): Likewise.
11107
11108 2013-08-01  Jan Hubicka  <jh@suse.cz>
11109
11110         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11111         partitions are not needed.
11112
11113 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11114
11115         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11116         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11117         MAYBE_NON_Q_CLASS_P where appropriate.
11118
11119 2013-08-01  Jan Hubicka  <jh@suse.cz>
11120
11121         * cgraph.h (release_function_body): Declare.
11122         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11123         of unused delcarations.
11124
11125 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11126
11127         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11128         RTL form when subtracting a constant.
11129
11130 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11131
11132         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11133         Generate canonical plus rtx with negated immediate instead of minus
11134         where appropriate.
11135         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11136
11137 2013-08-01  Jan Hubicka  <jh@suse.cz>
11138
11139         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11140         (cgraph_release_function_body): Likewise.
11141         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11142         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11143         to used_as_abstract_origin.
11144         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11145         symbols used as abstract origins.
11146         * cgraphunit.c (analyze_functions): Update.
11147         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11148         used_as_abstract_origin.
11149         * tree-inline.c (tree_function_versioning): Update
11150         used_as_abstract_origin; be ready for DECL_RESULT and
11151         DECL_ARGUMENTS to be NULL.
11152
11153         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11154         for abstract functions.
11155         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11156         real symbols.
11157
11158 2013-08-01  Jan Hubicka  <jh@suse.cz>
11159
11160         * profile.c (compute_value_histograms): Fix thinko.
11161
11162 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11163
11164         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11165         aarch-common-protos.h to extra_headers.
11166         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11167         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11168         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11169
11170 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11171
11172         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11173         Include "../arm/types.md".  Define "type" attribute for all patterns.
11174         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11175         attribute changes.
11176
11177 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11178
11179         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11180         to support power8 load fusion.
11181         (fusion_gpr_mem_load): Likewise.
11182
11183         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11184
11185         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11186         declarations for power8 load fusion.
11187         (emit_fusion_gpr_load): Likewise.
11188
11189         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11190         tuning for power8, turn on fusion mode by default.  Turn on sign
11191         extending fusion mode if normal fusion mode is on, and we are at
11192         -O2 or -O3.
11193         (fusion_gpr_load_p): New function, return true if we can fuse an
11194         addis instruction with a dependent load to a GPR.
11195         (emit_fusion_gpr_load): Emit the instructions for power8 load
11196         fusion to GPRs.
11197
11198         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11199         (VSX load fusion peepholes): New peepholes to fuse together an
11200         addi instruction with a VSX load instruction.
11201
11202         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11203         peepholes to fuse an addis instruction with a load to a GPR base
11204         register.  If we are supporting sign extending fusions, convert
11205         sign extending loads to zero extending loads and add an explicit
11206         sign extension.
11207
11208 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11209
11210         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11211         aarch-common-protos.h to extra_headers.
11212         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11213         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11214         (arm_early_store_addr_dep): Likewise.
11215         (arm_no_early_alu_shift_dep): Likewise.
11216         (arm_no_early_alu_shift_value_dep): Likewise.
11217         (arm_no_early_mul_dep): Likewise.
11218         (arm_no_early_store_addr_dep): Likewise.
11219         (arm_mac_accumulator_is_mul_result): Likewise.
11220         (arm_mac_accumulator_is_result): Likewise.
11221         * config/arm/aarch-common.c: ... here.  New file.
11222         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11223         here to ...
11224         (arm_early_store_addr_dep): Likewise.
11225         (arm_no_early_alu_shift_dep): Likewise.
11226         (arm_no_early_alu_shift_value_dep): Likewise.
11227         (arm_no_early_mul_dep): Likewise.
11228         (arm_no_early_store_addr_dep): Likewise.
11229         (arm_mac_accumulator_is_mul_result): Likewise.
11230         (arm_mac_accumulator_is_result): Likewise.
11231         * config/arm/aarch-common-protos.h: ... here.  New file.
11232         * config/arm/t-arm (aarch-common.o): Define.
11233
11234 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11235
11236         * config/arm/arm.md: Include new file "types.md".
11237         (define_attr "type"): Move from here to ...
11238         (define_attr "mul32"): Likewise.
11239         (define_attr "mul64"): Likewise.
11240         * config/arm/types.md: ... here.  New file.
11241
11242 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11243
11244         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11245         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11246
11247 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11248
11249         * gen-pass-instances.awk: Fix offset of substr().
11250
11251 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11252
11253         * Makefile.in (pass-instances.def): New.
11254         (passes.o): Replace dependency on passes.def with one on
11255         pass-instances.def
11256
11257         * gen-pass-instances.awk: New.
11258
11259         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11260         rather than passes.def, updating local definition of NEXT_PASS
11261         macro to add an extra NUM parameter (currently unused).
11262
11263 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11264
11265         * Makefile.in (PASS_MANAGER_H): New.
11266         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11267         (passes.o): Likewise.
11268         (statistics.o): Likewise.
11269         (cgraphunit.o): Likewise.
11270         (context.o): Depend on PASS_MANAGER_H.
11271
11272         * pass_manager.h: New.
11273
11274         * cgraphunit.c (cgraph_add_new_function): Update for moves
11275         of globals to fields of pass_manager.
11276         (analyze_function): Likewise.
11277         (expand_function): Likewise.
11278         (ipa_passes): Likewise.
11279         (compile): Likewise.
11280
11281         * context.c (context::context): New.
11282         * context.h  (context::context): New.
11283         (context::get_passes): New.
11284         (context::passes_): New.
11285
11286         * lto-cgraph.c (input_node): Update for moves of globals to
11287         fields of pass_manager.
11288
11289         * passes.c (all_passes): Remove, in favor of a field of the
11290         same name within the new class pass_manager.
11291         (all_small_ipa_passes): Likewise.
11292         (all_lowering_passes): Likewise.
11293         (all_regular_ipa_passes): Likewise.
11294         (all_late_ipa_passes): Likewise.
11295         (all_lto_gen_passes): Likewise.
11296         (passes_by_id): Likewise.
11297         (passes_by_id_size): Likewise.
11298         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11299         the new class pass_manager.
11300         (set_pass_for_id): Convert to...
11301         (pass_manager::set_pass_for_id): ...method.
11302         (get_pass_for_id): Convert to...
11303         (pass_manager::get_pass_for_id): ...method.
11304         (register_one_dump_file): Move body of implementation into...
11305         (pass_manager::register_one_dump_file): ...here.
11306         (register_dump_files_1): Convert to...
11307         (pass_manager::register_dump_files_1): ...method.
11308         (register_dump_files): Convert to...
11309         (pass_manager::register_dump_files): ...method.
11310         (create_pass_tab): Update for moves of globals to fields of
11311         pass_manager.
11312         (dump_passes): Move body of implementation into...
11313         (pass_manager::dump_passes): ...here.
11314         (register_pass): Move body of implementation into...
11315         (pass_manager::register_pass): ...here.
11316         (init_optimization_passes): Convert into...
11317         (pass_manager::pass_manager): ...constructor for new
11318         pass_manager class, and initialize the pass_lists array.
11319         (check_profile_consistency): Update for moves of globals to
11320         fields of pass_manager.
11321         (dump_profile_report): Move body of implementation into...
11322         (pass_manager::dump_profile_report): ...here.
11323         (ipa_write_summaries_1): Update for moves of pass lists from
11324         being globals to fields of pass_manager.
11325         (ipa_write_optimization_summaries): Likewise.
11326         (ipa_read_summaries):  Likewise.
11327         (ipa_read_optimization_summaries): Likewise.
11328         (execute_all_ipa_stmt_fixups): Likewise.
11329
11330         * statistics.c (statistics_fini): Update for moves of globals to
11331         fields of pass_manager.
11332
11333         * toplev.c (general_init): Replace call to
11334         init_optimization_passes with construction of the pass_manager
11335         instance.
11336
11337         * tree-pass.h (all_passes): Remove, in favor of a field of the
11338         same name within the new class pass_manager.
11339         (all_small_ipa_passes): Likewise.
11340         (all_lowering_passes): Likewise.
11341         (all_regular_ipa_passes): Likewise.
11342         (all_lto_gen_passes): Likewise.
11343         (all_late_ipa_passes): Likewise.
11344         (passes_by_id): Likewise.
11345         (passes_by_id_size): Likewise.
11346         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11347         the new class pass_manager.
11348         (get_pass_for_id): Remove.
11349
11350 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11351
11352         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11353         configs.
11354
11355 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11356
11357         * arm.md (mulhi3): New expand pattern.
11358
11359 2013-07-30  Jan Hubicka  <jh@suse.cz>
11360             Martin Liska  <marxin.liska@gmail.com>
11361
11362         * profile.c (compute_value_histograms): Do not ICE when
11363         there is mismatch only on some counters.
11364
11365 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11366
11367         PR rtl-optimization/57637
11368         * function.c (move_insn_for_shrink_wrap): Also check the
11369         GEN set of the LIVE problem for the liveness analysis
11370         if it exists, otherwise give up.
11371
11372 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11373
11374         PR tree-optimization/57993
11375         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11376         replaced statement in the candidate table.
11377         (phi_add_costs): Return infinite cost when the hidden basis does
11378         not dominate all phis on which the candidate is dependent.
11379         (replace_one_candidate): Record replaced statement in the
11380         candidate table.
11381
11382 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11383
11384         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11385         (ashlv2si3): New expander.
11386         (*ashlv2si3_i): New define_insn_and_split.
11387         * predicates.md (float_operation): Allow patterns with three
11388         basic sub-patterns.
11389
11390         PR rtl-optimization/58021
11391         * mode-switching.c (create_pre_exit): Always split off preceding
11392         insns if we are not at the basic block head.
11393
11394 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11395
11396         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11397         (UCLIBC_DYNAMIC_LINKER): New macro.
11398         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11399         `-mnan=2008'.
11400         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11401         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11402         `-mnan=2008'.
11403         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11404         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11405         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11406         for SF and DF modes.  Use ieee_quad_format for TF mode.
11407         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11408         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11409         (mips_option_override): Handle `-mnan=legacy'.
11410         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11411         `-mabs=2008' and `-mnan=2008'.
11412         (OPTION_DEFAULT_SPECS): Add "nan" default.
11413         (ASM_SPEC): Handle `-mnan='.
11414         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11415         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11416         comment accordingly.
11417         (neg<mode>2): Likewise.
11418         * config/mips/mips.opt (mabs, mnan): New options.
11419         * doc/install.texi (Configuration): Document `--with-nan=' option.
11420         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11421         `-mnan=' options.
11422         (MIPS Options): Document them.
11423         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11424         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11425         * configure: Regenerate.
11426         * config.in: Regenerate.
11427
11428 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11429
11430         * config/i386/i386.md (float post-reload splitters): Do not check
11431         for subregs of SSE registers.
11432
11433 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11434             H.J. Lu  <hongjiu.lu@intel.com>
11435
11436         PR target/57954
11437         PR target/57988
11438         * config/i386/i386.md (post-reload splitter
11439         to avoid partial SSE reg dependency stalls): New pattern.
11440
11441 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11442
11443         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11444
11445 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11446
11447         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11448         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11449
11450 2013-07-26  Cary Coutant  <ccoutant@google.com>
11451
11452         * dwarf2out.c (die_checksum_ordered): Don't include template
11453         instantiations in signature.
11454         (is_template_parameter): New function.
11455         (is_template_instantiation): New function.
11456         (generate_skeleton_bottom_up): Don't include template instantiations
11457         in type unit DIE.
11458         (generate_skeleton): Likewise.
11459         (break_out_comdat_types): Move recursive call to break out nested
11460         types earlier.
11461         (prune_unused_types_mark_generic_parms_dies): Call
11462         is_template_parameter.
11463
11464 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11465
11466         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11467         uses vector registers.
11468         * config/aarch64/iterators.md: Add attributes rtn and vas.
11469
11470 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11471             Richard Earnshaw  <richard.earnshaw@arm.com>
11472
11473         * combine.c (simplify_comparison): Re-canonicalize operands
11474         where appropriate.
11475         * config/arm/arm.md (movcond_addsi): New splitter.
11476
11477 2013-07-25  Sterling Augustine  <saugustine@google.com>
11478
11479         * dwarf2out.c (size_of_pubnames): Move code to...
11480         (include_pubname_in_output): ...here.  New.
11481         (want_pubnames): Rearrange.
11482         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11483
11484 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11485
11486         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11487
11488 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11489
11490         PR target/38836
11491         * doc/extend.texi: Remove obsolete builtins. Fix
11492         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11493
11494 2013-07-25  Jan Hubicka  <jh@suse.cz>
11495
11496         * cgraph.c (release_function_body): Break out from ...
11497         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11498         and DECL_ARGUMENTS.
11499         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11500         old_tree in the map.
11501         (create_specialized_node): Update.
11502         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11503         into index.
11504         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11505         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11506         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11507         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11508         DECL_RESULT.
11509
11510 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11511
11512         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11513         addsi3_carryin_alt2_<optab>): Correct output template.
11514
11515 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11516
11517         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11518         Adjust for arm_restrict_it.
11519         Remove trailing whitespace.
11520
11521 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
11522
11523         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11524         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11525
11526         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11527
11528 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11529
11530         PR rtl-optimization/57960
11531         * lra-constraints.c (process_alt_operands): Use the right mode
11532         when checking strict_low.
11533
11534 2013-07-25  Jan Hubicka  <jh@suse.cz>
11535
11536         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11537         * cgraph.c (cgraph_remove_node): Do not release function body
11538         when in cgraph streaming.
11539         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11540         in other partitions are not considered reachable; fix handling of
11541         clones.
11542
11543 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11544
11545         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11546
11547 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11548
11549         PR target/19599
11550         PR target/57731
11551         PR target/57837
11552         * config/arm/arm.md ("*sibcall_insn): Replace use of
11553         Ss with US. Adjust output for v5 and v4t.
11554         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11555
11556         * config/arm/constraints.md ("Ss"): Rename to US.
11557
11558 2013-07-25  Terry Guo  <terry.guo@arm.com>
11559
11560         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11561         shift_add/shift_sub0/shift_sub1 RTXs.
11562
11563 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11564             Anton Blanchard  <anton@au1.ibm.com>
11565
11566         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11567         (altivec_vpks<VI_char>ss): Likewise.
11568         (altivec_vpks<VI_char>us): Likewise.
11569         (altivec_vpku<VI_char>us): Likewise.
11570         (altivec_vpku<VI_char>um): Likewise.
11571
11572 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11573
11574         Introduce context class.
11575
11576         * Makefile.in (CONTEXT_H): New.
11577         (OBJS): Add context.o.
11578         (toplev.o): Add CONTEXT_H to dependencies.
11579         (context.o): New.
11580
11581         * toplev.c (general_init): Create the singleton gcc::context instance.
11582
11583         * context.c: New.
11584
11585         * context.h: New.
11586
11587 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11588
11589         PR rtl-optimization/57968
11590         * mode-switching.c (create_pre_exit): Allow instructions that
11591         don't set a return register to need a non-exit mode.
11592
11593 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11594             Anton Blanchard  <anton@au1.ibm.com>
11595
11596         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11597         operands to vperm for little endian.
11598         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11599         of lvsl to create the control mask for a vperm for little endian.
11600
11601 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11602             Anton Blanchard  <anton@au1.ibm.com>
11603
11604         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11605         two operands for little-endian.
11606
11607 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11608
11609         * config/mips/mips.c (mips_case_values_threshold): New.
11610         (TARGET_CASE_VALUES_THRESHOLD): Define.
11611
11612 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11613             Anton Blanchard  <anton@au1.ibm.com>
11614
11615         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11616         selection of field for vector splat in little endian mode.
11617
11618 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11619
11620         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11621         expanders to rs6000.md.
11622         (ior<mode>3): Likewise.
11623         (and<mode>3): Likewise.
11624         (one_cmpl<mode>2): Likewise.
11625         (nor<mode>3): Likewise.
11626         (andc<mode>3): Likewise.
11627         (eqv<mode>3): Likewise.
11628         (nand<mode>3): Likewise.
11629         (orc<mode>3): Likewise.
11630
11631         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11632         declaration.
11633
11634         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11635         to split multi-word logical operations.
11636         (rs6000_split_logical_di): Likewise.
11637         (rs6000_split_logical): Likewise.
11638
11639         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11640         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11641         and allow TImode operations in 32-bit.
11642         (vsx_and<mode>3_64bit): Likewise.
11643         (vsx_ior<mode>3_32bit): Likewise.
11644         (vsx_ior<mode>3_64bit): Likewise.
11645         (vsx_xor<mode>3_32bit): Likewise.
11646         (vsx_xor<mode>3_64bit): Likewise.
11647         (vsx_one_cmpl<mode>2_32bit): Likewise.
11648         (vsx_one_cmpl<mode>2_64bit): Likewise.
11649         (vsx_nor<mode>3_32bit): Likewise.
11650         (vsx_nor<mode>3_64bit): Likewise.
11651         (vsx_andc<mode>3_32bit): Likewise.
11652         (vsx_andc<mode>3_64bit): Likewise.
11653         (vsx_eqv<mode>3_32bit): Likewise.
11654         (vsx_eqv<mode>3_64bit): Likewise.
11655         (vsx_nand<mode>3_32bit): Likewise.
11656         (vsx_nand<mode>3_64bit): Likewise.
11657         (vsx_orc<mode>3_32bit): Likewise.
11658         (vsx_orc<mode>3_64bit): Likewise.
11659
11660         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11661         logical types in GPRs.
11662
11663         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11664         logical insns to rs6000.md, and allow TImode operations in
11665         32-bit.
11666         (altivec_ior<mode>3): Likewise.
11667         (altivec_xor<mode>3): Likewise.
11668         (altivec_one_cmpl<mode>2): Likewise.
11669         (altivec_nor<mode>3): Likewise.
11670         (altivec_andc<mode>3): Likewise.
11671
11672         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11673         attributes for moving the 128-bit logical operations into
11674         rs6000.md.
11675         (BOOL_REGS_OUTPUT): Likewise.
11676         (BOOL_REGS_OP1): Likewise.
11677         (BOOL_REGS_OP2): Likewise.
11678         (BOOL_REGS_UNARY): Likewise.
11679         (BOOL_REGS_AND_CR0): Likewise.
11680         (one_cmpl<mode>2): Add support for DI logical operations on
11681         32-bit, splitting the operations to 32-bit.
11682         (anddi3): Likewise.
11683         (iordi3): Likewise.
11684         (xordi3): Likewise.
11685         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11686         changes to combine the 32/64-bit code, allow logical operations on
11687         TI mode in 32-bit, and to use similar match_operator patterns like
11688         scalar mode uses.  Combine the Altivec and VSX code for logical
11689         operations, and move it here.
11690         (ior<mode>3, 128-bit types): Likewise.
11691         (xor<mode>3, 128-bit types): Likewise.
11692         (one_cmpl<mode>3, 128-bit types): Likewise.
11693         (nor<mode>3, 128-bit types): Likewise.
11694         (andc<mode>3, 128-bit types): Likewise.
11695         (eqv<mode>3, 128-bit types): Likewise.
11696         (nand<mode>3, 128-bit types): Likewise.
11697         (orc<mode>3, 128-bit types): Likewise.
11698         (and<mode>3_internal): Likewise.
11699         (bool<mode>3_internal): Likewise.
11700         (boolc<mode>3_internal1): Likewise.
11701         (boolc<mode>3_internal2): Likewise.
11702         (boolcc<mode>3_internal1): Likewise.
11703         (boolcc<mode>3_internal2): Likewise.
11704         (eqv<mode>3_internal1): Likewise.
11705         (eqv<mode>3_internal2): Likewise.
11706         (one_cmpl1<mode>3_internal): Likewise.
11707
11708 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11709
11710         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11711         Rename flag_stack_usage to flag_stack_usage_info.
11712
11713 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11714
11715         * config/microblaze/sync.md: New file.
11716         * config/microblaze/microblaze.md: Include sync.md
11717         * config/microblaze/microblaze.c: Add print_operand 'y'.
11718         * config/microblaze/constraints.md: Add memory_contraint
11719         'Q' which is a single register.
11720
11721 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11722
11723         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11724
11725 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11726
11727         * reload.c (find_reloads): Exit loop once we find this operand
11728         cannot be reloaded somehow for this alternative.
11729
11730         * reload.c (find_reloads): Exit loop once we find a hard register.
11731
11732         * rtlanal.c (computed_jump_p): Exit loop once we find label
11733         reference is used.
11734
11735         * i386.c (ix86_pad_returns): Exit loop after setting replace.
11736
11737         * cfgloopmanip.c (remove_path): Exit loop after setting
11738         irred_invalidated.
11739
11740         * gensupport.c (subst_dup): Avoid loop if code is not
11741         MATCH_DUP nor MATCH_OP_DUP.
11742
11743 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
11744
11745         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
11746
11747 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11748
11749         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
11750         true for SP_REGNUM if mode == ptr_mode.
11751         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
11752         with value R0_REGNUM + 31.
11753
11754 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11755
11756         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
11757         pad pointer-typed argument downward.
11758
11759 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11760
11761         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
11762         and __ILP32__ when the ILP32 model is in use.
11763
11764 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11765
11766         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
11767         (aarch64_load_symref_appropriately): In the case of
11768         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
11769         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
11770         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
11771         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
11772         if 'mode' doesn't equal to 'ptr_mode'.
11773         (aarch64_output_mi_thunk): Add an assertion on the alignment of
11774         'vcall_offset'; change to call aarch64_emit_move differently depending
11775         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
11776         to calculate the upper bound of 'vcall_offset'.
11777         (aarch64_cannot_force_const_mem): Change to also return true if
11778         mode != ptr_mode.
11779         (aarch64_legitimize_reload_address): In the case of large
11780         displacements, add new local variable 'xmode' and an assertion
11781         based on it; change to use 'xmode' to generate the new rtx and
11782         reload.
11783         (aarch64_asm_trampoline_template): Change to generate the template
11784         differently depending on TARGET_ILP32 or not; change to use
11785         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
11786         (aarch64_trampoline_size): Removed.
11787         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
11788         and replace immediate literals with it.  Change to use 'ptr_mode'
11789         instead of 'DImode' and call convert_memory_address if the mode
11790         of 'fnaddr' doesn't equal to 'ptr_mode'.
11791         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
11792         to output symbol.
11793         (aarch64_elf_asm_destructor): Likewise.
11794         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
11795         on TARGET_ILP32 instead of aarch64_trampoline_size.
11796         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
11797         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
11798         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
11799         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
11800         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
11801         (storewb_pair<GPI:mode>_<P:mode>): ... this.
11802         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
11803         depending on the value of 'mode'.
11804         (add_losym_<mode>): New.
11805         (ldr_got_small_<mode>): New, based on ldr_got_small.
11806         (ldr_got_small): Remove.
11807         (ldr_got_small_sidi): New.
11808         * config/aarch64/iterators.md (P): New.
11809         (PTR): Change to 'ptr_mode' in the condition.
11810
11811 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11812
11813         * config.gcc (aarch64*-*-*): Support --with-abi.
11814         (aarch64*-*-elf): Support --with-multilib-list.
11815         (aarch64*-*-linux*): Likewise.
11816         (supported_defaults): Add abi to aarch64*-*-*.
11817         * configure.ac: Mention AArch64 for --with-multilib-list.
11818         * configure: Re-generated.
11819         * config/aarch64/biarchilp32.h: New file.
11820         * config/aarch64/biarchlp64.h: New file.
11821         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
11822         (ABI_SPEC): Ditto.
11823         (MULTILIB_DEFAULTS): Ditto.
11824         (DRIVER_SELF_SPECS): Ditto.
11825         (ASM_SPEC): Update to also substitute -mabi.
11826         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
11827         file whose name depends on -mabi= and -mbig-endian.
11828         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
11829         TARGET_ILP32.
11830         (POINTER_SIZE): New define.
11831         (POINTERS_EXTEND_UNSIGNED): Ditto.
11832         (enum aarch64_abi_type): New enumeration tag.
11833         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
11834         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
11835         (TARGET_ILP32): New define.
11836         * config/aarch64/aarch64.opt (mabi): New.
11837         (aarch64_abi): New.
11838         (ilp32, lp64): New values for -mabi.
11839         * config/aarch64/t-aarch64 (comma): New define.
11840         (MULTILIB_OPTIONS): Ditto.
11841         (MULTILIB_DIRNAMES): Ditto.
11842         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
11843         * doc/invoke.texi: Document -mabi for AArch64.
11844
11845 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
11846
11847         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
11848
11849 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11850             Anton Blanchard  <anton@au1.ibm.com>
11851
11852         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
11853         endianness when selecting field to splat.
11854
11855 2013-07-22  Eric Christopher  <echristo@gmail.com>
11856
11857         * dwarf2out.c (die_odr_checksum): New function to use
11858         CHECKSUM_ macros and ULEB128 for DIE tag.
11859         (generate_type_signature): Use.
11860
11861 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
11862
11863         * config.gcc (sparc*-*-*): Accept leon3 processor.
11864         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
11865         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
11866         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
11867         * config/sparc/sparc.opt (enum processor_type): Add leon3.
11868         (mfix-ut699): Adjust comment.
11869         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
11870         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
11871         (CPP_CPU_SPEC): Likewise.
11872         (ASM_CPU_SPEC): Likewise.
11873         * config/sparc/sparc.c (leon3_cost): New constant.
11874         (sparc_option_override): Add leon3 support.
11875         (mem_ref): New function.
11876         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
11877         (sparc_do_work_around_errata): Look into the instruction in the delay
11878         slot and adjust accordingly.  Add fix for the data cache nullify issues
11879         of the UT699.  Change insertion position for the NOP.
11880         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
11881         (leon3_load): New reservation.
11882         (leon_store): Bump latency to 2.
11883         (grfpu): New automaton.
11884         (grfpu_alu): New unit.
11885         (grfpu_ds): Likewise.
11886         (leon_fp_alu): Adjust.
11887         (leon_fp_mult): Delete.
11888         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
11889         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
11890         * config/sparc/sparc.md (cpu): Add leon3.
11891         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
11892         (swapsi): Likewise.
11893         (atomic_test_and_set): Likewise.
11894         (ldstub): Likewise.
11895
11896 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
11897
11898         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
11899         default for R5900 targets.
11900         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
11901         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
11902         * config/mips/mips.c (mips_option_override): Report an error for
11903         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
11904         for -march=r5900 -mhard-float.
11905
11906 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11907
11908         * df-problems.c (can_move_insns_across): Exit loop once we
11909         find a non-fixed, non-global register.
11910
11911         * ipa-pure-const.c (propagate_nothrow): Exit loop after
11912         setting can_throw.
11913
11914         * omega.c (omega_eliminate_red): Break after setting red_found.
11915         (omega_problem_has_red_equations): Similarly after setting found.
11916         (omega_query_variable): Similarly after setting coupled.
11917
11918 2013-07-22  Marek Polacek  <polacek@redhat.com>
11919
11920         * gimplify.c: Don't include gimple.h twice.
11921
11922 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11923
11924         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
11925         instead of TARGET_THUMB1.
11926         (Pz): New constraint.
11927         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
11928         encodings.
11929         (compare_negsi_si): Likewise.
11930         (compare_addsi2_op0): Likewise.
11931         (compare_addsi2_op1): Likewise.
11932         (addsi3_carryin_<optab>): Likewise.
11933         (addsi3_carryin_alt2_<optab>): Likewise.
11934         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
11935         for arm_restrict_it.
11936         (subsi3_carryin): Likewise.
11937         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
11938         (minmax_arithsi): Disable for arm_restrict_it.
11939         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
11940         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
11941         (satsi_<SAT:code>_shift): Likewise.
11942         (arm_shiftsi3): Add alternative for 16-bit encoding.
11943         (arm32_movhf): Disable for arm_restrict_it.
11944         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
11945         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
11946
11947 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
11948
11949         * config/arm/arm.md (attribute "insn"): Delete.
11950         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
11951         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
11952         (not_shiftsi): Update for attribute change.
11953         (not_shiftsi_compare0): Likewise.
11954         (not_shiftsi_compare0_scratch): Likewise.
11955         (arm_one_cmplsi2): Likewise.
11956         (thumb1_one_cmplsi2): Likewise.
11957         (notsi_compare0): Likewise.
11958         (notsi_compare0_scratch): Likewise.
11959         (thumb1_movdi_insn): Likewise.
11960         (arm_movsi_insn): Likewise.
11961         (movhi_insn_arch4): Likewise.
11962         (movhi_bytes): Likewise.
11963         (arm_movqi_insn): Likewise.
11964         (thumb1_movqi_insn): Likewise.
11965         (arm32_movhf): Likewise.
11966         (thumb1_movhf): Likewise.
11967         (arm_movsf_soft_insn): Likewise.
11968         (thumb1_movsf_insn): Likewise.
11969         (thumb_movdf_insn): Likewise.
11970         (movsicc_insn): Likewise.
11971         (movsfcc_soft_insn): Likewise.
11972         (and_scc): Likewise.
11973         (cond_move): Likewise.
11974         (if_move_not): Likewise.
11975         (if_not_move): Likewise.
11976         (if_shift_move): Likewise.
11977         (if_move_shift): Likewise.
11978         (if_shift_shift): Likewise.
11979         (if_not_arith): Likewise.
11980         (if_arith_not): Likewise.
11981         (cond_move_not): Likewise.
11982         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
11983         (neon_mov<mode>): Likewise.
11984         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
11985         (thumb2_movsi_vfp): Likewise.
11986         (movsf_vfp): Likewise.
11987         (thumb2_movsf_vfp): Likewise.
11988         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
11989         change.
11990         (cortexa7_older_only): Likewise.
11991         (cortexa7_younger): Likewise.
11992         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
11993         (1020alu_shift_op): Likewise.
11994         (1020alu_shift_reg_op): Likewise.
11995         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
11996         (alu_shift_op): Likewise.
11997         (alu_shift_reg_op): Likewise.
11998         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
11999         (11_alu_shift_op): Likewise.
12000         (11_alu_shift_reg_op): Likewise.
12001         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12002         (9_alu_shift_reg_op): Likewise.
12003         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12004         change.
12005         (cortex_a15_alu_shift): Likewise.
12006         (cortex_a15_alu_shift_reg): Likewise.
12007         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12008         (cortex_a5_alu_shift): Likewise.
12009         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12010         change.
12011         (cortex_a53_alu_shift): Likewise.
12012         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12013         change.
12014         (cortex_a7_alu_reg): Likewise.
12015         (cortex_a7_alu_shift): Likewise.
12016         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12017         (cortex_a8_alu_shift): Likewise.
12018         (cortex_a8_alu_shift_reg): Likewise.
12019         (cortex_a8_mov): Likewise.
12020         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12021         (cortex_a9_dp_shift): Likewise.
12022         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12023         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12024         (cortex_r4_mov): Likewise.
12025         (cortex_r4_alu_shift): Likewise.
12026         (cortex_r4_alu_shift_reg): Likewise.
12027         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12028         (526_alu_shift_op): Likewise.
12029         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12030         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12031         (626te_alu_shift_op): Likewise.
12032         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12033         (726te_alu_op): Likewise.
12034         (726te_alu_shift_op): Likewise.
12035         (726te_alu_shift_reg_op): Likewise.
12036         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12037         (mp626_alu_shift_op): Likewise.
12038         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12039         (pj4_alu_e1_conds): Likewise.
12040         (pj4_alu): Likewise.
12041         (pj4_alu_conds): Likewise.
12042         (pj4_shift): Likewise.
12043         (pj4_shift_conds): Likewise.
12044         (pj4_alu_shift): Likewise.
12045         (pj4_alu_shift_conds): Likewise.
12046
12047 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12048
12049         * config/arm/predicates.md (shiftable_operator_strict_it):
12050         New predicate.
12051         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12052         Disable cond_exec version for arm_restrict_it.
12053         (thumb2_smaxsi3): Convert to generate cond_exec.
12054         (thumb2_sminsi3): Likewise.
12055         (thumb32_umaxsi3): Likewise.
12056         (thumb2_uminsi3): Likewise.
12057         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12058         (thumb2_neg_abssi2): Likewise.
12059         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12060         (thumb2_movsicc_insn): Adjust alternatives.
12061         (thumb2_mov_negscc): Disable for arm_restrict_it.
12062         (thumb2_mov_negscc_strict_it): New pattern.
12063         (thumb2_mov_notscc_strict_it): New pattern.
12064         (thumb2_mov_notscc): Disable for arm_restrict_it.
12065         (thumb2_ior_scc): Likewise.
12066         (thumb2_ior_scc_strict_it): New pattern.
12067         (thumb2_cond_move): Adjust for arm_restrict_it.
12068         (thumb2_cond_arith): Disable for arm_restrict_it.
12069         (thumb2_cond_arith_strict_it): New pattern.
12070         (thumb2_cond_sub): Adjust for arm_restrict_it.
12071         (thumb2_movcond): Likewise.
12072         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12073         (thumb2_zero_extendhisi2_v6): Likewise.
12074         (thumb2_zero_extendqisi2_v6): Likewise.
12075         (orsi_notsi_si): Likewise.
12076         (orsi_not_shiftsi_si): Likewise.
12077
12078 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12079
12080         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12081         instruction sequence is 1 byte shorter.
12082
12083 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12084
12085         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12086         it is not needed after split.
12087
12088 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12089
12090         PR target/51784
12091         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12092         second label for nonlocal goto receivers. Don't output pic base labels
12093         unless we're producing PIC; mark that action unreachable().
12094         (ix86_save_reg): If the function contains a nonlocal label, save the
12095         PIC base reg.
12096         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12097         * config/darwin.c (emitted_pic_label_num): New GTY.
12098         (update_pic_label_number_if_needed): New.
12099         (machopic_output_function_base_name): Adjust for nonlocal receiver
12100         case.
12101         (machopic_should_output_picbase_label): New.
12102         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12103         (nonlocal_goto_receiver): New insn and split.
12104
12105 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12106
12107         * config/aarch64/aarch64-builtins.c
12108         (aarch64_fold_builtin): Fold abs in all modes.
12109         * config/aarch64/aarch64-simd-builtins.def
12110         (abs): Enable for all modes.
12111         * config/aarch64/arm_neon.h
12112         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12113         (vabs_f64): Add missing intrinsic.
12114
12115 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12116
12117         * config/aarch64/arm_neon.h (vabs_s64): New function
12118
12119 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12120
12121         PR target/57516
12122         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12123         * config/avr/avr.md (adjust_len): Add `round'.
12124         * config/avr/avr-protos.h (avr_out_round): New prototype.
12125         (avr_out_plus): Add `out_label' argument.
12126         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12127         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12128         Handle the case where `insn' is just a pattern.
12129         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12130         (avr_out_round): New function.
12131         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12132
12133 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12134
12135         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12136         Add check for flag_stack_usage to handle -fstack-usage support
12137
12138 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12139
12140         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12141         interaction for new Power8 flags and VSX.
12142
12143 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12144
12145         PR middle-end/57698
12146         * tree-inline.c (expand_call_inline): Emit errors during
12147         early_inlining only if optimization is not turned on.
12148
12149 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12150
12151         * passes.def: New.
12152
12153         * passes.c (init_optimization_passes): Move the construction of
12154         the pass hierarchy into a new passes.def file.
12155
12156         * Makefile.in (passes.o): Add dependency on passes.def.
12157
12158 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12159
12160         * passes.c (init_optimization_passes): Introduce macros for
12161         constructing the tree of passes (INSERT_PASSES_AFTER,
12162         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12163         TERMINATE_PASS_LIST).
12164
12165 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12166             Wei Mi  <wmi@google.com>
12167
12168         PR rtl-optimization/57878
12169         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12170         top.
12171         (reload_pseudo_compare_func): Check nregs first for reload
12172         pseudos.
12173
12174 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12175
12176         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12177
12178 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12179
12180         * read-rtl.c (validate_const_int): Once an invalid character is
12181         seen, quit the loop.
12182
12183         * gengtype.c (write_roots): Similarly once we find the "deletable"
12184         or "if_marked" option.
12185
12186 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12187
12188         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12189         "xtab" and "sat".  Move value "clz" from here to ...
12190         (attriubte "type"): ... here.
12191         (satsi_<SAT:code>): Delete "insn" attribute.
12192         (satsi_<SAT:code>_shift): Likewise.
12193         (arm_zero_extendqisi2addsi): Likewise.
12194         (arm_extendqisi2addsi): Likewise.
12195         (clzsi2): Update for attribute changes.
12196         (rbitsi2): Likewise.
12197         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12198         attribute.
12199         (arm_usatsihi): Likewise.
12200         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12201
12202 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12203
12204         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12205         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12206         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12207         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12208         in alphabetical order.
12209         (attribute "core_cycles"): Update for attribute changes.
12210         (arm_addsi3): Likewise.
12211         (addsi3_compare0): Likewise.
12212         (addsi3_compare0_scratch): Likewise.
12213         (addsi3_compare_op1): Likewise.
12214         (addsi3_compare_op2): Likewise.
12215         (compare_addsi2_op0): Likewise.
12216         (compare_addsi2_op1): Likewise.
12217         (addsi3_carryin_shift_<optab>): Likewise.
12218         (subsi3_carryin_shift): Likewise.
12219         (rsbsi3_carryin_shift): Likewise.
12220         (arm_subsi3_insn): Likewise.
12221         (subsi3_compare0): Likewise.
12222         (subsi3_compare): Likewise.
12223         (arm_andsi3_insn): Likewise.
12224         (thumb1_andsi3_insn): Likewise.
12225         (andsi3_compare0): Likewise.
12226         (andsi3_compare0_scratch): Likewise.
12227         (zeroextractsi_compare0_scratch
12228         (andsi_not_shiftsi_si): Likewise.
12229         (iorsi3_insn): Likewise.
12230         (iorsi3_compare0): Likewise.
12231         (iorsi3_compare0_scratch): Likewise.
12232         (arm_xorsi3): Likewise.
12233         (thumb1_xorsi3_insn): Likewise.
12234         (xorsi3_compare0): Likewise.
12235         (xorsi3_compare0_scratch): Likewise.
12236         (satsi_<SAT:code>_shift): Likewise.
12237         (rrx): Likewise.
12238         (arm_shiftsi3): Likewise.
12239         (shiftsi3_compare0): Likewise.
12240         (not_shiftsi): Likewise.
12241         (not_shiftsi_compare0): Likewise.
12242         (not_shiftsi_compare0_scratch): Likewise.
12243         (arm_one_cmplsi2): Likewise.
12244         (thumb_one_complsi2): Likewise.
12245         (notsi_compare0): Likewise.
12246         (notsi_compare0_scratch): Likewise.
12247         (thumb1_zero_extendhisi2): Likewise.
12248         (arm_zero_extendhisi2): Likewise.
12249         (arm_zero_extendhisi2_v6): Likewise.
12250         (arm_zero_extendhisi2addsi): Likewise.
12251         (thumb1_zero_extendqisi2): Likewise.
12252         (thumb1_zero_extendqisi2_v6): Likewise.
12253         (arm_zero_extendqisi2): Likewise.
12254         (arm_zero_extendqisi2_v6): Likewise.
12255         (arm_zero_extendqisi2addsi): Likewise.
12256         (thumb1_extendhisi2): Likewise.
12257         (arm_extendhisi2): Likewise.
12258         (arm_extendhisi2_v6): Likewise.
12259         (arm_extendqisi): Likewise.
12260         (arm_extendqisi_v6): Likewise.
12261         (arm_extendqisi2addsi): Likewise.
12262         (thumb1_extendqisi2): Likewise.
12263         (thumb1_movdi_insn): Likewise.
12264         (arm_movsi_insn): Likewise.
12265         (movsi_compare0): Likewise.
12266         (movhi_insn_arch4): Likewise.
12267         (movhi_bytes): Likewise.
12268         (arm_movqi_insn): Likewise.
12269         (thumb1_movqi_insn): Likewise.
12270         (arm32_movhf): Likewise.
12271         (thumb1_movhf): Likewise.
12272         (arm_movsf_soft_insn): Likewise.
12273         (thumb1_movsf_insn): Likewise.
12274         (movdf_soft_insn): Likewise.
12275         (thumb_movdf_insn): Likewise.
12276         (arm_cmpsi_insn): Likewise.
12277         (cmpsi_shiftsi): Likewise.
12278         (cmpsi_shiftsi_swp): Likewise.
12279         (arm_cmpsi_negshiftsi_si): Likewise.
12280         (movsicc_insn): Likewise.
12281         (movsfcc_soft_insn): Likewise.
12282         (arith_shiftsi): Likewise.
12283         (arith_shiftsi_compare0
12284         (arith_shiftsi_compare0_scratch
12285         (sub_shiftsi): Likewise.
12286         (sub_shiftsi_compare0
12287         (sub_shiftsi_compare0_scratch
12288         (and_scc): Likewise.
12289         (cond_move): Likewise.
12290         (if_plus_move): Likewise.
12291         (if_move_plus): Likewise.
12292         (if_move_not): Likewise.
12293         (if_not_move): Likewise.
12294         (if_shift_move): Likewise.
12295         (if_move_shift): Likewise.
12296         (if_shift_shift): Likewise.
12297         (if_not_arith): Likewise.
12298         (if_arith_not): Likewise.
12299         (cond_move_not): Likewise.
12300         (thumb1_ashlsi3): Set type attribute.
12301         (thumb1_ashrsi3): Likewise.
12302         (thumb1_lshrsi3): Likewise.
12303         (thumb1_rotrsi3): Likewise.
12304         (shiftsi3_compare0_scratch): Likewise.
12305         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12306         (neon_mov<mode>): Likewise.
12307         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12308         attribute changes.
12309         (thumb2_movsi_insn): Likewise.
12310         (thumb2_cmpsi_neg_shiftsi): Likewise.
12311         (thumb2_extendqisi_v6): Likewise.
12312         (thumb2_zero_extendhisi2_v6): Likewise.
12313         (thumb2_zero_extendqisi2_v6): Likewise.
12314         (thumb2_shiftsi3_short): Likewise.
12315         (thumb2_addsi3_compare0_scratch): Likewise.
12316         (orsi_not_shiftsi_si): Likewise.
12317         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12318         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12319         changes.
12320         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12321         (1020alu_shift_op): Likewise.
12322         (1020alu_shift_reg_op): Likewise.
12323         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12324         (alu_shift_op): Likewise.
12325         (alu_shift_reg_op): Likewise.
12326         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12327         (11_alu_shift_op): Likewise.
12328         (11_alu_shift_reg_op): Likewise.
12329         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12330         (9_alu_shift_reg_op): Likewise.
12331         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12332         changes.
12333         (cortex_a15_alu_shift): Likewise.
12334         (cortex_a15_alu_shift_reg): Likewise.
12335         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12336         changes.
12337         (cortex_a5_alu_shift): Likewise.
12338         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12339         changes.
12340         (cortex_a53_alu_shift): Likewise.
12341         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12342         changes.
12343         (cortex_a7_alu_reg): Likewise.
12344         (cortex_a7_alu_shift): Likewise.
12345         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12346         changes.
12347         (cortex_a8_alu_shift): Likewise.
12348         (cortex_a8_alu_shift_reg): Likewise.
12349         (cortex_a8_mov): Likewise.
12350         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12351         (cortex_a9_dp_shift): Likewise.
12352         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12353         changes.
12354         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12355         changes.
12356         (cortex_r4_mov): Likewise.
12357         (cortex_r4_alu_shift): Likewise.
12358         (cortex_r4_alu_shift_reg): Likewise.
12359         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12360         (526_alu_shift_op): Likewise.
12361         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12362         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12363         (626te_alu_shift_op): Likewise.
12364         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12365         (726te_alu_op): Likewise.
12366         (726te_alu_shift_op): Likewise.
12367         (726te_alu_shift_reg_op): Likewise.
12368         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12369         (mp626_alu_shift_op): Likewise.
12370         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12371         (pj4_alu_e1_conds): Likewise.
12372         (pj4_alu): Likewise.
12373         (pj4_alu_conds): Likewise.
12374         (pj4_shift): Likewise.
12375         (pj4_shift_conds): Likewise.
12376         (pj4_alu_shift): Likewise.
12377         (pj4_alu_shift_conds): Likewise.
12378         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12379         changes.
12380         (cortexa7_older_only): Likewise.
12381         (cortexa7_younger): Likewise.
12382
12383 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12384
12385         * ipa-pure-const.c (generate_summary): Rename to...
12386         (pure_const_generate_summary): ... this.
12387
12388 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12389
12390         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12391
12392 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12393
12394         PR target/57909
12395         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12396         usage in HI mode.
12397
12398 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12399
12400         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12401         enabled without -march=zEC12.
12402         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12403         flags to be set.
12404
12405 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12406
12407         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12408         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12409         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12410         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12411         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12412         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12413         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12414         ISA_HAS_FP4.
12415         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12416         and ISA_HAS_NMADD3_NMSUB3.
12417         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12418         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12419         (nmsub4<mode>, nmsub3<mode>): Likewise.
12420         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12421
12422 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12423
12424         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12425         TARGET_MIPS5400 checking.
12426
12427 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12428             Peter Bergner  <bergner@vnet.ibm.com>
12429
12430         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12431         registers in the comment.
12432         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12433         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12434         rather than FIRST_PSEUDO_REGISTERS.
12435
12436 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12437
12438         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12439         enable extra ISA flags with TARGET_HTM.
12440
12441 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12442
12443         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12444         Fix comment typos.
12445
12446 2013-07-15  Cong Hou  <congh@google.com>
12447
12448         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12449         in compare function for sorting.
12450
12451 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12452
12453         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12454         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12455         * config/rs6000/rs6000.opt: Add -mhtm option.
12456         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12457         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12458         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12459         __HTM__ if the HTM instructions are available.
12460         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12461         htm_spr_reg_operand): New define_predicates.
12462         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12463         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12464         Include htm.md.
12465         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12466         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12467         HTM builtin functions.
12468         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12469         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12470         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12471         (rs6000_builtin_mask_calculate): Likewise.
12472         (rs6000_option_override_internal): Likewise.
12473         (bdesc_htm): Add new HTM builtin support.
12474         (htm_spr_num): New function.
12475         (htm_spr_regno): Likewise.
12476         (rs6000_htm_spr_icode): Likewise.
12477         (htm_expand_builtin): Likewise.
12478         (htm_init_builtins): Likewise.
12479         (rs6000_expand_builtin): Add support for HTM builtin functions.
12480         (rs6000_init_builtins): Likewise.
12481         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12482         option.
12483         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12484         (TARGET_HTM, MASK_HTM): Define macros.
12485         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12486         (FIXED_REGISTERS): Likewise.
12487         (CALL_USED_REGISTERS): Likewise.
12488         (CALL_REALLY_USED_REGISTERS): Likewise.
12489         (REG_ALLOC_ORDER): Likewise.
12490         (enum reg_class): Likewise.
12491         (REG_CLASS_NAMES): Likewise.
12492         (REG_CLASS_CONTENTS): Likewise.
12493         (REGISTER_NAMES): Likewise.
12494         (ADDITIONAL_REGISTER_NAMES): Likewise.
12495         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12496         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12497         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12498         * config/rs6000/htm.md: New file.
12499         * config/rs6000/htmintrin.h: New file.
12500         * config/rs6000/htmxlintrin.h: New file.
12501
12502 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12503
12504         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12505         Define SYMBOL_TINY_GOT, update comment.
12506         * config/aarch64/aarch64.c
12507         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12508         (aarch64_expand_mov_immediate): Likewise.
12509         (aarch64_print_operand): Likewise.
12510         (aarch64_classify_symbol): Likewise.
12511         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12512         (ldr_got_tiny): Define.
12513
12514 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12515
12516         PR tree-optimization/54094
12517         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12518           scheduling dimension for the parallelism check from the polyhedral
12519           information in the AST.
12520         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12521           in 2D + 1 form.
12522
12523 2013-07-13  Jason Merrill  <jason@redhat.com>
12524
12525         * print-tree.c (debug_vec_tree): Use debug_raw.
12526         (debug_raw (vec<tree, va_gc> &)): New.
12527         (debug_raw (vec<tree, va_gc> *)): New.
12528         * tree.h: Declare them.
12529
12530 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12531
12532         * ifcvt.c (ifcvt_after_combine): New static variable.
12533         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12534         for size.
12535         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12536         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12537         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12538
12539 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12540
12541         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12542
12543 2013-07-12  Michael Matz  <matz@suse.de>
12544
12545         PR middle-end/55771
12546         * convert.c (convert_to_real): Reject non-float inner types.
12547
12548 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12549
12550         * config/aarch64/aarch64-protos.h
12551         (aarch64_simd_immediate_valid_for_move): Remove.
12552         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12553         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12554         cases.
12555         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12556
12557 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12558
12559         * config/mips/mips.c (mips_conditional_register_usage): Do not
12560         use t[0-7] registers in MIPS16 mode when optimizing for size.
12561
12562 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12563
12564         * config/i386/i386.c (dispatch_function_versions): Fix array
12565         indexing of function_version_info to match actual_versions.
12566
12567 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12568
12569         * vec.h (struct va_gc): Move release out-of-line.
12570         (va_gc::release): Call ggc_free on released vec.
12571
12572 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12573
12574         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12575         Require GOT register as additional operand in UNSPEC.
12576         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12577         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12578         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12579         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12580         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12581         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12582         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12583
12584 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12585
12586         PR target/57631
12587         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12588         name seen by assembler/linker rather if available.
12589
12590 2013-07-11  Andreas Schwab  <schwab@suse.de>
12591
12592         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12593
12594 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12595
12596         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12597
12598 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12599
12600         * doc/tm.texi.in: Move hook documentation to ....
12601         * target.def: ... here.
12602
12603         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12604         text on @hook line.
12605         * doc/tm.texi: Regenerate.
12606
12607 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12608
12609         PR c++/57869
12610         * doc/invoke.texi: Document -Wconditionally-supported.
12611
12612 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12613
12614         PR target/57844
12615         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12616         of my_fp.
12617
12618 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12619
12620         PR target/57506
12621         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12622         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12623         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12624         Remove duplicate devices.
12625         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12626         * config/avr/t-multilib: Regenerate.
12627         * config/avr/avr-tables.opt: Regenerate.
12628         * doc/avr-mmcu.texi: Regenerate.
12629
12630 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12631
12632         PR target/56987
12633         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12634
12635 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12636
12637         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12638         the cost of MULT when optimizing for size.
12639
12640 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12641
12642         * config/cr16/cr16-protos.h: Don't include target.h.
12643
12644 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12645
12646         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12647         adjust register size for TDmode and TFmode for VSX registers.
12648
12649 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12650
12651         PR target/56892
12652         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12653         hook_bool_const_tree_true.
12654
12655 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12656
12657         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12658         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12659         * config/s390/s390.md: Define FPR*_REGNUM constants.
12660         Fix FPR2_REGNUM constant (18 -> 17).
12661         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12662         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12663         ("trunc<BFP:mode><DFP_ALL:mode>2")
12664         ("trunc<DFP_ALL:mode><BFP:mode>2")
12665         ("*extend<BFP:mode><DFP_ALL:mode>2")
12666         ("*extend<DFP_ALL:mode><BFP:mode>2")
12667         ("extend<BFP:mode><DFP_ALL:mode>2")
12668         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12669         FPR4_REGNUM.
12670
12671 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12672
12673         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12674
12675 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12676
12677         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12678         and cfun_fpr_bit_p to cfun_fpr_save_p.
12679         (s390_frame_area, s390_register_info, s390_frame_info)
12680         (s390_emit_prologue, s390_emit_epilogue)
12681         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12682         register numbers.
12683         * config/s390/s390.h: Define *_REGNUM macros for floating point
12684         register numbers.
12685
12686 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12687
12688         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12689
12690 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12691
12692         PR rtl-optimization/57786
12693         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12694         and break out of the loop when it is set to false.
12695
12696 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12697
12698         PR target/57819
12699         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12700         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12701         (const_int 63)) 0)).
12702         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12703         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12704         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12705
12706         PR rtl-optimization/57829
12707         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12708         mask bits outside of mode are just sign-extension from mode to HWI.
12709
12710 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12711
12712         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12713         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12714         adjust_address instead of change_address to keep info about alignment.
12715         (emit_strmov): Remove.
12716         (emit_memmov): New function.
12717         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12718         (expand_movmem_epilogue): Likewise and return updated rtx for
12719         destination.
12720         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12721         destination and source.
12722         (decide_alignment): Refactor, handle vector_loop.
12723         (ix86_expand_movmem): Likewise.
12724         (ix86_expand_setmem): Likewise.
12725         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12726
12727 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
12728
12729         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
12730         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
12731
12732 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12733
12734         * config/i386/sse.md (sse_movlhps): Change alternative 3
12735         of operand 2 to "m".
12736
12737 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12738
12739         PR target/57807
12740         * config/i386/sse.md (iptr): New mode attribute.
12741         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
12742         (<sse>_vm<plusminus_insn><mode>3): Ditto.
12743         (<sse>_vmmul<mode>3): Ditto.
12744         (<sse>_vmdiv<mode>3): Ditto.
12745         (sse_vmrcpv4sf2): Ditto.
12746         (<sse>_vmsqrt<mode>2): Ditto.
12747         (sse_vmrsqrtv4sf2): Ditto.
12748         (<sse>_vm<code><mode>3): Ditto.
12749         (avx_vmcmp<mode>3): Ditto.
12750         (<sse>_vmmaskcmp<mode>3): Ditto.
12751         (<sse>_comi): Ditto.
12752         (<sse>_ucomi): Ditto.
12753         (*xop_vmfrcz_<mode>): Ditto.
12754         (*fmai_fmadd_<mode>): Ditto.
12755         (*fmai_fmsub_<mode>): Ditto.
12756         (*fmai_fnmadd_<mode>): Ditto.
12757         (*fmai_fnmsub_<mode>): Ditto.
12758         (*fma4i_vmfmadd_<mode>): Ditto.
12759         (*fma4i_vmfmsub_<mode>): Ditto.
12760         (*fma4i_vmfnmadd_<mode>): Ditto.
12761         (*fma4i_vmfnmsub_<mode>): Ditto.
12762         (*xop_vmfrcz_<mode>): Ditto.
12763         (sse_cvtps2pi): Ditto.
12764         (sse_cvttps2pi): Ditto.
12765         (sse_cvtss2si): Ditto.
12766         (sse_cvtss2si_2): Ditto.
12767         (sse_cvtss2siq_2): Ditto.
12768         (sse_cvttss2si): Ditto.
12769         (sse_cvttss2siq): Ditto.
12770         (sse_cvtsd2si): Ditto.
12771         (sse_cvtsd2si_2): Ditto.
12772         (sse_cvtsd2siq_2): Ditto.
12773         (sse_cvttsd2si): Ditto.
12774         (sse_cvttsd2siq): Ditto.
12775         (sse_cvtsd2ss): Ditto.
12776         (sse_cvtss2sd): Ditto.
12777         (avx2_pbroadcast<mode>): Ditto.
12778         (avx2_pbroadcast<mode>_1): Ditto.
12779         (*avx_vperm_broadcast_v4sf): Ditto.
12780
12781         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
12782         (sse_movlhps): Ditto.
12783         (sse_storehps): Ditto.
12784         (sse_loadhps): Ditto.
12785         (sse_storelps): Ditto.
12786         (sse_loadlps): Ditto.
12787         (*vec_concatv4sf): Ditto.
12788         (*vec_interleave_highv2df): Ditto.
12789         (*vec_interleave_lowv2df): Ditto.
12790         (*vec_extractv2df_1_sse): Ditto.
12791         (*vec_extractv2df_0_sse): Ditto.
12792         (sse2_storelpd): Ditto.
12793         (sse2_loadlpd): Ditto.
12794         (sse2_movsd): Ditto.
12795         (*vec_concatv4si): Ditto.
12796         (vec_concatv2di): Ditto.
12797
12798         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
12799         for Intel asm dialect.
12800         (mmx_punpcklwd): Ditto.
12801         (mmx_punpckldq): Ditto.
12802
12803         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
12804         for intel assembler dialect.
12805
12806 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
12807
12808         PR target/29776
12809         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
12810         for BUILT_IN_C{LZ,LRSB}*.
12811         * tree.h (CASE_INT_FN): Add FN##IMAX case.
12812         * tree-vrp.c (extract_range_basic): Handle
12813         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
12814         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
12815         fall thru to code calling set_value*.
12816         * builtins.c (expand_builtin): Remove *IMAX cases.
12817         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
12818         if width is bigger than 2*HWI.
12819
12820 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
12821
12822         PR rtl-optimization/55342
12823         * lra-int.h (lra_subreg_reload_pseudos): New.
12824         * lra.c: Add undoing optional reloads to the block diagram.
12825         (lra_subreg_reload_pseudos): New.
12826         (lra_optional_reload_pseudos): Change comments.
12827         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
12828         lra_optional_reload_pseudos after undo transformations.
12829         * lra-assigns.c (pseudo_prefix_title): New.
12830         (lra_setup_reg_renumber): Use it.
12831         (spill_for): Ditto.  Check subreg reload pseudos too.
12832         (assign_by_spills): Consider subreg reload pseudos too.
12833         * lra-constraints.c (simplify_operand_subreg): Use
12834         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
12835         (curr_insn_transform): Recognize and do optional reloads.
12836         (undo_optional_reloads): New.
12837         (lra_undo_inheritance): Call undo_optional_reloads.
12838
12839 2013-07-05  Thomas Quinot  <quinot@adacore.com>
12840
12841         * tree-complex.c (expand_complex_operations_1): Fix typo.
12842
12843 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
12844
12845         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
12846         (tune_params): New member 'const vec_costs'.
12847         * config/aarch64/aarch64.c (generic_vector_cost): New.
12848         (generic_tunings): New member 'generic_vector_cost'.
12849         (aarch64_builtin_vectorization_cost): New.
12850         (aarch64_add_stmt_cost): New.
12851         (TARGET_VECTORIZE_ADD_STMT_COST): New.
12852         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
12853
12854 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
12855
12856         PR target/57777
12857         * config/i386/predicates.md (vsib_address_operand): Disallow
12858         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
12859
12860 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
12861
12862         PR middle-end/55030
12863         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
12864         expand_builtin_setjmp_receiver.
12865         (expand_label): Adjust, call expand_builtin_setjmp_receiver
12866         with NULL for the label parameter.
12867         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
12868         the frame-pointer.  Adjust comments.
12869         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
12870         only if LABEL is non-NULL.
12871
12872 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
12873
12874         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
12875         (ARM_ABI_AAPCS64): Ditto.
12876         (arm_abi): Ditto.
12877         (ARM_DEFAULT_ABI): Ditto.
12878
12879 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
12880
12881         * config/aarch64/aarch64-builtins.c
12882         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
12883         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
12884         (st1): Likewise.
12885         * config/aarch64/aarch64-simd.md
12886         (aarch64_ld1<VALL:mode>): New.
12887         (aarch64_st1<VALL:mode>): Likewise.
12888         * config/aarch64/arm_neon.h
12889         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
12890
12891 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
12892
12893         * config/i386/i386.c (gate_insert_vzeroupper): Check if
12894         target ISA is AVX.
12895         (ix86_option_override_internal):Turn on all -mavx target flags by
12896         default as they are dependent on AVX anyway.
12897
12898 2013-07-02  Cary Coutant  <ccoutant@google.com>
12899
12900         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
12901         deterministic hash.
12902         (loc_checksum_ordered): Likewise.
12903         (hash_loc_operands): Remove inline keyword.
12904
12905 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
12906
12907         PR tree-optimization/57741
12908         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
12909         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
12910         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
12911         Allow REAL_CST step_exprs if flag_associative_math.
12912         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
12913
12914 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
12915
12916         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
12917
12918 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
12919
12920         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
12921
12922 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12923
12924         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
12925         encoding.
12926         (iorsi3_insn): Likewise.
12927         (arm_xorsi3): Likewise.
12928
12929 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
12930
12931         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
12932         to ...
12933         (attribute "type"): ... here, and prefix with "wmmx_".
12934         (attribute "core_cycles"): Update for attribute changes.
12935         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
12936         (tbcstv4hi): Likewise.
12937         (tbcstv2si): Likewise.
12938         (iwmmxt_iordi3): Likewise.
12939         (iwmmxt_xordi3): Likewise.
12940         (iwmmxt_anddi3): Likewise.
12941         (iwmmxt_nanddi3): Likewise.
12942         (iwmmxt_arm_movdi): Likewise.
12943         (iwmmxt_movsi_insn): Likewise.
12944         (mov<mode>_internal): Likewise.
12945         (and<mode>3_iwmmxt): Likewise.
12946         (ior<mode>3_iwmmxt): Likewise.
12947         (xor<mode>3_iwmmxt): Likewise.
12948         (add<mode>3_iwmmxt): Likewise.
12949         (ssaddv8qi3): Likewise.
12950         (ssaddv4hi3): Likewise.
12951         (ssaddv2si3): Likewise.
12952         (usaddv8qi3): Likewise.
12953         (usaddv4hi3): Likewise.
12954         (usaddv2si3): Likewise.
12955         (sub<mode>3_iwmmxt): Likewise.
12956         (sssubv8qi3): Likewise.
12957         (sssubv4hi3): Likewise.
12958         (sssubv2si3): Likewise.
12959         (ussubv8qi3): Likewise.
12960         (ussubv4hi3): Likewise.
12961         (ussubv2si3): Likewise.
12962         (mulv4hi3_iwmmxt): Likewise.
12963         (smulv4hi3_highpart): Likewise.
12964         (umulv4hi3_highpart): Likewise.
12965         (iwmmxt_wmacs): Likewise.
12966         (iwmmxt_wmacsz): Likewise.
12967         (iwmmxt_wmacu): Likewise.
12968         (iwmmxt_wmacuz): Likewise.
12969         (iwmmxt_clrdi): Likewise.
12970         (iwmmxt_clrv8qi): Likewise.
12971         (iwmmxt_clr4hi): Likewise.
12972         (iwmmxt_clr2si): Likewise.
12973         (iwmmxt_uavgrndv8qi3): Likewise.
12974         (iwmmxt_uavgrndv4hi3): Likewise.
12975         (iwmmxt_uavgv8qi3): Likewise.
12976         (iwmmxt_uavgv4hi3): Likewise.
12977         (iwmmxt_tinsrb): Likewise.
12978         (iwmmxt_tinsrh): Likewise.
12979         (iwmmxt_tinsrw): Likewise.
12980         (iwmmxt_textrmub): Likewise.
12981         (iwmmxt_textrmsb): Likewise.
12982         (iwmmxt_textrmuh): Likewise.
12983         (iwmmxt_textrmsh): Likewise.
12984         (iwmmxt_textrmw): Likewise.
12985         (iwmxxt_wshufh): Likewise.
12986         (eqv8qi3): Likewise.
12987         (eqv4hi3): Likewise.
12988         (eqv2si3): Likewise.
12989         (gtuv8qi3): Likewise.
12990         (gtuv4hi3): Likewise.
12991         (gtuv2si3): Likewise.
12992         (gtv8qi3): Likewise.
12993         (gtv4hi3): Likewise.
12994         (gtv2si3): Likewise.
12995         (smax<mode>3_iwmmxt): Likewise.
12996         (umax<mode>3_iwmmxt): Likewise.
12997         (smin<mode>3_iwmmxt): Likewise.
12998         (umin<mode>3_iwmmxt): Likewise.
12999         (iwmmxt_wpackhss): Likewise.
13000         (iwmmxt_wpackwss): Likewise.
13001         (iwmmxt_wpackdss): Likewise.
13002         (iwmmxt_wpackhus): Likewise.
13003         (iwmmxt_wpackwus): Likewise.
13004         (iwmmxt_wpackdus): Likewise.
13005         (iwmmxt_wunpckihb): Likewise.
13006         (iwmmxt_wunpckihh): Likewise.
13007         (iwmmxt_wunpckihw): Likewise.
13008         (iwmmxt_wunpckilb): Likewise.
13009         (iwmmxt_wunpckilh): Likewise.
13010         (iwmmxt_wunpckilw): Likewise.
13011         (iwmmxt_wunpckehub): Likewise.
13012         (iwmmxt_wunpckehuh): Likewise.
13013         (iwmmxt_wunpckehuw): Likewise.
13014         (iwmmxt_wunpckehsb): Likewise.
13015         (iwmmxt_wunpckehsh): Likewise.
13016         (iwmmxt_wunpckehsw): Likewise.
13017         (iwmmxt_wunpckelub): Likewise.
13018         (iwmmxt_wunpckeluh): Likewise.
13019         (iwmmxt_wunpckeluw): Likewise.
13020         (iwmmxt_wunpckelsb): Likewise.
13021         (iwmmxt_wunpckelsh): Likewise.
13022         (iwmmxt_wunpckelsw): Likewise.
13023         (ror<mode>3): Likewise.
13024         (ashr<mode>3_iwmmxt): Likewise.
13025         (lshr<mode>3_iwmmxt): Likewise.
13026         (ashl<mode>3_iwmmxt): Likewise.
13027         (ror<mode>3_di): Likewise.
13028         (ashr<mode>3_di): Likewise.
13029         (lshr<mode>3_di): Likewise.
13030         (ashl<mode>3_di): Likewise.
13031         (iwmmxt_wmadds): Likewise.
13032         (iwmmxt_wmaddu): Likewise.
13033         (iwmmxt_tmia): Likewise.
13034         (iwmmxt_tmiaph): Likewise.
13035         (iwmmxt_tmiabb): Likewise.
13036         (iwmmxt_tmiatb): Likewise.
13037         (iwmmxt_tmiabt): Likewise.
13038         (iwmmxt_tmiatt): Likewise.
13039         (iwmmxt_tmovmskb): Likewise.
13040         (iwmmxt_tmovmskh): Likewise.
13041         (iwmmxt_tmovmskw): Likewise.
13042         (iwmmxt_waccb): Likewise.
13043         (iwmmxt_wacch): Likewise.
13044         (iwmmxt_waccw): Likewise.
13045         (iwmmxt_waligni): Likewise.
13046         (iwmmxt_walignr): Likewise.
13047         (iwmmxt_walignr0): Likewise.
13048         (iwmmxt_walignr1): Likewise.
13049         (iwmmxt_walignr2): Likewise.
13050         (iwmmxt_walignr3): Likewise.
13051         (iwmmxt_wsadb): Likewise.
13052         (iwmmxt_wsadh): Likewise.
13053         (iwmmxt_wsadbz): Likewise.
13054         (iwmmxt_wsadhz): Likewise.
13055         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13056         (iwmmxt_wabsdiffb): Likewise.
13057         (iwmmxt_wabsdiffh): Likewise.
13058         (iwmmxt_wabsdiffw): Likewise.
13059         (iwmmxt_waddsubhx): Likewise
13060         (iwmmxt_wsubaddhx): Likewise.
13061         (addc<mode>3): Likewise.
13062         (iwmmxt_avg4): Likewise.
13063         (iwmmxt_avg4r): Likewise.
13064         (iwmmxt_wmaddsx): Likewise.
13065         (iwmmxt_wmaddux): Likewise.
13066         (iwmmxt_wmaddsn): Likewise.
13067         (iwmmxt_wmaddun): Likewise.
13068         (iwmmxt_wmulwsm): Likewise.
13069         (iwmmxt_wmulwum): Likewise.
13070         (iwmmxt_wmulsmr): Likewise.
13071         (iwmmxt_wmulumr): Likewise.
13072         (iwmmxt_wmulwsmr): Likewise.
13073         (iwmmxt_wmulwumr): Likewise.
13074         (iwmmxt_wmulwl): Likewise.
13075         (iwmmxt_wqmulm): Likewise.
13076         (iwmmxt_wqmulwm): Likewise.
13077         (iwmmxt_wqmulmr): Likewise.
13078         (iwmmxt_wqmulwmr): Likewise.
13079         (iwmmxt_waddbhusm): Likewise.
13080         (iwmmxt_waddbhusl): Likewise.
13081         (iwmmxt_wqmiabb): Likewise.
13082         (iwmmxt_wqmiabt): Likewise.
13083         (iwmmxt_wqmiatb): Likewise.
13084         (iwmmxt_wqmiatt): Likewise.
13085         (iwmmxt_wqmiabbn): Likewise.
13086         (iwmmxt_wqmiabtn): Likewise.
13087         (iwmmxt_wqmiatbn): Likewise.
13088         (iwmmxt_wqmiattn): Likewise.
13089         (iwmmxt_wmiabb): Likewise.
13090         (iwmmxt_wmiabt): Likewise.
13091         (iwmmxt_wmiatb): Likewise.
13092         (iwmmxt_wmiatt): Likewise.
13093         (iwmmxt_wmiabbn): Likewise.
13094         (iwmmxt_wmiabtn): Likewise.
13095         (iwmmxt_wmiatbn): Likewise.
13096         (iwmmxt_wmiattn): Likewise.
13097         (iwmmxt_wmiawbb): Likewise.
13098         (iwmmxt_wmiawbt): Likewise.
13099         (iwmmxt_wmiawtb): Likewise.
13100         (iwmmxt_wmiawtt): Likewise.
13101         (iwmmxt_wmiawbbn): Likewise.
13102         (iwmmxt_wmiawbtn): Likewise.
13103         (iwmmxt_wmiawtbn): Likewise.
13104         (iwmmxt_wmiawttn): Likewise.
13105         (iwmmxt_wmerge): Likewise.
13106         (iwmmxt_tandc<mode>3): Likewise.
13107         (iwmmxt_torc<mode>3): Likewise.
13108         (iwmmxt_torvsc<mode>3): Likewise.
13109         (iwmmxt_textrc<mode>3): Likewise.
13110         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13111         (wmmxt_pack): Likewise.
13112         (wmmxt_mult_c1): Likewise.
13113         (wmmxt_mult_c2): Likewise.
13114         (wmmxt_alu_c1): Likewise.
13115         (wmmxt_alu_c2): Likewise.
13116         (wmmxt_alu_c3): Likewise.
13117         (wmmxt_transfer_c1): Likewise.
13118         (wmmxt_transfer_c2): Likewise.
13119         (wmmxt_transfer_c3): Likewise.
13120         (marvell_f_iwmmxt_wstr): Likewise.
13121         (marvell_f_iwmmxt_wldr): Likewise.
13122
13123 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13124
13125         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13126
13127 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13128
13129         Revert:
13130         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13131         * lra-constraints.c (need_for_split_p): Check call used hard regs
13132         living through calls.
13133
13134         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13135         call used regs for call insn.
13136
13137 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13138
13139         PR target/57736
13140         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13141         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13142
13143 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13144
13145         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13146
13147 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13148
13149         * lra-constraints.c (need_for_split_p): Check call used hard regs
13150         living through calls.
13151
13152 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13153
13154         PR target/57744
13155         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13156         to tie with any other modes.  Eliminate Altivec vector mode tests,
13157         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13158         code, to return 0 if testing MODE2 for a condition, if we've
13159         already tested MODE1 for the same condition.
13160
13161 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13162
13163         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13164         layout.
13165
13166 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13167
13168         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13169         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13170
13171 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13172
13173         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13174         Define.
13175         (aarch64_symbolic_constant_p): Remove.
13176         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13177         static.  Fix line length and white space.
13178         (aarch64_symbolic_constant_p): Remove.
13179         * config/aarch64/predicates.md (aarch64_valid_symref):
13180         Use aarch64_classify_symbol_expression.
13181
13182 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13183
13184         * config/arm/constraints.md (Ts): New constraint.
13185         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13186         16-bit encodings.
13187         (compare_scc): Use "Ts" constraint for operand 0.
13188         (ior_scc_scc): Likewise.
13189         (and_scc_scc): Likewise.
13190         (and_scc_scc_nodom): Likewise.
13191         (ior_scc_scc_cmp): Likewise for operand 7.
13192         (and_scc_scc_cmp): Likewise.
13193         * config/arm/thumb2.md (thumb2_movsi_insn):
13194         Add alternatives for 16-bit encodings.
13195         (thumb2_movhi_insn): Likewise.
13196         (thumb2_movsicc_insn): Likewise.
13197         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13198         (thumb2_negscc): Use "Ts" constraint.
13199         Move mvn instruction outside cond_exec block.
13200         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13201         for 16-bit encodings.
13202
13203 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13204
13205         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13206         encoding.
13207         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13208         (mulsi3subsi): Likewise.
13209         (mulsidi3adddi): Likewise.
13210         (mulsidi3_v6): Likewise.
13211         (umulsidi3_v6): Likewise.
13212         (umulsidi3adddi_v6): Likewise.
13213         (smulsi3_highpart_v6): Likewise.
13214         (umulsi3_highpart_v6): Likewise.
13215         (mulhisi3tb): Likewise.
13216         (mulhisi3bt): Likewise.
13217         (mulhisi3tt): Likewise.
13218         (maddhisi4): Likewise.
13219         (maddhisi4tb): Likewise.
13220         (maddhisi4tt): Likewise.
13221         (maddhidi4): Likewise.
13222         (maddhidi4tb): Likewise.
13223         (maddhidi4tt): Likewise.
13224         (zeroextractsi_compare0_scratch): Likewise.
13225         (insv_zero): Likewise.
13226         (insv_t2): Likewise.
13227         (anddi_notzesidi_di): Likewise.
13228         (anddi_notsesidi_di): Likewise.
13229         (andsi_notsi_si): Likewise.
13230         (iordi_zesidi_di): Likewise.
13231         (xordi_zesidi_di): Likewise.
13232         (andsi_iorsi3_notsi): Likewise.
13233         (smax_0): Likewise.
13234         (smax_m1): Likewise.
13235         (smin_0): Likewise.
13236         (not_shiftsi): Likewise.
13237         (unaligned_loadsi): Likewise.
13238         (unaligned_loadhis): Likewise.
13239         (unaligned_loadhiu): Likewise.
13240         (unaligned_storesi): Likewise.
13241         (unaligned_storehi): Likewise.
13242         (extv_reg): Likewise.
13243         (extzv_t2): Likewise.
13244         (divsi3): Likewise.
13245         (udivsi3): Likewise.
13246         (arm_zero_extendhisi2addsi): Likewise.
13247         (arm_zero_extendqisi2addsi): Likewise.
13248         (compareqi_eq0): Likewise.
13249         (arm_extendhisi2_v6): Likewise.
13250         (arm_extendqisi2addsi): Likewise.
13251         (arm_movt): Likewise.
13252         (thumb2_ldrd): Likewise.
13253         (thumb2_ldrd_base): Likewise.
13254         (thumb2_ldrd_base_neg): Likewise.
13255         (thumb2_strd): Likewise.
13256         (thumb2_strd_base): Likewise.
13257         (thumb2_strd_base_neg): Likewise.
13258         (arm_negsi2): Add alternative for 16-bit encoding.
13259         (arm_one_cmplsi2): Likewise.
13260
13261 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13262
13263         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13264         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13265         (movdfcc): Likewise.
13266         * config/arm/vfp.md (*thumb2_movsf_vfp):
13267         Disable predication for arm_restrict_it.
13268         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13269         (*thumb2_movdfcc_vfp): Likewise.
13270         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13271         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13272         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13273         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13274         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13275         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13276         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13277         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13278         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13279         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13280         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13281         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13282         Disable predication for arm_restrict_it.
13283
13284 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13285
13286         * config/i386/bmiintrin.h (_bextr_u32): New.
13287         (_bextr_u64): Ditto.
13288
13289 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13290
13291         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13292         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13293         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13294         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13295         * config/mips/n32-elf.h: ...this new file.
13296
13297 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13298
13299         PR target/57224
13300         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13301         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13302
13303 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13304
13305         * config/mips/mips-tables.opt: Regenerate.
13306         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13307         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13308         * doc/invoke.texi: Add -m14kc.
13309
13310 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13311
13312         PR target/57623
13313         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13314         constraints of operand 1 and 2.
13315
13316         PR target/57623
13317         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13318         to match RTL canonicalization.  Swap predicates and
13319         constraints of operand 1 and 2.
13320
13321 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13322
13323         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13324         Process OP_INOUT regs for splitting too.
13325
13326 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13327
13328         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13329         decl before the loop, initialize to NULL.
13330         (vectorizable_load): Initialize ptr_incr to NULL.
13331
13332 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13333
13334         PR lto/57208
13335         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13336         * ipa-ref.c (ipa_maybe_record_reference): New function.
13337         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13338         * ipa-cp.c (create_specialized_node): Record potential references from
13339         aggvals.
13340         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13341
13342 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13343
13344         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13345         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13346         'mode' to force_reg.
13347         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13348         (aarch64_expand_mov_immediate): Likewise.
13349
13350 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13351
13352         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13353         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13354
13355 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13356
13357         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13358         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13359         (struct machine_function): Add tbegin_p.
13360         (s390_canonicalize_comparison): Fold CC mode compares to
13361         conditional jump if possible.
13362         (s390_emit_jump): Return the emitted jump.
13363         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13364         Handle CCRAWmode compares.
13365         (s390_option_override): Default to -mhtm if available.
13366         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13367         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13368         FPRs instead of df_regs_ever_live_p.
13369         (s390_optimize_nonescaping_tx): New function.
13370         (s390_init_frame_layout): Extend clobbered_regs array to cover
13371         FPRs as well.
13372         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13373         (s390_expand_tbegin): New function.
13374         (enum s390_builtin): New enum definition.
13375         (code_for_builtin): New array definition.
13376         (s390_init_builtins): New function.
13377         (s390_expand_builtin): New function.
13378         (TARGET_INIT_BUILTINS): Define.
13379         (TARGET_EXPAND_BUILTIN): Define.
13380         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13381         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13382         (s390_alc_comparison): Likewise.
13383         * config/s390/s390-modes.def: Add CCRAWmode.
13384         * config/s390/s390.h (processor_flags): Add PF_TX.
13385         (TARGET_CPU_HTM): Define macro.
13386         (TARGET_HTM): Define macro.
13387         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13388         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13389         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13390         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13391         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13392         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13393         ("*cc_to_int"): Move up.
13394         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13395         constants other than 0.
13396         ("*ccraw_to_int"): New insn and splitter definition.
13397         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13398         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13399         ("tx_assist"): New expander.
13400         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13401         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13402         * config/s390/s390.opt: Add -mhtm option.
13403         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13404         * config/s390/htmxlintrin.h: New file.
13405         * config/s390/htmintrin.h: New file.
13406         * config/s390/s390intrin.h: New file.
13407         * doc/extend.texi: Document htm builtins.
13408         * config.gcc: Add the new header files to extra_headers.
13409
13410 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13411
13412         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13413         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13414
13415 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13416             Pat Haugen  <pthaugen@us.ibm.com>
13417             Peter Bergner  <bergner@vnet.ibm.com>
13418
13419         * config/rs6000/power8.md: New.
13420         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13421         setting for power8 entry.
13422         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13423         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13424         test for Power4/Power5 only.
13425         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13426         support.
13427         (force_new_group): Adjust comment.
13428         * config/rs6000/rs6000.md: Include power8.md.
13429
13430 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13431
13432         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13433         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13434         declaration.
13435         (tune_params): Update comment.
13436         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13437         (arm_max_conditional_execute): New function.
13438         (thumb2_final_prescan_insn): Use max_insn_skipped and
13439         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13440
13441 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13442
13443         PR tree-optimization/57705
13444         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13445         SSA_NAME step, provided that it is not defined inside the loop.
13446         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13447         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13448
13449 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13450
13451         PR middle-end/57670
13452         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13453         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13454         calls in the dump.
13455         (ipa_note_param_call): Initialize member_ptr flag.
13456         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13457         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13458         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13459         (ipa_read_indirect_edge_info): Likewise.
13460
13461 2013-06-25  Richard Biener  <rguenther@suse.de>
13462
13463         PR middle-end/56977
13464         * passes.c (init_optimization_passes): Move pass_fold_builtins
13465         and pass_dce earlier with -Og.
13466
13467 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13468
13469         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13470         <BIT_FIELD_REF>: Remove trailing TAB.
13471         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13472         remove blank line.
13473
13474 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13475
13476         PR tree-optimization/57358
13477         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13478         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13479         (ipa_analyze_params_uses): Generate pessimistic info when true.
13480
13481 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13482
13483         PR tree-optimization/57539
13484         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13485         global.inlined_to of the new node to it.  All callers changed.
13486         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13487         inlining_into, pass it to cgraph_clone_node.
13488         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13489         ipa_free_edge_args_substructures.
13490         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13491         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13492         Assert we have found an rdesc in the rdesc list.
13493
13494 2013-06-24  Richard Biener  <rguenther@suse.de>
13495
13496         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13497         (pointer_set_lookup): Declare.
13498         (class pointer_map): New template class implementing a
13499         generic pointer to T map.
13500         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13501         pointer_map<T>::contains, pointer_map<T>::insert,
13502         pointer_map<T>::traverse): New functions.
13503         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13504         (pointer_set_lookup): New function.
13505         (pointer_set_contains): Use pointer_set_lookup.
13506         (pointer_set_insert): Likewise.
13507         (insert_aux): Remove.
13508         (struct pointer_map_t): Embed a pointer_set_t.
13509         (pointer_map_create): Adjust.
13510         (pointer_map_destroy): Likewise.
13511         (pointer_map_contains): Likewise.
13512         (pointer_map_insert): Likewise.
13513         (pointer_map_traverse): Likewise.
13514         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13515         pointer_map<unsigned> instead of a pointer_map_t.
13516         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13517         (streamer_tree_cache_lookup): Likewise.
13518         (streamer_tree_cache_create): Likewise.
13519         (streamer_tree_cache_delete): Likewise.
13520         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13521         pointer_map<unsigned> instead of a pointer_map_t.
13522         (lto_init_tree_ref_encoder): Adjust.
13523         (lto_destroy_tree_ref_encoder): Likewise.
13524         * lto-section-out.c (lto_output_decl_index): Likewise.
13525         (lto_record_function_out_decl_state): Likewise.
13526         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13527
13528 2013-06-24  Richard Biener  <rguenther@suse.de>
13529
13530         PR tree-optimization/57488
13531         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13532
13533 2013-06-24  Alan Modra  <amodra@gmail.com>
13534
13535         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13536         (gen_easy_altivec_constant): Likewise.
13537         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13538         easy_vector_constant_msb): Likewise.
13539
13540 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13541
13542         PR target/57688
13543         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13544         add missing return true.
13545
13546 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13547
13548         PR target/52483
13549         * config/sh/predicates.md (general_extend_operand): Invoke
13550         general_movsrc_operand for memory operands.
13551         (general_movsrc_operand): Allow reg+reg addressing, do not use
13552         general_operand for memory operands.
13553
13554 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13555
13556         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13557         when current target options does not apply.
13558         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13559         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13560         * config/i386/bmiintrin.h: Pass appropriate target
13561         attributes to header.
13562         * config/i386/mmintrin.h: Ditto.
13563         * config/i386/nmmintrin.h: Ditto.
13564         * config/i386/avx2intrin.h: Ditto.
13565         * config/i386/fxsrintrin.h: Ditto.
13566         * config/i386/tbmintrin.h: Ditto.
13567         * config/i386/xsaveintrin.h: Ditto.
13568         * config/i386/f16cintrin.h: Ditto.
13569         * config/i386/xtestintrin.h: Ditto.
13570         * config/i386/xsaveoptintrin.h: Ditto.
13571         * config/i386/bmi2intrin.h: Ditto.
13572         * config/i386/lzcntintrin.h: Ditto.
13573         * config/i386/smmintrin.h: Ditto.
13574         * config/i386/wmmintrin.h: Ditto.
13575         * config/i386/x86intrin.h: Remove all header include guards.
13576         * config/i386/prfchwintrin.h: Ditto.
13577         * config/i386/pmmintrin.h: Ditto.
13578         * config/i386/tmmintrin.h: Ditto.
13579         * config/i386/xmmintrin.h: Ditto.
13580         * config/i386/popcntintrin.h: Ditto.
13581         * config/i386/rdseedintrin.h: Ditto.
13582         * config/i386/ammintrin.h: Ditto.
13583         * config/i386/emmintrin.h: Ditto.
13584         * config/i386/immintrin.h: Remove all header include guards.
13585         * config/i386/fma4intrin.h: Ditto.
13586         * config/i386/lwpintrin.h: Ditto.
13587         * config/i386/xopintrin.h: Ditto.
13588         * config/i386/ia32intrin.h: Ditto.
13589         * config/i386/avxintrin.h: Ditto.
13590         * config/i386/rtmintrin.h: Ditto.
13591         * config/i386/fmaintrin.h: Ditto.
13592         * config/i386/mm3dnow.h: Ditto.
13593
13594 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13595
13596         * common/config/i386/i386-common.c: Handle LZCNT.
13597
13598 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13599
13600         * doc/extend.texi: Use __atomic_store_n instead of
13601         __atomic_store in HLE example.
13602
13603 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13604
13605         * config/sh/sh.c: Remove <cstdlib> workaround.
13606
13607 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13608
13609         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13610
13611 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13612
13613         * doc/extend.texi: Document that __atomic_clear and
13614           __atomic_test_and_set should only be used with bool.
13615
13616 2013-06-20  Jan Hubicka  <jh@suse.cz>
13617
13618         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13619         types_same_for_odr.
13620         * tree.c (decls_same_for_odr): New function.
13621         (same_for_edr): New function.
13622         (types_same_for_odr): New function.
13623         (get_binfo_at_offset): Use it.
13624         * tree.h (types_same_for_odr): Declare.
13625
13626 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13627             Jason Merrill  <jason@redhat.com>
13628
13629         * system.h: Include <cstdlib> as well as <stdlib.h>.
13630
13631 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13632
13633         PR target/57655
13634         * config/i386/i386.c (construct_container): Report error if
13635         long double is used with disabled x87 float returns.
13636
13637 2013-06-20  Jan Hubicka  <jh@suse.cz>
13638
13639         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13640
13641 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13642
13643         PR rtl-optimization/57425
13644         PR rtl-optimization/57569
13645         * alias.c (write_dependence_p): Remove parameters mem_mode and
13646         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13647         Changed all callers.
13648         (canon_anti_dependence): Get comments and semantics in sync.
13649         Add parameter mem_canonicalized.  Changed all callers.
13650         * rtl.h (canon_anti_dependence): Update prototype.
13651
13652 2013-06-20  Richard Biener  <rguenther@suse.de>
13653
13654         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13655         byte case, inline streamer_read_uchar and defer section
13656         overrun check.
13657
13658 2013-06-20  Richard Biener  <rguenther@suse.de>
13659
13660         PR tree-optimization/57584
13661         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13662         SSA names into the expanded expression that take part in
13663         abnormal coalescing.
13664
13665 2013-06-19  Sharad Singhai  <singhai@google.com>
13666
13667         * gcov.c (print_usage): Handle new option.
13668         (process_args): Ditto.
13669         (get_gcov_intermediate_filename): New function.
13670         (output_intermediate_file): New function.
13671         (output_gcov_file): New function
13672         (generate_results): Handle new option.
13673         (release_function): Relase demangled name.
13674         (read_graph_file): Handle demangled name.
13675         (output_lines): Ditto.
13676         * doc/gcov.texi: Document gcov intermediate format.
13677
13678 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13679
13680         PR bootstrap/57604
13681         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13682         (lra_emit_add): Use the functions.  Add comment about Y as an
13683         address segment.
13684
13685 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13686
13687         PR driver/57652
13688         * collect2.c (collect_atexit): New.
13689         (collect_exit): Delete.
13690         (main): Register collect_atexit with atexit.
13691         (collect_wait): Change collect_exit to exit.
13692         (do_wait): Same.
13693         * collect2.h (collect_exit): Delete.
13694         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13695
13696 2013-06-19  Wei Mi  <wmi@google.com>
13697
13698         PR rtl-optimization/57518
13699         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13700         if regno is used in paradoxical subreg.
13701         (update_equiv_regs): Check pdx_subregs[regno] before
13702         set a reg to be equivalent with a mem.
13703
13704 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13705
13706         PR driver/57651
13707         * file-find.h (find_a_file): Add a mode parameter.
13708         * file-find.c (find_a_file): Likewise.
13709         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13710         with X_OK for the executables.
13711         * collect2.c (main): Call find_a_file with X_OK.
13712
13713 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13714
13715         PR target/56942
13716         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13717         Use NEXT_INSN instead of next_real_insn.
13718
13719 2013-06-19  Jan Hubicka  <jh@suse.cz>
13720
13721         * cgraph.h (const_value_known_p): Replace by ...
13722         (ctor_for_folding): .. this one.
13723         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13724         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13725         * expr.c (expand_expr_real_1): Likewise.
13726         (string_constant): Likewise.
13727         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
13728         * ipa.c (process_references): Likewise.
13729         (symtab_remove_unreachable_nodes): Likewise.
13730         * ipa-inline-analysis.c (param_change_prob): Likewise.
13731         * gimple-fold.c (canonicalize_constructor_val): Likewise.
13732         (get_base_constructor): Likwise.
13733         * varpool.c (varpool_remove_node): Likewise.
13734         (varpool_remove_initializer): LIkewise.
13735         (dump_varpool_node): LIkwise.
13736         (const_value_known_p): Rewrite to ...
13737         (ctor_for_folding): ... this one.
13738
13739 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
13740
13741         PR driver/57651
13742         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
13743         PERSONALITY in $PATH derived prefixes.
13744
13745 2013-06-19  Jeff Law  <law@redhat.com>
13746
13747         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
13748         in comment.
13749
13750         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
13751         (simplify_bitwise_binary): Use it to simpify certain binary ops on
13752         booleans.
13753
13754 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
13755
13756         * config/arm/vfp.md: Move VFP instruction classification documentation
13757         to ...
13758         * config/arm/arm.md: ... here.  Update instruction classification
13759         documentation.
13760
13761 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
13762
13763         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
13764         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
13765         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
13766
13767 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
13768
13769         PR target/57609
13770         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13771         with NEXT_INSN.  Use tablejump_p to check for jump table data
13772         insns.
13773
13774 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
13775
13776         PR c++/56544
13777         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
13778         that now in C++ the value is correct per the C++ standards.
13779
13780 2013-06-19  Richard Biener  <rguenther@suse.de>
13781
13782         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
13783         for global context.
13784
13785 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13786
13787         Revert:
13788         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13789
13790         PR target/57609
13791         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13792         with next_active_insn.
13793
13794 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
13795
13796         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
13797         functions are inlined during failures to flag an error.
13798         * tree-inline.c (expand_call_inline): Allow the error to be flagged
13799         in early inline pass.
13800
13801 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
13802
13803         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
13804         in comments.
13805
13806 2013-06-18  Julian Brown  <julian@codesourcery.com>
13807
13808         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
13809         Permit virtual register pre-reload if !strict.
13810         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
13811         change.
13812         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
13813         prototype.
13814         * config/arm/neon.md (movmisalign<mode>): Use
13815         neon_perm_struct_or_reg_operand instead of
13816         neon_struct_or_register_operand.
13817         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
13818         neon_permissive_struct_operand instead of neon_struct_operand.
13819         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
13820         neon_vector_mem_operand.
13821         * config/arm/predicates.md (neon_struct_operand): Adjust call to
13822         neon_vector_mem_operand.
13823         (neon_permissive_struct_operand): New.
13824         (neon_struct_or_register_operand): Rename to...
13825         (neon_perm_struct_or_reg_operand): This. Adjust call to
13826         neon_vector_mem_operand.
13827
13828 2013-06-18  Richard Biener  <rguenther@suse.de>
13829
13830         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
13831         * lto-streamer.h: Include pointer-set.h.
13832         (struct lto_decl_slot): Remove.
13833         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
13834         Remove next_index entry.
13835         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
13836         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
13837         (lto_init_tree_ref_encoder): Adjust.
13838         (lto_destroy_tree_ref_encoder): Likewise.
13839         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
13840         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
13841         (lto_output_decl_index): Adjust.
13842         (lto_new_out_decl_state): Likewise.
13843         (lto_record_function_out_decl_state): Likewise.
13844         * lto-streamer-out.c (copy_function): Likewise.
13845
13846 2013-06-18  Richard Biener  <rguenther@suse.de>
13847
13848         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
13849         * cgraphunit.c: Include cfgloop.h.
13850         (init_lowered_empty_function): Initialize the loop tree.
13851         (assemble_thunk): Insert new BBs into loops.
13852
13853 2013-06-18  Richard Biener  <rguenther@suse.de>
13854
13855         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
13856         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
13857         the map from cache entry to cache index optional.
13858         (streamer_tree_cache_replace_tree): Adjust accordingly.
13859         (streamer_tree_cache_append): Likewise.
13860         (streamer_tree_cache_delete): Likewise.
13861         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
13862         streamer cache map from cache entry to cache index.
13863         * lto-streamer-out.c (create_output_block): Adjust.
13864
13865 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
13866
13867         * config/arm/arm.md (attribute "insn"): Move multiplication and
13868         division attributes to...
13869         (attribute "type"): ... here.  Remove mult.
13870         (attribute "mul32"): New attribute.
13871         (attribute "mul64"): Add umaal.
13872         (*arm_mulsi3): Update attributes.
13873         (*arm_mulsi3_v6): Likewise.
13874         (*thumb_mulsi3): Likewise.
13875         (*thumb_mulsi3_v6): Likewise.
13876         (*mulsi3_compare0): Likewise.
13877         (*mulsi3_compare0_v6): Likewise.
13878         (*mulsi_compare0_scratch): Likewise.
13879         (*mulsi_compare0_scratch_v6): Likewise.
13880         (*mulsi3addsi): Likewise.
13881         (*mulsi3addsi_v6): Likewise.
13882         (*mulsi3addsi_compare0): Likewise.
13883         (*mulsi3addsi_compare0_v6): Likewise.
13884         (*mulsi3addsi_compare0_scratch): Likewise.
13885         (*mulsi3addsi_compare0_scratch_v6): Likewise.
13886         (*mulsi3subsi): Likewise.
13887         (*mulsidi3adddi): Likewise.
13888         (*mulsi3addsi_v6): Likewise.
13889         (*mulsidi3adddi_v6): Likewise.
13890         (*mulsidi3_nov6): Likewise.
13891         (*mulsidi3_v6): Likewise.
13892         (*umulsidi3_nov6): Likewise.
13893         (*umulsidi3_v6): Likewise.
13894         (*umulsidi3adddi): Likewise.
13895         (*umulsidi3adddi_v6): Likewise.
13896         (*smulsi3_highpart_nov6): Likewise.
13897         (*smulsi3_highpart_v6): Likewise.
13898         (*umulsi3_highpart_nov6): Likewise.
13899         (*umulsi3_highpart_v6): Likewise.
13900         (mulhisi3): Likewise.
13901         (*mulhisi3tb): Likewise.
13902         (*mulhisi3bt): Likewise.
13903         (*mulhisi3tt): Likewise.
13904         (maddhisi4): Likewise.
13905         (*maddhisi4tb): Likewise.
13906         (*maddhisi4tt): Likewise.
13907         (maddhidi4): Likewise.
13908         (*maddhidi4tb): Likewise.
13909         (*maddhidi4tt): Likewise.
13910         (divsi3): Likewise.
13911         (udivsi3): Likewise.
13912         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
13913         (thumb2_mulsi_short_compare0): Likewise.
13914         (thumb2_mulsi_short_compare0_scratch): Likewise.
13915         * config/arm/arm1020e.md (1020mult1): Update attribute change.
13916         (1020mult2): Likewise.
13917         (1020mult3): Likewise.
13918         (1020mult4): Likewise.
13919         (1020mult5): Likewise.
13920         (1020mult6): Likewise.
13921         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
13922         change.
13923         (cortex_a15_mult64): Likewise.
13924         (cortex_a15_sdiv): Likewise.
13925         (cortex_a15_udiv): Likewise.
13926         * config/arm/arm1026ejs.md (mult1): Update attribute change.
13927         (mult2): Likewise.
13928         (mult3): Likewise.
13929         (mult4): Likewise.
13930         (mult5): Likewise.
13931         (mult6): Likewise.
13932         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
13933         (pj4_ir_div): Likewise.
13934         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
13935         (11_mult2): Likewise.
13936         (11_mult3): Likewise.
13937         (11_mult4): Likewise.
13938         (11_mult5): Likewise.
13939         (11_mult6): Likewise.
13940         (11_mult7): Likewise.
13941         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
13942         (cortex_a8_mla): Likewise.
13943         (cortex_a8_mull): Likewise.
13944         (cortex_a8_smulwy): Likewise.
13945         (cortex_a8_smlald): Likewise.
13946         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
13947         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
13948         (cortex_r4_mul_3): Likewise.
13949         (cortex_r4_mla_4): Likewise.
13950         (cortex_r4_mla_3): Likewise.
13951         (cortex_r4_smlald): Likewise.
13952         (cortex_r4_mull): Likewise.
13953         (cortex_r4_sdiv): Likewise.
13954         (cortex_r4_udiv): Likewise.
13955         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
13956         (cortex_a7_idiv): Likewise.
13957         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
13958         (9_mult2): Likewise.
13959         (9_mult3): Likewise.
13960         (9_mult4): Likewise.
13961         (9_mult5): Likewise.
13962         (9_mult6): Likewise.
13963         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
13964         (cortex_a53_sdiv): Likewise.
13965         (cortex_a53_udiv): Likewise.
13966         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
13967         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
13968         (mp626_mult2): Likewise.
13969         (mp626_mult3): Likewise.
13970         (mp626_mult4): Likewise.
13971         * config/arm/fa526.md (526_mult1): Update attribute change.
13972         (526_mult2): Likewise.
13973         * config/arm/arm-generic.md (mult): Update attribute change.
13974         (mult_ldsched_strongarm): Likewise.
13975         (mult_ldsched): Likewise.
13976         (multi_cycle): Likewise.
13977         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
13978         * config/arm/fa606te.md (606te_mult1): Update attribute change.
13979         (606te_mult2): Likewise.
13980         (606te_mult3): Likewise.
13981         (606te_mult4): Likewise.
13982         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
13983         (cortex_a9_mac16): Likewise.
13984         (cortex_a9_multiply): Likewise.
13985         (cortex_a9_mac): Likewise.
13986         (cortex_a9_multiply_long): Likewise.
13987         * config/arm/fa626te.md (626te_mult1): Update attribute change.
13988         (626te_mult2): Likewise.
13989         (626te_mult3): Likewise.
13990         (626te_mult4): Likewise.
13991
13992 2013-06-18  Richard Biener  <rguenther@suse.de>
13993
13994         PR lto/57334
13995         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
13996
13997 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13998
13999         PR target/57609
14000         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14001         with next_active_insn.
14002
14003 2013-06-18  Alan Modra  <amodra@gmail.com>
14004
14005         * config/rs6000/rs6000.h (enum data_align): New.
14006         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14007         (DATA_ABI_ALIGNMENT): Define.
14008         (CONSTANT_ALIGNMENT): Correct comment.
14009         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14010         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14011
14012 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14013
14014         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14015         ATTRIBUTE_UNUSED marking.
14016
14017 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14018
14019         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14020         alternative and update.
14021         (aarch64_dup_lanedi): Delete.
14022         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14023         * config/aarch64/aarch64-simd-builtins.def: Update.
14024
14025 2013-06-17  Richard Biener  <rguenther@suse.de>
14026
14027         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14028         (lto_input_scc): Declare.
14029         (lto_input_tree_1): Likewise.
14030         (struct lto_stats_d): Add num_tree_bodies_output and
14031         num_pickle_refs_output.
14032         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14033         (lto_read_tree_1): Split out from ...
14034         (lto_read_tree): ... this.
14035         (lto_input_scc): New function.
14036         (lto_input_tree_1): Split out from ...
14037         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14038         (lto_data_in_create): Create the streamer cache without hashes.
14039         * lto-streamer-out.c (create_output_block): Create the streamer
14040         cache with hashes when not doing WPA.
14041         (lto_write_tree_1): Split out from ...
14042         (lto_write_tree): ... this.
14043         (get_symbol_initial_value): New function.
14044         (lto_output_tree_1): Split out from ...
14045         (lto_output_tree): ... this.  Write trees as series of SCCs
14046         using a DFS walk via DFS_write_tree.
14047         (struct sccs, struct scc_entry): New types.
14048         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14049         (DFS_write_tree_body): New function.
14050         (DFS_write_tree): Likewise.
14051         (hash_tree): Likewise.
14052         (scc_entry_compare): Likewise.
14053         (hash_scc): Likewise.
14054         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14055         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14056         TREE_CHAIN as regular reference.
14057         (streamer_read_integer_cst): Remove.
14058         (streamer_get_pickled_tree): Adjust.
14059         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14060         of DECL_EXTERNALs in BLOCK_VARS for now.
14061         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14062         reference.
14063         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14064         Add hash value argument and record that if hashes are recorded
14065         in the cache.
14066         (streamer_tree_cache_insert_1): Adjust.
14067         (streamer_tree_cache_insert): Likewise.
14068         (streamer_tree_cache_insert_at): Rename to ...
14069         (streamer_tree_cache_replace_tree): ... this and adjust.
14070         (streamer_tree_cache_append): Adjust.
14071         (record_common_node): Likewise.
14072         (streamer_tree_cache_create): Add argument whether to
14073         record hash values together with trees.
14074         (streamer_tree_cache_delete): Adjust.
14075         * tree-streamer.h (struct streamer_tree_cache_d): Add
14076         vector of hashes.
14077         (streamer_read_integer_cst): Remove.
14078         (streamer_tree_cache_insert): Adjust.
14079         (streamer_tree_cache_append): Likewise.
14080         (streamer_tree_cache_insert_at): Rename to ...
14081         (streamer_tree_cache_replace_tree): ... this and adjust.
14082         (streamer_tree_cache_create): Add argument whether to record hashes.
14083         (streamer_tree_cache_get): Rename to ...
14084         (streamer_tree_cache_get_tree): ... this.
14085         (streamer_tree_cache_get_hash): New function.
14086         * tree.c (cache_integer_cst): New function.
14087         * tree.h (cache_integer_cst): Declare.
14088         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14089         * lto-symtab.c (lto_varpool_replace_node): Only release
14090         DECL_INITIAL of non-prevailing decls.
14091         * varpool.c (varpool_remove_initializer): Do not release
14092         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14093
14094 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14095
14096         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14097         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14098         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14099         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14100         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14101         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14102         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14103         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14104         instead of TARGET_64BIT.
14105         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14106         Require ISA_HAS_<D>DIV.
14107
14108 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14109
14110         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14111         (mips*-*-linux*): Move default with_llsc setting to where other
14112         defaults are set.
14113         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14114         with_arch block.
14115         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14116         Likewise.  Remove default with_tune setting.  Move default float
14117         setting to its own block.  Handle with_llsc in the same block as above.
14118
14119 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14120
14121         PR rtl-optimization/57425
14122         PR rtl-optimization/57569
14123         * alias.c (write_dependence_p): Add new parameters mem_mode,
14124         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14125         Changed all callers.
14126         (canon_anti_dependence): New function.
14127         * cse.c (check_dependence): Use canon_anti_dependence.
14128         * cselib.c (cselib_invalidate_mem): Likewise.
14129         * rtl.h (canon_anti_dependence): Declare.
14130
14131 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14132
14133         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14134         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14135         ".set mips3" for 64-bit targets.
14136
14137 2013-06-15  Dehao Chen  <dehao@google.com>
14138
14139         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14140         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14141         (gimple_check_call_args): Likewise.
14142         * value-prof.c (check_ic_target): Likewise.
14143         * ipa-inline.c (early_inliner): Likewise.
14144         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14145         * cgraph.c (cgraph_create_edge_1): Likewise.
14146         (cgraph_make_edge_direct): Likewise.
14147
14148 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14149
14150         PR target/57615
14151         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14152         rs6000_output_move_128bit to handle emitting quad memory
14153         operations.  Set attribute length to 8 bytes.
14154
14155 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14156
14157         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14158         New pattern.
14159         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14160         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14161         (aarch64_<su>mlsl<mode>): Likewise.
14162
14163 2013-06-14  Mike Stump  <mikestump@comcast.net>
14164
14165         * Makefile.in (TARGET_H): Add insn-codes.h.
14166
14167 2013-06-14  Alan Modra  <amodra@gmail.com>
14168
14169         PR middle-end/57134
14170         PR middle-end/57586
14171         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14172         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14173         bitfield expansion when EXPAND_MEMORY.
14174         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14175
14176 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14177
14178         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14179         test for clearing quad memory on 32-bit later.
14180
14181 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14182
14183         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14184         (fold_negate_expr): Likewise.
14185         (fold_real_zero_addition_p): Handle vectors.
14186         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14187
14188 2013-06-14  Alan Modra  <amodra@gmail.com>
14189
14190         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14191
14192 2013-06-13  Jan Hubicka  <jh@suse.cz>
14193
14194         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14195         Local comdats are not externally visible.
14196         * symtab.c (dump_symtab_base): Dump externally visible.
14197         (verify_symtab_base): Verify back links in the symtab hash.
14198
14199 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14200
14201         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14202         CONVERT_EXPR as equal nodes.
14203
14204 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14205
14206         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14207
14208 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14209
14210         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14211         Generalize to complex and vector.
14212         * tree.c (build_all_ones_cst): New function.
14213         * tree.h (build_all_ones_cst): Declare it.
14214
14215 2013-06-13  Alan Modra  <amodra@gmail.com>
14216
14217         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14218         * config/rs6000/rs6000.md (signbittf2): New insn.
14219         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14220         (abstf2_internal, cmptf_internal2): Likewise.
14221         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14222
14223 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14224             Pat Haugen  <pthaugen@us.ibm.com>
14225             Peter Bergner  <bergner@vnet.ibm.com>
14226
14227         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14228         power8 byte, half-word, and quad-word atomic instructions.
14229         (emit_store_conditional): Likewise.
14230         (rs6000_expand_atomic_compare_and_swap): Likewise.
14231         (rs6000_expand_atomic_op): Likewise.
14232
14233         * config/rs6000/sync.md (larx): Add new modes for power8.
14234         (stcx): Likewise.
14235         (AINT): New mode iterator to include TImode as well as normal
14236         integer modes on power8.
14237         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14238         that VSX registers are not considered.  Use AINT mode iterator
14239         instead of INT1 to allow inclusion of quad word atomic operations
14240         on power8.
14241         (load_locked<mode>): Likewise.
14242         (store_conditional<mode>): Likewise.
14243         (atomic_compare_and_swap<mode>): Likewise.
14244         (atomic_exchange<mode>): Likewise.
14245         (atomic_nand<mode>): Likewise.
14246         (atomic_fetch_<fetchop_name><mode>): Likewise.
14247         (atomic_nand_fetch<mode>): Likewise.
14248         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14249         each type.
14250         (ATOMIC): On power8, add QImode, HImode modes.
14251         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14252         modes that promote to SImode.
14253         (load_lockedti): Convert TImode arguments to PTImode, so that we
14254         get a guaranteed even/odd register pair.
14255         (load_lockedpti): Likewise.
14256         (store_conditionalti): Likewise.
14257         (store_conditionalpti): Likewise.
14258
14259         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14260         atomic load/store instructions.
14261         (HSI): Likewise.
14262
14263 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14264
14265         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14266         loads.
14267         (insn_count): New attribute, with most cases extracted from...
14268         (length): ...here.  Redefine most cases in terms of insn_count.
14269         (single_insn): Delete.
14270         (can_delay): Use insn_count to check for single instructions.
14271         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14272         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14273         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14274         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14275         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14276         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14277         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14278         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14279         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14280         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14281         rather than "length".
14282         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14283         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14284         Use "insn_count" rather than "length".
14285         * config/mips/mips-dsp.md
14286         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14287         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14288         length attributes.
14289
14290 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14291
14292         PR tree-optimization/57361
14293         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14294
14295 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14296
14297         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14298         to split.
14299         (aarch64_simd_combine<mode>): New instruction expansion.
14300         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14301         function prototype.
14302         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14303         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14304
14305 2013-06-12  Jan Hubicka  <jh@suse.cz>
14306
14307         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14308         decl has when in streaming stage.
14309         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14310         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14311
14312 2013-06-12  Roland Stigge  <stigge@antcom.de>
14313
14314         PR target/57578
14315         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14316
14317 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14318
14319         PR tree-optimization/57537
14320         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14321         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14322
14323 2013-06-12  Richard Biener  <rguenther@suse.de>
14324
14325         * data-streamer.h (streamer_write_char_stream): CSE
14326         obs->current_pointer.
14327         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14328         streamer_write_char_stream manually and optimize the resulting loop.
14329         (streamer_write_hwi_stream): Likewise.
14330
14331 2013-06-12  Jan Hubicka  <jh@suse.cz>
14332
14333         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14334         * cgraph.h (varpool_create_empty_node): Declare.
14335         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14336         duplicated nodes.
14337         * symtab.c (symtab_unregister_node): Be lax about missin entries
14338         in node hash.
14339         (symtab_get_node): Update comment.
14340         * varpool.c (varpool_create_empty_node): Break out from ...
14341         (varpool_node_for_decl): ... here.
14342         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14343
14344 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14345
14346         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14347         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14348         part.  Use straight-line flow at the end.
14349         <COMPONENT_REF>: Remove superfluous else.
14350         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14351
14352 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14353
14354         PR target/56564
14355         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14356         target hook even for !TREE_PUBLIC decls.  If no resolution info
14357         is available, return false for common and external decls.
14358
14359 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14360
14361         * config/rl78/constraints.md (U): New constraint.
14362         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14363         valloc attribute.
14364
14365 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14366
14367         PR target/57589
14368         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14369         to allow returning address to AT_PLATFORM name.
14370
14371 2013-06-11  Jan Hubicka  <jh@suse.cz>
14372
14373         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14374         * cgraph.h (symtab_node_base): Add weakref flag.
14375         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14376         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14377         (output_weakrefs): Use weakref flag.
14378         * fold-const.c (simple_operand_p): Handle WEAK.
14379         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14380         * ipa.c (varpool_externally_visible_p): Drop weakref.
14381         (function_and_variable_visibility): Update comment; fix weakref
14382         sanity checks; do not clear DECL_WEAK on them.
14383         * lto-cgraph.c (lto_output_node): update.
14384         (lto_output_varpool_node): Update.
14385         (input_overwrite_node): Update.
14386         (input_node): Update.
14387         (input_varpool_node): Update.
14388         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14389         (lto_symtab_merge_symbols): Add sanity check.
14390         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14391         * passes.c (rest_of_decl_compilation): Set static flag, too.
14392         * symtab.c (dump_symtab_base): Dump weakref.
14393         (verify_symtab_base): Sanity check weakrefs.
14394         (symtab_make_decl_local): Remove duplicated code.
14395         (symtab_alias_ultimate_target): Simplify.
14396         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14397
14398 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14399
14400         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14401         sequence_vect == NULL.
14402
14403 2013-06-11  DJ Delorie  <dj@redhat.com>
14404
14405         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14406         (rl78_unwind_word_mode): New.
14407
14408 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14409
14410         * final.c (debug_prefix_maps): Make static.
14411
14412 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14413
14414         * function.c (initial_trampoline): Remove stray copy.
14415
14416 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14417
14418         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14419
14420 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14421
14422         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14423         within bounds at the beginning of the function.
14424
14425 2013-06-11  Alan Modra  <amodra@gmail.com>
14426
14427         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14428         reporting.
14429         (get_named_section): Don't NULL !DECL_P decl.
14430
14431 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14432
14433         * doc/invoke.texi (core-avx2): Document.
14434         (slm): Likewise.
14435         (atom): Updated with MOVBE.
14436
14437 2013-06-11  Richard Biener  <rguenther@suse.de>
14438
14439         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14440
14441 2013-06-11  Anton Blanchard  <anton@samba.org>
14442
14443         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14444         correct shift value in little-endian mode.
14445
14446 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14447
14448         PR target/56564
14449         * varasm.c (get_variable_align): Move #endif to the right place.
14450
14451 2013-06-10  Cary Coutant  <ccoutant@google.com>
14452
14453         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14454         for hash so that hash table traversal order is deterministic.
14455
14456 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14457             Pat Haugen  <pthaugen@us.ibm.com>
14458             Peter Bergner  <bergner@vnet.ibm.com>
14459
14460         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14461         of vectors in GPRS if they are direct moves or quad word load or
14462         store moves.
14463
14464         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14465         declaration.
14466         (direct_move_p): Likewise.
14467         (quad_load_store_p): Likewise.
14468
14469         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14470         classes into bins based on the physical register type.
14471         (reg_class_to_reg_type): Likewise.
14472         (IS_STD_REG_TYPE): Likewise.
14473         (IS_FP_VECT_REG_TYPE): Likewise.
14474         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14475         use direct move instructions.
14476         (reload_gpr_vsx): Likewise.
14477         (reload_vsx_gpr): Likewise.
14478         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14479         information that is a simplification of register classes.  Also
14480         precalculate direct move reload helpers.
14481         (direct_move_p): New function to return true if the operation can
14482         be done as a direct move instruciton.
14483         (quad_load_store_p): New function to return true if the operation
14484         is a quad memory operation.
14485         (rs6000_legitimize_address): If quad memory, only allow register
14486         indirect for TImode addresses.
14487         (rs6000_legitimate_address_p): Likewise.
14488         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14489         (rs6000_reload_register_type): Likewise.
14490         (register_to_reg_type): Return register type.
14491         (rs6000_secondary_reload_simple_move): New helper function for
14492         secondary reload and secondary memory needed to identify anything
14493         that is a simple move, and does not need reloading.
14494         (rs6000_secondary_reload_direct_move): New helper function for
14495         secondary reload to identify cases that can be done with several
14496         instructions via the direct move instructions.
14497         (rs6000_secondary_reload_move): New helper function for secondary
14498         reload to identify moves between register types that can be done.
14499         (rs6000_secondary_reload): Add support for quad memory operations
14500         and for direct move.
14501         (rs6000_secondary_memory_needed): Likewise.
14502         (rs6000_debug_secondary_memory_needed): Change argument names.
14503         (rs6000_output_move_128bit): New function to return the move to
14504         use for 128-bit moves, including knowing about the various
14505         limitations of quad memory operations.
14506
14507         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14508         memory operations.  call rs6000_output_move_128bit for the actual
14509         instruciton(s) to generate.
14510         (vsx_movti_64bit): Likewise.
14511
14512         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14513         (UNSPEC_P8V_MTVSRWZ): Likewise.
14514         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14515         (UNSPEC_P8V_MTVSRD): Likewise.
14516         (UNSPEC_P8V_XXPERMDI): Likewise.
14517         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14518         (UNSPEC_FUSION_GPR): Likewise.
14519         (FMOVE128_GPR): New iterator for direct move.
14520         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14521         (f32_sv): Likewise.
14522         (f32_dm): Likewise.
14523         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14524         loads and direct move instructions.
14525         (zero_extendsidi2_lfiwzx): Likewise.
14526         (extendsidi2_lfiwax): Likewise.
14527         (extendsidi2_nocell): Likewise.
14528         (floatsi<mode>2_lfiwax): Likewise.
14529         (lfiwax): Likewise.
14530         (floatunssi<mode>2_lfiwzx): Likewise.
14531         (lfiwzx): Likewise.
14532         (fix_trunc<mode>_stfiwx): Likewise.
14533         (fixuns_trunc<mode>_stfiwx): Likewise.
14534         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14535         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14536         (parity<mode>2_cmpb): Set length/type attr.
14537         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14538         for 'mr.' to fast_compare.
14539         (bpermd_<mode>): Change type attr to popcnt.
14540         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14541         (p8_mtvsrwz_1): Likewise.
14542         (p8_mtvsrwz_2): Likewise.
14543         (reload_fpr_from_gpr<mode>): Likewise.
14544         (p8_mtvsrd_1): Likewise.
14545         (p8_mtvsrd_2): Likewise.
14546         (p8_xxpermdi_<mode>): Likewise.
14547         (reload_vsx_from_gpr<mode>): Likewise.
14548         (reload_vsx_from_gprsf): Likewise.
14549         (p8_mfvsrd_3_<mode>): LIkewise.
14550         (reload_gpr_from_vsx<mode>): Likewise.
14551         (reload_gpr_from_vsxsf): Likewise.
14552         (p8_mfvsrd_4_disf): Likewise.
14553         (multi-word GPR splits): Do not split direct moves or quad memory
14554         operations.
14555
14556 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14557
14558         * tree-into-ssa.c (interesting_blocks): Make static.
14559
14560 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14561
14562         PR target/56564
14563         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14564         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14565         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14566         (get_variable_align): New function.
14567         (get_variable_section, emit_bss, emit_common,
14568         assemble_variable_contents, place_block_symbol): Use
14569         get_variable_align instead of DECL_ALIGN.
14570         (assemble_noswitch_variable): Add align argument, use it
14571         instead of DECL_ALIGN.
14572         (assemble_variable): Adjust caller.  Use get_variable_align
14573         instead of DECL_ALIGN.
14574         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14575         caller.
14576         (DATA_ABI_ALIGNMENT): Define.
14577         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14578         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14579         opt is false, only return the psABI mandated alignment increase.
14580         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14581         (DATA_ABI_ALIGNMENT): ... this.
14582         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14583         (DATA_ABI_ALIGNMENT): ... this.
14584         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14585         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14586         (DATA_ABI_ALIGNMENT): ... this.
14587         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14588         * doc/tm.texi: Regenerated.
14589
14590 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14591
14592         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14593         cmp_code to construct REG_EQUAL note.
14594
14595 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14596
14597         PR target/57568
14598         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14599         that operands[2] doesn't overlap with operands[0].
14600
14601 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14602             Jan Hubicka  <jh@suse.cz>
14603
14604         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14605         hack to mark symbols as used.
14606
14607 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14608
14609         PR rtl-optimization/57559
14610         * lra-constraints.c (process_alt_operands): Don't discourage
14611         memory with known offset for offsetable memory constraint.
14612         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14613
14614 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14615
14616         * varasm.c (struct oc_local_state): Reorder fields.
14617         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14618         and adjust accordingly.
14619         (output_constructor): Reorder initialization code and adjust call to
14620         output_constructor_bitfield.
14621
14622 2013-06-07  Jan Hubicka  <jh@suse.cz>
14623
14624         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14625
14626 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14627
14628         * tree-object-size.c (unknown): Make const.
14629
14630 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14631
14632         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14633         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14634         for last alternative in the cpu_facility attribute.
14635
14636 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14637
14638         PR target/56315
14639         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14640         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14641         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14642         * config/arm/constraints.md (Dg): New constraint.
14643         * config/arm/neon.md (xordi3_neon): Remove.
14644         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14645         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14646
14647 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14648
14649         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14650         Clean up alternatives.
14651
14652 2013-06-07  Alan Modra  <amodra@gmail.com>
14653
14654         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14655         va_list_gpr_size.
14656
14657 2013-06-07  Alan Modra  <amodra@gmail.com>
14658
14659         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14660
14661 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14662
14663         * config/arm/constraints.md (Df): New constraint.
14664         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14665         Correct length attribute for last two alternatives.
14666
14667 2013-06-07  Alan Modra  <amodra@gmail.com>
14668
14669         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14670         override user -mfp-in-toc.
14671         (offsettable_ok_by_alignment): Consider just the current access
14672         rather than the whole object, unless BLKmode.  Handle
14673         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14674         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14675         for -mcmodel=medium.
14676         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14677         override user -mfp-in-toc or -msum-in-toc.  Default to
14678         -mno-fp-in-toc for -mcmodel=medium.
14679
14680 2013-06-06  DJ Delorie  <dj@redhat.com>
14681
14682         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14683         TARGET_VALID_POINTER_MODE.
14684
14685 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14686             Pat Haugen  <pthaugen@us.ibm.com>
14687             Peter Bergner  <bergner@vnet.ibm.com>
14688
14689         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14690         Document new power8 builtins.
14691
14692         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14693         condition code register, to allow 128-bit logical operations to be
14694         done in the VSX or GPR registers.
14695         (nor<mode>3): Use the canonical form for nor.
14696         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14697         vclz*, and vpopcnt* vector instructions.
14698         (nand<mode>3): Likewise.
14699         (orc<mode>3): Likewise.
14700         (clz<mode>2): LIkewise.
14701         (popcount<mode>2): Likewise.
14702
14703         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14704         that only the GPRs are recognized.
14705
14706         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14707         support for new power8 builtins.
14708
14709         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14710         builtin functions.
14711         (xscvdpspn): Likewise.
14712         (vclz): Likewise.
14713         (vclzb): Likewise.
14714         (vclzh): Likewise.
14715         (vclzw): Likewise.
14716         (vclzd): Likewise.
14717         (vpopcnt): Likewise.
14718         (vpopcntb): Likewise.
14719         (vpopcnth): Likewise.
14720         (vpopcntw): Likewise.
14721         (vpopcntd): Likewise.
14722         (vgbbd): Likewise.
14723         (vmrgew): Likewise.
14724         (vmrgow): Likewise.
14725         (eqv): Likewise.
14726         (eqv_v16qi3): Likewise.
14727         (eqv_v8hi3): Likewise.
14728         (eqv_v4si3): Likewise.
14729         (eqv_v2di3): Likewise.
14730         (eqv_v4sf3): Likewise.
14731         (eqv_v2df3): Likewise.
14732         (nand): Likewise.
14733         (nand_v16qi3): Likewise.
14734         (nand_v8hi3): Likewise.
14735         (nand_v4si3): Likewise.
14736         (nand_v2di3): Likewise.
14737         (nand_v4sf3): Likewise.
14738         (nand_v2df3): Likewise.
14739         (orc): Likewise.
14740         (orc_v16qi3): Likewise.
14741         (orc_v8hi3): Likewise.
14742         (orc_v4si3): Likewise.
14743         (orc_v2di3): Likewise.
14744         (orc_v4sf3): Likewise.
14745         (orc_v2df3): Likewise.
14746
14747         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
14748         allow power8 quad mode in 64-bit.
14749         (rs6000_builtin_vectorized_function): Add support to vectorize
14750         ISA 2.07 count leading zeros, population count builtins.
14751         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
14752         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
14753         (builtin_function_type): Add vgbbd builtin function which takes an
14754         unsigned argument.
14755         (altivec_expand_vec_perm_const): Add support for new power8 merge
14756         instructions.
14757
14758         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
14759         that does not include TImdoe for use with 32-bit.
14760         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
14761         instructions.
14762         (UNSPEC_VSX_CVDPSPN): Likewise.
14763         (vsx_xscvdpspn): Likewise.
14764         (vsx_xscvspdpn): Likewise.
14765         (vsx_xscvdpspn_scalar): Likewise.
14766         (vsx_xscvspdpn_directmove): Likewise.
14767         (vsx_and<mode>3): Split logical operations into 32-bit and
14768         64-bit. Add support to do logical operations on TImode as well as
14769         VSX vector types.  Allow logical operations to be done in either
14770         VSX registers or in general purpose registers in 64-bit mode.  Add
14771         splitters if GPRs were used. For AND, add clobber of CCmode to
14772         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
14773         encoding.
14774         (vsx_and<mode>3_32bit): Likewise.
14775         (vsx_and<mode>3_64bit): Likewise.
14776         (vsx_ior<mode>3): Likewise.
14777         (vsx_ior<mode>3_32bit): Likewise.
14778         (vsx_ior<mode>3_64bit): Likewise.
14779         (vsx_xor<mode>3): Likewise.
14780         (vsx_xor<mode>3_32bit): Likewise.
14781         (vsx_xor<mode>3_64bit): Likewise.
14782         (vsx_one_cmpl<mode>2): Likewise.
14783         (vsx_one_cmpl<mode>2_32bit): Likewise.
14784         (vsx_one_cmpl<mode>2_64bit): Likewise.
14785         (vsx_nor<mode>3): Likewise.
14786         (vsx_nor<mode>3_32bit): Likewise.
14787         (vsx_nor<mode>3_64bit): Likewise.
14788         (vsx_andc<mode>3): Likewise.
14789         (vsx_andc<mode>3_32bit): Likewise.
14790         (vsx_andc<mode>3_64bit): Likewise.
14791         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
14792         and xxlorc instructions.
14793         (vsx_eqv<mode>3_64bit): Likewise.
14794         (vsx_nand<mode>3_32bit): Likewise.
14795         (vsx_nand<mode>3_64bit): Likewise.
14796         (vsx_orc<mode>3_32bit): Likewise.
14797         (vsx_orc<mode>3_64bit): Likewise.
14798
14799         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
14800
14801         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
14802         instruction.
14803         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
14804         (p8_vmrgow): Likewise.
14805         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
14806         GPRs to be split under VSX.
14807         (p8v_clz<mode>2): Add power8 count leading zero support.
14808         (p8v_popcount<mode>2): Add power8 population count support.
14809         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
14810         support.
14811
14812         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
14813         instruction.
14814
14815         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
14816         builtin functions.
14817         (vec_nand): Likewise.
14818         (vec_vclz): Likewise.
14819         (vec_vclzb): Likewise.
14820         (vec_vclzd): Likewise.
14821         (vec_vclzh): Likewise.
14822         (vec_vclzw): Likewise.
14823         (vec_vgbbd): Likewise.
14824         (vec_vmrgew): Likewise.
14825         (vec_vmrgow): Likewise.
14826         (vec_vpopcnt): Likewise.
14827         (vec_vpopcntb): Likewise.
14828         (vec_vpopcntd): Likewise.
14829         (vec_vpopcnth): Likewise.
14830         (vec_vpopcntw): Likewise.
14831
14832 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14833
14834         PR rtl-optimization/57468
14835         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
14836         spilled pseudos.
14837
14838 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14839
14840         PR rtl-optimization/57459
14841         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
14842         type when setting live regs.
14843
14844 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
14845
14846         * config/s390/s390.opt (mlra): New option.
14847         * config/s390/s390.c (s390_decompose_address): Check displacement
14848         for all registers for LRA.
14849         (s390_secondary_reload): Don't used secondary reloads for LRA.
14850         (s390_lra_p): New function.
14851         (TARGET_LRA_P): Define.
14852         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
14853         of attribute cpu_facility to zarch for the last alternative.
14854         (*cmpmem_short): Ditto.
14855
14856 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
14857
14858         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
14859         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
14860         (arm_expand_prologue): Likewise.
14861
14862 2013-06-06  Teresa Johnson  <tejohnson@google.com>
14863
14864         PR c++/53743
14865         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
14866         as this is now done by redirect_edge_and_branch_force.
14867         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
14868         barriers, and fix interaction with splitting.
14869         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
14870         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
14871         reflect changes made in the routine.
14872         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
14873         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
14874         since this is called in cfglayout mode, and replace partition fixup
14875         with assert as that is now done by force_nonfallthru_and_redirect.
14876         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
14877         already be marked with region crossing note.
14878         (insert_section_boundary_note): Make non-static, gate on flag
14879         has_bb_partition, rewrite to also check for multiple partitions.
14880         (rest_of_handle_reorder_blocks): Remove call to
14881         insert_section_boundary_note, now done later during free_cfg.
14882         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
14883         * bb-reorder.h (insert_section_boundary_note): Declare.
14884         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
14885         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
14886         invoke insert_section_boundary_note.
14887         (try_redirect_by_replacing_jump): Remove unnecessary
14888         check for region crossing note.
14889         (fixup_partition_crossing): New function.
14890         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
14891         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
14892         in non-cfglayout mode.
14893         (force_nonfallthru_and_redirect): Fixup partition boundaries,
14894         remove old code that tried to do this. Emit barrier correctly
14895         when we are in cfglayout mode.
14896         (last_bb_in_partition): New function.
14897         (rtl_split_edge): Correctly fixup partition boundaries.
14898         (commit_one_edge_insertion): Remove old code that tried to
14899         fixup region crossing edge since this is now handled in
14900         split_block, and set up insertion point correctly since
14901         block may now end in a jump.
14902         (verify_hot_cold_block_grouping): Guard against checking when not in
14903         linearized RTL mode.
14904         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
14905         notes.
14906         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
14907         rtl_verify_flow_info, so not called in cfglayout mode.
14908         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
14909         (fixup_reorder_chain): Remove old code that attempted to fixup region
14910         crossing note as this is now handled in force_nonfallthru_and_redirect.
14911         (duplicate_insn_chain): Don't duplicate switch section notes.
14912         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
14913         note.
14914         * basic-block.h (emit_barrier_after_bb): Declare.
14915
14916 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14917
14918         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
14919         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
14920         arm_usatsihi): Adjust alternatives for arm_restrict_it.
14921
14922 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14923
14924         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
14925         where appropriate.
14926         * config/arm/ldmstm.md: Regenerate.
14927
14928 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14929
14930         * config/arm/sync.md (atomic_loaddi_1):
14931         Disable predication for arm_restrict_it.
14932         (arm_load_exclusive<mode>): Likewise.
14933         (arm_load_exclusivesi): Likewise.
14934         (arm_load_exclusivedi): Likewise.
14935         (arm_load_acquire_exclusive<mode>): Likewise.
14936         (arm_load_acquire_exclusivesi): Likewise.
14937         (arm_load_acquire_exclusivedi): Likewise.
14938         (arm_store_exclusive<mode>): Likewise.
14939         (arm_store_exclusive<mode>): Likewise.
14940         (arm_store_release_exclusivedi): Likewise.
14941         (arm_store_release_exclusive<mode>): Likewise.
14942
14943 2013-06-06  Richard Biener  <rguenther@suse.de>
14944
14945         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
14946         after LTO_null.
14947         (lto_tag_is_tree_code_p): Adjust.
14948         (lto_tag_is_gimple_code_p): Likewise.
14949         (lto_gimple_code_to_tag): Likewise.
14950         (lto_tag_to_gimple_code): Likewise.
14951         (lto_tree_code_to_tag): Likewise.
14952         (lto_tag_to_tree_code): Likewise.
14953         * data-streamer.h (streamer_write_hwi_in_range): Use
14954         uhwi streaming to stream the normalized range.
14955         (streamer_read_hwi_in_range): Likewise.
14956
14957 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14958
14959         * config/arm/arm.md (enabled_for_depr_it): New attribute.
14960         (predicable_short_it): Likewise.
14961         (predicated): Likewise.
14962         (enabled): Handle above.
14963         (define_cond_exec): Set predicated attribute to yes.
14964
14965 2013-06-05  Mike Stump  <mikestump@comcast.net>
14966
14967         * gdbinit.in (__FUNCTION__): Add.
14968
14969 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
14970
14971         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
14972         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
14973
14974 2013-06-05  Jan Hubicka  <jh@suse.cz>
14975
14976         * varasm.c (mark_decl_referenced): Revert the removal until targets
14977         are fixed.
14978
14979 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
14980
14981         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
14982         instead of mark_decl_referenced.
14983
14984 2013-06-05  Jan Hubicka  <jh@suse.cz>
14985
14986         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
14987         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
14988         and symtab_used_from_object_file_p.
14989         (cgraph_make_node_local_1): Clear forced_by_abi.
14990         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
14991         * cgraph.h (symtab_node_base): Add forced_by_abi.
14992         (decide_is_variable_needed): Remove.
14993         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
14994         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
14995         (decide_is_symbol_needed): ... this one; handle symbols in general;
14996         always analyze virtuals; honnor forced_by_abi.
14997         (cgraph_finalize_function): Update.
14998         (varpool_finalize_decl): Update.
14999         (symbol_defined_and_needed): Remove.
15000         (analyze_functions): Update.
15001         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15002         output_refs, input_overwrite_node): Handle forced_by_abi.
15003         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15004         (address_taken_from_non_vtable_p): ... this one.
15005         (comdat_can_be_unshared_p_1): New function.
15006         (cgraph_comdat_can_be_unshared_p): Rename to ...
15007         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15008         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15009         (function_and_variable_visibility): Clear forced_by_abi as needed.
15010         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15011         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15012         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15013         * varpool.c (decide_is_variable_needed): Remove.
15014
15015 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15016
15017         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15018         (arm_option_override): Override arm_restrict_it where appropriate.
15019         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15020         * config/arm/arm.opt (mrestrict-it): New command-line option.
15021         * doc/invoke.texi: Document -mrestrict-it.
15022
15023 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15024
15025         * tsan.c (tsan_atomic_table): Make const.
15026
15027 2013-06-05  Richard Biener  <rguenther@suse.de>
15028
15029         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15030         index associated with the tree we are supposed to replace.
15031         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15032         TREE_ASM_WRITTEN as zero for everything but SSA names.
15033
15034 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15035
15036         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15037
15038 2013-06-04  Jan Hubicka  <jh@suse.cz>
15039
15040         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15041         (input_node, input_varpool_node): Handle correctly external same
15042         body aliases.
15043         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15044         nodes at ltrans stage.
15045
15046 2013-06-04  Jan Hubicka  <jh@suse.cz>
15047
15048         * ipa-inline.c (update_caller_keys): Fix availability test.
15049         (update_callee_keys): Likewise.
15050         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15051         to follow ELF standard.
15052
15053 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15054
15055         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15056         (mips64r5900el-*-elf*): New configurations.
15057         * config/mips/mips-cpus.def (r5900): New processor.
15058         * config/mips/mips-tables.opt: Regenerate.
15059         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15060         (mips_issue_rate): Handle PROCESSOR_R5900.
15061         (mips_reorg_process_insns): Force reorder mode for the R5900.
15062         * config/mips/mips.h (TARGET_MIPS5900): Define.
15063         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15064         TARGET_MIPS5900.
15065         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15066         TARGET_MIPS5900.
15067         * config/mips/mips.md (processor): Add r5900.
15068         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15069
15070 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15071
15072         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15073         into function to generate MOVI instruction.
15074         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15075         (aarch64_preferred_simd_mode): Turn into wrapper.
15076         (aarch64_output_scalar_simd_mov_immediate): New function.
15077         * config/aarch64/aarch64-protos.h: Add prototype for above.
15078
15079 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15080
15081         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15082         element_char member.
15083         (sizetochar): Return signed char.
15084         (aarch64_simd_valid_immediate): Remove elchar and other
15085         unnecessary variables.
15086         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15087         Calculate element_char as required.
15088         * config/aarch64/aarch64-protos.h: Update and move prototype
15089         for aarch64_output_simd_mov_immediate.
15090         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15091         Update arguments.
15092
15093 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15094
15095         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15096         information completed by aarch64_simd_valid_immediate.
15097         (aarch64_legitimate_constant_p): Update arguments.
15098         (aarch64_simd_valid_immediate): Work with struct rather than many
15099         pointers.
15100         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15101         (aarch64_simd_make_constant): Update arguments.
15102         (aarch64_output_simd_mov_immediate): Work with struct rather than
15103         many pointers.  Output immediate directly rather than as operand.
15104         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15105         Update prototype.
15106         * config/aarch64/constraints.md (Dn): Update arguments.
15107
15108 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15109
15110         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15111         longer static.
15112         (aarch64_simd_immediate_valid_for_move): Remove.
15113         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15114         (aarch64_simd_make_constant): Update call.
15115         (aarch64_output_simd_mov_immediate): Update call.
15116         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15117         Add prototype.
15118         * config/aarch64/constraints.md (Dn): Update call.
15119
15120 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15121
15122         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15123         return type to bool for prototype.
15124         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15125         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15126         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15127
15128 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15129
15130         * config/mips/mips.opt (meva): New.
15131         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15132         (ASM_SPEC): Handle -meva.
15133         * doc/invoke.texi (meva):  Document.
15134
15135 2013-06-04  Alan Modra  <amodra@gmail.com>
15136
15137         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15138         constant output.
15139
15140 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15141
15142         * rtl.def: Add extra fourth optional field to define_cond_exec.
15143         * gensupport.c (process_one_cond_exec): Process attributes from
15144         define_cond_exec.
15145         * doc/md.texi: Document fourth field in define_cond_exec.
15146
15147 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15148
15149         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15150         out the processing order as in store_bit_field_1.
15151
15152 2013-06-04  Jan Hubicka  <jh@suse.cz>
15153
15154         PR middle-end/57500
15155         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15156         non-VAR_DECL node if it does not exist yet.
15157
15158 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15159
15160         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15161         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15162         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15163         target_cpu_default setting.
15164
15165 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15166
15167         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15168         default to -fopt-info=optimized instead of all.
15169         * doc/invoke.texi: Ditto.
15170         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15171         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15172         (execute_vect_slp): Emit BB vectorization success under
15173         MSG_OPTIMIZED_LOCATIONS.
15174         * tree-vect-slp.c (vect_slp_transform_bb): Change
15175         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15176         * tree-vect-loop.c (vect_transform_loop): Ditto.
15177
15178 2013-06-03  Jason Merrill  <jason@redhat.com>
15179
15180         PR c++/57415
15181         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15182         Use TARGET_EXPR for C++.
15183
15184 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15185
15186         PR rtl-optimization/57268
15187         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15188         if DEBUG_INSN_P (insn).
15189
15190         Reapply
15191         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15192
15193         PR rtl-optimization/57268
15194         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15195         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15196
15197 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15198
15199         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15200         (ix86_avoid_lea_for_addr): Likewise.
15201         (exact_dependency_1): Likewise.
15202         (ix86_adjust_cost): Likewise.
15203         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15204         removed.
15205         (do_reorder_for_imul): Fix typo, formatting and
15206         !reload_completed check removed.
15207         (ix86_sched_reorder): Fix typo and formatting.
15208         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15209         list.
15210
15211 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15212
15213         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15214
15215 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15216
15217         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15218         <STRING_CST>: Likewise.
15219         <VECTOR_CST>: Likewise.
15220
15221 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15222             Mikael Morin  <mikael@gcc.gnu.org>
15223
15224         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15225         * config.in: Regenerated.
15226         * configure: Regenerated.
15227
15228 2013-06-01  Jan Hubicka  <jh@suse.cz>
15229
15230         PR middle-end/57366
15231         * cgraphunit.c (compile): When weakref is not supported,
15232         set up transparent aliases before final output pass.
15233         * varasm.c (assemble_alias): Do not try to do it here.
15234
15235 2013-06-01  Jan Hubicka  <jh@suse.cz>
15236
15237         PR middle-end/57467
15238         * passes.c (for_per_function): Skip unanalyzed functions.
15239
15240 2013-06-01  Jan Hubicka  <jh@suse.cz>
15241
15242         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15243         (lto_symtab_merge_symbols_1): ... this one.
15244         (lto_symtab_merge_cgraph_nodes): Rename to ...
15245         (lto_symtab_merge_symbols): ... this one; simplify.
15246         * cgraph.c (same_body_aliases_done): Rename to ...
15247         (cpp_implicit_aliases_done): ... this one.
15248         (cgraph_create_function_alias): Update.
15249         (cgraph_same_body_alias): Update.
15250         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15251         (verify_edge_corresponds_to_fndecl): Simplify.
15252         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15253         (cgraph_node): Remove same_body_alias.
15254         (varpool_node): Remove alias_of and extra_name_alias.
15255         (same_body_aliases_done): Rename to ..
15256         (cpp_implicit_aliases_done): ... this one.
15257         (symtab_alias_ultimate_target): Add default parameter.
15258         (symtab_resolve_alias): New function.
15259         (fixup_same_cpp_alias_visibility): Declare.
15260         (cgraph_function_node): Add default parameter.
15261         (cgraph_node_asm_name): Likewise.
15262         (cgraph_function_or_thunk_node): Add default parameter; do
15263         not ICE when it is NULL.
15264         (varpool_variable_node): Likewise.
15265         * tree-emutls.c (create_emultls_var): Update.
15266         (ipa_lower_emutls): Update.
15267         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15268         (cgraph_reset_node): Reset alias info.
15269         (cgraph_finalize_function): Update.
15270         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15271         (analyze_function): Simplify.
15272         (cgraph_process_same_body_aliases): Simplify.
15273         (analyze_functions): Fixup same body aliases.
15274         (handle_alias_pairs): Simplify.
15275         (assemble_thunk): Update.
15276         (assemble_thunks_and_aliases): Update.
15277         (output_weakrefs): Rewrite.
15278         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15279         (lto_output_varpool_node): Likewise.
15280         (compute_ltrans_boundary): Remve assert.
15281         (get_alias_symbol): New functoin.
15282         (input_node): Rewrite alias handling.
15283         (input_varpool_node): Likewise.
15284         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15285         * ipa.c (process_references): Handle weakrefs correctly.
15286         (symtab_remove_unreachable_nodes): Likewise.
15287         * trans-mem.c (get_cg_data): Update.
15288         (ipa_tm_create_version_alias): Update.
15289         (ipa_tm_execute): Update.
15290         * symtab.c (dump_symtab_base): Dump aliases.
15291         (verify_symtab_base): Verify aliases.
15292         (symtab_node_availability): New function.
15293         (symtab_alias_ultimate_target): Simplify.
15294         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15295         handle all the fixup cases.
15296         (symtab_resolve_alias): New function.
15297         * passes.c (ipa_write_summaries): Handle weakrefs.
15298         * varpool.c (varpool_analyze_node): Simplify.
15299         (assemble_aliases): Update.
15300         (varpool_create_variable_alias): Simplify.
15301         (varpool_extra_name_alias): Simplify.
15302         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15303         (lto_symtab_merge_symbols): ... this one.
15304
15305 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15306
15307         Revert
15308         PR rtl-optimization/57268
15309         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15310         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15311
15312 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15313
15314         Partially reverted:
15315         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15316
15317         PR middle-end/57073
15318         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15319         further up.
15320
15321 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15322
15323         PR rtl-optimization/57268
15324         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15325         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15326
15327 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15328
15329         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15330         unordered comparison operators when -fno-trapping-math is in effect
15331         on the e500.
15332         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15333         and implement unordered comparison operators properly on the e500.
15334
15335 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15336
15337         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15338         for constant scalar integers.
15339         (simplify_relational_operation_1): Likewise.
15340
15341 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15342
15343         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15344         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15345         Fix comment.
15346
15347 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15348             Igor Zamyatin  <igor.zamyatin@intel.com>
15349
15350         Silvermont (SLM) architecture performance tuning.
15351         * config/i386/i386.h (enum ix86_tune_indices): Add
15352         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15353         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15354
15355         * config/i386/i386.c (initial_ix86_tune_features)
15356         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15357         (ix86_lea_outperforms): Handle Silvermont tuning.
15358         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15359         call.
15360         (ix86_use_lea_for_mov): Likewise.
15361         (ix86_avoid_lea_for_addr): Likewise.
15362         (ix86_lea_for_add_ok): Likewise.
15363         (exact_dependency_1): New function.
15364         (exact_store_load_dependency): Likewise.
15365         (ix86_adjust_cost): Handle Silvermont tuning.
15366         (do_reoder_for_imul): Likewise.
15367         (swap_top_of_ready_list): New function.
15368         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15369
15370         * config/i386/i386.md (peepholes that split memory operand in fp
15371         converts): New.
15372
15373 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15374
15375         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15376         Remove un-necessary braces.
15377
15378 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15379
15380         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15381         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15382
15383 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15384
15385         PR middle-end/57073
15386         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15387
15388 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15389
15390         PR target/56315
15391         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15392         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15393         * config/arm/neon.md (iordi3_neon): Remove.
15394         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15395         * config/arm/predicates.md (imm_for_neon_logic_operand):
15396         Move to earlier in the file.
15397         (neon_logic_op2): Likewise.
15398         (arm_iordi_operand_neon): New predicate.
15399
15400 2013-05-31  Richard Biener  <rguenther@suse.de>
15401
15402         PR tree-optimization/57478
15403         PR tree-optimization/57453
15404         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15405         are life as well.
15406
15407 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15408
15409         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15410         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15411
15412 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15413             Thomas Koenig  <tkoenig@gcc.gnu.org>
15414
15415         PR middle-end/57073
15416         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15417         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15418
15419 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15420
15421         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15422
15423 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15424
15425         * target.def (register_usage_leveling_p): New hook.
15426         * targhooks.c (default_register_usage_leveling_p): New.
15427         * targhooks.h (default_register_usage_leveling_p): New prototype.
15428         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15429         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15430         * doc/tm.texi: Update.
15431         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15432
15433 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15434
15435         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15436         (*insv_reg<mode>): New define_insn.
15437
15438 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15439
15440         PR rtl-optimization/57439
15441         * postreload.c (move2add_valid_value_p): Check that we have
15442         a zero subreg_regno_offset when accessing the register in
15443         the requested mode.
15444
15445 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15446             Igor Zamyatin  <igor.zamyatin@intel.com>
15447
15448         Silvermont (SLM) architecture pipeline model, tuning and
15449         insn selection.
15450         * config.gcc: Add slm config options and target.
15451
15452         * config/i386/slm.md: New.
15453
15454         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15455
15456         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15457         PROCESSOR_SLM.
15458         (ix86_target_macros_internal): Likewise.
15459
15460         * config/i386/i386.c (slm_cost): New cost.
15461         (m_SLM): New macro flag.
15462         (initial_ix86_tune_features): Set m_SLM.
15463         (x86_accumulate_outgoing_args): Likewise.
15464         (x86_arch_always_fancy_math_387): Likewise.
15465         (processor_target_table): Add slm cost.
15466         (cpu_names): Add slm cpu name.
15467         (x86_option_override_internal): Set SLM ISA.
15468         (ix86_issue_rate): New case PROCESSOR_SLM.
15469         (ia32_multipass_dfa_lookahead): Likewise.
15470         (fold_builtin_cpu): Add slm.
15471
15472         * config/i386/i386.h (TARGET_SLM): New target macro.
15473         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15474         (processor_type): Add PROCESSOR_SLM.
15475
15476         * config/i386/i386.md (cpu): Add new value "slm".
15477         (slm.md): Include slm.md.
15478
15479 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15480             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15481
15482         * config/arm/arm-protos.h: Add and update function protos.
15483         * config/arm/arm.c (use_simple_return_p): New added.
15484         (thumb2_expand_return): Check simple_return flag.
15485         * config/arm/arm.md: Add simple_return and conditional simple_return.
15486         * config/arm/iterators.md: Add iterator for return and simple_return.
15487
15488 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15489
15490         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15491         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15492         (arm_emit_vfp_multi_reg_pop): Likewise.
15493         (thumb2_emit_ldrd_pop): Likewise.
15494         (arm_expand_epilogue): Add misc REG_CFA notes.
15495         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15496
15497 2013-05-29  Lawrence Crowl  <crowl@google.com>
15498
15499         * config/arm/t-arm: Update for below.
15500
15501         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15502         Change type to hash_table.  Update dependent calls and types.
15503
15504         * config/i386/t-cygming: Update for below.
15505
15506         * config/i386/t-interix: Update for below.
15507
15508         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15509         Change type to hash_table.  Update dependent calls and types.
15510         (i386_find_on_wrapper_list::wrappers): Likewise.
15511
15512         * config/ia64/t-ia64: Update for below.
15513
15514         * config/ia64/ia64.c (bundle_state_table):
15515         Change type to hash_table.  Update dependent calls and types.
15516
15517         * config/mips/mips.c (mips_reorg_process_insns::htab):
15518         Change type to hash_table.  Update dependent calls and types.
15519
15520         * config/sol2.c (solaris_comdat_htab):
15521         Change type to hash_table.  Update dependent calls and types.
15522
15523         * config/t-sol2: Update for above.
15524
15525 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15526
15527         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15528         functions are not yet marked as defined.
15529
15530 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15531             Pat Haugen  <pthaugen@us.ibm.com>
15532             Peter Bergner  <bergner@vnet.ibm.com>
15533
15534         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15535         instructions.
15536         (VEC_A): Likewise.
15537         (VEC_C): Likewise.
15538         (vrotl<mode>3): Likewise.
15539         (vashl<mode>3): Likewise.
15540         (vlshr<mode>3): Likewise.
15541         (vashr<mode>3): Likewise.
15542
15543         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15544         support for power8 V2DI builtins.
15545
15546         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15547         power8 V2DI builtins.
15548         (vupkhsw): Likewise.
15549         (vupklsw): Likewise.
15550         (vaddudm): Likewise.
15551         (vminsd): Likewise.
15552         (vmaxsd): Likewise.
15553         (vminud): Likewise.
15554         (vmaxud): Likewise.
15555         (vpkudum): Likewise.
15556         (vpksdss): Likewise.
15557         (vpkudus): Likewise.
15558         (vpksdus): Likewise.
15559         (vrld): Likewise.
15560         (vsld): Likewise.
15561         (vsrd): Likewise.
15562         (vsrad): Likewise.
15563         (vsubudm): Likewise.
15564         (vcmpequd): Likewise.
15565         (vcmpgtsd): Likewise.
15566         (vcmpgtud): Likewise.
15567         (vcmpequd_p): Likewise.
15568         (vcmpgtsd_p): Likewise.
15569         (vcmpgtud_p): Likewise.
15570         (vupkhsw): Likewise.
15571         (vupklsw): Likewise.
15572         (vaddudm): Likewise.
15573         (vmaxsd): Likewise.
15574         (vmaxud): Likewise.
15575         (vminsd): Likewise.
15576         (vminud): Likewise.
15577         (vpksdss): Likewise.
15578         (vpksdus): Likewise.
15579         (vpkudum): Likewise.
15580         (vpkudus): Likewise.
15581         (vrld): Likewise.
15582         (vsld): Likewise.
15583         (vsrad): Likewise.
15584         (vsrd): Likewise.
15585         (vsubudm): Likewise.
15586
15587         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15588         support for power8 V2DI instructions.
15589
15590         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15591         power8 V2DI instructions.  Combine pack and unpack insns to use an
15592         iterator for each mode.  Check whether a particular mode supports
15593         Altivec instructions instead of just checking TARGET_ALTIVEC.
15594         (UNSPEC_VPKUWUM): Likewise.
15595         (UNSPEC_VPKSHSS): Likewise.
15596         (UNSPEC_VPKSWSS): Likewise.
15597         (UNSPEC_VPKUHUS): Likewise.
15598         (UNSPEC_VPKSHUS): Likewise.
15599         (UNSPEC_VPKUWUS): Likewise.
15600         (UNSPEC_VPKSWUS): Likewise.
15601         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15602         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15603         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15604         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15605         (UNSPEC_VUPKHSB): Likewise.
15606         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15607         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15608         (UNSPEC_VUPKHSH): Likewise.
15609         (UNSPEC_VUPKLSB): Likewise.
15610         (UNSPEC_VUPKLSH): Likewise.
15611         (VI2): Likewise.
15612         (VI_char): Likewise.
15613         (VI_scalar): Likewise.
15614         (VI_unit): Likewise.
15615         (VP): Likewise.
15616         (VP_small): Likewise.
15617         (VP_small_lc): Likewise.
15618         (VU_char): Likewise.
15619         (add<mode>3): Likewise.
15620         (altivec_vaddcuw): Likewise.
15621         (altivec_vaddu<VI_char>s): Likewise.
15622         (altivec_vadds<VI_char>s): Likewise.
15623         (sub<mode>3): Likewise.
15624         (altivec_vsubcuw): Likewise.
15625         (altivec_vsubu<VI_char>s): Likewise.
15626         (altivec_vsubs<VI_char>s): Likewise.
15627         (altivec_vavgs<VI_char>): Likewise.
15628         (altivec_vcmpbfp): Likewise.
15629         (altivec_eq<mode>): Likewise.
15630         (altivec_gt<mode>): Likewise.
15631         (altivec_gtu<mode>): Likewise.
15632         (umax<mode>3): Likewise.
15633         (smax<mode>3): Likewise.
15634         (umin<mode>3): Likewise.
15635         (smin<mode>3): Likewise.
15636         (altivec_vpkuhum): Likewise.
15637         (altivec_vpkuwum): Likewise.
15638         (altivec_vpkshss): Likewise.
15639         (altivec_vpkswss): Likewise.
15640         (altivec_vpkuhus): Likewise.
15641         (altivec_vpkshus): Likewise.
15642         (altivec_vpkuwus): Likewise.
15643         (altivec_vpkswus): Likewise.
15644         (altivec_vpks<VI_char>ss): Likewise.
15645         (altivec_vpks<VI_char>us): Likewise.
15646         (altivec_vpku<VI_char>us): Likewise.
15647         (altivec_vpku<VI_char>um): Likewise.
15648         (altivec_vrl<VI_char>): Likewise.
15649         (altivec_vsl<VI_char>): Likewise.
15650         (altivec_vsr<VI_char>): Likewise.
15651         (altivec_vsra<VI_char>): Likewise.
15652         (altivec_vsldoi_<mode>): Likewise.
15653         (altivec_vupkhsb): Likewise.
15654         (altivec_vupkhs<VU_char>): Likewise.
15655         (altivec_vupkls<VU_char>): Likewise.
15656         (altivec_vupkhsh): Likewise.
15657         (altivec_vupklsb): Likewise.
15658         (altivec_vupklsh): Likewise.
15659         (altivec_vcmpequ<VI_char>_p): Likewise.
15660         (altivec_vcmpgts<VI_char>_p): Likewise.
15661         (altivec_vcmpgtu<VI_char>_p): Likewise.
15662         (abs<mode>2): Likewise.
15663         (vec_unpacks_hi_v16qi): Likewise.
15664         (vec_unpacks_hi_v8hi): Likewise.
15665         (vec_unpacks_lo_v16qi): Likewise.
15666         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15667         (vec_unpacks_lo_v8hi): Likewise.
15668         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15669         (vec_pack_trunc_v8h): Likewise.
15670         (vec_pack_trunc_v4si): Likewise.
15671         (vec_pack_trunc_<mode>): Likewise.
15672
15673         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15674         V2DI builtins.
15675         (vec_vmaxsd): Likewise.
15676         (vec_vmaxud): Likewise.
15677         (vec_vminsd): Likewise.
15678         (vec_vminud): Likewise.
15679         (vec_vpksdss): Likewise.
15680         (vec_vpksdus): Likewise.
15681         (vec_vpkudum): Likewise.
15682         (vec_vpkudus): Likewise.
15683         (vec_vrld): Likewise.
15684         (vec_vsld): Likewise.
15685         (vec_vsrad): Likewise.
15686         (vec_vsrd): Likewise.
15687         (vec_vsubudm): Likewise.
15688         (vec_vupkhsw): Likewise.
15689         (vec_vupklsw): Likewise.
15690
15691 2013-05-29  Jan Hubicka  <jh@suse.cz>
15692
15693         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15694         flags; reorder rest of fields in more consistent way.
15695         (varpool_node): Remove analyzed, finalized and alias.
15696         (cgraph_ndoe): Likewise.
15697         (symtab_alias_ultimate_target): New function.
15698         (cgraph_function_node): Move offline.
15699         (cgraph_reset_node): Declare.
15700         (cgraph_comdat_can_be_unshared_p): Remove.
15701         (varpool_remove_initializer): Declare.
15702         (varpool_first_defined_variable, varpool_next_defined_variable
15703         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15704         (cgraph_function_with_gimple_body_p): Update.
15705         (varpool_all_refs_explicit_p): Update.
15706         (symtab_alias_target): New function.
15707         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15708         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15709         (cgraph_function_or_thunk_node): Simplify using
15710         symtab_alias_ultimate_target.
15711         (varpool_variable_node): Likewise.
15712         * cgraph.c (cgraph_create_function_alias): Update.
15713         (cgraph_add_thunk): Update.
15714         (cgraph_remove_node): Update.
15715         (dump_cgraph_node): Do not dump removed flags.
15716         (cgraph_function_body_availability): Update.
15717         (cgraph_propagate_frequency): Update.
15718         (verify_cgraph_node): Check sanity of local flag.
15719         (cgraph_function_node): Move here from cgraph.h; revamp for
15720         cgraph_function_or_thunk_node.
15721         * lto-symtab.c (lto_varpool_replace_node): Update.
15722         (lto_symtab_resolve_can_prevail_p): Update.
15723         (lto_symtab_merge_cgraph_nodes): Update.
15724         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15725         propagate_constants_accross_call, devirtualization_time_bonus,
15726         ipcp_propagate_stage): Update.
15727         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
15728         * ipa-inline-transform.c (clone_inlined_nodes,
15729         preserve_function_body_p): Update.
15730         * ipa-reference.c (propagate): Update.
15731         (write_node_summary_p): Update.
15732         * toplev.c (wrapup_global_declaration_2): Update.
15733         * cgraphunit.c (cgraph_analyze_function): Rename to ...
15734         (analyze_function) ... this one.
15735         (cgraph_process_new_functions): Update.
15736         (cgraph_reset_node): Export.
15737         (cgraph_finalize_function): Update.
15738         (cgraph_add_new_function): Update.
15739         (process_function_and_variable_attributes): Update.
15740         (varpool_finalize_decl): Update.
15741         (symbol_finalized): Remove.
15742         (symbol_finalized_and_needed): Rename to ...
15743         (symbol_defined_and_needed): ... update.
15744         (cgraph_analyze_functions): Update.
15745         (handle_alias_pairs): Update.
15746         (mark_functions_to_output): Update.
15747         (assemble_thunk): Update.
15748         (output_in_order): Update.
15749         (output_weakrefs): Update.
15750         (finalize_compilation_unit): Update.
15751         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
15752         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
15753         input_node, input_varpool_node): Update.
15754         * dbxout.c (dbxout_expand_expr): Update.
15755         * cgraphclones.c (cgraph_clone_node): Update.
15756         (cgraph_copy_node_for_versioning): Update.
15757         (cgraph_materialize_clone): Update.
15758         (cgraph_materialize_all_clones): Update.
15759         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
15760         propagate_pure_const, propagate_nothrow): Update.
15761         * lto-streamer-out.c (lto_output, write_symbol): Update.
15762         * ipa-utils.c (ipa_reverse_postorder): Update.
15763         * ipa-inline.c (can_inline_edge_p): Update.
15764         (update_caller_keys, ipa_inline): Update.
15765         * dwarf2out.c (reference_to_unused,
15766         premark_types_used_by_global_vars_helper): Update.
15767         * tree-eh.c (tree_could_trap_p): Update.
15768         * ipa-split.c (consider_split, execute_split_functions): Update.
15769         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
15770         has_addr_references_p): Update; move ahead in file for better
15771         readability.
15772         (process_references): Simplify.
15773         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
15774         bodies are removed.
15775         (cgraph_comdat_can_be_unshared_p): Make static.
15776         (cgraph_externally_visible_p): Update.
15777         (varpool_externally_visible_p): Update.
15778         (function_and_variable_visibility): Update.
15779         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
15780         ipa_tm_mark_force_output_node): Update.
15781         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
15782         estimate_edge_devirt_benefit, inline_generate_summary,
15783         inline_write_summary): Update.
15784         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
15785         * ipa-prop.c (ipa_compute_jump_functions): Update.
15786         (ipa_print_node_params, ipa_prop_read_section,
15787         ipa_update_after_lto_read, read_replacements_section): Update.
15788         * varasm.c (mark_decl_referenced): Update.
15789         (assemble_alias, dump_tm_clone_pairs): Update.
15790         * tree-inline.c (copy_bb): Update.
15791         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
15792         Update.
15793         * symtab.c (dump_symtab_base): Print new flags.
15794         (verify_symtab_base): Verify new flags.
15795         (symtab_alias_ultimate_target): New function.
15796         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
15797         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
15798         Update.
15799         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
15800         Update.
15801         * i386.c (ix86_get_function_versions_dispatcher,
15802         ix86_generate_version_dispatcher_body): Update.
15803         (fold_builtin_cpu): Use varpool_add_new_variable.
15804         * varpool.c (varpool_remove_initializer): Break out from ...
15805         (varpool_remove_node): ... this one.
15806         (dump_varpool_node, varpool_node_for_asm,
15807         cgraph_variable_initializer_availability, varpool_analyze_node,
15808         varpool_assemble_decl, varpool_remove_unreferenced_decls,
15809         varpool_finalize_named_section_flags, varpool_create_variable_alias):
15810         Update.
15811
15812 2013-05-29  Jan Hubicka  <jh@suse.cz>
15813
15814         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
15815
15816 2013-05-29  Easwaran Raman  <eraman@google.com>
15817
15818         PR tree-optimization/57442
15819         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
15820         when control exits the main loop.
15821
15822 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
15823
15824         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
15825         and RX600.
15826         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
15827         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
15828         * rx/t-rx: Add rx100 under multi library matches option for nofpu
15829         option.
15830
15831 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15832
15833         PR tree-optimization/57441
15834         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
15835         Don't limit size of incr_vec to number of candidates.
15836
15837 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
15838
15839         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
15840         and mips16 directories.
15841         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
15842         (MULTILIB_DIRNAMES): Ditto.
15843         (MULTILIB_EXCEPTIONS): Add new exceptions.
15844         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
15845         (MULTILIB_DIRNAMES): Ditto.
15846         (MULTILIB_EXCEPTIONS): Add new exceptions.
15847
15848 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
15849             Marcus Shawcroft  <marcus.shawcroft@arm.com>
15850
15851         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
15852         SYMBOL_TINY_ABSOLUTE.
15853         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
15854         SYMBOL_TINY_ABSOLUTE.
15855         (aarch64_expand_mov_immediate): Likewise.
15856         (aarch64_classify_symbol): Likewise.
15857         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
15858         Permit SYMBOL_TINY_ABSOLUTE.
15859         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
15860
15861 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
15862             Marcus Shawcroft  <marcus.shawcroft@arm.com>
15863
15864         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
15865         Refactor if/switch.  Replace gcc_assert with if.
15866
15867 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
15868
15869         * config/i386/i386.c (initial_ix86_tune_features): Enable
15870         FP Reassociation for AMD bdver1 and bdver2.
15871
15872 2013-05-29  Martin Jambor  <mjambor@suse.cz>
15873
15874         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
15875         and IMAGPART_EXPR do not occur within other handled_components.
15876
15877 2013-05-29  Richard Biener  <rguenther@suse.de>
15878
15879         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
15880         access on whether the use is in the BB we currently try to
15881         vectorize.
15882         (vect_bb_vectorization_profitable_p): Pass the BB we currently
15883         vectorize to vect_bb_slp_scalar_cost.
15884
15885 2013-05-29  Richard Biener  <rguenther@suse.de>
15886
15887         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
15888         computing scalar cost offsetted by stmts that are kept live
15889         by scalar uses.
15890         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
15891         for computation of scalar cost.
15892
15893 2013-05-28  Steve Ellcey  <sellcey@mips.com>
15894
15895         * config/mips/mips-cpus.def (mips32r2): Change processor type.
15896
15897 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
15898
15899         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
15900         array notation built-in reduction functions.
15901         * doc/passes.texi (Passes): Added documentation about changes done
15902         for Cilk Plus.
15903         * doc/invoke.texi (C Dialect Options): Added documentation about
15904         the -fcilkplus flag.
15905         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
15906         (BUILTINS_DEF): Depend on cilkplus.def.
15907         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
15908         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
15909         * cilkplus.def: New file.
15910
15911 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
15912
15913         PR rtl-optimization/57439
15914         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
15915
15916 2013-05-28  Easwaran Raman  <eraman@google.com>
15917
15918         PR tree-optimization/57337
15919         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
15920         (find_insert_point): Correctly identify the insertion point
15921         when two statements with the same UID is compared.
15922
15923 2013-05-28  Richard Biener  <rguenther@suse.de>
15924
15925         PR tree-optimization/56787
15926         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
15927         from the list of data references.
15928         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
15929         clobbers.
15930         (vect_analyze_loop_operations): Likewise.
15931         (vect_transform_loop): Remove clobbers.
15932
15933 2013-05-28  Martin Jambor  <mjambor@suse.cz>
15934
15935         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
15936         and REALPART_EXPRs have scalar type.
15937
15938 2013-05-28  Richard Biener  <rguenther@suse.de>
15939
15940         PR tree-optimization/57411
15941         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
15942         virtual operands.
15943         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
15944         virtual operand propagation.
15945
15946 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
15947
15948         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
15949         destination register for bmasksi_vis.
15950         (vector_init_bshuffle): Likewise.
15951         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
15952
15953 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
15954
15955         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
15956         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
15957         mode if the instruction isn't available in the original mode.
15958         * config/sparc/sparc.opt (mfix-ut699): New option.
15959         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
15960         (divdf3): Turn into expander.
15961         (divdf3_nofix): New insn.
15962         (divdf3_fix): Likewise.
15963         (divsf3): Disable if -mfix-ut699.
15964         (sqrtdf2): Turn into expander.
15965         (sqrtdf2_nofix): New insn.
15966         (sqrtdf2_fix): Likewise.
15967         (sqrtsf2): Disable if -mfix-ut699.
15968
15969 2013-05-27  Richard Biener  <rguenther@suse.de>
15970
15971         PR middle-end/57412
15972         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
15973         block for the new loop.
15974
15975 2013-05-27  Richard Biener  <rguenther@suse.de>
15976
15977         PR tree-optimization/57343
15978         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
15979         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
15980         (number_of_iterations_cond): Do not build the folded tree.
15981
15982 2013-05-27  Richard Biener  <rguenther@suse.de>
15983
15984         Revert
15985         PR middle-end/57381
15986         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
15987         OEP_CONSTANT_ADDRESS_OF retained.
15988
15989         PR tree-optimization/57417
15990         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
15991         for unchanged base.
15992         (set_ssa_val_to): Compare addresses using
15993         get_addr_base_and_unit_offset.
15994
15995 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
15996
15997         PR rtl-optimization/56833
15998         * postreload.c (move2add_record_mode): New function.
15999         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16000         (move2add_use_add2_insn): Use move2add_record_sym_value.
16001         (move2add_use_add3_insn): Likewise.
16002         (reload_cse_move2add): Use move2add_valid_value_p and
16003         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16004         affected regs by setting reg_mode to VOIDmode.
16005         (move2add_note_store): Don't pretend the inside of a SUBREG is
16006         the actual destination.  Invalidate single/leading registers by
16007         setting reg_mode to VOIDmode.
16008         Use move2add_record_sym_value, move2add_valid_value_p and
16009         move2add_record_mode.
16010
16011 2013-05-27  Richard Biener  <rguenther@suse.de>
16012
16013         PR tree-optimization/57396
16014         * tree-affine.c (double_int_constant_multiple_p): Properly
16015         return false for val == 0 and div != 0.
16016
16017 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16018
16019         * config/mips/mips.h: Use #elif in preprocessor conditions.
16020
16021 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16022
16023         PR target/53916
16024         * config/mips/constraints.md (kl): New constraint.
16025         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16026         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16027         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16028         from LO for MIPS16.
16029         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16030
16031 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16032
16033         PR target/55777
16034         * config/mips/mips.c (mips_can_inline_p): New function.
16035         (TARGET_CAN_INLINE_P): Define.
16036
16037 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16038
16039         * sched-int.h (ds_t, dw_t): Make unsigned int.
16040         Fix documentation that describes how all the ds_t bits are used.
16041         Reserve the last bit for delayed-branch scheduling.
16042         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16043         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16044         (gen_dep_weak_1): Remove prototype.
16045         * sched-deps.c (get_dep_weak_1): Make static.
16046         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16047         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16048         * doc/tm.texi: Regenerate.
16049         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16050
16051 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16052
16053         PR debug/56950
16054         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16055
16056 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16057             Sandra Loosemore  <sandra@codesourcery.com>
16058
16059         * config.gcc (powerpc-*): Allow native for with-cpu.
16060
16061 2013-05-24  Jeff Law  <law@redhat.com>
16062
16063         PR tree-optimization/57124
16064         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16065         conversion feeding a condition if the range has an overflow
16066         if -fstrict-overflow.  Add warnings for when we do make the
16067         transformation.
16068
16069 2013-05-24  Dehao Chen  <dehao@google.com>
16070
16071         * tree-cfg.c (locus_discrim_map): Fix the typo.
16072         (locus_discrim_hasher): Likewise.
16073         (locus_discrim_hasher::hash): Likewise.
16074         (locus_discrim_hasher::equal): Likewise.
16075
16076 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16077
16078         PR tree-optimization/57294
16079         * cgraph.h (ipa_record_stmt_references): Declare.
16080         * cgraphbuild.c (ipa_record_stmt_references): New function.
16081         (build_cgraph_edges): Use ipa_record_stmt_references.
16082         (rebuild_cgraph_edges): Likewise.
16083         (cgraph_rebuild_references): Likewise.
16084         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16085         associated with the old statement and build references from the
16086         newly built statements.
16087         * ipa-ref.c (ipa_remove_stmt_references): New function.
16088         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16089
16090 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16091
16092         * lra-constraints.c (emit_spill_move): Use smaller mode for
16093         mem-mem moves.
16094         (check_and_process_move): Consider mem-reg moves for secondary
16095         too.
16096         (curr_insn_transform): Don't lose insns emitted before for
16097         secondary memory moves.
16098         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16099         reg set up in the current insn.
16100
16101 2013-05-24  Dehao Chen  <dehao@google.com>
16102
16103         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16104         hash function.
16105         (locus_descrim_hasher::equal): Likewise.
16106         (build_gimple_cfg): New discriminator assignment algorithm.
16107         (make_edges): Likewise.
16108         (next_discriminator_for_locus): Likewise.
16109         (same_line_p): Likewise.
16110         (assign_discriminators): Likewise.
16111         (make_cond_expr_edges): Likewise.
16112         (make_gimple_switch_edges): Likewise.
16113         (make_goto_expr_edges): Likewise.
16114         (make_gimple_asm_edges): Likewise.
16115
16116 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16117
16118         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16119         X format specifier to only display bottom 16 bits.
16120         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16121         immediate to match for operand 2, since it will be masked.
16122
16123 2013-05-24  Richard Biener  <rguenther@suse.de>
16124
16125         PR tree-optimization/57287
16126         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16127         all SSA names that occur in abnormal PHIs.
16128
16129 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16130
16131         PR tree-ssa/57385
16132         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16133         that index is not negative.
16134
16135 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16136
16137         PR rtl-optimization/55177
16138         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16139         (simplify_byte_swapping_operation): New.
16140         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16141         (simplify_relational_operation_1): Deal with BSWAP.
16142
16143 2013-05-23  Richard Henderson  <rth@redhat.com>
16144
16145         PR target/56742
16146         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16147         (ix86_reorg): Call it.
16148
16149 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16150
16151         PR target/57379
16152         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16153         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16154         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16155
16156 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16157
16158         PR debug/57351
16159         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16160
16161 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16162             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16163
16164         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16165         * config/aarch64/constraints.md (Usa): Remove.
16166         * doc/md.texi (AArch64 Usa): Remove.
16167
16168 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16169             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16170
16171         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16172         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16173         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16174         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16175
16176 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16177
16178         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16179         instruction (AdvSIMD).
16180         * config/aarch64/aarch64-builtins.c
16181         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16182         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16183
16184 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16185
16186         PR middle-end/57347
16187         * tree.h (contains_bitfld_component_ref_p): Declare.
16188         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16189         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16190         caller.
16191         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16192         not access a bit-field.  Assert all final offsets are byte-aligned.
16193
16194 2013-05-23  Richard Biener  <rguenther@suse.de>
16195
16196         PR tree-optimization/57380
16197         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16198         least one invariant or re-used load.
16199         * passes.c (init_optimization_passes): Move pass_phiprop before
16200         pass_forwprop.
16201
16202 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16203
16204         * config/aarch64/aarch64-simd.md
16205         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16206
16207 2013-05-23  Richard Biener  <rguenther@suse.de>
16208
16209         PR middle-end/57381
16210         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16211         OEP_CONSTANT_ADDRESS_OF retained.
16212
16213 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16214
16215         PR middle-end/57344
16216         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16217         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16218
16219 2013-05-23  Richard Biener  <rguenther@suse.de>
16220
16221         PR rtl-optimization/57341
16222         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16223         instead of true_dependence.
16224
16225 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16226
16227         * bb-reorder.c (branch_threshold): Make const.
16228         (exec_threshold): Ditto.
16229
16230 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16231             Pat Haugen  <pthaugen@us.ibm.com>
16232             Peter Bergner  <bergner@vnet.ibm.com>
16233
16234         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16235         documentation for the power8 crypto builtins.
16236
16237         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16238
16239         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16240         macros for defining power8 builtin functions.
16241         (BU_P8V_AV_2): Likewise.
16242         (BU_P8V_AV_P): Likewise.
16243         (BU_P8V_VSX_1): Likewise.
16244         (BU_P8V_OVERLOAD_1): Likewise.
16245         (BU_P8V_OVERLOAD_2): Likewise.
16246         (BU_CRYPTO_1): Likewise.
16247         (BU_CRYPTO_2): Likewise.
16248         (BU_CRYPTO_3): Likewise.
16249         (BU_CRYPTO_OVERLOAD_1): Likewise.
16250         (BU_CRYPTO_OVERLOAD_2): Likewise.
16251         (XSCVSPDP): Fix typo, point to the correct instruction.
16252         (VCIPHER): Add power8 crypto builtins.
16253         (VCIPHERLAST): Likewise.
16254         (VNCIPHER): Likewise.
16255         (VNCIPHERLAST): Likewise.
16256         (VPMSUMB): Likewise.
16257         (VPMSUMH): Likewise.
16258         (VPMSUMW): Likewise.
16259         (VPERMXOR_V2DI): Likewise.
16260         (VPERMXOR_V4SI: Likewise.
16261         (VPERMXOR_V8HI: Likewise.
16262         (VPERMXOR_V16QI: Likewise.
16263         (VSHASIGMAW): Likewise.
16264         (VSHASIGMAD): Likewise.
16265         (VPMSUM): Likewise.
16266         (VPERMXOR): Likewise.
16267         (VSHASIGMA): Likewise.
16268
16269         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16270         __CRYPTO__ if the crypto instructions are available.
16271         (altivec_overloaded_builtins): Add support for overloaded power8
16272         builtins.
16273
16274         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16275         support for power8 crypto builtins.
16276         (builtin_function_type): Likewise.
16277         (altivec_init_builtins): Add support for builtins that take vector
16278         long long (V2DI) arguments.
16279
16280         * config/rs6000/crypto.md: New file, define power8 crypto
16281         instructions.
16282
16283 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16284             Pat Haugen  <pthaugen@us.ibm.com>
16285             Peter Bergner  <bergner@vnet.ibm.com>
16286
16287         * doc/invoke.texi (Option Summary): Add power8 options.
16288         (RS/6000 and PowerPC Options): Likewise.
16289
16290         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16291         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16292         wm, wn, wr documentation.
16293
16294         * config/rs6000/constraints.md (wm): New constraint for VSX
16295         registers if direct move instructions are enabled.
16296         (wn): New constraint for no registers.
16297         (wq): New constraint for quad word even GPR registers.
16298         (wr): New constraint if 64-bit instructions are enabled.
16299         (wv): New constraint if power8 vector instructions are enabled.
16300         (wQ): New constraint for quad word memory locations.
16301
16302         * config/rs6000/predicates.md (const_0_to_15_operand): New
16303         constraint for 0..15 for crypto instructions.
16304         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16305         as GPR and floating point registers.
16306         (int_reg_operand): New predicate to match only GPR registers.
16307         (base_reg_operand): New predicate to match base registers.
16308         (quad_int_reg_operand): New predicate to match even GPR registers
16309         for quad memory operations.
16310         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16311         operations in both GPR and VSX registers.
16312         (quad_memory_operand): New predicate for quad memory operations.
16313         (reg_or_indexed_operand): New predicate for direct move support.
16314
16315         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16316         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16317         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16318         (POWERPC_MASKS): Add power8 options.
16319         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16320         various options.
16321
16322         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16323         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16324
16325         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16326         (-mpower8-fusion): New power8 options.
16327         (-mpower8-fusion-sign): Likewise.
16328         (-mpower8-vector): Likewise.
16329         (-mcrypto): Likewise.
16330         (-mdirect-move): Likewise.
16331         (-mquad-memory): Likewise.
16332
16333         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16334         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16335         registers.
16336         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16337         (rs6000_debug_vector_unit): Add p8_vector.
16338         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16339         definitions.  Also print fusion state.
16340         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16341         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16342         (rs6000_option_override_internal): Add support for power8.
16343         (rs6000_common_init_builtins): Add debugging for skipped builtins
16344         if -mdebug=builtin.
16345         (rs6000_adjust_cost): Add power8 support.
16346         (rs6000_issue_rate): Likewise.
16347         (insn_must_be_first_in_group): Likewise.
16348         (insn_must_be_last_in_group): Likewise.
16349         (force_new_group): Likewise.
16350         (rs6000_register_move_cost): Likewise.
16351         (rs6000_opt_masks): Likewise.
16352
16353         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16354         power8 capable assembler, default to power7 options.
16355         (TARGET_DIRECT_MOVE): Likewise.
16356         (TARGET_CRYPTO): Likewise.
16357         (TARGET_P8_VECTOR): Likewise.
16358         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16359         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16360         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16361         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16362         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16363         (TARGET_XSCVDPSPN): Likewise.
16364         (TARGET_XSCVSPDPN): Likewsie.
16365         (TARGET_SYNC_HI_QI): Likewise.
16366         (TARGET_SYNC_TI): Likewise.
16367         (MASK_CRYPTO): Likewise.
16368         (MASK_DIRECT_MOVE): Likewise.
16369         (MASK_P8_FUSION): Likewise.
16370         (MASK_P8_VECTOR): Likewise.
16371         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16372         temporary used by some of the direct move instructions to get two FP
16373         temporary registers does not force creation of a stack frame.
16374         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16375         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16376         that any VSX registers are tieable, even if they are also an
16377         Altivec vector mode.
16378         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16379         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16380         (RS6000_BTM_CRYPTO): Likewise.
16381         (RS6000_BTM_COMMON): Likewise.
16382
16383         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16384         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16385         (enum rs6000_vector): Add power8 vector support.
16386
16387 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16388
16389         PR target/19599
16390         PR target/57340
16391         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16392         (any_sibcall_could_use_r3): this and handle indirect calls.
16393         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16394
16395 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16396
16397         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16398
16399 2013-05-22  Richard Biener  <rguenther@suse.de>
16400
16401         PR middle-end/57349
16402         * profile.c (branch_prob): Do not split blocks that are
16403         abnormally receiving from ECF_RETURNS_TWICE functions.
16404
16405 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16406
16407         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16408         address mode.  Move pointer mode initialization to the same place.
16409
16410 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16411
16412         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16413         while it has any effect.
16414
16415 2013-05-21  Easwaran Raman  <eraman@google.com>
16416
16417         PR tree-optimization/57322
16418         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16419         UID of the statement added to the BB to be 1.
16420
16421 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16422
16423         PR tree-optimization/57331
16424         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16425         of conversion from pointer type to integral type with integer.
16426
16427 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16428
16429         PR lto/57289
16430         * ipa-prop.c (ipa_read_node_info): Process param_used and
16431         controlled_uses in the same order as when writing.
16432
16433 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16434
16435         PR plugins/56754
16436         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16437
16438 2013-05-21  Richard Biener  <rguenther@suse.de>
16439
16440         PR tree-optimization/57318
16441         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16442         estimate stmts with side-effects as likely eliminated.
16443
16444 2013-05-21  Richard Biener  <rguenther@suse.de>
16445
16446         PR tree-optimization/57330
16447         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16448         preserve the call stmts fntype.
16449
16450 2013-05-21  Richard Biener  <rguenther@suse.de>
16451
16452         PR tree-optimization/57303
16453         * tree-ssa-sink.c (statement_sink_location): Improve killing
16454         stmt detection and properly handle self-assignments.
16455
16456 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16457
16458         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16459         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16460         registers. Set register size out of the PARALLEL loop.
16461
16462 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16463
16464         PR target/56547
16465         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16466         (*fmasf4, *fmasf4_media): New insns.
16467
16468 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16469
16470         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16471         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16472         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16473         (mips_idiv_insns): Update the comments to say that the returned
16474         instruction counts are in units of BASE_INSN_LENGTH.
16475         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16476         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16477         using 2 rather than 4 as the length of indirect MIPS16 and
16478         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16479         length of a NOP.  Don't divide MIPS16 lengths by 2.
16480         (mips16_split_long_branches): Assume a branch is long if the
16481         length is greater than 4 rather than 8.
16482         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16483         rather than multiplying them by 2.  Multiply instruction counts
16484         by BASE_INSN_LENGTH rather than 4.
16485         (*jump_mips16, tls_get_tp_mips16_<mode>)
16486         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16487
16488 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16489
16490         * config/mips/mips.md (extended_mips16): Remove branch case.
16491         (length): Remove duplicated extended_mips16 test.
16492
16493 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16494
16495         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16496
16497 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16498
16499         * recog.h (Recog_data): Rename to...
16500         (recog_data_d): ...this.
16501         (recog_data): Update accordingly.
16502         * recog.c (recog_data): Likewise.
16503         * reload.c (save_recog_data): Likewise.
16504         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16505         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16506
16507 2013-05-17  Julian Brown  <julian@codesourcery.com>
16508
16509         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16510         found in a REG_EQUAL note, invalidate it.
16511
16512 2013-05-17  Easwaran Raman  <eraman@google.com>
16513
16514         * tree-ssa-reassoc.c (find_insert_point): New function.
16515         (insert_stmt_after): Likewise.
16516         (get_def_stmt): Likewise.
16517         (ensure_ops_are_available): Likewise.
16518         (not_dominated_by): Likewise.
16519         (rewrite_expr_tree): Do not move statements beyond what is
16520         necessary. Remove call to swap_ops_for_binary_stmt...
16521         (reassociate_bb): ... and move it here.
16522         (build_and_add_sum): Assign UIDs for new statements.
16523         (linearize_expr): Likewise.
16524         (do_reassoc): Renumber gimple statement UIDs.
16525
16526 2013-05-17  Jan Hubicka  <jh@suse.cz>
16527
16528         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16529         weakrefs.
16530         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16531         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16532         weakrefs.
16533         (output_weakrefs): Update.
16534
16535 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16536             Martin Jambor  <mjambor@suse.cz>
16537
16538         PR middle-end/57276
16539         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16540         value that corresponds to the given aggval is found in values vector.
16541
16542 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16543
16544         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16545         sse, sse2, sse3, ssse3 and sse4a flags to options.
16546
16547 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16548
16549         * gengtype-state.c: (s_expr_writer): New class, to handle
16550         prettifying of output layout of s-expressions.
16551         (state_writer): New class, to write out gtype.state.
16552         (state_written_type_count): Move this variable into member data of
16553         state_writer.
16554         (s_expr_writer::s_expr_writer): New code: constructor for new class
16555         (state_writer::state_writer(): ditto
16556         (s_expr_writer::write_new_line): New function
16557         (s_expr_writer::write_any_indent): ditto
16558         (s_expr_writer::begin_s_expr): ditto
16559         (s_expr_writer::end_s_expr): ditto
16560         (write_state_fileloc): convert to method of state_writer...
16561         (state_writer:: write_state_fileloc): ...and use methods of
16562         s_expr_writer to write indentation into the gtype.state output file
16563         to visually represent the hierarchical structure of the list
16564         structures
16565         (write_state_fields): ditto, renaming to...
16566         (state_writer::write_state_fields)
16567         (write_state_a_string): ditto, renaming to...
16568         (state_writer::write_state_a_string)
16569         (write_state_string_option): ditto, renaming to...
16570         (state_writer::write_state_string_option)
16571         (write_state_type_option): ditto, renaming to...
16572         (state_writer::write_state_type_option)
16573         (write_state_nested_option): ditto, renaming to...
16574         (state_writer::write_state_nested_option)
16575         (write_state_option): ditto, renaming to...
16576         (state_writer::write_state_option)
16577         (write_state_options): ditto, renaming to...
16578         (state_writer::write_state_options)
16579         (write_state_lang_bitmap): ditto, renaming to...
16580         (state_writer::write_state_lang_bitmap)
16581         (write_state_version): ditto, renaming to...
16582         (state_writer::write_state_version)
16583         (write_state_scalar_type): ditto, renaming to...
16584         (state_writer::write_state_scalar_type)
16585         (write_state_string_type): ditto, renaming to...
16586         (state_writer::write_state_string_type)
16587         (write_state_undefined_type): ditto, renaming to...
16588         (state_writer::write_state_undefined_type)
16589         (write_state_struct_union_type): ditto, renaming to...
16590         (state_writer::write_state_struct_union_type)
16591         (write_state_struct_type): ditto, renaming to...
16592         (state_writer::write_state_struct_type)
16593         (write_state_user_struct_type): ditto, renaming to...
16594         (state_writer::write_state_user_struct_type)
16595         (write_state_lang_struct_type): ditto, renaming to...
16596         (state_writer::write_state_lang_struct_type)
16597         (write_state_param_struct_type): ditto, renaming to...
16598         (state_writer::write_state_param_struct_type)
16599         (write_state_pointer_type): ditto, renaming to...
16600         (state_writer::write_state_pointer_type)
16601         (write_state_array_type): ditto, renaming to...
16602         (state_writer::write_state_array_type)
16603         (write_state_gc_used): ditto, renaming to...
16604         (state_writer::write_state_gc_used)
16605         (write_state_common_type_content): ditto, renaming to...
16606         (state_writer::write_state_common_type_content)
16607         (write_state_type): ditto, renaming to...
16608         (state_writer::write_state_type)
16609         (write_state_pair_list): ditto, renaming to...
16610         (state_writer::write_state_pair_list)
16611         (write_state_pair): ditto, renaming to...
16612         (state_writer::write_state_pair)
16613         (write_state_typedefs): ditto, renaming to...
16614         (state_writer::write_state_typedefs)
16615         (write_state_structures): ditto, renaming to...
16616         (state_writer::write_state_structures)
16617         (write_state_param_structs): ditto, renaming to...
16618         (state_writer::write_state_param_structs)
16619         (write_state_variables): ditto, renaming to...
16620         (state_writer::write_state_variables)
16621         (write_state_srcdir): ditto, renaming to...
16622         (state_writer::write_state_srcdir)
16623         (write_state_files_list): ditto, renaming to...
16624         (state_writer::write_state_files_list)
16625         (write_state_languages): ditto, renaming to...
16626         (state_writer::write_state_languages)
16627         (write_state): create a state_writer instance and use it when
16628         writing out the state file
16629
16630 2013-05-17  Mike Stump  <mikestump@comcast.net>
16631
16632         PR rtl-optimization/57304
16633         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16634         accessing DF_REF_REAL_LOC.
16635
16636 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16637
16638         PR rtl-optimization/57281
16639         PR rtl-optimization/57300
16640         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16641         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16642         what the other splitter did if the registers are dead.
16643
16644 2013-05-17  Richard Biener  <rguenther@suse.de>
16645
16646         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16647         MEM_REF offsets.
16648
16649 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16650
16651         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16652         linking.
16653
16654 2013-05-17  Marek Polacek  <polacek@redhat.com>
16655
16656         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16657         length when doing non-zero store of storing '\0' to '\0'.
16658
16659 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16660
16661         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16662         vect_external_def oprnd1 with loop_vinfo, try to emit
16663         optional cast, negation and and stmts on the loop preheader
16664         edge instead of into the pattern def seq.
16665
16666         PR tree-optimization/57051
16667         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16668         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16669
16670 2013-05-16  Nick Clifton  <nickc@redhat.com>
16671
16672         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16673         (rl78_is_naked_func): New function.
16674         (rl78_expand_prologue): Skip prologue generation for naked functions.
16675         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16676         * doc/extend.texi (naked): Add RL78 to the list of processors
16677         that supports this attribute.
16678
16679 2013-05-16  Jeff Law  <law@redhat.com>
16680
16681         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16682
16683 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16684
16685         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16686         cache parameters using detect_caches_amd also for CYRIX,
16687         NSC and TM2 signatures.
16688
16689 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16690             Dzianis Kahanovich  <mahatma@eu.by>
16691
16692         PR target/45359
16693         PR target/46396
16694         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16695         VIA/Centaur processors and determine their cache parameters
16696         using detect_caches_amd.
16697
16698 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16699
16700         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16701         (rtl_verify_edges): New function.
16702         (rtl_verify_bb_insns): Ditto.
16703         (rtl_verify_bb_pointers): Ditto.
16704         (rtl_verify_bb_insn_chain): Ditto.
16705         (rtl_verify_fallthru): Ditto.
16706         (rtl_verify_bb_layout): Ditto.
16707         (rtl_verify_flow_info_1): Outline checks into new functions.
16708         (rtl_verify_flow_info): Ditto.
16709
16710 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16711
16712         * cfghooks.c (copy_bbs): Add update_dominance argument.
16713         * cfghooks.h (copy_bbs): Update prototype.
16714         * tree-cfg.c (gimple_duplicate_sese_region):
16715         Add update_dominance argument.
16716         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16717         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16718         gimple_duplicate_sese_region call.
16719         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16720         Update copy_bbs call.
16721         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16722         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16723
16724 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16725
16726         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16727         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
16728         vect_recog_rotate_pattern.
16729         (vect_recog_rotate_pattern): New function.
16730
16731 2013-05-16  Jason Merrill  <jason@redhat.com>
16732
16733         * Makefile.in (LLINKER): New variable.
16734         (mostlyclean): Remove link mutex.
16735         * configure.ac: Handle --enable-link-mutex.
16736         * lock-and-run.sh: New script.
16737
16738 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16739
16740         PR target/19599
16741         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
16742         for NULL decl.
16743
16744 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16745
16746         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
16747
16748 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16749
16750         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
16751         * config/arm/arm.c (next_consecutive_mem): New function.
16752         (gen_movmem_ldrd_strd): Likewise.
16753         * config/arm/arm.md (movmemqi): Update condition and code.
16754         (unaligned_loaddi, unaligned_storedi): New patterns.
16755
16756 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16757
16758         * config.gcc: Obsolete *-*-solaris2.9*.
16759         * doc/install.texi (Specific, *-*-solaris2*): Document it.
16760
16761 2013-05-16  Richard Biener  <rguenther@suse.de>
16762
16763         * passes.c (init_optimization_passes): Move pass_parallelize_loops
16764         earlier, after GRAPHITE transforms and IV canonicalization.
16765
16766 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16767
16768         * omp-low.c (extract_omp_for_data): For collapsed loops,
16769         if at least one of the loops is known at compile time to
16770         iterate zero times, set count to 0.
16771         (expand_omp_regimplify_p): New function.
16772         (expand_omp_for_generic): For collapsed loops, if at least
16773         one of the loops isn't known to iterate at least once,
16774         add runtime check with setting count to 0.
16775         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
16776         For unsigned types if it isn't known at compile time that
16777         the loop will iterate at least once, add runtime check to bypass
16778         the whole loop if initial condition isn't true.
16779
16780 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
16781
16782         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
16783
16784 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
16785
16786         PR middle-end/57286
16787         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
16788         transformations to avoid an infinite loop.
16789
16790 2013-05-16  Marek Polacek  <polacek@redhat.com>
16791
16792         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
16793
16794 2013-05-15  Leif Ekblad  <leif@rdos.net>
16795
16796         * config/i386/i386.c (ix86_decompose_address): Use
16797         DEFAULT_TLS_SEG_REG to access TLS segment register.
16798         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
16799         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
16800         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
16801
16802 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
16803
16804         PR target/57260
16805         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
16806         sibling calls to functions that would normally be lazily bound,
16807         unless $gp is call-clobbered.
16808
16809 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
16810
16811         * config/i386/i386.c (ix86_option_override_internal): Update
16812         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
16813         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
16814         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
16815         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
16816         of TARGET_3DNOW.
16817         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
16818
16819 2013-05-15  Andreas Schwab  <schwab@suse.de>
16820
16821         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
16822         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
16823         third operand.
16824
16825 2013-05-15  Teresa Johnson  <tejohnson@google.com>
16826
16827         * loop-unroll.c (report_unroll_peel): Check decision before
16828         emitting unroll/peel message.
16829
16830 2013-05-15  Teresa Johnson  <tejohnson@google.com>
16831
16832         * function.h (has_bb_partition): New rtl_data flag.
16833         (bb_reorder_complete): Ditto.
16834         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
16835         instead of flag_reorder_blocks_and_partition.
16836         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
16837         with some enhancements.
16838         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
16839         * bb-reorder.c (connect_traces): Check for has_bb_partition
16840         instead of flag_reorder_blocks_and_partition.
16841         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
16842         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
16843         verify_hot_cold_block_grouping.
16844         (partition_hot_cold_basic_blocks): Set has_bb_partition.
16845
16846 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16847
16848         PR target/19599
16849         * config/arm/predicates.md (call_insn_operand): New predicate.
16850         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
16851         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
16852         if insn is not a tail call.
16853         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
16854         registers.
16855         * config/arm/arm.h (enum reg_class): New caller save register class.
16856         (REG_CLASS_NAMES): Likewise.
16857         (REG_CLASS_CONTENTS): Likewise.
16858         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
16859         without decls.
16860
16861 2013-05-15  Richard Biener  <rguenther@suse.de>
16862
16863         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
16864         of MSG_OPTIMIZED_LOCATIONS.
16865         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
16866         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
16867         message.
16868         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
16869         of MSG_OPTIMIZED_LOCATIONS.
16870         (execute_vect_slp): Likewise.
16871         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
16872         (vect_create_cond_for_alias_checks): Likewise.
16873         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
16874         (vect_recog_widen_mult_pattern): Likewise.
16875         (vect_recog_widen_sum_pattern): Likewise.
16876         (vect_recog_over_widening_pattern): Likewise.
16877         (vect_recog_widen_shift_pattern): Likewise.
16878         (vect_recog_vector_vector_shift_pattern): Likewise.
16879         (vect_recog_divmod_pattern): Likewise.
16880         (vect_recog_mixed_size_cond_pattern): Likewise.
16881         (vect_recog_bool_pattern): Likewise.
16882         (vect_pattern_recog_1): Likewise.
16883
16884 2013-05-15  Martin Jambor  <mjambor@suse.cz>
16885
16886         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
16887         non-functions to builtin_unreachable.
16888         * ipa-inline-transform.c (inline_call): Do not assert estimates were
16889         correct when new direct edges were discovered.
16890
16891 2013-05-15  Martin Jambor  <mjambor@suse.cz>
16892
16893         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
16894         header, print symbol order instead of node uid, print more information
16895         about indirect edge targets.
16896         (ipa_make_edge_direct_to_target): Print symbol order instead of node
16897         uids.
16898         (ipa_make_edge_direct_to_target): Likewise.
16899         (remove_described_reference): Likewise.
16900         (propagate_controlled_uses): Likewise.
16901         (ipa_print_node_params): Also print symbol order.
16902         (ipcp_transform_function): Print symbol order instead of node uids.
16903         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
16904         (cgraph_get_create_real_symbol_node): Likewise.
16905         * ipa-cp.c (print_lattice): Likewise.
16906         (print_all_lattices): Likewise.
16907         (determine_versionability): Likewise.
16908         (initialize_node_lattices): Likewise.
16909         (estimate_local_effects): Likewise.
16910         (update_profiling_info): Likewise.
16911         (create_specialized_node): Likewise.
16912         (perhaps_add_new_callers): Likewise.
16913         (decide_about_value): Likewise.
16914         (decide_whether_version_node): Likewise.
16915         (identify_dead_nodes): Likewise.
16916         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
16917         (dump_inline_summary): Likewise.
16918         (estimate_node_size_and_time): Likewise.
16919         (inline_analyze_function): Likewise.
16920         * ipa-inline.c (report_inline_failed_reason): Likewise.
16921         (want_early_inline_function_p): Likewise.
16922         (edge_badness): Likewise.
16923         (update_edge_key): Likewise.
16924         (inline_small_functions): Likewise.  Add dumping of order to two other
16925         dumps.
16926         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
16927         instead of node uids.
16928         (propagate_pure_const): Likewise.
16929         (propagate_pure_const): Likewise.
16930         * ipa-utils.c (dump_cgraph_node_set): Likewise.
16931         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
16932         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
16933         of node uids.
16934         * tree-pretty-print.c (dump_function_header): Likewise.
16935         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
16936         Print symbol order instead of node uids.
16937
16938 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16939
16940         * config/s390/s390.c (s390_register_move_cost): Don't impose the
16941         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
16942
16943 2013-05-15  Richard Biener  <rguenther@suse.de>
16944
16945         PR tree-optimization/57275
16946         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
16947         return value for fail to do runtime alias checks for gather loads.
16948
16949 2013-05-15  Jan Hubicka  <jh@suse.cz>
16950
16951         PR lto/57038
16952         PR lto/47375
16953         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
16954         weakrefs are not external.
16955         (lto_symtab_merge_decls): Fix thinko when dealing with
16956         non-lto_symtab decls.
16957         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
16958         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
16959         * varpool.c (dump_varpool_node): Dump more flags.
16960
16961 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16962
16963         * config/i386/i386.c (processor_alias_table): Add instruction
16964         FSGSBASE for AMD bdver3 architecture.
16965
16966 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
16967
16968         * tree.c (warn_deprecated_use): Print file:line using locus color.
16969         * diagnostic.c (diagnostic_report_current_module): Print file:line
16970         and file:line:column using locus color.
16971
16972 2013-05-14  Mike Stump  <mikestump@comcast.net>
16973
16974         * gdbinit.in: Add __null.
16975
16976 2013-05-14  Mike Stump  <mikestump@comcast.net>
16977
16978         * recog.h: Rename struct recog_data to Recog_data.
16979         * recog.c: Likewise.
16980         * reload.c (can_reload_into): Likewise.
16981         * config/picochip/picochip.c: Likewise.
16982
16983 2013-05-14  Mike Stump  <mikestump@comcast.net>
16984
16985         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
16986
16987 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
16988
16989         * resource.h (struct resources): Remove unch_memory member.
16990         (CLEAR_RESOURCE): Don't clear unch_memory.
16991         * resource.c (mark_referenced_resources): Don't set it.
16992         (mark_set_resources): Likewise.
16993         (mark_target_live_regs): Don't clear it.
16994         (init_resource_info): Likewise.
16995         * reorg.c (resource_conflicts_p): Don't compare it.
16996         (redundant_insn): Don't set it.
16997
16998         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
16999         Remove prototypes.
17000         * emit-rtl.c (next_label): Remove unused function.
17001         (skip_consecutive_labels, link_cc0_insns): Move to ...
17002         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17003         only place where these functions are used, and make them static.
17004
17005 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17006
17007         * fold-const.c (fold_negate_expr): Handle vectors.
17008         (fold_truth_not_expr): Make it static.
17009         (fold_invert_truthvalue): New static function.
17010         (invert_truthvalue_loc): Handle vectors. Do not call
17011         fold_truth_not_expr directly.
17012         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17013         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17014         (fold_comparison): Handle vector constants.
17015         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17016         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17017         * tree.h (fold_truth_not_expr): Remove declaration.
17018
17019 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17020
17021         * config/aarch64/aarch64-simd.md
17022         (aarch64_vcond_internal<mode>): Rename to...
17023         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17024         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17025         float modes. Clarify all iterator modes.
17026         (vcond<mode><mode>): Use new name for vcond expanders.
17027         (vcond<v_cmp_result><mode>): Likewise.
17028         (vcondu<mode><mode>: Likewise.
17029         * config/aarch64/iterators.md (VDQF_COND): New.
17030
17031 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17032
17033         PR bootstrap/57266
17034         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17035         variable for the shift amount. Check that we shift by non-negative
17036         amounts.
17037
17038 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17039
17040         PR target/42017
17041         * config/arm/arm.h (EPILOGUE_USES): Only return true
17042         for LR_REGNUM after epilogue_completed.
17043
17044 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17045
17046         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17047         is error_mark_node.
17048
17049 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17050
17051         PR target/57261
17052         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17053         and Solaris 11+/x86 with gld.
17054         * configure: Regenerate.
17055
17056 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17057
17058         * expmed.c (expand_shift_1): Canonicalize rotates by
17059         constant bitsize / 2 to bitsize - 1.
17060         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17061         case ROTATERT>: Likewise.
17062
17063         Revert:
17064         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17065
17066         * config/i386/i386.md (rotateinv): New code attr.
17067         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17068         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17069         roll $31, %eax, etc.
17070
17071 2013-05-14  Richard Biener  <rguenther@suse.de>
17072
17073         PR middle-end/57235
17074         * tree-eh.c (sink_clobbers): Give up for successors with
17075         multiple predecessors and no virtual uses.
17076
17077 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17078
17079         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17080         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17081
17082 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17083
17084         PR middle-end/57251
17085         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17086         the case when both op0 and op1 have VOIDmode.
17087
17088 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17089
17090         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17091         in multiply-accumulate mode.
17092
17093 2013-05-13  Guozhi Wei  <carrot@google.com>
17094
17095         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17096
17097 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17098
17099         PR target/56975
17100         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17101         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17102         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17103         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17104         (ix86_expand_prologue): Likewise.
17105         (ix86_expand_split_stack_prologue): Likewise.
17106         (legitimate_pic_address_disp_p): Likewise.
17107         (legitimize_pic_address): Likewise.
17108         (legitimize_tls_address): Likewise.
17109         (legitimize_pe_coff_symbol): Likewise.
17110         (output_pic_addr_const): Likewise.
17111         (construct_plt_address): Likewise.
17112         (ix86_expand_call): Likewise.
17113         (x86_output_mi_thunk): Likewise.
17114         (x86_function_profiler): Likewise.
17115
17116 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17117
17118         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17119         similar switch cases.
17120         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17121         (aarch64_simd_mov_to_<mode>low): Delete.
17122         (aarch64_simd_mov_to_<mode>high): Delete.
17123         (move_lo_quad_<mode>): Add w<-r alternative.
17124         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17125         (aarch64_simd_mov_from_*): Update type attribute.
17126         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17127         statement.
17128
17129 2013-05-13  Jan Hubicka  <jh@suse.cz>
17130
17131         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17132         * config/i386/i386.c (ix86_compute_frame_layout,
17133         ix86_expand_epilogue, emit_i387_cw_initialization,
17134         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17135         ix86_local_alignment): Fix use of size/speed predicates.
17136
17137 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17138
17139         PR tree-optimization/45216
17140         PR tree-optimization/57157
17141         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17142         the (-Y) & (B - 1) variant if OP is |.
17143         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17144         return shifted.  Use (-op1) & (prec - 1) as other_amount
17145         instead of prec - op1.
17146
17147 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17148
17149         PR middle-end/42371
17150         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17151         (ipa_constant_data): New type.
17152         (ipa_jump_func): Use ipa_constant_data to hold information about
17153         constant jump functions.
17154         (ipa_get_jf_constant): Adjust to jump function type changes.
17155         (ipa_get_jf_constant_rdesc): New function.
17156         (ipa_param_descriptor): New field controlled_uses.
17157         (ipa_get_controlled_uses): New function.
17158         (ipa_set_controlled_uses): Likewise.
17159         * ipa-ref.h (ipa_find_reference): Declare.
17160         * ipa-prop.c (ipa_cst_ref_desc): New type.
17161         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17162         changes.
17163         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17164         New parameter cs.  Adjust all callers.
17165         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17166         (remove_described_reference): New function.
17167         (jfunc_rdesc_usable): Likewise.
17168         (try_make_edge_direct_simple_call): Decrement controlled use count,
17169         attempt to remove reference if it hits zero.
17170         (combine_controlled_uses_counters): New function.
17171         (propagate_controlled_uses): Likewise.
17172         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17173         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17174         (ipa_print_node_params): Print described use counter.
17175         (ipa_write_jump_function): Adjust to jump function type changes.
17176         (ipa_read_jump_function): New parameter CS, pass it to
17177         ipa_set_jf_constant.  Adjust caller.
17178         (ipa_write_node_info): Stream controlled use count
17179         (ipa_read_node_info): Likewise.
17180         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17181         asserting.
17182         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17183         count.  Remove cloning-added reference if it reaches zero.
17184         * ipa-ref.c (ipa_find_reference): New function.
17185
17186 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17187
17188         * config/i386/i386.c (processor_target_table): Modified default
17189         alignment values for AMD BD and BT architectures.
17190
17191 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17192
17193         * tree-vect-generic.c (uniform_vector_p): Move ...
17194         * tree.c (uniform_vector_p): ... here.
17195         * tree.h (uniform_vector_p): Declare it.
17196         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17197         into a scalar.
17198
17199 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17200
17201         PR tree-optimization/57230
17202         * tree-ssa-strlen.c (handle_char_store): Record length for
17203         array store from STRING_CST.
17204
17205         PR tree-optimization/57230
17206         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17207         check.
17208
17209 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17210
17211         * config/epiphany/epiphany.c (epiphany_init): Check size of
17212         NUM_MODES_FOR_MODE_SWITCHING.
17213         (epiphany_expand_prologue):
17214         Remove CONFIG_REGNUM initial value handling code.
17215         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17216         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17217         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17218         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17219         Don't return 1 for FP_MODE_NONE.
17220         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17221         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17222         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17223         * config/epiphany/epiphany.md (save_config): New pattern.
17224
17225 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17226
17227         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17228
17229 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17230
17231         * config/i386/i386.md (memory): Handle sseishft1.
17232         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17233         (*vec_extractv2di_1): Ditto.
17234
17235 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17236
17237         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17238         saved registers.
17239
17240 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17241
17242         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17243         Add mthumb/march=armv7-a multilib.
17244         Add mthumb/march=armv7-r multilib.
17245         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17246
17247 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17248
17249         * config/v850/t-rtems: Add more multilibs.
17250
17251 2013-05-10  Richard Biener  <rguenther@suse.de>
17252
17253         PR tree-optimization/57214
17254         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17255         not propagate from SSA names that occur in abnormal PHI nodes.
17256
17257 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17258
17259         * stor-layout.c (element_precision): New function.
17260         * machmode.h (element_precision): Declare it.
17261         * tree.c (build_minus_one_cst): New function.
17262         (element_precision): Likewise.
17263         * tree.h (build_minus_one_cst): Declare new function.
17264         (element_precision): Likewise.
17265         * fold-const.c (operand_equal_p): Use element_precision.
17266         (fold_binary_loc): Handle vector types.
17267         * convert.c (convert_to_integer): Use element_precision.
17268         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17269         separately.
17270
17271 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17272
17273         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17274         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17275         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17276         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17277         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17278         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17279         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17280         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17281         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17282         (Uuw8): New constraints.
17283         (Usb4): Move into alphabetical order.
17284         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17285         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17286         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17287         previously unnamed patterns.
17288         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17289         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17290         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17291         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17292         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17293         of specific lengths.
17294
17295 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17296
17297         * config/i386/i386.md (rotateinv): New code attr.
17298         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17299         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17300         roll $31, %eax, etc.
17301
17302         PR tree-optimization/45216
17303         PR tree-optimization/57157
17304         * tree-ssa-forwprop.c (simplify_rotate): New function.
17305         (ssa_forward_propagate_and_combine): Call it.
17306
17307 2013-05-10  Richard Biener  <rguenther@suse.de>
17308
17309         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17310         disable peeling when we version for aliasing.
17311         (vector_alignment_reachable_p): Honor explicit user alignment.
17312         (vect_supportable_dr_alignment): Likewise.
17313         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17314         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17315         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17316         then peeling to arrange for the cost-model check to come first.
17317
17318 2013-05-10  Alan Modra  <amodra@gmail.com>
17319
17320         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17321         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17322         * configure: Regenerate.
17323
17324 2013-05-10  Alan Modra  <amodra@gmail.com>
17325
17326         PR target/55033
17327         * varasm.c (default_elf_select_section): Move !DECL_P check..
17328         (get_named_section): ..to here before calling get_section_name.
17329         Adjust assertion.
17330         (default_section_type_flags): Add DECL_P check.
17331         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17332         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17333
17334 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17335
17336         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17337         When using gen_stack_adjust_str with a register offset, add a
17338         REG_FRAME_RELATED_EXPR note.
17339
17340 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17341
17342         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17343         (*vec_extractv4si_zext_mem): Ditto.
17344         (*vec_extractv2di): Add 0->x and x->x alternatives.
17345         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17346         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17347
17348 2013-05-09  Jason Merrill  <jason@redhat.com>
17349
17350         N3639 C++1y VLA support
17351         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17352         DECL_VALUE_EXPR.
17353
17354         * tree.c (build_constructor_va): New.
17355         * tree.h: Declare it.
17356
17357 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17358
17359         PR lto/57084
17360         * gimple-fold.c (canonicalize_constructor_val): Call
17361         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17362
17363 2013-05-09  Jan Hubicka  <jh@suse.cz>
17364             Richard Biener  <rguenther@suse.de>
17365
17366         PR lto/54095
17367         * symtab.c (symtab_make_decl_local): Do not add private names.
17368
17369 2013-05-09  Jan Hubicka  <jh@suse.cz>
17370
17371         PR lto/54095
17372         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17373         (unlink_from_assembler_name_hash): Likewise.
17374         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17375         symtab_unregister_node, symtab_initialize_asm_name_hash,
17376         change_decl_assembler_name): Update.
17377
17378 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17379
17380         * config/aarch64/aarch64.md: New movtf split.
17381         (*movtf_aarch64): Update.
17382         (aarch64_movdi_tilow): Handle TF modes and rename to
17383         aarch64_movdi_<mode>low.
17384         (aarch64_movdi_tihigh): Handle TF modes and rename to
17385         aarch64_movdi_<mode>high
17386         (aarch64_movtihigh_di): Handle TF modes and rename to
17387         aarch64_mov<mode>high_di
17388         (aarch64_movtilow_di): Handle TF modes and rename to
17389         aarch64_mov<mode>low_di
17390         (aarch64_movtilow_tilow): Remove spurious whitespace.
17391         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17392         splits.
17393         (aarch64_print_operand): Update.
17394
17395 2013-05-09  Alan Modra  <amodra@gmail.com>
17396
17397         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17398         powerpc64le.
17399         * configure: Regenerate.
17400
17401 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17402
17403         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17404         splitter preparation statements.
17405         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17406         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17407         adjust_address_nv.
17408
17409 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17410
17411         * gimple-ssa-strength-reduction.c (count_candidates): Change
17412         return value to int.
17413         (analyze_candidates_and_replace): Change type of length to int.
17414
17415 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17416
17417         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17418         (*vec_extract<mode>): Use VI12_128 mode iterator.
17419         (*vec_extract<mode>_mem): Ditto.
17420         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17421         attribute.
17422
17423 2013-05-08  Diego Novillo  <dnovillo@google.com>
17424
17425         PR bootstrap/54659
17426
17427         Revert:
17428         2012-08-17  Diego Novillo  <dnovillo@google.com>
17429
17430         PR bootstrap/54281
17431         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17432         * config.in: Regenerate.
17433         * configure: Regenerate.
17434         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17435
17436 2013-05-08  Jan Hubicka  <jh@suse.cz>
17437
17438         PR lto/54095
17439         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17440         * cgraph.h (symtab_node_base): Add unique_name.
17441         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17442         input_overwrite_node, input_varpool_node): Stream unique_name.
17443         * cgraphclones.c (cgraph_create_virtual_clone,
17444         cgraph_function_versioning): Set unique_name.
17445         * ipa.c (function_and_variable_visibility): Set unique_name.
17446
17447 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17448
17449         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17450         (alloc_cand_and_find_basis): Restrict conditional candidate
17451         processing to CAND_MULTs.
17452
17453 2013-05-08  Jan Hubicka  <jh@suse.cz>
17454
17455         PR lto/54095
17456         lto-symtab.c (lto_symtab_symbol_p): New function.
17457         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17458         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17459         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17460         Skip static symbols.
17461
17462 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17463
17464         PR tree-optimization/57200
17465         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17466         Only call inform if the preceding warning_at returns true.
17467
17468 2013-05-07  Han Shen  <shenhan@google.com>
17469
17470         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17471         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17472         * common.opt (fstack-protector-strong): New option.
17473         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17474         * doc/invoke.texi (Optimization Options): Document
17475         "-fstack-protector-strong".
17476         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17477
17478 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17479
17480         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17481
17482 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17483
17484         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17485         it is not lhs of the insn.
17486         (setup_reg_equiv): Remove insn having equiv memory which it is not
17487         lhs of the insn.
17488         * lra-constraints.c (process_address): Try to improve generation
17489         code for address base + disp.
17490         (lra_constraints): Make correct the code for checking insn setting
17491         up backward equivalence.  Remove insn only if it is in the init
17492         insn list.
17493         * lra-eliminations.c (update_reg_eliminate): Change return value.
17494         (lra_eliminate): Use the result.
17495
17496 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17497
17498         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17499         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17500         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17501         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17502         register target operands.
17503         (*vec_extractv8hi_sse2): New pattern.
17504         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17505         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17506         (*vec_extract<mode>_mem): New insn and split pattern.
17507
17508 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17509
17510         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17511         (TARGET_ASAN_SHADOW_OFFSET): Define.
17512         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17513         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17514
17515 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17516
17517         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17518         (incr_vec_index): Return -1 if increment not found.
17519         (create_add_on_incoming_edge): Assert if increment not found.
17520         (record_increment): Limit number of increments recorded.
17521         (all_phi_incrs_profitable): Return false if an increment not found.
17522         (replace_profitable_candidates): Don't process increments that were
17523         not recorded.
17524         (analyze_candidates_and_replace): Limit size of incr_vec.
17525
17526 2013-05-07  Richard Biener  <rguenther@suse.de>
17527
17528         * calls.c (special_function_p): setjmp-like functions are leaf.
17529         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17530         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17531
17532 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17533
17534         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17535         (aarch64_simd_mov<mode>): New expander.
17536         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17537         (aarch64_simd_mov_to_<mode>high): Likewise.
17538         (aarch64_simd_mov_from_<mode>low): Likewise.
17539         (aarch64_simd_mov_from_<mode>high): Likewise.
17540         (aarch64_dup_lane<mode>): Update.
17541         (aarch64_dup_lanedi): New instruction pattern.
17542         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17543         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17544
17545 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17546
17547         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17548         (replace_mult_candidate): Remove unnecessary argument; remove
17549         unnecessary parameter from call to introduce_cast_before_cand.
17550         (replace_unconditional_candidate): Remove unnecessary parameter
17551         from call to replace_mult_candidate.
17552         (replace_conditional_candidate): Likewise.
17553         (insert_initializers): Use make_temp_ssa_name.
17554         (introduce_cast_before_cand): Remove unnecessary argument; use
17555         make_temp_ssa_name.
17556         (replace_one_candidate): Remove unnecessary argument; remove
17557         unnecessary parameter from calls to introduce_cast_before_cand.
17558         (replace_profitable_candidates): Remove unnecessary parameters
17559         from calls to replace_one_candidate.
17560
17561 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17562
17563         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17564         phi def as possibly hiding a basis for a CAND_ADD whose operands
17565         have been commuted in the analysis.
17566         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17567
17568 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17569
17570         * config/aarch64/aarch64.md
17571         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17572         shift value between 0-4.
17573
17574 2013-05-07  Richard Biener  <rguenther@suse.de>
17575
17576         * double-int.h (rshift): New overload.
17577         * double-int.c (rshift): New function.
17578         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17579         (create_reference_ops_from_ref): Remove.
17580         (vn_reference_insert): Use shared ops for constructing the
17581         reference and copy it.
17582
17583 2013-05-07  Richard Biener  <rguenther@suse.de>
17584
17585         PR middle-end/57190
17586         * tree-eh.c (sink_clobbers): Properly propagate
17587         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17588
17589 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17590
17591         PR tree-optimization/57149
17592         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17593         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17594         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17595         uninit_undefined_value_p instead of ssa_undefined_value_p.
17596
17597         PR debug/57184
17598         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17599         for modifier == EXPAND_INITIALIZER.
17600
17601 2013-05-07  Anton Blanchard  <anton@samba.org>
17602
17603         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17604         for powerpc64 little endian.
17605         * configure: Regenerate.
17606
17607 2013-05-06  Graham Stott  <grahams@btinternet.com>
17608
17609         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17610         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17611         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17612         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17613
17614 2013-05-06  Graham Stott  <grahams@btinternet.com>
17615
17616         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17617         codes which allow non-lvalues.
17618
17619 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17620
17621         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17622         components are all 1s.
17623         (integer_minus_onep): New function.
17624         * tree.h (integer_minus_onep): Declare it.
17625         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17626         integer_minus_onep instead of integer_all_onesp.
17627
17628 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17629
17630         PR target/52933
17631         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17632         variations of these patterns.
17633
17634 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17635
17636         * config/i386/i386.md (isa): Add x64_sse4 member.
17637         (enabled): Handle x64_sse4.
17638         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17639         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17640         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17641         instruction for SSE4_1 targets.  Update insn attributes.
17642         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17643         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17644         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17645         const_1 selector.
17646         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17647         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17648         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17649
17650 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17651
17652         PR target/57108
17653         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17654
17655 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17656
17657         * final.c (do_assembler_dialects): Don't handle curly braces and
17658         vertical bar escaped by % as dialect delimiters.
17659         (output_asm_insn): Print curly braces and vertical bar if escaped
17660         by % and ASSEMBLER_DIALECT defined.
17661         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17662         * doc/tm.texi: Regenerated.
17663
17664 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17665
17666         * config/mips/mips.c: Include tree-pass.h.
17667         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17668         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17669         (pass_mips_machine_reorg2): New machine specific pass.
17670         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17671         (mips_option_override): Register the new pass.
17672         * rtl.h (cleanup_barriers): Remove prototype.
17673         (dbr_schedule): Likewise.
17674         * jump.c (cleanup_barriers): Make static.
17675         * reorg.c (dbr_schedule): Likewise.
17676
17677 2013-05-06  Richard Biener  <rguenther@suse.de>
17678
17679         PR tree-optimization/57185
17680         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17681         SSA names properly.
17682
17683 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17684
17685         PR target/57106
17686         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17687         between "first_arg" and "insn", not "last" and "insn".
17688
17689 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17690
17691         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17692         (find_candidates_in_block): Re-enable slsr_process_phi.
17693         (create_phi_basis): Fix double counting of candidate adjustment.
17694
17695 2013-05-06  Richard Biener  <rguenther@suse.de>
17696
17697         PR middle-end/57147
17698         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17699         the edge is also fallthru, preserve it and just clear the
17700         abnormal flag.
17701         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17702         also complex, preserve that and just clear the fallthru flag.
17703         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17704         update virtual operands.
17705
17706 2013-05-06  Alan Modra  <amodra@gmail.com>
17707
17708         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17709         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17710         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17711         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17712         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17713         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17714
17715 2013-05-06  Alan Modra  <amodra@gmail.com>
17716
17717         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17718         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17719         (DEFAULT_ASM_ENDIAN): Define.
17720         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17721         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17722         Update -K PIC clause from sysv4.h.
17723         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17724         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17725
17726 2013-05-06  Alan Modra  <amodra@gmail.com>
17727
17728         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
17729         twice for little-endian.
17730         (ashrdi3_no_power, ashrdi3): Support little-endian.
17731
17732 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17733
17734         PR target/55303
17735         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
17736         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
17737         related expanders.
17738         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
17739         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
17740         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
17741         New predicates.
17742
17743 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
17744             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17745
17746         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
17747         * config/pa/pa.opt: Make mbig-switch a no-op.
17748         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
17749         (CASE_VECTOR_MODE): Always return SImode.
17750         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
17751         for the !TARGET_BIG_SWITCH case.
17752         * config/pa/pa-linux.h: Likewise.
17753         * config/pa/pa-openbsd.h: Likewise.
17754         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
17755         * config/pa/pa.md (short_jump): Remove define_insn.
17756         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
17757         (casesi0): Remove define_insn.
17758         (type): Remove btable_branch.
17759         (pa_combine_type): Likewise.
17760         (in_nullified_branch_delay): Likewise.
17761         (in_call_delay): Likewise.
17762         (define_delay): Likewise.
17763         (define_insn_reservation "Z3"): Likewise.
17764         (define_insn_reservation "Z4"): Likewise.
17765         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
17766         (pa_adjust_insn_length): Remove adjustment for btable branches.
17767         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
17768         and mno-big-switch
17769
17770 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
17771
17772         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17773         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
17774         Add m->r,x alternatives.
17775         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
17776         splitters using SWI48x mode iterator.
17777         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
17778         TARGET_64BIT.  Add m->x alternative.
17779         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
17780         Add o->x alternative.  Enable for TARGET_SSE.
17781         (sse_storeq): Remove expander.
17782         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
17783         with memory input operand.
17784         (*vec_extractv2di_1 splitter): New.
17785         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
17786         * config/i386/i386.md (ssevecmodelower): New mode attribute.
17787
17788 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
17789
17790         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
17791         (INT_LOWPART): Delete.
17792         (extract_MB): Adjust.
17793         (extract_ME): Adjust.
17794         (print_operand): Adjust.
17795
17796 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
17797
17798         * config/rs6000/predicates.md (reg_or_add_cint_operand,
17799         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
17800         (reg_or_logical_cint_operand, easy_fp_constant,
17801         logical_const_operand): Delete "CONST_DOUBLE" case.
17802         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
17803         "HOST_BITS_PER_WIDE_INT == 64" test.
17804         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
17805         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
17806         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
17807         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
17808         test.
17809         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
17810         CONST_DOUBLE DImode/VOIDmode case.
17811         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
17812         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
17813         CONST_DOUBLE VOIDmode case.
17814         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
17815         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
17816         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
17817         Delete CONST_DOUBLE case.
17818         (splitters for mov FMOVE64 const_double): Delete
17819         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
17820         "HOST_BITS_PER_WIDE_INT >= 64" test.
17821         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
17822         case.
17823         (mov DI const_double): Delete.
17824
17825 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
17826
17827         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
17828         on op shows all bits zero in mode of a lowpart subreg, return zero.
17829
17830 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
17831
17832         PR target/57150
17833         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
17834         to save TFmode registers and DImode to save TImode registers for
17835         caller save operations.
17836         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
17837         mark being partially clobbered since they only use the first
17838         double word.
17839
17840         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
17841         and TDmode only use the upper 64-bits of each VSX register.
17842
17843 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17844
17845         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
17846         (find_candidates_in_block): Disable slsr_process_phi.
17847
17848 2013-05-03  Guozhi Wei  <carrot@google.com>
17849
17850         * coverage.c (coverage_obj_init): Move the construction of gcov
17851         constructor to ...
17852         (build_init_ctor): ... here.
17853
17854 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17855
17856         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
17857         (slsr_cand_d): Redefine def_phi.
17858         (stride_status, phi_adjust_status, count_phis_status): New enums.
17859         (find_phi_def): New.
17860         (find_basis_for_base_expr): New.
17861         (find_basis_for_candidate): Handle hidden bases.
17862         (alloc_cand_and_find_basis): Handle phi candidates.
17863         (slsr_process_phi): New.
17864         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
17865         (create_mul_imm_cand): Likewise.
17866         (create_add_ssa_cand): Exclude phi base candidates.
17867         (create_add_imm_cand): Likewise.
17868         (slsr_process_cast): Likewise.
17869         (slsr_process_copy): Likewise.
17870         (find_candidates_in_block): Handle phi candidates.
17871         (dump_candidate): Likewise.
17872         (unconditional_cands): Delete.
17873         (unconditional_cands_with_known_stride_p): Delete.
17874         (phi_dependent_cand_p): New.
17875         (cand_increment): Handle phi-dependent candidates.
17876         (replace_dependent): Delete.
17877         (replace_mult_candidate): New.
17878         (replace_unconditional_candidate): New.
17879         (incr_vec_index): Move to avoid forward reference.
17880         (create_add_on_incoming_edge): New.
17881         (create_phi_basis): New.
17882         (replace_dependents): Delete.
17883         (replace_conditional_candidate): New.
17884         (phi_add_costs): New.
17885         (replace_uncond_cands_and_profitable_phis): New.
17886         (record_increment): Handle phi adjustments.
17887         (record_phi_increments): New.
17888         (record_increments): Handle phi adjustments.
17889         (phi_incr_cost): New.
17890         (lowest_cost_path): Handle phis.
17891         (total_savings): Likewise.
17892         (analyze_increments): Likewise.
17893         (ncd_with_phi): New.
17894         (ncd_of_cand_and_phis): New.
17895         (nearest_common_dominator_for_cands): Handle phi increments.
17896         (all_phi_incrs_profitable): New.
17897         (replace_profitable_candidates): Handle phi-dependent candidates.
17898         (analyze_candidates_and_replace): Likewise.
17899
17900 2013-05-03  Teresa Johnson  <tejohnson@google.com>
17901
17902         PR bootstrap/57154
17903         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
17904         do not exceed REG_BR_PROB_BASE.
17905
17906 2013-05-03  Jeff Law  <law@redhat.com>
17907
17908         PR tree-optimization/57144
17909         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
17910         operand of the condition will bit into the new type when eliminating
17911         a cast feeding a condition.
17912
17913 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
17914
17915         PR rtl-optimization/57130
17916         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
17917         of COMPARE as in_code to the recursive call if needed.
17918
17919 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
17920
17921         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
17922         (enabled): Handle new members.
17923         * config/i386/sse.md (*vec_concatv2si): Merge from
17924         *vec_concatv2si_sse2 and vec_concatv2si_sse.
17925         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
17926
17927 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
17928
17929         PR tree-optimization/57027
17930         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
17931         for fnms opportunity, check we got the prerequisite kind
17932         of tree / gimple before using accessor functions.
17933
17934 2013-05-03  Richard Biener  <rguenther@suse.de>
17935
17936         * double-int.h (lshift): New overload without precision
17937         and arith argument.
17938         (operator *=, operator +=, operator -=): Move ...
17939         * double-int.c (operator *=, operator +=, operator -=): ... here
17940         and implement more efficiently.
17941         (mul_double_with_sign): Remove.
17942         (lshift_double): Adjust to take unsinged shift argument, push
17943         dispatching code to callers.
17944         (mul_double_wide_with_sign): Add early out for callers that
17945         are not interested in high parts or overflow.
17946         (lshift): New function.
17947         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
17948         dispatch code here.
17949         (lrotate, rrotate): Use logical shifts.
17950         * expr.c (get_inner_reference): Use lshift.
17951         * fixed-value.c (do_fixed_divide): Likewise.
17952         * tree-dfa.c (get_ref_base_and_extent): Likewise.
17953         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
17954         (indirect_refs_may_alias_p): Likewise.
17955         (stmt_kills_ref_p_1): Likewise.
17956
17957 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
17958
17959         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
17960
17961 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
17962
17963         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
17964         scalar form of FABD instruction.
17965
17966 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
17967
17968         * lra-constraints.c (process_alt_operands): Add checking alt
17969         number to choose the best alternative.
17970
17971 2013-05-02  Richard Biener  <rguenther@suse.de>
17972
17973         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
17974         bitmap and its handling.
17975         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
17976
17977 2013-05-02  Richard Biener  <rguenther@suse.de>
17978
17979         PR middle-end/57140
17980         * tree-inline.c (copy_loops): Properly handle removed loops.
17981         (copy_cfg_body): Mark destination loops for fixup if source
17982         loops needed fixup.
17983
17984 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
17985
17986         PR target/56732
17987         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
17988         generating simple_return for naked functions.
17989
17990 2013-05-02  Martin Jambor  <mjambor@suse.cz>
17991
17992         PR middle-end/56988
17993         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
17994         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
17995         flags match.
17996         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
17997         ipa_agg_replacement_value structures.
17998         (known_aggs_to_agg_replacement_list): Likewise.
17999         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18000         (read_agg_replacement_chain): Likewise.
18001         (ipcp_transform_function): Also check that by_ref flags match.
18002
18003 2013-05-02  Richard Biener  <rguenther@suse.de>
18004
18005         * graphds.h (struct graph): Add obstack member.
18006         * graphds.c (new_graph): Initialize obstack and allocate
18007         vertices from it.
18008         (add_edge): Allocate edge from the obstack.
18009         (free_graph): Free the obstack instead of all edges and vertices.
18010
18011 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18012
18013         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18014         divides.
18015         * cfg.c (update_bb_profile_for_threading): Ditto.
18016         * tree-inline.c (copy_bb): Ditto.
18017         (copy_edges_for_bb): Ditto.
18018         (initialize_cfun): Ditto.
18019         (copy_cfg_body): Ditto.
18020         (expand_call_inline): Ditto.
18021         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18022         (estimate_node_size_and_time): Ditto.
18023         (inline_merge_summary): Ditto.
18024         * cgraphclones.c (cgraph_clone_edge): Ditto.
18025         (cgraph_clone_node): Ditto.
18026         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18027         (compute_trg_info): Ditto.
18028
18029 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18030
18031         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18032         S reg when fp attribute set.
18033         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18034
18035 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18036
18037         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18038         New pattern.
18039         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18040         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18041         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18042
18043 2013-05-02  Richard Biener  <rguenther@suse.de>
18044
18045         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18046         (struct instantiate_cache_entry): New type.
18047         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18048         (struct instantiate_cache_type): New type.
18049         (set_instantiated_value, get_instantiated_value): Remove.
18050         (get_instantiated_value_entry): New function.
18051         (instantiate_scev_name): Use the new cache and adjust.
18052         (instantiate_scev_poly): Adjust.
18053         (instantiate_scev_binary): Likewise.
18054         (instantiate_array_ref): Likewise.
18055         (instantiate_scev_convert): Likewise.
18056         (instantiate_scev_not): Likewise.
18057         (instantiate_scev_3): Likewise.
18058         (instantiate_scev_2): Likewise.
18059         (instantiate_scev_r): Likewise.
18060         (instantiate_scev): Likewise.
18061         (resolve_mixers): Likewise.
18062
18063 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18064
18065         PR target/57091
18066         * lra-constraints.c (best_small_class_operands_num): Remove.
18067         (process_alt_operands): Remove small_class_operands_num.  Take
18068         small classes operands into losers and only if the operand is not
18069         matched.  Modify debugging output.
18070         (curr_insn_transform): Remove best_small_class_operands_num.
18071         Print insn name.
18072
18073 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18074
18075         * config/aarch64/aarch64-builtins.c
18076         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18077         * config/aarch64/aarch64-simd-builtins.def
18078         (reduc_splus_): Add new modes.
18079         (reduc_uplus_): New.
18080         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18081         (reduc_uplus_v4sf): Likewise.
18082         (reduc_splus_v4sf): Likewise.
18083         (aarch64_addv<mode>): Likewise.
18084         (reduc_uplus_<mode>): Likewise.
18085         (reduc_splus_<mode>): Likewise.
18086         (aarch64_addvv2di): Likewise.
18087         (reduc_uplus_v2di): Likewise.
18088         (reduc_splus_v2di): Likewise.
18089         (aarch64_addvv2si): Likewise.
18090         (reduc_uplus_v2si): Likewise.
18091         (reduc_splus_v2si): Likewise.
18092         (reduc_<sur>plus_<mode>): New.
18093         (reduc_<sur>plus_v2di): Likewise.
18094         (reduc_<sur>plus_v2si): Likewise.
18095         (reduc_<sur>plus_v4sf): Likewise.
18096         (aarch64_addpv4sf): Likewise.
18097         * config/aarch64/arm_neon.h
18098         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18099         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18100         add UNSPEC_SADDV, UNSPEC_UADDV.
18101         (SUADDV): New.
18102         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18103
18104 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18105
18106         * config/aarch64/arm_neon.h
18107         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18108
18109 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18110
18111         * config/aarch64/aarch64-builtins
18112         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18113
18114 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18115
18116         * config/aarch64/aarch64-simd-builtins.def
18117         (reduc_smax_): New.
18118         (reduc_smin_): Likewise.
18119         (reduc_umax_): Likewise.
18120         (reduc_umin_): Likewise.
18121         (reduc_smax_nan_): Likewise.
18122         (reduc_smin_nan_): Likewise.
18123         (fmax): Remove.
18124         (fmin): Likewise.
18125         (smax): Update for V2SF, V4SF and V2DF modes.
18126         (smin): Likewise.
18127         (smax_nan): New.
18128         (smin_nan): Likewise.
18129         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18130         (<su><maxmin><mode>3): ...This, refactor.
18131         (s<maxmin><mode>3): New.
18132         (<maxmin_uns><mode>3): Likewise.
18133         (reduc_<maxmin_uns>_<mode>): Refactor.
18134         (reduc_<maxmin_uns>_v4sf): Likewise.
18135         (reduc_<maxmin_uns>_v2si): Likewise.
18136         (aarch64_<fmaxmin><mode>: Remove.
18137         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18138         new builtin names.
18139         (vmin<q>_f<32,64>): Likewise.
18140         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18141         (FMAXMIN): New.
18142         (su): Add mappings for smax, smin, umax, umin.
18143         (maxmin): New.
18144         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18145         (FMAXMIN): Rename as...
18146         (FMAXMIN_UNS): ...This.
18147         (maxminv): Remove.
18148         (fmaxminv): Likewise.
18149         (fmaxmin): Likewise.
18150         (maxmin_uns): New.
18151         (maxmin_uns_op): Likewise.
18152
18153 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18154
18155         * config/aarch64/arm_neon.h
18156         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18157         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18158         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18159
18160 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18161
18162         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18163         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18164
18165 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18166
18167         * config/aarch64/aarch64-simd.md
18168         (vcond<mode>_internal): Handle special cases for constant masks.
18169         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18170         (vcondu<mode><mode>): Likewise.
18171         (vcond<v_cmp_result><mode>): New.
18172
18173 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18174
18175         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18176         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18177         * config/aarch64/aarch64-simd-builtins.def
18178         (cmeq): Update to BUILTIN_VALLDI.
18179         (cmgt): Likewise.
18180         (cmge): Likewise.
18181         (cmle): Likewise.
18182         (cmlt): Likewise.
18183         * config/aarch64/arm_neon.h
18184         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18185         to builtins or C as appropriate.
18186
18187 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18188
18189         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18190         (cmgeu): ...This.
18191         (cmhi): Rename to...
18192         (cmgtu): ...This.
18193         * config/aarch64/aarch64-simd.md
18194         (simd_mode): Add SF.
18195         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18196         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18197         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18198         (cstore<mode>_neg): ...This.
18199         * config/aarch64/iterators.md
18200         (VALLF): new.
18201         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18202         (COMPARISONS): New.
18203         (UCOMPARISONS): Likewise.
18204         (optab): Add missing comparisons.
18205         (n_optab): New.
18206         (cmp_1): Likewise.
18207         (cmp_2): Likewise.
18208         (CMP): Likewise.
18209         (cmp): Remove.
18210         (VCMP_S): Likewise.
18211         (VCMP_U): Likewise.
18212         (V_cmp_result): Add DF, SF modes.
18213         (v_cmp_result): Likewise.
18214         (v): Likewise.
18215         (vmtype): Likewise.
18216         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18217
18218 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18219
18220         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18221         define_insn to define_insn_and_split.
18222         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18223         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18224         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18225         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18226         (thumb2_negscc): Likewise.
18227
18228 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18229
18230         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18231
18232 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18233
18234         * config/arm/thumb2.md: Remove trailing whitespaces.
18235
18236 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18237
18238         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18239         Use gen_int_mode rather than GEN_INT.
18240
18241 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18242
18243         * value-prof.c (stream_in_histogram_value): Remove the strayed
18244         debug_gimple_stmt.
18245
18246 2013-04-30  Richard Biener  <rguenther@suse.de>
18247
18248         PR middle-end/57122
18249         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18250
18251 2013-04-30  Richard Biener  <rguenther@suse.de>
18252
18253         PR middle-end/57107
18254         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18255
18256 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18257
18258         PR rtl-optimization/56957
18259         PR rtl-optimization/57105
18260         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18261         variable.  Use just INSN_UID for determining whether an insn
18262         should be only disconnected from the insn stream.
18263         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18264
18265 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18266
18267         PR tree-optimization/57104
18268         * tsan.c (instrument_expr): Don't instrument accesses to
18269         DECL_HARD_REGISTER VAR_DECLs.
18270
18271 2013-04-30  Richard Biener  <rguenther@suse.de>
18272
18273         * function.h (loops_for_fn): New inline function.
18274         (set_loops_for_fn): Likewise.
18275         * cfgloop.h (place_new_loop): Add struct function parameter.
18276         (get_loop): Likewise.
18277         (get_loops): Likewise.
18278         (number_of_loops): Likewise.
18279         (fel_next): Adjust.
18280         (fel_init): Likewise.
18281         * cfg.c (get_loop_copy): Adjust.
18282         * cfgloop.c (flow_loops_dump): Likewise.
18283         (record_loop_exits): Likewise.
18284         (verify_loop_structure): Likewise.
18285         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18286         (estimate_reg_pressure_cost): Likewise.
18287         (mark_loop_exit_edges): Likewise.
18288         * cfgloopmanip.c (place_new_loop): Likewise.
18289         (add_loop): Likewise.
18290         (duplicate_loop): Likewise.
18291         * graph.c (draw_cfg_nodes): Likewise.
18292         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18293         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18294         (extract_affine_chrec): Likewise.
18295         (build_scop_iteration_domain): Likewise.
18296         * graphite.c (graphite_initialize): Likewise.
18297         * ira-build.c (create_loop_tree_nodes): Likewise.
18298         (more_one_region_p): Likewise.
18299         (rebuild_regno_allocno_maps): Likewise.
18300         (mark_loops_for_removal): Likewise.
18301         (mark_all_loops_for_removal): Likewise.
18302         (remove_unnecessary_regions): Likewise.
18303         (ira_build): Likewise.
18304         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18305         * loop-init.c (fix_loop_structure): Likewise.
18306         (gate_rtl_move_loop_invariants): Likewise.
18307         (gate_rtl_unswitch): Likewise.
18308         (gate_rtl_unroll_and_peel_loops): Likewise.
18309         (rtl_doloop): Likewise.
18310         * lto-streamer-in.c (input_cfg): Likewise.
18311         * lto-streamer-out.c (output_cfg): Likewise.
18312         * modulo-sched.c (sms_schedule): Likewise.
18313         * predict.c (tree_estimate_probability): Likewise.
18314         (tree_estimate_probability_driver): Likewise.
18315         (estimate_loops): Likewise.
18316         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18317         (move_sese_region_to_fn): Likewise.
18318         (debug_loop_num): Likewise.
18319         * tree-chrec.c (chrec_evaluate): Likewise.
18320         (hide_evolution_in_other_loops_than_loop): Likewise.
18321         (chrec_component_in_loop_num): Likewise.
18322         (reset_evolution_in_loop): Likewise.
18323         (evolution_function_is_invariant_rec_p): Likewise.
18324         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18325         * tree-inline.c (copy_loops): Likewise.
18326         (copy_cfg_body): Likewise.
18327         (tree_function_versioning): Likewise.
18328         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18329         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18330         Likewise.
18331         (add_to_evolution_1): Likewise.
18332         (scev_const_prop): Likewise.
18333         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18334         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18335         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18336         (tree_ssa_lim_initialize): Likewise.
18337         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18338         (verify_loop_closed_ssa): Likewise.
18339         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18340         (tree_ssa_loop_im): Likewise.
18341         (tree_ssa_loop_unswitch): Likewise.
18342         (tree_vectorize): Likewise.
18343         (check_data_deps): Likewise.
18344         (tree_ssa_loop_ivcanon): Likewise.
18345         (tree_ssa_loop_bounds): Likewise.
18346         (tree_complete_unroll): Likewise.
18347         (tree_complete_unroll_inner): Likewise.
18348         (tree_parallelize_loops): Likewise.
18349         (tree_ssa_loop_prefetch): Likewise.
18350         (tree_ssa_loop_ivopts): Likewise.
18351         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18352         * tree-vectorizer.c (vectorize_loops): Likewise.
18353
18354 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18355
18356         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18357         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18358         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18359         with EABI_LINK_SPEC.
18360
18361 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18362
18363         PR target/44578
18364         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18365         alternative.
18366
18367 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18368
18369         PR target/57097
18370         * lra-constraints.c (process_alt_operands): Discourage a bit more
18371         using memory for pseudos.  Print cost dump for alternatives.
18372         Modify cost values for conflicts with early clobbers.
18373         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18374
18375 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18376
18377         PR target/57098
18378         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18379
18380 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18381
18382         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18383         from/to S register.
18384         (movdi_aarch64): Support LDR/STR from/to D register.
18385
18386 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18387
18388         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18389         or higher by default.
18390
18391 2013-04-29  Richard Biener  <rguenther@suse.de>
18392
18393         PR middle-end/57075
18394         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18395         even if not adding abnormal edges for calls that can make
18396         abnormal gotos.
18397
18398 2013-04-29  Richard Biener  <rguenther@suse.de>
18399
18400         PR middle-end/57103
18401         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18402         TREE_BLOCK.
18403         (move_stmt_r): Remove redundant checking.
18404
18405 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18406
18407         PR bootstrap/57077
18408         * basic-block.h (apply_scale): New function.
18409         (apply_probability): Use apply_scale.
18410         * gimple-streamer-in.c (input_bb): Ditto.
18411         * lto-streamer-in.c (input_cfg): Ditto.
18412         * lto-cgraph.c (merge_profile_summaries): Ditto.
18413         * tree-optimize.c (execute_fixup_cfg): Ditto.
18414         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18415         (copy_edges_for_bb): Ditto.
18416         (copy_cfg_body): Ditto.
18417
18418 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18419
18420         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18421         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18422         (tail_merge_optimize): Handle current_loops == NULL.
18423
18424 2013-04-26  Jeff Law  <law@redhat.com>
18425
18426         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18427         (simplify_cond_using_ranges): Generalize code to simplify
18428         COND_EXPRs where one argument is a constant and the other
18429         is an SSA_NAME created by an integral type conversion.
18430
18431 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18432
18433         * config/arm/arm.md (store_minmaxsi): Use only when
18434         optimize_insn_for_size_p.
18435
18436 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18437
18438         PR target/57108
18439         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18440
18441 2013-04-29  Richard Biener  <rguenther@suse.de>
18442
18443         PR middle-end/57089
18444         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18445         loop tree make sure to schedule a fixup for the child as well.
18446         (expand_omp_for_generic): Properly add loops.
18447         (expand_omp_for_static_nochunk): Likewise.
18448         (expand_omp_for_static_chunk): Likewise.
18449         (expand_omp_for): For the degenerate case fixup loops.
18450         (expand_omp_sections): Fix default bb placement in loops.
18451         (expand_omp_atomic_pipeline): Properly add loops.
18452
18453 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18454
18455         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18456
18457 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18458
18459         * tree-ssa-tail-merge.c: Update header comment.
18460
18461 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18462
18463         * config/aarch64/arm_neon.h
18464         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18465         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18466         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18467         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18468         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18469         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18470         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18471         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18472
18473 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18474
18475         * config/aarch64/aarch64-simd.md
18476         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18477         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18478         fix_trunc, fixuns_trunc.
18479         (ftrunc<VDQF:mode>2): New.
18480         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18481         (fix_trunc_optab): New.
18482
18483 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18484
18485         * config/aarch64/aarch64-builtins.c
18486         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18487         iceilf, lround, iroundf.
18488
18489 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18490
18491         PR target/54349
18492         * config/i386/i386.h (enum ix86_tune_indices)
18493         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18494         New, split from X86_TUNE_INTER_UNIT_MOVES.
18495         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18496         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18497         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18498         (TARGET_INTER_UNIT_MOVES): Remove.
18499         * config/i386/i386.c (initial_ix86_tune_features): Update.
18500         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18501         (ix86_expand_convert_uns_didf_sse): Use
18502         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18503         (ix86_expand_vector_init_one_nonzero): Ditto.
18504         (ix86_expand_vector_init_interleave): Ditto.
18505         (inline_secondary_memory_needed): Return true for moves from SSE class
18506         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18507         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18508         * config/i386/constraints.md (Yi, Ym): Depend on
18509         TARGET_INTER_UNIT_MOVES_TO_VEC.
18510         (Yj, Yn): New constraints.
18511         * config/i386/i386.md (*movdi_internal): Change constraints of
18512         operand 1 from Yi to Yj and from Ym to Yn.
18513         (*movsi_internal): Ditto.
18514         (*movdf_internal): Ditto.
18515         (*movsf_internal): Ditto.
18516         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18517         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18518         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18519         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18520         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18521         * config/i386/sse.md (movdi_to_sse): Ditto.
18522         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18523         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18524         TARGET_INTER_UNIT_MOVES.
18525         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18526         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18527         instead of TARGET_INTER_UNIT_MOVES.
18528         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18529         operand 1 from Yi to Yj and from Ym to Yn.
18530
18531 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18532
18533         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18534         (float_truncate_hi_): Likewise.
18535         (float_extend_lo_): Likewise.
18536         (float_truncate_lo_): Likewise.
18537         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18538         (aarch64_float_extend_lo_v2df): Likewise.
18539         (vec_unpacks_hi_v4sf): Likewise.
18540         (aarch64_float_truncate_lo_v2sf): Likewise.
18541         (aarch64_float_truncate_hi_v4sf): Likewise.
18542         (vec_pack_trunc_v2df): Likewise.
18543         (vec_pack_trunc_df): Likewise.
18544
18545 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18546
18547         * config/aarch64/aarch64-builtins.c
18548         (aarch64_fold_builtin): Fold float conversions.
18549         * config/aarch64/aarch64-simd-builtins.def
18550         (floatv2si, floatv4si, floatv2di): New.
18551         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18552         * config/aarch64/aarch64-simd.md
18553         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18554         * config/aarch64/iterators.md (FLOATUORS): New.
18555         (optab): Add float, floatuns.
18556         (su_optab): Likewise.
18557
18558 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18559
18560         * config/aarch64/aarch64-builtins.c
18561         (aarch64_builtin_vectorized_function): Use new names for
18562         fcvt builtins.
18563         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18564         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18565         (fcvtzu): Split as...
18566         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18567         (fcvtas): Split as...
18568         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18569         (fcvtau): Split as...
18570         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18571         (fcvtps): Split as...
18572         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18573         (fcvtpu): Split as...
18574         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18575         (fcvtms): Split as...
18576         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18577         (fcvtmu): Split as...
18578         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18579         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18580         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18581         (lfrintnusf, lfrintnudf): Likewise.
18582         * config/aarch64/aarch64-simd.md
18583         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18584         define_insn.
18585         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18586         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18587         (fcvt_pattern): Likewise.
18588
18589 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18590
18591         * config/aarch64/aarch64-simd.md
18592         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18593         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18594
18595 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18596
18597         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18598         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18599         (vrnd<a,m,n,p>_f32): Implement using builtins.
18600         (vrnd<i,x><q>_f<32, 64>): New.
18601
18602 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18603
18604         * config/aarch64/aarch64-builtins.c
18605         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18606         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18607         (frintz): Rename to...
18608         (btrunc): ...this.
18609         (frintp): Rename to...
18610         (ceil): ...this.
18611         (frintm): Rename to...
18612         (floor): ...this.
18613         (frinti): Rename to...
18614         (nearbyint): ...this.
18615         (frintx): Rename to...
18616         (rint): ...this.
18617         (frinta): Rename to...
18618         (round): ...this.
18619         * config/aarch64/aarch64-simd.md
18620         (aarch64_frint<frint_suffix><mode>): Delete.
18621         (<frint_pattern><mode>2): Convert to insn.
18622         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18623         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18624         (frint_pattern): Likewise.
18625         (frint_suffix): Likewise.
18626
18627 2013-04-29  Richard Biener  <rguenther@suse.de>
18628
18629         PR tree-optimization/57081
18630         * loop-init.c: Include tree-flow.h.
18631         (loop_optimizer_finalize): Free number of iteration estimates.
18632         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18633
18634 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18635
18636         PR tree-optimization/57083
18637         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18638         non-singleton shift count range, zero extend low_bound for uns case.
18639
18640         * config/i386/predicates.md (general_vector_operand): New predicate.
18641         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18642         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18643         if they aren't nonimmediate operands.  If their original values
18644         satisfy const_vector_equal_evenodd_p, don't shift them.
18645         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18646         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18647         if not nonimmediate_operand.
18648         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18649         instead of register_operand.
18650         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18651
18652 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18653
18654         * stor-layout.c (finalize_size_functions): Allocate a structure and
18655         reset cfun before dumping the functions.
18656
18657 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18658
18659         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18660
18661         PR target/56866
18662         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18663         use xop_pmacsdqh if uns_p.
18664         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18665         the immediate rotate count.
18666
18667 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18668
18669         * rtl.h (struct rtx_def): Add comment for field jump.
18670         (LRA_SUBREG_P): New macro.
18671         * recog.c (register_operand): Check LRA_SUBREG_P.
18672         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18673         stack frame.
18674         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18675         (lra_final_code_change): Skip subreg change for operators.
18676         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18677         if there are no operand changes.
18678         * lra-constraints.c (curr_insn_set): New.
18679         (match_reload): Set LRA_SUBREG_P.
18680         (emit_spill_move): Ditto.
18681         (check_and_process_move): Use curr_insn_set. Process only single
18682         set insns.  Don't initialize sec_mem_p and change_p.
18683         (simplify_operand_subreg): Use LRA_SUBREG_P.
18684         (reg_in_class_p): New function.
18685         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18686         of #ifdef.  Add code to remove cycling.
18687         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18688         non-null disp.  Reload inner instead of disp when base and index
18689         are null.  Try to put lo_sum into register.
18690         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18691         (check_and_process_move): Move code for move cost check to
18692         simple_move_p.  Remove equiv_substitution.
18693         (simple_move_p): New function.
18694         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18695         curr_insn_set.  Call check_and_process_move only for single set
18696         insns.  Use the new function.  Move call of check_and_process_move
18697         after operand equiv substitution and address process.
18698
18699 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18700
18701         PR go/57045
18702         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18703         with nonlocal goto receivers or returns twice calls, ignore
18704         unininitialized values from abnormal edges to nl goto receiver
18705         or returns twice call.
18706
18707 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18708
18709         PR tree-optimization/57051
18710         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18711         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18712         bitsize.
18713
18714 2013-04-26  Richard Biener  <rguenther@suse.de>
18715
18716         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18717         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18718         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18719         (fixup_loop_arrays_after_move): New function.
18720         (move_sese_region_to_fn): Properly outline the loop tree parts
18721         of the SESE region.
18722
18723 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18724
18725         * config/i386/i386.md (type, unit): Fix long lines.
18726
18727 2013-04-26  Richard Biener  <rguenther@suse.de>
18728
18729         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
18730         (lto-streamer-out.o): Likewise.
18731         * cfgloop.c (init_loops_structure): Export, add struct function
18732         argument and adjust.
18733         (flow_loops_find): Adjust.
18734         * cfgloop.h (enum loop_estimation): Add EST_LAST.
18735         (init_loops_structure): Declare.
18736         * lto-streamer-in.c: Include cfgloop.h.
18737         (input_cfg): Input the loop tree.
18738         * lto-streamer-out.c: Include cfgloop.h.
18739         (output_cfg): Output the loop tree.
18740         (output_struct_function_base): Do not drop PROP_loops.
18741
18742 2013-03-26  Richard Biener  <rguenther@suse.de>
18743
18744         * tree-cfg.c (execute_build_cfg): Build the loop tree.
18745         (pass_build_cfg): Provide PROP_loops.
18746         (move_sese_region_to_fn): Remove loops that are outlined into fn
18747         for now.
18748         * tree-inline.c: Include cfgloop.h.
18749         (initialize_cfun): Do not drop PROP_loops.
18750         (copy_loops): New function.
18751         (copy_cfg_body): Copy loop structure.
18752         (tree_function_versioning): Initialize destination loop tree.
18753         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
18754         (pass_parallelize_loops): Do IL verification.
18755         * loop-init.c (loop_optimizer_init): Fixup loops if required.
18756         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
18757         the CFG make sure we fixup loops as well.
18758         * tree-ssa-tail-merge.c: Include cfgloop.h.
18759         (replace_block_by): When merging loop latches mark loops for fixup.
18760         * lto-streamer-out.c (output_struct_function_base): Drop
18761         PROP_loops for now.
18762         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
18763         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
18764         * ipa-split.c: Include cfgloop.h.
18765         (split_function): Add the new return block to the loop tree root.
18766         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
18767         whether we have removed the forwarder block.
18768         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
18769         * cfgloop.h (place_new_loop): Declare.
18770         * cfgloopmanip.c (place_new_loop): Export.
18771         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
18772         (tree-switch-conversion.o): Likewise.
18773         (tree-complex.o): Likewise.
18774         (tree-inline.o): Likewise.
18775         (tree-ssa-tailmerge.o): Likewise.
18776         (ipa-split.o): Likewise.
18777         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
18778         (tree-ssa-copy.o): Likewise.
18779         * tree-switch-conversion.c: Include cfgloop.h
18780         (process_switch): If we emit a bit-test cascade, schedule loops
18781         for fixup.
18782         * tree-complex.c: Include cfgloop.h.
18783         (expand_complex_div_wide): Properly add new basic-blocks to loops.
18784         * asan.c: Include cfgloop.h.
18785         (create_cond_insert_point): Properly add new basic-blocks to
18786         loops, schedule loop fixup.
18787         * cfgloop.c (verify_loop_structure): Check that looks are not
18788         marked for fixup.
18789         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
18790         to loops.
18791         (expand_omp_for_generic): Likewise.
18792         (expand_omp_sections): Likewise.
18793         (expand_omp_atomic_pipeline): Schedule loops for fixup.
18794         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
18795         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
18796         is initialized, not when loops are present.
18797         * tree-parloops.c (parallelize_loops): Remove checking here.
18798         * passes.c (init_optimization_passes): Schedule a copy-propagation
18799         pass before complete unrolling of inner loops.
18800
18801 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18802
18803         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
18804         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
18805         (colorize_init): Add argument to _WIN32 version.
18806         * toplev.c: Include diagnostic-color.h.
18807         (process_options): Default to -fdiagnostics-color=auto if
18808         GCC_COLORS env var is in the environment.
18809         * common.opt (fdiagnostics-color=): Add Var and Init.
18810         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
18811         env var is in the environment, the default is auto rather than never.
18812
18813         * diagnostic.h (file_name_as_prefix): Add context argument.
18814         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
18815         the string as locus.
18816         * langhooks.c (lhd_print_error_function): Adjust caller.
18817
18818 2013-04-25  Lawrence Crowl  <crowl@google.com>
18819
18820         * var-tracking.c (shared_hash_def::htab):
18821         Change type to hash_table.  Update dependent calls and types.
18822
18823 2013-04-25  Lawrence Crowl  <crowl@google.com>
18824
18825         * Makefile.in: Update as needed below.
18826
18827         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
18828         Move declaration to after the type's method definitons.
18829
18830         * attribs.c (htab_t scoped_attributes::attribute_hash):
18831         Change type to hash_table.  Update dependent calls and types.
18832
18833         * bitmap.c (htab_t bitmap_desc_hash):
18834         Change type to hash_table.  Update dependent calls and types.
18835
18836         * cselib.c (htab_t cselib_hash_table):
18837         Change type to hash_table.  Update dependent calls and types.
18838
18839         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
18840         (hash_string_slot_node): Move implementation into lto-streamer.h
18841         struct string_slot_hasher.
18842         (eq_string_slot_node): Likewise.
18843
18844         * data-streamer-out.c: Update output_block::string_hash_table
18845         dependent calls and types.
18846
18847         * dwarf2cfi.c (htab_t trace_index):
18848         Change type to hash_table.  Update dependent calls and types.
18849
18850         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
18851         Change type to hash_table.  Update dependent calls and types.
18852         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
18853         (htab_t optimize_external_refs::map): Likewise.
18854         (htab_t output_comp_unit::extern_map): Likewise.
18855         (htab_t output_comdat_type_unit::extern_map): Likewise.
18856         (htab_t output_macinfo::macinfo_htab): Likewise.
18857         (htab_t optimize_location_lists::htab): Likewise.
18858         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
18859
18860         * except.c (htab_t ehspec_hash_type):
18861         Change type to hash_table.  Update dependent calls and types.
18862         (assign_filter_values::ttypes): Likewise.
18863         (assign_filter_values::ehspec): Likewise.
18864         (sjlj_assign_call_site_values::ar_hash): Likewise.
18865         (convert_to_eh_region_ranges::ar_hash): Likewise.
18866
18867         * gcse.c (htab_t pre_ldst_table):
18868         Change type to hash_table.  Update dependent calls and types.
18869
18870         * ggc-common.c (htab_t saving_htab):
18871         Change type to hash_table.  Update dependent calls and types.
18872         (htab_t loc_hash): Likewise.
18873         (htab_t ptr_hash): Likewise.
18874         (call_count): Rename ggc_call_count.
18875         (call_alloc): Rename ggc_call_alloc.
18876         (loc_descriptor): Rename make_loc_descriptor.
18877         (add_statistics): Rename ggc_add_statistics.
18878
18879         * ggc-common.c (saving_htab):
18880         Change type to hash_table.  Update dependent calls and types.
18881
18882         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
18883         (push_gimplify_context): Likewise.
18884         (pop_gimplify_context): Likewise.
18885         (struct gimple_temp_hash_elt): Added.
18886         (struct gimplify_hasher): Likewise.
18887         (struct gimplify_ctx.temp_htab):
18888         Change type to hash_table.  Update dependent calls and types.
18889
18890         * gimple-fold.c: Include gimplify-ctx.h.
18891
18892         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
18893         Change type to hash_table.  Update dependent calls and types.
18894         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
18895         avoid potential global name collision.
18896
18897         * gimplify.c: Include gimplify-ctx.h.
18898         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
18899         (htab_t gimplify_ctx::temp_htab):
18900         Update dependent calls and types for new type hash_table.
18901         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
18902         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
18903
18904         * gimplify-ctx.h: New.
18905         (struct gimple_temp_hash_elt): Move from gimplify.c.
18906         (class gimplify_hasher): New.
18907         (struct gimplify_ctx): Move from gimple.h.
18908         (htab_t gimplify_ctx::temp_htab):
18909         Change type to hash_table.  Update dependent calls and types.
18910
18911         * graphite-clast-to-gimple.c: Include graphite-htab.h.
18912         (htab_t ivs_params::newivs_index):
18913         Change type to hash_table.  Update dependent calls and types.
18914         (htab_t ivs_params::params_index): Likewise.
18915         (htab_t print_generated_program::params_index): Likewise.
18916         (htab_t gloog::newivs_index): Likewise.
18917         (htab_t gloog::params_index): Likewise.
18918
18919         * graphite.c: Include graphite-htab.h.
18920         4htab_t graphite_transform_loops::bb_pbb_mapping):
18921         Change type to hash_table.  Update dependent calls and types.
18922
18923         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
18924         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
18925         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
18926
18927         * graphite-dependences.c: Include graphite-htab.h.
18928         (loop_is_parallel_p): Change hash table type of parameter.
18929
18930         * graphite-htab.h: New.
18931         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
18932         (extern find_pbb_via_hash): Move from graphite-poly.h.
18933         (extern loop_is_parallel_p): Move from graphite-poly.h.
18934         (extern get_loop_body_pbbs): Move from graphite-poly.h.
18935
18936         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
18937         (extern loop_is_parallel_p): Move to graphite-htab.h.
18938         (extern get_loop_body_pbbs): Move to graphite-htab.h.
18939
18940         * haifa-sched.c (htab_t delay_htab):
18941         Change type to hash_table.  Update dependent calls and types.
18942         (htab_t delay_htab_i2): Likewise.
18943
18944         * ira-color.c (htab_t allocno_hard_regs_htab):
18945         Change type to hash_table.  Update dependent calls and types.
18946
18947         * ira-costs.c (htab_t cost_classes_htab):
18948         Change type to hash_table.  Update dependent calls and types.
18949
18950         * loop-invariant.c (htab_t merge_identical_invariants::eq):
18951         Change type to hash_table.  Update dependent calls and types.
18952
18953         * loop-iv.c (htab_t bivs):
18954         Change type to hash_table.  Update dependent calls and types.
18955
18956         * loop-unroll.c (htab_t opt_info::insns_to_split):
18957         Change type to hash_table.  Update dependent calls and types.
18958         (htab_t opt_info::insns_with_var_to_expand): Likewise.
18959
18960         * lto-streamer.h (struct string_slot): Move from data-streamer.h
18961         (struct string_slot_hasher): New.
18962         (htab_t output_block::string_hash_table):
18963         Change type to hash_table.  Update dependent calls and types.
18964
18965         * lto-streamer-in.c (freeing_string_slot_hasher): New.
18966         (htab_t file_name_hash_table):
18967         Change type to hash_table.  Update dependent calls and types.
18968
18969         * lto-streamer-out.c: Update output_block::string_hash_table dependent
18970         calls and types.
18971
18972         * lto-streamer.c (htab_t tree_htab):
18973         Change type to hash_table.  Update dependent calls and types.
18974
18975         * omp-low.c: Include gimplify-ctx.h.
18976
18977         * passes.c (htab_t name_to_pass_map):
18978         Change type to hash_table.  Update dependent calls and types.
18979         (pass_traverse): Rename to passes_pass_traverse.
18980
18981         * plugin.c (htab_t event_tab):
18982         Change type to hash_table.  Update dependent calls and types.
18983
18984         * postreload-gcse.c (htab_t expr_table):
18985         Change type to hash_table.  Update dependent calls and types.
18986         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
18987
18988         * sese.c (debug_rename_map_1): Make extern.
18989         (htab_t copy_bb_and_scalar_dependences::rename_map):
18990         Change type to hash_table.  Update dependent calls and types.
18991
18992         * sese.h (extern debug_rename_map): Move to .c file.
18993
18994         * store-motion.c (htab_t store_motion_mems_table):
18995         Change type to hash_table.  Update dependent calls and types.
18996
18997         * trans-mem.c (htab_t tm_new_mem_hash):
18998         Change type to hash_table.  Update dependent calls and types.
18999
19000         * tree-browser.c (htab_t TB_up_ht):
19001         Change type to hash_table.  Update dependent calls and types.
19002
19003         * tree-cfg.c (htab_t discriminator_per_locus):
19004         Change type to hash_table.  Update dependent calls and types.
19005
19006         * tree-complex.c: Include tree-hasher.h
19007         (htab_t complex_variable_components):
19008         Change type to hash_table.  Update dependent calls and types.
19009
19010         * tree-eh.c (htab_t finally_tree):
19011         Change type to hash_table.  Update dependent calls and types.
19012
19013         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19014         struct int_tree_hasher.
19015         (extern int_tree_map_eq): Likewise.
19016         (uid_decl_map_hash): Removed.
19017         (extern decl_tree_map_eq): Likewise.
19018
19019         * tree-hasher.h: New.
19020         (struct int_tree_hasher): New.
19021         (typedef int_tree_htab_type): New.
19022
19023         * tree-inline.c: Include gimplify-ctx.h.
19024
19025         * tree-mudflap.c: Include gimplify-ctx.h.
19026
19027         * tree-parloops.c: Include tree-hasher.h.
19028         (htab_t eliminate_local_variables_stmt::decl_address):
19029         Change type to hash_table.  Update dependent calls and types.
19030         (htab_t separate_decls_in_region::decl_copies): Likewise.
19031
19032         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19033         Change type to hash_table.  Update dependent calls and types.
19034
19035         * tree-sra.c (candidates):
19036         Change type to hash_table.  Update dependent calls and types.
19037
19038         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19039         in tree-flow.h.
19040         (int_tree_map_hash): Likewise.
19041
19042         * tree-ssa-dom.c (htab_t avail_exprs):
19043         Change type to hash_table.  Update dependent calls and types.
19044
19045         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19046         Change type to hash_table.  Update dependent calls and types.
19047
19048         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19049         Change type to hash_table.  Update dependent calls and types.
19050
19051         * tree-ssa-phiopt.c (seen_ssa_names):
19052         Change type to hash_table.  Update dependent calls and types.
19053
19054         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19055         Change type to hash_table.  Update dependent calls and types.
19056
19057         * tree-ssa-uncprop.c (equiv):
19058         Change type to hash_table.  Update dependent calls and types.
19059
19060 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19061
19062         PR rtl-optimization/57003
19063         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19064         call note_stores with kill_clobbered_value callback again after
19065         killing regs_invalidated_by_call.
19066
19067 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19068
19069         * config/aarch64/aarch64-simd.md
19070         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19071         (aarch64_simd_bsl<mode>): Likewise.
19072         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19073
19074 2013-04-25  Marek Polacek  <polacek@redhat.com>
19075
19076         PR tree-optimization/57066
19077         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19078
19079 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19080
19081         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19082
19083 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19084
19085         * config/aarch64/aarch64-builtins.c
19086         (aarch64_fold_builtin): New.
19087         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19088         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19089         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19090         * config/aarch64/arm_neon.h
19091         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19092
19093 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19094             Tejas Belagod  <tejas.belagod@arm.com>
19095
19096         * config/aarch64/aarch64-builtins.c
19097         (aarch64_gimple_fold_builtin): New.
19098         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19099         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19100         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19101         (addvv4sf): Update.
19102         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19103
19104 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19105
19106         * config/aarch64/aarch64.md
19107         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19108
19109 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19110
19111         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19112         (*ngcsi_uxtw): New pattern.
19113
19114 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19115             Julian Brown  <julian@codesourcery.com>
19116
19117         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19118         (TB_DREG): Add T_V4HF.
19119         (v4hf_UP): New macro.
19120         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19121         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19122         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19123         built-ins.
19124         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19125         (arm_vector_mode_supported_p): Handle V4HF.
19126         (arm_mangle_map): Handle V4HFmode.
19127         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19128         * config/arm/arm_neon_builtins.def: Add entries for
19129         vcvtv4hfv4sf, vcvtv4sfv4hf.
19130         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19131         (neon_vcvtv4hfv4sf): Likewise.
19132         * config/arm/neon-gen.ml: Handle half-precision floating point
19133         features.
19134         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19135         * config/arm/arm_neon.h: Regenerate.
19136         * config/arm/neon.ml (type elts): Add F16.
19137         (type vectype): Add T_float16x4, T_floatHF.
19138         (type vecmode): Add V4HF.
19139         (type features): Add Requires_FP_bit feature.
19140         (elt_width): Handle F16.
19141         (elt_class): Likewise.
19142         (elt_of_class_width): Likewise.
19143         (mode_of_elt): Refactor.
19144         (type_for_elt): Handle F16, fix error messages.
19145         (vectype_size): Handle T_float16x4.
19146         (vcvt_sh): New function.
19147         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19148         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19149         (string_of_mode): Handle V4HF.
19150         * doc/arm-neon-intrinsics.texi: Regenerate.
19151
19152 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19153
19154         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19155         format specifier in 'X' case.
19156
19157 2013-04-25  Alan Modra  <amodra@gmail.com>
19158
19159         PR target/57052
19160         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19161         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19162         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19163         Repeat for many other rotate/shift and mask patterns using subregs.
19164         Name lshiftrt insns.
19165         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19166         on WORDS_BIG_ENDIAN.
19167
19168 2013-04-25  Alan Modra  <amodra@gmail.com>
19169
19170         * config.gcc: Support little-endian powerpc-linux targets.
19171         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19172         (LINK_OS_LINUX_SPEC): Define.
19173         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19174         Preserve MASK_LITTLE_ENDIAN.
19175         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19176         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19177         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19178         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19179         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19180         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19181         Correct fp word order for little-endian.  Don't shift toc entries
19182         smaller than a word for little-endian.
19183         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19184         (bswapdi2 splits): Correct low-part subreg for little-endian.
19185         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19186         low/high where such is correct only for be.
19187         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19188         little-endian for -mcall-aixdesc.
19189
19190 2013-04-25  Alan Modra  <amodra@gmail.com>
19191
19192         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19193         replace_equiv_address_nv.
19194
19195 2013-04-25  Alan Modra  <amodra@gmail.com>
19196
19197         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19198
19199 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19200
19201         Revert:
19202         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19203         * rtl.h (struct rtx_def): ...
19204
19205 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19206
19207         PR rtl-optimizations/57046
19208         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19209         for multi-reg splits.
19210
19211 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19212
19213         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19214
19215 2013-04-24  Sterling Augustine  <saugustine@google.com>
19216
19217         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19218         (comp_dir_string, debug_str_dwo_section): New.
19219         (DEBUG_STR_DWO_SECTION): Rename to ...
19220         (DEBUG_DWO_STR_SECTION): ... this.
19221         (DEBUG_NORM_STR_SECTION): Delete.
19222         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19223         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19224         (find_AT_string): Move most logic to ...
19225         (find_AT_string_in_table): ... here.  New.
19226         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19227         add_skeleton_AT_string.  Delete logic.
19228         (output_skeleton_debug_sections): Remove call to
19229         add_top_level_skeleton_die_attrs.
19230         (add_comp_dir_attribute): Move logic to comp_dir_string.
19231         (dwarf2out_init): Initialize debug_str_dwo_section.
19232         (output_indirect_string): Call find_string_form.
19233         (output_indirect_strings): Rewrite.
19234         (prune_unused_types): Empty skeleton_debug_str_hash.
19235         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19236         (dwarf2out_finish):  Call output_indirect_strings.
19237
19238 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19239
19240         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19241
19242 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19243
19244         * rtl.h (struct rtx_def): Add comment for field jump.
19245         (LRA_SUBREG_P): New macro.
19246         * recog.c (register_operand): Check LRA_SUBREG_P.
19247         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19248         stack frame.
19249         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19250         (lra_final_code_change): Skip subreg change for operators.
19251         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19252         if there are no operand changes.
19253         * lra-constraints.c (curr_insn_set): New.
19254         (match_reload): Set LRA_SUBREG_P.
19255         (emit_spill_move): Ditto.
19256         (check_and_process_move): Use curr_insn_set. Process only single
19257         set insns.  Don't initialize sec_mem_p and change_p.
19258         (simplify_operand_subreg): Use LRA_SUBREG_P.
19259         (reg_in_class_p): New function.
19260         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19261         of #ifdef.  Add code to remove cycling.
19262         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19263         non-null disp.  Reload inner instead of disp when base and index
19264         are null.  Try to put lo_sum into register.
19265         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19266         (check_and_process_move): Move code for move cost check to
19267         simple_move_p.  Remove equiv_substitution.
19268         (simple_move_p): New function.
19269         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19270         curr_insn_set.  Call check_and_process_move only for single set
19271         insns.  Use the new function.  Move call of check_and_process_move
19272         after operand equiv substitution and address process.
19273
19274 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19275
19276         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19277         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19278         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19279
19280 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19281
19282         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19283
19284 2013-04-24  Marek Polacek  <polacek@redhat.com>
19285
19286         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19287         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19288         (select_loops_exit_conditions): Likewise.
19289         (number_of_iterations_for_all_loops): Likewise.
19290         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19291         (scev_analysis): Likewise.
19292
19293 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19294             Chao-ying Fu  <fu@mips.com>
19295
19296         * config/mips/micromips.md (jraddiusp): New pattern.
19297         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19298         instruction if possible.
19299
19300 2013-04-24  Alan Modra  <amodra@gmail.com>
19301
19302         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19303
19304 2013-04-24  Julian Brown  <julian@codesourcery.com>
19305             Chung-Lin Tang  <cltang@codesourcery.com>
19306
19307         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19308         dependency behavior in enumeration type DIE generation. Add TODO note
19309         to comments about future DW_FORM_sdata/udata re-work of related code.
19310
19311 2013-04-23  Lawrence Crowl  <crowl@google.com>
19312
19313         * Makefile.in: Update as needed below.
19314
19315         * hash-table.h (class hash_table):
19316         Correct many methods with parameter types compare_type to the correct
19317         value_type.  (Correct code was unlikely to notice the change.)
19318         (hash_table::elements_with_deleted) New.
19319         (class hashtable::iterator): New.
19320         (hashtable::begin()): New.
19321         (hashtable::end()): New.
19322         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19323
19324         * statistics.c (statistics_hashes):
19325         Change type to hash_table.  Update dependent calls and types.
19326
19327         * tree-into-ssa.c (var_infos):
19328         Change type to hash_table.  Update dependent calls and types.
19329
19330         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19331         Change type to hash_table.  Update dependent calls and types.
19332
19333         * tree-ssa-loop-im.c (struct mem_ref.refs):
19334         Change type to hash_table.  Update dependent calls and types.
19335
19336         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19337         Change type to hash_table.  Update dependent calls and types.
19338
19339         * tree-ssa-sccvn.c (vn_tables_s::nary):
19340         Change type to hash_table.  Update dependent calls and types.
19341         (vn_tables_s::phis): Likewise.
19342         (vn_tables_s::references): Likewise.
19343
19344         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19345         (vn_reference_eq): Update parameter and return types.
19346
19347         * tree-ssa-structalias.c (pointer_equiv_class_table):
19348         Change type to hash_table.  Update dependent calls and types.
19349         (location_equiv_class_table): Likewise.
19350
19351         * tree-vect-data-refs.c: Consequential changes for making
19352         peeling a hash_table.
19353
19354         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19355         (destroy_loop_vec_info): Dependent hash_table update.
19356
19357         * tree-vectorizer.h (peeling_htab):
19358         Change type to hash_table.  Update dependent calls and types.
19359
19360 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19361
19362         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19363         to check the register content is equal or not.
19364         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19365         to assign register content record.
19366         * lra-eliminations.c (update_reg_eliminate): Use
19367         lra_update_reg_val_offset to update register content offset.
19368         * lra-int.h (struct lra_reg): Add offset member.
19369         (lra_reg_val_equal_p): New static inline function.
19370         (lra_update_reg_val_offset): New static inline function.
19371         (lra_assign_reg_val): New static inline function.
19372         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19373         to assign register content record.
19374         (initialize_lra_reg_info_element): Initial offset to zero.
19375
19376 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19377
19378         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19379         operands.  Record compression.
19380
19381 2013-04-23  Xinliang David Li  <davidxl@google.com>
19382
19383         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19384
19385 2013-04-23  Richard Biener  <rguenther@suse.de>
19386
19387         PR middle-end/57036
19388         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19389         parameter, only add abnormal goto edges from the copied body
19390         if the call could perform abnormal gotos.
19391         (copy_cfg_body): Adjust.
19392
19393 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19394
19395         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19396
19397 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19398
19399         * coretypes.h (gimple_stmt_iterator): Add struct to make
19400         compatible with C.
19401
19402 2013-04-23  Richard Biener  <rguenther@suse.de>
19403
19404         PR tree-optimization/57026
19405         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19406         from SSA names occuring in abnormal PHI nodes.
19407
19408 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19409
19410         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19411
19412 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19413
19414         * lto/lto.c (print_lto_report_1): Declare early.
19415         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19416
19417 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19418
19419         * common.opt (-flto-report-wpa): Add.
19420         * doc/invoke.texi (-flto-report-wpa): Add.
19421         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19422         (lto_main): dito.
19423
19424 2013-04-22  Xinliang David Li  <davidxl@google.com>
19425
19426         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19427         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19428         * Makefile.in: New dependency
19429
19430             David Daney <ddaney.cavm@gmail.com>
19431
19432         * configure.ac (gcc_cv_as_micromips_support):  Use the
19433         --fatal-warnings option.
19434         * configure: Regenerate.
19435
19436 2013-04-22  Marek Polacek  <polacek@redhat.com>
19437
19438         PR sanitizer/56990
19439         * tsan.c (instrument_expr): Don't instrument expression
19440         in case its size is zero.
19441
19442 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19443
19444         PR target/57032
19445         Revert:
19446         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19447
19448         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19449
19450 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19451
19452         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19453         (gimple_stmt_iterator): New typedef.
19454         * gimple.h (gimple_stmt_iterator): Rename to...
19455         (gimple_stmt_iterator_d): ... This.
19456         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19457         trees be valid for GIMPLE and GENERIC.
19458         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19459         * gimple-fold.c (gimple_fold_call): Call target hook
19460         gimple_fold_builtin.
19461         * hooks.c (hook_bool_gsiptr_false): New.
19462         * hooks.h (hook_bool_gsiptr_false): New.
19463         * target.def (fold_stmt): New.
19464         * doc/tm.texi: Regenerate.
19465
19466 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19467
19468         PR target/57018
19469         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19470         a set sp if no stack realignment.
19471
19472 2013-04-22  Nick Clifton  <nickc@redhat.com>
19473
19474         * config.gcc (tilegx-linux): Extend extra_objs rather than
19475         overwriting it.
19476         (tilepro-linux): Likewise.
19477
19478 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19479
19480         * config/aarch64/aarch64-builtins.c
19481         (CF): Remove.
19482         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19483         (VAR<1-12>): Add MAP parameter.
19484         (BUILTIN_*): Likewise.
19485         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19486         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19487         (aarch64_ushl_n<mode>): Likewise.
19488         (aarch64_sshr_n<mode>): Likewise.
19489         (aarch64_ushr_n<mode>): Likewise.
19490         (aarch64_<maxmin><mode>): Likewise.
19491         (aarch64_sqrt<mode>): Likewise.
19492         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19493         (vshr<q>_n_*): Likewise.
19494
19495 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19496
19497         * config/aarch64/aarch64-builtins.c
19498         (aarch64_simd_builtin_type_mode): Handle SF types.
19499         (sf_UP): Define.
19500         (BUILTIN_GPF): Define.
19501         (aarch64_init_simd_builtins): Handle SF types.
19502         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19503         (frecps): Likewise.
19504         (frecpx): Likewise.
19505         * config/aarch64/aarch64-simd.md
19506         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19507         (aarch64_frecpe<mode>): New.
19508         (aarch64_frecps<mode>): Likewise.
19509         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19510         (v8type): Add frecp<esx>.
19511         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19512         (aarch64_frecps<mode>): Likewise.
19513         * config/aarch64/iterators.md (FRECP): New.
19514         (frecp_suffix): Likewise.
19515         * config/aarch64/arm_neon.h
19516         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19517
19518 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19519
19520         PR target/56995
19521         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19522         (REG_CLASS_NAMES): Idem.
19523         (REG_CLASS_CONTENTS): Idem.
19524         (REGCLASS_HAS_FP_REG): Idem.
19525         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19526         (sh_conditional_register_usage): Idem.
19527
19528 2013-04-21  Jeff Law  <law@redhat.com>
19529
19530         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19531         (ssa_forward_propagate_and_combine): Use it.
19532
19533 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19534
19535         * lra.c: Update the flow chart diagram.
19536
19537 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19538
19539         PR rtl-optimization/56847
19540         * lra-constraints.c (process_alt_operands): Discourage alternative
19541         with non-matche doffsettable memory constraint fro memory with
19542         known offset.
19543
19544 2013-04-19  Richard Biener  <rguenther@suse.de>
19545
19546         PR tree-optimization/56982
19547         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19548         function.
19549         * gimplify.c (gimplify_call_expr): Notice special calls.
19550         (gimplify_modify_expr): Likewise.
19551         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19552         abnormal control flow receivers.
19553         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19554         in the same way as cfun->has_nonlocal_labels.
19555         (gimple_purge_dead_abnormal_call_edges): Likewise.
19556         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19557         receivers start a basic-block.
19558
19559 2013-04-19  Richard Biener  <rguenther@suse.de>
19560
19561         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19562         member ...
19563         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19564         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19565         (SLP_TREE_LOAD_PERMUTATION): Add.
19566         (vect_transform_slp_perm_load): Adjust prototype.
19567         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19568         (vect_free_slp_instance): Likewise.
19569         (vect_create_new_slp_node): Likewise.
19570         (vect_supported_slp_permutation_p): Remove.
19571         (vect_slp_rearrange_stmts): Adjust.
19572         (vect_supported_load_permutation_p): Likewise.  Inline
19573         vect_supported_slp_permutation_p here.
19574         (vect_analyze_slp_instance): Compute load permutations per
19575         slp node instead of per instance.
19576         (vect_get_slp_defs): Adjust.
19577         (vect_transform_slp_perm_load): Likewise.
19578         (vect_schedule_slp_instance): Remove redundant code.
19579         (vect_schedule_slp): Remove hack for PR56270, add it ...
19580         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19581         CSE loads for SLP.  Adjust.
19582
19583 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19584
19585         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19586         spelling in two comments.
19587
19588 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19589
19590         PR target/56797
19591         * config/arm/arm.c (load_multiple_sequence): Require SP
19592         as base register for loads if SP is in the register list.
19593
19594 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19595
19596         PR tree-optimization/56718
19597         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19598         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19599         and made public.  Adjusted all callers.
19600         (ipa_intraprocedural_devirtualization): New function.
19601         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19602         (ipa_intraprocedural_devirtualization): Likewise.
19603         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19604
19605 2013-04-19  Richard Biener  <rguenther@suse.de>
19606
19607         PR tree-optimization/57000
19608         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19609
19610 2013-04-19  Terry Guo  <terry.guo@arm.com>
19611
19612         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19613         Replace with ...
19614         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19615         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19616         (cortex_m4_fmacs): Use new reservations.
19617         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19618
19619 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19620
19621         PR rtl-optimization/56999
19622         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19623         related code.
19624         (lra_coalesce): Remove split_origin_bitmap and related code.
19625         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19626         ranges if necessary.
19627
19628 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19629
19630         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19631         New array.
19632         (ix86_expand_call): Remove clobbered_registers array and use
19633         x86_64_ms_sysv_extra_clobbered_registers instead.
19634         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19635         Declare here.
19636         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19637         predicate.
19638         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19639         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19640         (*call_value_rex64_ms_sysv): Ditto.
19641
19642 2013-04-18  Cary Coutant  <ccoutant@google.com>
19643
19644         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19645         parent instead of die_mark.
19646
19647 2013-04-18  Diego Novillo  <dnovillo@google.com>
19648
19649         * gimple.c (create_gimple_tmp): New.
19650         (get_expr_type): New.
19651         (build_assign): New.
19652         (build_type_cast): New.
19653         * gimple.h (enum ssa_mode): Define.
19654         (gimple_seq_set_location): New.
19655         * asan.c (build_check_stmt): Change some gimple_build_* calls
19656         to use build_assign and build_type_cast.
19657
19658 2013-04-18  Richard Biener  <rguenther@suse.de>
19659
19660         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19661         handle negative step.  Remove redundant checks.
19662         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19663         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19664         for negative step and grouped loads fail to vectorize.
19665
19666 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19667
19668         * emit-rtl.c (reset_insn_used_flags): New function.
19669         (reset_all_used_flags): Use it.
19670         (verify_insn_sharing): New function.
19671         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19672
19673 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19674
19675         PR tree-optimization/56984
19676         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19677         and (x >> M) >= N don't register any assertion if N << M is the
19678         minimum value.
19679
19680 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19681
19682         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19683         do not delete_insn insns that have not yet been emitted, only
19684         unlink them with remove_insn.
19685         * df-scan.c (df_insn_delete): Revert r197492.
19686
19687 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19688
19689         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19690         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19691
19692 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19693
19694         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19695         define_insn_and_split.
19696         (and_scc,ior_scc,negscc): Likewise.
19697         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19698
19699 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19700
19701         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19702         can benefit from using a sequence of LDRD instructions in epilogue
19703         instead of a single LDM instruction.
19704
19705 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19706
19707         PR 45688
19708         * doc/extend.texi: Fix typo.
19709
19710 2013-04-17  Richard Biener  <rguenther@suse.de>
19711
19712         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19713         (vect_build_slp_tree): ... here.
19714         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19715         match.  Remove special-casing of mismatched complex loads.
19716         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19717         re-try the match with swapped commutative operands.
19718         (vect_supported_load_permutation_p): Remove special-casing of
19719         mismatched complex loads.
19720         (vect_analyze_slp_instance): Adjust.
19721
19722 2013-04-17  Richard Biener  <rguenther@suse.de>
19723
19724         PR rtl-optimization/56921
19725         * cfgloop.h (struct loop): Add simple_loop_desc member.
19726         (struct niter_desc): Mark with GTY(()).
19727         (simple_loop_desc): Do not use aux field but simple_loop_desc.
19728         * loop-iv.c (get_simple_loop_desc): Likewise.
19729         (free_simple_loop_desc): Likewise.
19730
19731         Revert
19732         2013-04-16  Richard Biener  <rguenther@suse.de>
19733
19734         PR rtl-optimization/56921
19735         * loop-init.c (pass_rtl_move_loop_invariants): Add
19736         TODO_do_not_ggc_collect to todo_flags_finish.
19737         (pass_rtl_unswitch): Same.
19738         (pass_rtl_unroll_and_peel_loops): Same.
19739         (pass_rtl_doloop): Same.
19740
19741 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
19742
19743         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
19744         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
19745         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
19746         references.
19747         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
19748         * tree-streamer.c (record_common_node): Adjust reference in comment.
19749
19750 2013-04-17  Terry Guo  <terry.guo@arm.com>
19751
19752         * config/arm/cortex-m4.md: Add a new bypass.
19753
19754 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19755
19756         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
19757         New pattern.
19758         (*subs_<optab><mode>_multp2): New pattern.
19759         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19760         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19761
19762 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19763
19764         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
19765         (*subs_mul_imm_<mode>): New pattern.
19766
19767 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
19768
19769         PR target/56948
19770         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
19771         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
19772         (vsx_movti_32bit): Change j->wa to O->wa.
19773
19774 2013-04-16  Richard Biener  <rguenther@suse.de>
19775
19776         PR rtl-optimization/56921
19777         * loop-init.c (pass_rtl_move_loop_invariants): Add
19778         TODO_do_not_ggc_collect to todo_flags_finish.
19779         (pass_rtl_unswitch): Same.
19780         (pass_rtl_unroll_and_peel_loops): Same.
19781         (pass_rtl_doloop): Same.
19782
19783 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
19784
19785         * config/arm/arm.c (emit_multi_reg_push): New declaration
19786         for an existing function.
19787         (arm_emit_strd_push): New function.
19788         (arm_expand_prologue): Used here.
19789         (arm_emit_ldrd_pop): New function.
19790         (arm_expand_epilogue): Used here.
19791         (arm_get_frame_offsets): Update condition.
19792         (arm_emit_multi_reg_pop): Add a special case for load of a single
19793         register with writeback.
19794
19795 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
19796
19797         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
19798         description.
19799
19800 2013-04-16  Richard Biener  <rguenther@suse.de>
19801
19802         PR tree-optimization/56756
19803         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
19804         (first_mem_ref_loc): New.
19805         (execute_sm): Place the load temporarily before a previous
19806         access instead of in the latch edge to ensure its SSA dependencies
19807         are defined at points dominating the load.
19808
19809 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
19810
19811         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
19812         correct fix by moving header and footer insn to the footer of
19813         the merged basic block.  Clear BB_END of the merged-away block.
19814
19815         PR middle-end/43631
19816         * emit-rtl.c (make_note_raw): New function.
19817         (link_insn_into_chain): New static inline function.
19818         (add_insn): Use it.
19819         (add_insn_before, add_insn_after): Factor insn chain linking code...
19820         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
19821         using link_insn_into_chain.
19822         (note_outside_basic_block_p): New helper function for emit_note_after
19823         and emit_note_before.
19824         (emit_note_after): Use nobb variant of add_insn_after if the note
19825         should not be contained in a basic block.
19826         (emit_note_before): Use nobb variant of add_insn_before if the note
19827         should not be contained in a basic block.
19828         (emit_note_copy): Use make_note_raw.
19829         (emit_note): Likewise.
19830         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
19831         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
19832         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
19833         the moved barrier the tail of the basic block it follows.
19834         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
19835
19836 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
19837
19838         PR tree-optimization/56962
19839         * gimple-ssa-strength-reduction.c (record_increment): Only set
19840         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
19841         either rhs1 or rhs2 is equal to c->base_expr.
19842
19843 2013-04-15  Richard Biener  <rguenther@suse.de>
19844
19845         PR tree-optimization/56933
19846         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
19847         member.
19848         (GROUP_READ_WRITE_DEPENDENCE): Remove.
19849         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
19850         * tree-vect-data-refs.c (vect_analyze_group_access): Move
19851         dependence check ...
19852         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
19853         ... here.
19854         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
19855         GROUP_READ_WRITE_DEPENDENCE.
19856
19857 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19858
19859         * emit-rtl.c (reset_all_used_flags): New function.
19860         (verify_rtl_sharing): Call reset_all_used_flags before and after
19861         performing the checks.
19862
19863 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19864
19865         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
19866         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
19867         * config/arm/constraints.md (De): New constraint.
19868         * config/arm/neon.md (anddi3_neon): Delete.
19869         (neon_vand<mode>): Expand to standard anddi3 pattern.
19870         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
19871         Move earlier in the file.
19872         (neon_inv_logic_op2): Likewise.
19873         (arm_anddi_operand_neon): New predicate.
19874
19875 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19876
19877         * configure.ac (gcc_cv_ld_as_needed): Set
19878         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
19879         Use -z ignore, -z record on *-*-solaris2*.
19880         (HAVE_LD_AS_NEEDED): Update comment.
19881         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
19882         * configure: Regenerate.
19883         * config.in: Regenerate.
19884         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
19885         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
19886         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
19887         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
19888         equivalents.  Fix markup.
19889         * doc/tm.texi: Regenerate.
19890
19891 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
19892
19893         * config/i386/i386.opt: New option mstack-protector-guard=.
19894         * config/i386/i386-opts.h: Add enum stack_protector_guard.
19895         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
19896         TARGET_SSP_TLS_GUARD.
19897         * config/i386/i386.c (ix86_option_override_internal): Set
19898         ix86_stack_protector_guard.
19899         * config/i386/i386.md (stack_protect_set): Enable for
19900         TARGET_SSP_TLS_GUARD only.
19901         (stack_protect_set_<mode>): Ditto.
19902         (stack_protect_test): Ditto.
19903         (stack_protect_test_<mode>): Ditto.
19904         * doc/invoke.texi (i386 Option): Document.
19905
19906 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
19907
19908         PR target/56890
19909         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
19910         (S_MODES): Set H_MODE bit.
19911         (SF_MODES): Set only S_MODE and SF_MODE bits.
19912         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
19913         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
19914         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
19915         <MODE_FLOAT>: Likewise.
19916
19917 2013-04-15  Joey Ye  <joey.ye@arm.com>
19918
19919         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
19920
19921 2013-04-15  Joey Ye  <joey.ye@arm.com>
19922
19923         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
19924         for real far jump.
19925         (thumb_far_jump_used_p): Count instruction size and set
19926         far_jump_used.
19927
19928 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
19929
19930         * reorg.c (fill_simple_delay_slots): Reindent block of code.
19931         * resource.c (mark_target_live_regs): Reformat conditional block.
19932
19933 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
19934
19935         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
19936         notes, they are emitted only just before final.
19937         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
19938
19939 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
19940
19941         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
19942         * cfgrtl.c (delete_insn): Call it here instead.
19943         * lra-spills.c (lra_final_code_change): Use delete_insn.
19944         * haifa-sched.c (sched_remove_insn): Likewise.
19945         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
19946         returning to the nop pool.
19947         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
19948         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
19949
19950 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
19951
19952         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
19953         * doc/tm.texi: Regenerated.
19954
19955 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
19956
19957         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
19958         QImode checks.
19959
19960 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
19961
19962         * df-core.c (df_find_def): Compare register numbers.
19963         (df_find_use): Likewise.
19964
19965 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
19966
19967         PR target/56903
19968         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
19969         lra_in_progress for return.
19970
19971 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
19972
19973         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
19974         define_insn into define_insn_and_split and emit movsicc patterns.
19975
19976 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
19977
19978         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
19979
19980 2013-04-12  Richard Biener  <rguenther@suse.de>
19981
19982         * tree-pass.h (TODO_do_not_ggc_collect): New.
19983         * passes.c (execute_one_ipa_transform_pass): Honor
19984         TODO_do_not_ggc_collect.
19985         (execute_one_pass): Likewise.
19986
19987         Revert
19988         2013-04-10  Richard Biener  <rguenther@suse.de>
19989
19990         * passes.c (init_optimization_passes): Remove reload pass.
19991         * ira.c (do_reload): Merge into ...
19992         (ira): ... this.
19993         (rest_of_handle_reload): Remove.
19994         (pass_reload): Likewise.
19995         * config/i386/i386.c (ix86_option_override): Refer to ira instead
19996         of reload for vzeroupper pass placement.
19997
19998 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
19999
20000         PR tree-optimization/56918
20001         PR tree-optimization/56920
20002         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20003         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20004         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20005         use wide_mul_with_sign method.
20006
20007 2013-04-12  Richard Biener  <rguenther@suse.de>
20008
20009         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20010         not be considered a gimple constant.
20011
20012 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20013
20014         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20015         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20016         operations.
20017
20018 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20019             Jakub Jelinek  <jakub@redhat.com>
20020
20021         * opts.c: Include diagnostic-color.h.
20022         (common_handle_option): Handle OPT_fdiagnostics_color_.
20023         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20024         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20025         (diagnostic-color.o): New.
20026         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20027         (diagnostic_color_rule): New enum.
20028         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20029         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20030         the location string.
20031         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20032         either NULL, or color kind.
20033         * diagnostic-color.c: New file.
20034         * diagnostic-color.h: New file.
20035         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20036         arguments.
20037         * doc/invoke.texi (-fdiagnostics-color): Document.
20038         * pretty-print.h (pp_show_color): Define.
20039         (struct pretty_print_info): Add show_color field.
20040         * diagnostic.c: Include diagnostic-color.h.
20041         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20042         macros.  Colorize error:, warning: etc. strings and also the location
20043         string.
20044         (diagnostic_show_locus): Colorize the caret line.
20045         * pretty-print.c: Include diagnostic-color.h.
20046         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20047         inside of %< %> quotes or quoted through q format modifier.
20048
20049 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20050
20051         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20052
20053 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20054
20055         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20056         code in CC_NZ mode.
20057         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20058         pattern.
20059
20060 2013-04-11  Marek Polacek  <polacek@redhat.com>
20061
20062         PR tree-optimization/48184
20063         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20064
20065 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20066
20067         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20068         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20069         (skip_simple_arithmetic): Tidy up.
20070         * tree.h (skip_simple_constant_arithmetic): Declare.
20071
20072 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20073
20074         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20075
20076 2013-04-11  Richard Biener  <rguenther@suse.de>
20077
20078         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20079         generate vector constants.
20080
20081 2013-04-11  Richard Biener  <rguenther@suse.de>
20082
20083         PR tree-optimization/56878
20084         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20085         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20086         New function.
20087         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20088         Prefer to align the DR with the most invariant base address.
20089
20090 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20091
20092         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20093         comment.
20094
20095 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20096
20097         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20098         floating-point vector comparisons against 0.
20099
20100 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20101
20102         PR tree-optimization/56899
20103         * fold-const.c (extract_muldiv_1): Apply distributive law
20104         only if TYPE_OVERFLOW_WRAPS (ctype).
20105
20106 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20107
20108         PR target/56124
20109         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20110         loading has side effect.
20111
20112 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20113
20114         * config/sparc/sparc.c: Include tree-pass.h.
20115         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20116         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20117         head of file.  Change return type.  Split off gate function.
20118         (sparc_gate_work_around_errata): New function.
20119         (pass_work_around_errata): New pass definition.
20120         (insert_pass_work_around_errata) New pass insert definition to
20121         insert pass_work_around_errata just after delayed-branch scheduling.
20122         (sparc_option_override): Insert the pass.
20123         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20124
20125 2013-04-10  David S. Miller  <davem@davemloft.net>
20126
20127         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20128         or -mcpu=hypersparc.
20129
20130         * target.def (cstore_mode): New hook.
20131         * target.h: Include insn-codes.h
20132         * targhooks.c: Likewise.
20133         (default_cstore_mode): New function.
20134         * targhooks.h: Declare it.
20135         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20136         * doc/tm.texi: Rebuild.
20137         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20138         target hook, rather than inspecting the insn_data.
20139         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20140         (TARGET_CSTORE_MODE): Redefine.
20141         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20142         result patterns.
20143         * config/sparc/predicates.md (cstore_result_operand): New special
20144         predicate.
20145         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20146         Use it for operand 0.
20147         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20148         (*snesi_special): Likewise.
20149         (*snesi_zero): Likewise.
20150         (*seqsi_zero): Likewise.
20151         (*sltu_insn): Likewise.
20152         (*sgeu_insn): Likewise.
20153         (*seqdi_special): Make operand 0 and comparison operation be of
20154         DImode.
20155         (*snedi_special): Likewise.
20156         (*snedi_special_vis3): Likewise.
20157         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20158         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20159         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20160         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20161         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20162         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20163         (*sltu_extend_sp64): Likewise.
20164         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20165         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20166         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20167         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20168         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20169
20170 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20171
20172         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20173         (aarch64_start_file): Use the new function.
20174
20175 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20176             Jason Merrill  <jason@redhat.com>
20177
20178         * common.opt: Add -gdwarf.
20179         * opts.c (common_handle_option): Handle it.
20180         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20181
20182 2013-04-10  Richard Biener  <rguenther@suse.de>
20183
20184         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20185         (execute_one_ipa_transform_pass): But unconditionally here.
20186         (execute_one_pass): And here.
20187         (init_optimization_passes): Remove reload pass.
20188         * tree-pass.h (TODO_ggc_collect): Remove.
20189         (pass_reload): Likewise.
20190         * ira.c (do_reload): Merge into ...
20191         (ira): ... this.
20192         (rest_of_handle_reload): Remove.
20193         (pass_reload): Likewise.
20194         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20195         of reload for vzeroupper pass placement.
20196         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20197         and todo_flags_finish of all passes.
20198
20199 2013-04-10  Richard Biener  <rguenther@suse.de>
20200
20201         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20202         first_const_oprnd field, rename first_def_type to first_op_type.
20203         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20204         (vect_get_and_check_slp_defs): Always use the type of the
20205         operand.  Allow mixed vect_external_def, vect_constant_def types.
20206         (vect_get_constant_vectors): Handle mixed vect_external_def,
20207         vect_constant_def types.
20208
20209 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20210
20211         PR tree-optimization/55524
20212         * tree-ssa-math-opts.c
20213         (convert_mult_to_fma): Don't use an fms construct
20214         when we don't have an fms operation, but fnma, and it looks
20215         likely that we'll be able to use the latter.
20216
20217 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20218
20219         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20220         function.
20221         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20222         inline fail caused by overwritable functions.
20223
20224 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20225
20226         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20227         unnecessary bits in the constant power of two case.
20228
20229 2013-04-10  Richard Biener  <rguenther@suse.de>
20230
20231         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20232         broken code swapping operands.
20233         (vect_build_slp_tree): Do not compute load permutations here.
20234         (vect_analyze_slp_instance): Compute load permutations here,
20235         after building the SLP tree.
20236
20237 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20238
20239         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20240         of next/prev_real_insn.
20241
20242 2013-04-09  Jan Hubicka  <jh@suse.cz>
20243
20244         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20245         Drop aliased parameter.
20246         (function_and_variable_visibility): Do not handle alias pairs.
20247         * cgraph.c (varpool_externally_visible_p): Update prototype.
20248         * varpool.c (varpool_add_new_variable): Update.
20249
20250 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20251
20252         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20253
20254 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20255
20256         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20257
20258         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20259
20260 2013-04-09  Marek Polacek  <polacek@redhat.com>
20261
20262         PR tree-optimization/48762
20263         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20264
20265 2013-04-09  Richard Biener  <rguenther@suse.de>
20266
20267         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20268         dealing with cost.
20269         (vect_build_slp_tree): Likewise.
20270         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20271         calculating the cost of a SLP instance.
20272         (vect_analyze_slp_instance): Use it from here, after building
20273         the SLP tree.
20274
20275 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20276
20277         PR middle-end/56883
20278         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20279         expand_omp_for_static_chunk): Use simple_p = true in
20280         force_gimple_operand_gsi calls when assigning to addressable decls.
20281
20282 2013-04-09  Jeff Law  <law@redhat.com>
20283
20284         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20285         when the boolean was created by converting a wider object which
20286         had a boolean range.
20287
20288 2013-04-09  Richard Biener  <rguenther@suse.de>
20289
20290         * tree-vectorizer.h (slp_void_p): Remove.
20291         (slp_tree): Typedef before _slp_tree declaration.
20292         (struct _slp_tree): Use a vector of slp_tree as children.
20293         (vect_get_place_in_interleaving_chain): Remove.
20294         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20295         Move ...
20296         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20297         and make static.
20298         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20299         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20300         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20301         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20302         Use slp_node instead of slp_void_p and adjust.
20303
20304 2013-04-09  Richard Biener  <rguenther@suse.de>
20305
20306         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20307         work that is not necessary.
20308
20309 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20310
20311         PR tree-optimization/56854
20312         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20313         forward into clobber stmts if it would change MEM_REF lhs into
20314         non-MEM_REF.
20315
20316 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20317
20318         * tree.c (type_hash_lookup, type_hash_add): Make static.
20319         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20320
20321 2013-04-09  Richard Biener  <rguenther@suse.de>
20322
20323         * tree.h (unsave_expr_now): Remove.
20324         * tree-inline.c (mark_local_for_remap_r): Remove.
20325         (unsave_expr_1): Likewise.
20326         (unsave_r): Likewise.
20327         (unsave_expr_now): Likewise.
20328         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20329         (propagate_tree_value): Likewise.
20330
20331 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20332
20333         * doc/rtl.texi (sequence): Rewrite documentation to match the
20334         current use of SEQUENCE rtl objects.
20335         * rtl.def (SEQUENCE): Likewise.
20336
20337         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20338         Update documentation.
20339         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20340         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20341
20342         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20343
20344 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20345
20346         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20347         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20348         methods.
20349         (estimate_edge_size_and_time): Add comment to suggest using rounding
20350         methods.
20351         (estimate_node_size_and_time): Ditto.
20352         (remap_edge_change_prob): Use helper rounding divide methods.
20353         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20354         (gimple_mod_pow2_value_transform): Ditto.
20355         (gimple_mod_subtract_transform): Ditto.
20356         (gimple_ic_transform): Ditto.
20357         (gimple_stringops_transform): Ditto.
20358         * stmt.c (conditional_probability): Ditto.
20359         (emit_case_dispatch_table): Ditto.
20360         * lto-cgraph.c (merge_profile_summaries): Ditto.
20361         * tree-optimize.c (execute_fixup_cfg): Ditto.
20362         * cfgcleanup.c (try_forward_edges): Ditto.
20363         * cfgloopmanip.c (scale_loop_profile): Ditto.
20364         (loopify): Ditto.
20365         (duplicate_loop_to_header_edge): Ditto.
20366         (lv_adjust_loop_entry_edge): Ditto.
20367         * tree-vect-loop.c (vect_transform_loop): Ditto.
20368         * profile.c (compute_branch_probabilities): Ditto.
20369         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20370         * lto-streamer-in.c (input_cfg): Ditto.
20371         * gimple-streamer-in.c (input_bb): Ditto.
20372         * ipa-cp.c (update_profiling_info): Ditto.
20373         (update_specialized_profile): Ditto.
20374         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20375         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20376         rounding methods.
20377         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20378         (compute_trg_info): Ditto.
20379         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20380         (purge_dead_edges): Ditto.
20381         * loop-unswitch.c (unswitch_loop): Ditto.
20382         * cgraphclones.c (cgraph_clone_edge): Ditto.
20383         (cgraph_clone_node): Ditto.
20384         * tree-inline.c (copy_bb): Ditto.
20385         (copy_edges_for_bb): Ditto.
20386         (initialize_cfun): Ditto.
20387         (copy_cfg_body): Ditto.
20388         (expand_call_inline): Ditto.
20389
20390 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20391
20392         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20393         TARGET_CYGWIN64 by TARGET_64BIT.
20394
20395 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20396
20397         * config/epiphany/epiphany.md (GPR_1): New constant.
20398         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20399         * config/epiphany/epiphany.c (gen_compare_reg):
20400         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20401         is already in place.
20402         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20403         Don't require being called during rtl expansion; If y operlaps r0,
20404         return 0.
20405         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20406         (epiphany_expand_epilogue): Likewise.
20407
20408         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20409         Don't use CC_FPmode for ORDERED / UNORDERED.
20410         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20411
20412         * config/epiphany/constraints.md (CnL): New constraint.
20413         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20414         * config/epiphany/predicates.md (add_operand): Allow 1024.
20415
20416         * config/epiphany/epiphany.md (logical_op): New code iterator.
20417         (op_mnc): New code attribute.
20418         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20419         (mov_f+1, mov_f+2): New peephole2 patterns.
20420
20421         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20422         (cstoresi4): Also allow re-use of zero result when doing a NE
20423         comparison to a non-zero operand.
20424         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20425
20426         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20427         Use gen_addsi3_i / gen_subsi3_i.
20428
20429 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20430
20431         PR c++/34949
20432         PR c++/50243
20433         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20434         contain anything but clobbers, at most one __builtin_stack_restore,
20435         optionally debug stmts and final resx, and if it has at least one
20436         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20437         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20438         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20439         which isn't defaut definition, remove them.
20440         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20441         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20442         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20443         with MEM_REF LHS with SSA_NAME address.
20444
20445 2013-04-08  Jeff Law  <law@redhat.com>
20446
20447         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20448
20449 2013-04-08  Richard Biener  <rguenther@suse.de>
20450
20451         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20452         extra newline.
20453         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20454         determined vector type.
20455         (vect_analyze_data_refs): Likewise.
20456         (vect_get_new_vect_var): Adjust.
20457         (vect_create_destination_var): Preserve SSA name versions.
20458         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20459         not dump anything here.
20460
20461 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20462
20463         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20464         Add member lr_slot_known.
20465         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20466         if necessary.
20467         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20468         Remove code that sets lr_slot_offset according to what a previous
20469         version of epiphany_emit_save_restore used to do.
20470         (epiphany_emit_save_restore): When doing an lr save or restore,
20471         set/verify lr_slot_known and lr_slot_offset.
20472
20473 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20474
20475         PR target/54338
20476         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20477         in ALL_REGS.
20478
20479 2013-04-08  Richard Biener  <rguenther@suse.de>
20480
20481         * alias.c (find_base_term): Fix thinko in previous change.
20482
20483 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20484
20485         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20486         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20487         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20488         if possible to compute val.
20489         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20490         For QImode integers don't require anything about precision.  Use
20491         const_with_all_bytes_same to find out if the constant doesn't have
20492         repeated bytes in it.
20493
20494 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20495
20496         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20497         within mode size.
20498
20499 2013-04-08  Marek Polacek  <polacek@redhat.com>
20500
20501         PR rtl-optimization/48182
20502         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20503         value to 1.
20504
20505 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20506
20507         PR target/55487
20508         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20509         nuses, make sure we have a label.
20510
20511 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20512
20513         PR target/56843
20514         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20515         (rs6000_emit_swdiv_low_precision): Remove.
20516         (rs6000_emit_swdiv): Rewrite to handle between one and four
20517         iterations of Newton-Raphson generally; modify required number of
20518         iterations for some cases.
20519         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20520
20521 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20522
20523         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20524         set-but-unused variable.
20525
20526         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20527         basic blocks of released function bodies garbage-collectable.
20528
20529         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20530         (struct rtl_opt_pass): Add TODO_df_finish.
20531
20532         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20533
20534 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20535
20536         * config/arm/constraints.md (q): New constraint.
20537         * config/arm/ldrdstrd.md: New file.
20538         * config/arm/arm.md (ldrdstrd.md) New include.
20539         (arm_movdi): Use "q" instead of "r" constraint
20540         for double-word memory access.
20541         (movdf_soft_insn): Likewise.
20542         * config/arm/vfp.md (movdi_vfp): Likewise.
20543         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20544         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20545         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20546         (mem_ok_for_ldrd_strd): Likewise.
20547         (output_move_double): Update assertion.
20548
20549 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20550
20551         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20552
20553 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20554
20555         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20556         define_insn_and_split.
20557         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20558
20559 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20560
20561         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20562         define_insn_and_split.
20563         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20564         (shiftsi3_compare): New pattern.
20565         (rrx): New pattern.
20566         * config/arm/unspecs.md (UNSPEC_RRX): New.
20567
20568 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20569
20570         * config/arm/arm.md (negdi_extendsidi): New pattern.
20571         (negdi_zero_extendsidi): Likewise.
20572
20573 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20574
20575         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20576         define_insn_and_split.
20577         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20578         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20579
20580 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20581
20582         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20583         define_insn_and_split.
20584         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20585         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20586
20587 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20588
20589         * config/arm/arm.md (subsi3_carryin): New pattern.
20590         (subsi3_carryin_const): Likewise.
20591         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20592         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20593
20594 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20595
20596         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20597
20598 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20599
20600         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20601         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20602
20603 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20604
20605         * config/arm/arm.c (arm_expand_builtin): Change fcode
20606         type to unsigned int.
20607
20608 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20609
20610         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20611
20612 2013-04-04  Ian Lance Taylor  <iant@google.com>
20613
20614         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20615         language standard.
20616
20617 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20618
20619         PR middle-end/56729
20620         * df-scan.c (df_insn_delete): Disable failing assert.
20621
20622 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20623
20624         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20625         New function prototype.
20626         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20627         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20628         (arm_builtin_vectorized_function): New function.
20629
20630 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20631
20632         * config/arm/arm_neon_builtins.def: New file.
20633         * config/arm/arm.c (neon_builtin_data): Move contents to
20634         arm_neon_builtins.def.
20635         (enum arm_builtins): Include neon builtin definitions.
20636         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20637         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20638
20639 2013-04-04  Marek Polacek  <polacek@redhat.com>
20640
20641         PR tree-optimization/48186
20642         * predict.c (maybe_hot_frequency_p): Return false if
20643         HOT_BB_FREQUENCY_FRACTION is 0.
20644         (cgraph_maybe_hot_edge_p): Likewise.
20645
20646 2013-04-04  Richard Biener  <rguenther@suse.de>
20647
20648         PR tree-optimization/56826
20649         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20650         more accurately.
20651
20652 2013-04-04  Richard Biener  <rguenther@suse.de>
20653
20654         PR tree-optimization/56213
20655         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20656         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20657
20658 2013-04-04  Richard Biener  <rguenther@suse.de>
20659
20660         PR tree-optimization/56837
20661         * tree-loop-distribution.c (classify_partition): For non-zero
20662         values require that the value has the same precision as its
20663         mode to be useful as memset value.
20664
20665 2013-04-03  Nick Clifton  <nickc@redhat.com>
20666
20667         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20668         (fmssf4): Use fmsf.s on E3V5 architectures.
20669         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20670         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20671
20672 2013-04-03  Jeff Law  <law@redhat.com>
20673
20674         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20675         (lra-eliminations.o): Likewise.
20676
20677 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20678
20679         * gcov-io.c (compute_working_sets): Moved most of body of old
20680         compute_working_sets here from profile.c.
20681         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20682         (gcov_working_set_t): Moved typedef here from basic-block.h
20683         (compute_working_set): Declare.
20684         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20685         (get_working_sets): Renamed from compute_working_set,
20686         replace most of body with call to new compute_working_sets.
20687         (get_exec_counts): Replace call to compute_working_sets
20688         to get_working_sets.
20689         * profile.h (get_working_sets): Renamed from compute_working_set.
20690         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20691         to get_working_sets.
20692         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20693         * gcov-dump.c (dump_working_sets): New function.
20694
20695 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20696
20697         * hwint.c (sext_hwi, zext_hwi): New functions.
20698         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20699         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20700         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20701         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20702         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20703         (sext_hwi, zext_hwi): New functions.
20704
20705 2013-04-03  Jeff Law  <law@redhat.com>
20706
20707         PR tree-optimization/56799
20708         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20709         back test for widening conversion erroneously dropped in prior change.
20710
20711 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20712
20713         PR target/56809
20714         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20715         instead of next_real_insn.
20716
20717 2013-04-03  Marek Polacek  <polacek@redhat.com>
20718
20719         PR sanitizer/55702
20720         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20721
20722 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20723
20724         PR target/56809
20725         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20726         next_real_insn.
20727         (thumb1_output_casesi): Likewise.
20728         (thumb2_output_casesi): Likewise.
20729
20730 2013-04-03  Richard Biener  <rguenther@suse.de>
20731
20732         PR tree-optimization/56817
20733         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
20734         Split out ...
20735         (tree_unroll_loops_completely_1): ... new function to manually
20736         walk the loop tree, properly defering outer loops of unrolled
20737         loops to later iterations.
20738
20739 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20740
20741         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
20742         (vectorizable_load): Likewise.
20743         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
20744         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
20745
20746 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20747
20748         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
20749         BIT_FIELD_REF.
20750
20751 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
20752
20753         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
20754
20755 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
20756
20757         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
20758
20759 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20760
20761         PR tree-optimization/56790
20762         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
20763         folding.
20764
20765 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20766
20767         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
20768         Handle VEC_MERGE.
20769         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
20770         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
20771         equal arguments.
20772
20773 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
20774
20775         PR c/19449
20776         * tree.h (force_folding_builtin_constant_p): New decl.
20777         * builtins.c (force_folding_builtin_constant_p): New variable.
20778         (fold_builtin_constant_p): Fold immediately also if
20779         force_folding_builtin_constant_p.
20780
20781 2013-04-03  Richard Biener  <rguenther@suse.de>
20782
20783         PR tree-optimization/56812
20784         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
20785         DRs of the same interleaving chain are independent.
20786
20787 2013-04-02  Jason Merrill  <jason@redhat.com>
20788
20789         * gdbinit.in (pbb): Use debug fn.
20790
20791 2013-04-02  Lawrence Crowl  <crowl@google.com>
20792
20793         * sese.h (struct ivtype_map_elt_s): Remove unused.
20794         (extern debug_ivtype_map): Remove unused.
20795         (extern eq_ivtype_map_elts): Remove unused.
20796         * sese.c (debug_ivtype_map): Removed unused.
20797         (debug_ivtype_map_1): Removed unused.
20798         (debug_ivtype_elt): Remove unused.
20799         (eq_ivtype_map_elts): Remove unused.
20800
20801
20802 2013-04-02  Kai Tietz  <ktietz@redhat.com>
20803
20804         PR target/52790
20805         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
20806         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
20807         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
20808         function.
20809         (legitimize_pe_coff_symbol): Likewise.
20810         (is_imported_p): New helper-function.
20811         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
20812         for Windows x64 targets.
20813         (ix86_expand_prologue): Optimize for pe-coff targets.
20814         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
20815         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
20816         medium/large code-model.
20817         (legitimize_pic_address): Likewise.
20818         (legitimize_tls_address): Likewise.
20819         (ix86_expand_call): Likewise.
20820         (x86_output_mi_thunk): Likewise.
20821         (get_dllimport_decl): Add new beimport argument.
20822         (construct_plt_address): Don't assert for x64 pe-coff targets.
20823         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
20824         targets.
20825         (SYMBOL_FLAG_STUBVAR): New macro.
20826         (SYMBOL_REF_STUBVAR_P): Likewise.
20827         * config/i386/winnt.c (stub_list): New structure.
20828         (stub_head): New local variable.
20829         (i386_pe_record_stub): New function.
20830         (i386_pe_file_end): Emit refptr-stubs.
20831
20832 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
20833
20834         PR rtl-optimization/56745
20835         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
20836         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
20837
20838         PR c++/34949
20839         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
20840         and both of them are MEM_REFs, just compare first argument for
20841         equality and attempt to deal even with differing offsets.
20842
20843         PR c++/34949
20844         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
20845         of gimple_clobber_p to be MEM_REF.
20846         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
20847         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
20848         after gimplification.
20849         * asan.c (get_mem_ref_of_assignment): Don't instrument
20850         gimple_clobber_p stmts.
20851         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
20852         gimple_clobber_p stmt if they have MEM_REF lhs and
20853         are dead because of another gimple_clobber_p stmt.
20854         * tree-ssa-live.c (clear_unused_block_pointer): Treat
20855         gimple_clobber_p stmts like debug stmts.
20856         (remove_unused_locals): Remove clobbers with MEM_REF lhs
20857         that refer to unused VAR_DECLs or uninitialized values.
20858         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
20859         gimple_clobber_p stmts if they refer to removed parameters.
20860         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
20861         formatting.
20862
20863 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
20864
20865         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
20866         using SWI48 mode attribute.
20867
20868 2013-04-02  Wei Mi  <wmi@google.com>
20869
20870         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
20871         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
20872         *<rotate_insn><mode>3_mask in i386.md.
20873
20874 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
20875
20876         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
20877
20878 2013-04-02  Richard Biener  <rguenther@suse.de>
20879
20880         PR tree-optimization/56778
20881         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
20882         Runtime alias tests are not supported for gather loads.
20883         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
20884         stmts referenced from SSA operands before updating SSA form.
20885
20886 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
20887             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20888
20889         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
20890         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
20891         * config/arm/cortex-a53.md: New file.
20892         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
20893         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
20894         * config/arm/arm.c (arm_issue_rate): Likewise.
20895         * config/arm/arm-tune.md: Regenerate
20896         * config/arm/arm-tables.opt: Regenerate.
20897         * config/arm/arm-cores.def: Add cortex-a53.
20898
20899 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
20900
20901         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
20902         non-static link.
20903
20904 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
20905
20906         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
20907         scalar load/store operations using B/H registers.
20908         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
20909
20910 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
20911
20912         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
20913         scalar move.
20914         * config/aarch64/aarch64.c
20915         (aarch64_simd_scalar_immediate_valid_for_move): New.
20916         * config/aarch64/aarch64-protos.h
20917         (aarch64_simd_scalar_immediate_valid_for_move): New.
20918         * config/aarch64/constraints.md (Dh, Dq): New.
20919         * config/aarch64/iterators.md (hq): New.
20920
20921 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
20922
20923         * reorg.c (get_branch_condition): Deal with conditional returns.
20924         (fill_simple_delay_slots): Remove dead code dealing with jumps.
20925
20926 2013-04-01  Wei Mi  <wmi@google.com>
20927
20928         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
20929         Truncate operand 2 using %b asm operand modifier.
20930         (*<shift_insn><mode>3_mask): Ditto.
20931         (*<rotate_insn><mode>3_mask): Ditto.
20932
20933 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
20934
20935         PR middle-end/56798
20936         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
20937
20938 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
20939
20940         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
20941         of next_real_insn.
20942         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
20943
20944 2013-03-30  Lawrence Crowl  <crowl@google.com>
20945
20946         * dse.c (clear_alias_sets): Remove never set.
20947         (disqualified_clear_alias_sets): Remove never set.
20948         (clear_alias_mode_pool): Remove never set.
20949         (dse_step0): Remove condition that is never true.
20950         (canon_address): Remove condition that is never true.
20951         (dse_step7): Remove condition that is never true.
20952         (rest_of_handle_dse): Remove condition that is never true.
20953         (rest_of_handle_dse::did_global): Remove never read from above.
20954         (dse_step2_spill): Remove never called from above.
20955         (dse_step5_spill): Remove never called from above.
20956
20957 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
20958
20959         * doc/md.texi (Standard Names) <casesi>: Update documentation for
20960         JUMP_TABLE_DATA changes.
20961         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
20962         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
20963         (Insns) <jump_table_data>: New entry.
20964         * doc/tm.texi: Regenerate.
20965
20966         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
20967
20968         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
20969         for table jump at the end of a basic block using tablejump_p.
20970         * targhooks.c (default_invalid_within_doloop): Likewise.
20971         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
20972         target hook implementation that is identical to the default hook.
20973         (rs6000_invalid_within_doloop): Remove.
20974
20975         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
20976         unused variable from tablejump_p call.
20977
20978         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
20979         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
20980         (INSN_DELETED_P): Likewise.
20981         (emit_jump_table_data): New prototype.
20982         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
20983         after 4th as unused.
20984         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
20985         * sched-vis.c (print_insn): Likewise.
20986         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
20987         insn for compatibility with back ends that use next_active_insn to
20988         identify jump table data.
20989         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
20990         (remove_insn): Likewise.
20991         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
20992         to be emitted.
20993         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
20994         (emit_jump_table_data): New function.
20995
20996         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
20997         basic block, a JUMP_TABLE_DATA never is.
20998         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
20999         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21000         off from code handling real insns.
21001         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21002         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21003         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21004         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21005         is not a NONDEBUG_INSN_P.
21006         * ira-costs.c (scan_one_insn): Likewise.
21007         * jump.c (mark_all_labels): Likewise.
21008         (mark_jump_label_1): Likewise.
21009         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21010         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21011         a basic block.
21012         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21013         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21014         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21015         JUMP_TABLE_DATA_P insns.
21016         (calculate_elim_costs_all_insns): Likewise.
21017         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21018         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21019         (delete_output_reload): Code style fixups.
21020         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21021         insn flags on this non-insn.
21022         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21023         as scheduling barriers, for pre-change compatibility.
21024         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21025         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21026
21027         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21028         redundant JUMP_TABLE_DATA_P test.
21029         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21030         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21031         (frv_for_each_packet): Likewise.
21032         * config/i386/i386.c (min_insn_size): Likewise.
21033         (ix86_avoid_jump_mispredicts): Likewise.
21034         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21035         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21036         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21037         (mips16_insn_length): Robustify.
21038         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21039         (mips16_split_long_branches): Likewise.
21040         * config/pa/pa.c (pa_combine_instructions): Likewise.
21041         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21042         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21043         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21044         as contributing to pool range lengths.
21045         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21046         Remove redundant JUMP_TABLE_DATA_P test.
21047         (sh_loop_align): Likewise.
21048         (split_branches): Likewise.
21049         (sh_insn_length_adjustment): Likewise.
21050         * config/spu/spu.c (get_branch_target): Likewise.
21051
21052 2013-03-29  Jan Hubicka  <jh@suse.cz>
21053
21054         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21055         gcov streaming; stream hot bb threshold to ltrans.
21056         * predict.c (get_hot_bb_threshold): Break out from ....
21057         (maybe_hot_count_p): ... here.
21058         (set_hot_bb_threshold): New function.
21059         * lto-section-in.c (lto_section_name): Add profile.
21060         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21061         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21062         and data-streamer.h
21063         (histogram_entry): New structure.
21064         (histogram, histogram_pool): New global vars.
21065         (histogram_hash): New structure.
21066         (histogram_hash::hash): New method.
21067         (histogram_hash::equal): Likewise.
21068         (account_time_size): New function.
21069         (cmp_counts): New function.
21070         (dump_histogram): New function.
21071         (ipa_profile_generate_summary): New function.
21072         (ipa_profile_write_summary): New function.
21073         (ipa_profile_read_summary): New function.
21074         (ipa_profile): Decide on threshold.
21075         (pass_ipa_profile): Add ipa_profile_write_summary and
21076         ipa_profile_read_summary.
21077         * Makefile.in (ipa.o): Update dependencies.
21078         * lto-streamer.h (LTO_section_ipa_profile): New section.
21079
21080 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21081
21082         * tree.h (VAR_P): New.
21083
21084 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21085
21086         PR lto/56777
21087         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21088
21089 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21090
21091         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21092         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21093         (control_flow_insn_p): Likewise.
21094         * cfgrtl.c (duplicate_insn_chain): Likewise.
21095         * final.c (get_attr_length_1): Likewise.
21096         (shorten_branches): Likewise.
21097         (final_scan_insn): Likewise.
21098         * function.c (instantiate_virtual_regs): Likewise.
21099         * gcse.c (insert_insn_end_basic_block): Likewise.
21100         * ira-costs.c (scan_one_insn): Likewise.
21101         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21102         * lra.c (check_rtl): Likewise.
21103         * reload1.c (elimination_costs_in_insn): Likewise.
21104         * reorg.c (follow_jumps): Likewise.
21105
21106         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21107         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21108         (thumb_far_jump_used_p): Likewise.
21109         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21110         (workaround_speculation): Likewise.
21111         (add_sched_insns_for_speculation): Likewise.
21112         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21113         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21114         (frv_for_each_packet): Likewise.
21115         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21116         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21117         (final_emit_insn_group_barriers): Likewise.
21118         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21119         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21120         (mips16_insn_length): Likewise.
21121         * config/pa/pa.c (pa_reorg): Likewise.
21122         (pa_combine_instructions): Likewise.
21123         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21124         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21125         (sh_reorg): Likewise.
21126         (split_branches): Likewise.
21127         * config/spu/spu.c (get_branch_target): Likewise.
21128
21129         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21130         JUMP_TABLE_DATA_P.
21131
21132 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21133
21134         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21135         Fix declaration name.
21136
21137 2013-03-28  Lawrence Crowl  <crowl@google.com>
21138
21139         * graphds.h (struct graph.indicies): Remove unused.
21140         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21141         (SCOP_ORIGINAL_PDDRS): Remove unused.
21142         * sese.h (extern insert_loop_close_phis): Removed unused.
21143         (extern insert_guard_phis): Removed unused.
21144         (extern ivtype_map_elt_info): Removed unused.
21145         (new_ivtype_map_elt): Removed unused.
21146         * sese.c (ivtype_map_elt_info): Removed unused.
21147
21148 2013-03-28  Lawrence Crowl  <crowl@google.com>
21149
21150         * Makefile.in: Add several missing include dependences.
21151         (DUMPFILE_H): New.
21152         (test-dump.o): New.  This object is not added to any executable,
21153         but is present for ad-hoc testing.
21154         * bitmap.c
21155         (debug (const bitmap_head_def &)): New.
21156         (debug (const bitmap_head_def *)): New.
21157         * bitmap.h
21158         (extern debug (const bitmap_head_def &)): New.
21159         (extern debug (const bitmap_head_def *)): New.
21160         * cfg.c
21161         (debug (edge_def &)): New.
21162         (debug (edge_def *)): New.
21163         * cfghooks.c
21164         (debug (basic_block_def &)): New.
21165         (debug (basic_block_def *)): New.
21166         * dumpfile.h
21167         (dump_node (const_tree, int, FILE *)): Correct source file.
21168         * dwarf2out.c
21169         (debug (die_struct &)): New.
21170         (debug (die_struct *)): New.
21171         * dwarf2out.h
21172         (extern debug (die_struct &)): New.
21173         (extern debug (die_struct *)): New.
21174         * gimple-pretty-print.c
21175         (debug (gimple_statement_d &)): New.
21176         (debug (gimple_statement_d *)): New.
21177         * gimple-pretty-print.h
21178         (extern debug (gimple_statement_d &)): New.
21179         (extern debug (gimple_statement_d *)): New.
21180         * ira-build.c
21181         (debug (ira_allocno_copy &)): New.
21182         (debug (ira_allocno_copy *)): New.
21183         (debug (ira_allocno &)): New.
21184         (debug (ira_allocno *)): New.
21185         * ira-int.h
21186         (extern debug (ira_allocno_copy &)): New.
21187         (extern debug (ira_allocno_copy *)): New.
21188         (extern debug (ira_allocno &)): New.
21189         (extern debug (ira_allocno *)): New.
21190         * ira-lives.c
21191         (debug (live_range &)): New.
21192         (debug (live_range *)): New.
21193         * lra-int.h
21194         (debug (lra_live_range &)): New.
21195         (debug (lra_live_range *)): New.
21196         * lra-lives.c
21197         (debug (lra_live_range &)): New.
21198         (debug (lra_live_range *)): New.
21199         * omega.c
21200         (debug (omega_pb_d &)): New.
21201         (debug (omega_pb_d *)): New.
21202         * omega.h
21203         (extern debug (omega_pb_d &)): New.
21204         (extern debug (omega_pb_d *)): New.
21205         * print-rtl.c
21206         (debug (const rtx_def &)): New.
21207         (debug (const rtx_def *)): New.
21208         * print-tree.c
21209         (debug_tree (tree): Move within file.
21210         (debug_raw (const tree_node &)): New.
21211         (debug_raw (const tree_node *)): New.
21212         (dump_tree_via_hooks (const tree_node *, int)): New.
21213         (debug (const tree_node &)): New.
21214         (debug (const tree_node *)): New.
21215         (debug_verbose (const tree_node &)): New.
21216         (debug_verbose (const tree_node *)): New.
21217         (debug_head (const tree_node &)): New.
21218         (debug_head (const tree_node *)): New.
21219         (debug_body (const tree_node &)): New.
21220         (debug_body (const tree_node *)): New.
21221         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21222         (debug (vec<tree, va_gc> &)): New.
21223         (debug (vec<tree, va_gc> *)): New.
21224         * rtl.h
21225         (extern debug (const rtx_def &)): New.
21226         (extern debug (const rtx_def *)): New.
21227         * sbitmap.c
21228         (debug_raw (simple_bitmap_def &)): New.
21229         (debug_raw (simple_bitmap_def *)): New.
21230         (debug (simple_bitmap_def &)): New.
21231         (debug (simple_bitmap_def *)): New.
21232         * sbitmap.h
21233         (extern debug (simple_bitmap_def &)): New.
21234         (extern debug (simple_bitmap_def *)): New.
21235         (extern debug_raw (simple_bitmap_def &)): New.
21236         (extern debug_raw (simple_bitmap_def *)): New.
21237         * sel-sched-dump.c
21238         (debug (vinsn_def &)): New.
21239         (debug (vinsn_def *)): New.
21240         (debug_verbose (vinsn_def &)): New.
21241         (debug_verbose (vinsn_def *)): New.
21242         (debug (expr_def &)): New.
21243         (debug (expr_def *)): New.
21244         (debug_verbose (expr_def &)): New.
21245         (debug_verbose (expr_def *)): New.
21246         (debug (vec<rtx> &)): New.
21247         (debug (vec<rtx> *)): New.
21248         * sel-sched-dump.h
21249         (extern debug (vinsn_def &)): New.
21250         (extern debug (vinsn_def *)): New.
21251         (extern debug_verbose (vinsn_def &)): New.
21252         (extern debug_verbose (vinsn_def *)): New.
21253         (extern debug (expr_def &)): New.
21254         (extern debug (expr_def *)): New.
21255         (extern debug_verbose (expr_def &)): New.
21256         (extern debug_verbose (expr_def *)): New.
21257         (extern debug (vec<rtx> &)): New.
21258         (extern debug (vec<rtx> *)): New.
21259         * sel-sched-ir.h
21260         (_list_iter_cond_expr): Make inline instead of static.
21261         * sreal.c
21262         (debug (sreal &)): New.
21263         (debug (sreal *)): New.
21264         * sreal.h
21265         (extern debug (sreal &)): New.
21266         (extern debug (sreal *)): New.
21267         * tree.h
21268         (extern debug_raw (const tree_node &)): New.
21269         (extern debug_raw (const tree_node *)): New.
21270         (extern debug (const tree_node &)): New.
21271         (extern debug (const tree_node *)): New.
21272         (extern debug_verbose (const tree_node &)): New.
21273         (extern debug_verbose (const tree_node *)): New.
21274         (extern debug_head (const tree_node &)): New.
21275         (extern debug_head (const tree_node *)): New.
21276         (extern debug_body (const tree_node &)): New.
21277         (extern debug_body (const tree_node *)): New.
21278         (extern debug (vec<tree, va_gc> &)): New.
21279         (extern debug (vec<tree, va_gc> *)): New.
21280         * tree-cfg.c
21281         (debug (struct loop &)): New.
21282         (debug (struct loop *)): New.
21283         (debug_verbose (struct loop &)): New.
21284         (debug_verbose (struct loop *)): New.
21285         * tree-dump.c: Add header dependence.
21286         * tree-flow.h
21287         (extern debug (struct loop &)): New.
21288         (extern debug (struct loop *)): New.
21289         (extern debug_verbose (struct loop &)): New.
21290         (extern debug_verbose (struct loop *)): New.
21291         * tree-data-ref.c
21292         (debug (data_reference &)): New.
21293         (debug (data_reference *)): New.
21294         (debug (vec<data_reference_p> &)): New.
21295         (debug (vec<data_reference_p> *)): New.
21296         (debug (vec<ddr_p> &)): New.
21297         (debug (vec<ddr_p> *)): New.
21298         * tree-data-ref.h
21299         (extern debug (data_reference &)): New.
21300         (extern debug (data_reference *)): New.
21301         (extern debug (vec<data_reference_p> &)): New.
21302         (extern debug (vec<data_reference_p> *)): New.
21303         (extern debug (vec<ddr_p> &)): New.
21304         (extern debug (vec<ddr_p> *)): New.
21305         * tree-ssa-alias.c
21306         (debug (pt_solution &)): New.
21307         (debug (pt_solution *)): New.
21308         * tree-ssa-alias.h
21309         (extern debug (pt_solution &)): New.
21310         (extern debug (pt_solution *)): New.
21311         * tree-ssa-alias.c
21312         (debug (_var_map &)): New.
21313         (debug (_var_map *)): New.
21314         (debug (tree_live_info_d &)): New.
21315         (debug (tree_live_info_d *)): New.
21316         * tree-ssa-alias.h
21317         (extern debug (_var_map &)): New.
21318         (extern debug (_var_map *)): New.
21319         (extern debug (tree_live_info_d &)): New.
21320         (extern debug (tree_live_info_d *)): New.
21321
21322 2013-03-28  Jan Hubicka  <jh@suse.cz>
21323
21324         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21325
21326 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21327
21328         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21329         record only when desired or required.
21330
21331 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21332
21333         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21334         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21335
21336 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21337
21338         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21339         (*andsi3_compare0_uxtw): New pattern.
21340         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21341         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21342
21343 2013-03-28  Jan Hubicka  <jh@suse.cz>
21344
21345         * data-streamer-in.c (streamer_read_gcov_count): New function.
21346         * gimple-streamer-out.c: Include value-prof.h.
21347         (output_gimple_stmt): Output histogram.
21348         (output_bb): Use streamer_write_gcov_count.
21349         * value-prof.c: Include data-streamer.h
21350         (dump_histogram_value): Add HIST_TYPE_MAX.
21351         (stream_out_histogram_value): New function.
21352         (stream_in_histogram_value): New function.
21353         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21354         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21355         * data-streamer-out.c (streamer_write_gcov_count): New function.
21356         (streamer_write_gcov_count_stream): New function.
21357         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21358         (lto_output_node): Likewise.
21359         (input_node, input_edge): Likewise.
21360         * lto-streamer-out.c (output_cfg): Update streaming.
21361         * lto-streamer-in.c (input_cfg): Likewise.
21362         * data-streamer.h (streamer_write_gcov_count,
21363         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21364         * gimple-streamer-in.c: Include value-prof.h
21365         (input_gimple_stmt): Input histograms.
21366         (input_bb): Update profile streaming.
21367
21368 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21369
21370         * genmodes.c (emit_max_int): New function.
21371         (emit_insn_modes_h): Added call to emit_max_function.
21372         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21373         Added doc.
21374         * machmode.def: Fixed comment.
21375
21376 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21377
21378         * combine.c (try_combine): Removed useless assert.
21379         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21380
21381 2013-03-28  Marek Polacek  <polacek@redhat.com>
21382             Richard Biener  <rguenther@suse.de>
21383
21384         PR tree-optimization/56695
21385         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21386         build signed result of a vector comparison.
21387         * tree-cfg.c (verify_gimple_comparison): Check that a result
21388         of a vector comparison has signed type.
21389
21390 2013-03-28  Richard Biener  <rguenther@suse.de>
21391
21392         PR tree-optimization/37021
21393         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21394         do not restrict gaps between groups.
21395         * tree-vect-stmts.c (vectorizable_load): Properly account for
21396         a gap between groups.
21397
21398 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21399
21400         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21401         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21402         is not enabled.
21403
21404 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21405
21406         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21407         * doc/extend.texi (Named Address Spaces): Ditto.
21408         (Variable Attributes): Ditto.
21409
21410 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21411
21412         * config.build: Add support for cygwin x64 target.
21413         * config.gcc: Likewise.
21414         * config.host: Likewise.
21415         * configure.ac: Likewise
21416         * configure: Regenerated.
21417
21418 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21419
21420         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21421         * config/i386/t-cygwin-w64: New file.
21422         * config/i386/cygwin-w64.h: New file.
21423         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21424         and add support for x64-cygwin target.
21425         (CPP_SPEC): Likewise.
21426         (CXX_WRAP_SPEC_LIST): Undefine before define.
21427         (LIBGCJ_SONAME): Use 15 as version.
21428
21429 2013-03-27  Richard Biener  <rguenther@suse.de>
21430
21431         PR tree-optimization/56716
21432         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21433         dumping for ref nodes.
21434
21435 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21436
21437         PR tree-optimization/55334
21438         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21439         restricted pointers to arrays.
21440
21441 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21442
21443         * Makefile.in (.SUFFIXES): Add .cc.
21444         (.c.o): Apply same recipe for implicit rule .cc.o.
21445
21446 2013-03-27  Richard Biener  <rguenther@suse.de>
21447
21448         PR tree-optimization/37021
21449         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21450         REALPART/IMAGPART_EXPRs around the supported refs.
21451         * tree-ssa-structalias.c (find_func_aliases): Assume that
21452         floating-point values are not used to transfer pointers.
21453
21454 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21455
21456         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21457         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21458         * doc/tm.texi: Regenerate.
21459         * targhooks.h (default_has_ifunc_p): New.
21460         * targhooks.c (default_has_ifunc_p): Ditto.
21461         * config/linux-protos.h: New file.
21462         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21463         hook for linux which disables support of indirect functions in android.
21464         * config/linux-android.c: New file.
21465         * config/t-linux-android.c: Ditto.
21466         * config.gcc: Added new object file linux-android.o.
21467         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21468         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21469         * varasm.c (do_assemble_alias): Likewise.
21470         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21471         doesn't support indirect functions.
21472         * configure: Regenerate.
21473
21474 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21475
21476         PR target/56102
21477         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21478         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21479         mult-word mode.
21480
21481 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21482
21483         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21484
21485 2013-03-27  Terry Guo  <terry.guo@arm.com>
21486
21487         * config/arm/arm-cores.def: Added core cortex-r7.
21488         * config/arm/arm-tune.md: Regenerated.
21489         * config/arm/arm-tables.opt: Regenerated.
21490         * doc/invoke.texi: Added entry for core cortex-r7.
21491
21492 2013-03-27  Walter Lee  <walt@tilera.com>
21493
21494         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21495         double-decrement of next_scratch_regno.
21496
21497 2013-03-27  Walter Lee  <walt@tilera.com>
21498
21499         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21500         input operands.
21501         (insn_v1mulus): Ditto.
21502         (insn_v2muls): Ditto.
21503
21504 2013-03-27  Walter Lee  <walt@tilera.com>
21505
21506         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21507         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21508
21509 2013-03-27  Walter Lee  <walt@tilera.com>
21510
21511         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21512         (*sibcall_value): Ditto.
21513
21514 2013-03-27  Walter Lee  <walt@tilera.com>
21515
21516         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21517         (insn_mnz_v8qi): ... this ...
21518         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21519         vector equivalent.
21520         (insn_v<n>mnz): Replaced by ...
21521         (insn_v1mnz): ... this ...
21522         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21523         equivalent.
21524         (insn_mz_<mode>): Replaced by ...
21525         (insn_mz_v8qi): ... this ...
21526         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21527         vector equivalent.
21528         (insn_v<n>mz): Replaced by ...
21529         (insn_v1mz): ... this ...
21530         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21531         equivalent.
21532
21533 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21534
21535         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21536
21537 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21538
21539         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21540         than fprintf with a non-constant, non-format string.
21541
21542 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21543
21544         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21545         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21546         operand 0 predicate.
21547         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21548         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21549         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21550         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21551         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21552         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21553         (mov<mode>_insv_1): Remove expander.  Merge insn with
21554         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21555         Use general_x64nomem_operand as operand 1 predicate.
21556         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21557         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21558         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21559         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21560         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21561         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21562         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21563         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21564         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21565         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21566         (general_x64nomem_operand): Ditto.
21567
21568 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21569
21570         * config/rtems.opt: Add -pthread option.
21571
21572 2013-03-26  Richard Biener  <rguenther@suse.de>
21573
21574         * alias.c (find_base_term): Avoid redundant and not used recursion.
21575         (base_alias_check): Get the initial base term from the caller.
21576         (true_dependence_1): Compute and pass base terms to base_alias_check.
21577         (write_dependence_p): Likewise.
21578         (may_alias_p): Likewise.
21579
21580 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21581
21582         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21583         PC-relative load in SI modes and above only.
21584
21585 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21586
21587         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21588         * config/arm/iwmmxt.md (WCGR0): Update.
21589         (WCGR1, WCGR2, WCGR3): Likewise.
21590
21591 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21592
21593         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21594         Use x64 and nox64 isa attributes.
21595
21596 2013-03-26  Richard Biener  <rguenther@suse.de>
21597
21598         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21599         alignment computations and rely on get_object_alignment_1
21600         for the !TYPE_P case.
21601         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21602
21603 2013-03-26  Walter Lee  <walt@tilera.com>
21604
21605         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21606         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21607
21608 2013-03-25  Jeff Law  <law@redhat.com>
21609
21610         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21611         check for INTEGRAL_TYPE_P that was missing due to checking in
21612         wrong version of prior patch.
21613
21614 2013-03-25  Walter Lee  <walt@tilera.com>
21615
21616         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21617         TILEGX_INSN_SHUFFLEBYTES1.
21618         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21619         shufflebytes1.
21620         (tilegx_builtins): Ditto.
21621         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21622
21623 2013-03-25  Walter Lee  <walt@tilera.com>
21624
21625         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21626         (floatunssisf2): New pattern.
21627         (floatsidf2): New pattern.
21628         (floatunssidf2): New pattern.
21629
21630 2013-03-25  Walter Lee  <walt@tilera.com>
21631
21632         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21633         tests for constraint J, K, N, P.
21634
21635 2013-03-25  Walter Lee  <walt@tilera.com>
21636
21637         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21638         Use indirect/pcrel encoding.
21639         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21640         Ditto.
21641
21642 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21643
21644         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21645         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21646         * config/mips/mips.c (mips_option_override): Set IMADD default.
21647         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21648         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21649         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21650         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21651         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21652
21653 2013-03-25  Jeff Law  <law@redhat.com>
21654
21655         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21656         slightly to avoid creating and folding useless trees.  Simplify
21657         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21658
21659 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21660
21661         * config/i386/i386.md (*zero_extendsidi2): Merge with
21662         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21663         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21664         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21665         targets.  Clarify comment.
21666
21667 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21668
21669         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21670         pass-through jump functions differently.
21671         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21672         up jump functions.
21673
21674 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21675
21676         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21677         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21678         process it.
21679         (ipa_get_indirect_edge_target): New function.
21680         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21681         ipa_get_indirect_edge_target.  Update all callers.
21682         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21683         ipa_get_indirect_edge_target_1 instead of calling
21684         ipa_get_indirect_edge_target.
21685         (create_specialized_node): Pass aggvlas to
21686         ipcp_discover_new_direct_edges.
21687
21688 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21689
21690         * config/arm/arm.md (f_sels, f_seld): New types.
21691         (*cmov<mode>): New pattern.
21692         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21693         predicate.
21694
21695 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21696
21697         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21698         POSIX-printf for mingw-hosted  builds.
21699
21700 2013-03-25  Richard Biener  <rguenther@suse.de>
21701
21702         PR middle-end/56694
21703         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21704         must-not-throw stmt location.
21705
21706 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21707
21708         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21709         Emit load-acquire versions when acq is true.
21710         (arm_emit_store_exclusive): Add rel parameter.
21711         Emit store-release versions when rel is true.
21712         (arm_split_compare_and_swap): Use acquire-release instructions
21713         instead.
21714         of barriers when appropriate.
21715         (arm_split_atomic_op): Likewise.
21716         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21717         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21718         (VUNSPEC_SLX): Likewise.
21719         (VUNSPEC_LDA): Likewise.
21720         (VUNSPEC_STL): Likewise.
21721         * config/arm/sync.md (atomic_load<mode>): New pattern.
21722         (atomic_store<mode>): Likewise.
21723         (arm_load_acquire_exclusive<mode>): Likewise.
21724         (arm_load_acquire_exclusivesi): Likewise.
21725         (arm_load_acquire_exclusivedi): Likewise.
21726         (arm_store_release_exclusive<mode>): Likewise.
21727
21728 2013-03-25  Catherine Moore  <clm@codesourcery.com>
21729
21730         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
21731         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
21732         * config/mip/predicates.md (lwsp_swsp_operand,
21733         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
21734         sb16_operand, db4_operand, db7_operand, ib3_operand,
21735         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
21736         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
21737         andi16_operand): New predicates.
21738         * config/mips/mips.md (compression): New attribute.
21739         (enabled): New attribute.
21740         (length): Consider compression in computing length.
21741         (shift_compression): New code attribute.
21742         (*add<mode>3): New operands. Record compression.
21743         (sub<mode>3): Likewise.
21744         (one_cmpl<mode>2): Likewise.
21745         (*and<mode>3): Likewise.
21746         (*ior<mode>3): Likewise.
21747         (unnamed pattern for xor): Likewise.
21748         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
21749         (*<optab><mode>3): Likewise.
21750         (*mov<mode>_internal: Likewise.
21751         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
21752         (mips_unsigned_immediate_p): New.
21753         (umips_lwsp_swsp_address_p): New.
21754         (m16_based_address_p): New.
21755         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
21756         (mips_unsigned_immediate_p): New prototype.
21757         (lwsp_swsp_address_p): New prototype.
21758         (m16_based_address_p): New prototype.
21759         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
21760         (mips_signed_immediate_p): New function.
21761         (m16_based_address_p): New function.
21762         (lwsp_swsp_address_p): New function.
21763         (mips_print_operand_punctuation): Recognize short delay slot insns
21764         for microMIPS.add<mode>3"
21765
21766 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21767
21768         PR target/56720
21769         * config/arm/iterators.md (v_cmp_result): New mode attribute.
21770         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
21771
21772 2013-03-25  Richard Biener  <rguenther@suse.de>
21773
21774         PR tree-optimization/56689
21775         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
21776         any edge.
21777
21778 2013-03-25  Richard Biener  <rguenther@suse.de>
21779
21780         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
21781         of bitmap.
21782         (memory_references): Likewise.
21783         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
21784         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
21785         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
21786         (gather_mem_refs_in_loops): Fold into ...
21787         (analyze_memory_references): ... this.  Move initialization
21788         to tree_ssa_lim_initialize.
21789         (fill_always_executed_in): Rename to ...
21790         (fill_always_executed_in_1): ... this.
21791         (fill_always_executed_in): Move contains_call computation to
21792         this new function from ...
21793         (tree_ssa_lim_initialize): ... here.
21794         (tree_ssa_lim): Call fill_always_executed_in.
21795
21796 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
21797
21798         * postreload.c (reload_combine): Fix code detecting returns.
21799
21800 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
21801
21802         * function.c (emit_use_return_register_into_block): On cc0 targets,
21803         do not emit the sequence between cc0 setter and user.
21804
21805 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21806
21807         * config/i386/predicates.md (local_symbolic_operand): Interpret
21808         dll-imported symbols as none-local.
21809
21810 2013-03-25  Richard Biener  <rguenther@suse.de>
21811
21812         * tree-ssa-loop-im.c (struct depend): Remove.
21813         (struct lim_aux_data): Make depends a vec of gimples.
21814         (free_lim_aux_data): Adjust.
21815         (add_dependency): Likewise.
21816         (set_level): Likewise.
21817
21818 2013-03-25  Richard Biener  <rguenther@suse.de>
21819
21820         PR middle-end/56434
21821         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
21822         the pointer returned by calls with ECF_MALLOC set.
21823
21824 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
21825
21826         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
21827
21828 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
21829
21830         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
21831         using MMXMODE mode iterator.
21832         (*move<mode>_internal): Merge with *movv2sf_internal and
21833         *movv2sf_internal_rex64 using MMXMODE mode iterator.
21834
21835 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
21836
21837         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
21838         (record_last_mem_set_info): Likewise.
21839
21840         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
21841         of XNEWVEC followed by memset.
21842         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
21843
21844 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
21845
21846         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
21847         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
21848         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
21849         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
21850         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
21851         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
21852         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
21853         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
21854         BARRIER_P instead of GET_CODE.
21855
21856 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21857
21858         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
21859         inaccuracy in the probing code.
21860
21861         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
21862         (ctrapdi4): Likewise.
21863
21864 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21865
21866         * calls.c (expand_call): Add missing guard to code handling return
21867         of non-BLKmode structures in MSB.
21868         * function.c (expand_function_end): Likewise.
21869
21870 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
21871
21872         * combine.c (try_combine): Adjust comment.  Do not add the set of
21873         insn #0 if the destination indirectly is set or dies in insn #2.
21874         Tidy up code to distribute a new note.
21875
21876 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
21877
21878         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
21879         also for alternatives 16 and 17.
21880
21881 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
21882
21883         * config/i386/sse.md (*mov<mode>_internal): Merge with
21884         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
21885         Emit insn template depending on type attribute.  Use
21886         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
21887         movd instead of movq mnemonic for interunit moves.  Rewrite mode
21888         attribute calculation.  Remove unit attribute calculation.
21889         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
21890         Set prefix_data16 attribute for DImode ssemov types.
21891         Use Ym instead of y for SSE-MMX conversion alternatives.
21892         Reorder operand constraints.
21893
21894 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
21895
21896         * df.h (df_insn_delete): Adjust prototype.
21897         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
21898         and let it decide whether mark the basic block dirty.
21899         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
21900         * df-scan.c (df_insn_info_delete): New helper function, split
21901         off from df_insn_delete.
21902         (df_scan_free_bb_info): Use it.
21903         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
21904         Likewise.
21905         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
21906         that the insn is actually an insn and it has a non-NULL basic block.
21907         Do not mark basic block dirty if only deleting a DEBUG_INSN.
21908
21909 2013-03-22  Richard Biener  <rguenther@suse.de>
21910
21911         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
21912         dep_ref members.
21913         (mem_ref_alloc): Do not allocate them.
21914         (refs_independent_p): Do not query or maintain a cache.
21915
21916 2013-03-22  Richard Biener  <rguenther@suse.de>
21917
21918         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
21919         (gather_mem_refs_in_loops): Do not compute it.
21920         (analyze_memory_references): Do not allocate it.
21921         (tree_ssa_lim_finalize): Do not free it.
21922         (for_all_locs_in_loop): Do not query all_refs_in_loop.
21923
21924 2013-03-22  Richard Biener  <rguenther@suse.de>
21925
21926         * is-a.h (as_a): Use gcc_checking_assert.
21927
21928 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
21929
21930         * config/aarch64/aarch64.c (aarch64_print_operand): New
21931         format specifier for printing a constant in hex.
21932         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
21933         format specifier for printing second operand.
21934
21935 2013-03-22  Richard Biener  <rguenther@suse.de>
21936
21937         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
21938         bitmaps.
21939         (gather_mem_refs_in_loops): Perform store accumulation here.
21940         (create_vop_ref_mapping_loop): Remove.
21941         (create_vop_ref_mapping): Likewise.
21942         (analyze_memory_references): Initialize refs_stored_in_loop.
21943         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
21944         (record_indep_loop): Remove.
21945         (record_dep_loop): New function.
21946         (ref_indep_loop_p_1): Adjust to only walk over references
21947         in the loop, not its subloops.
21948         (ref_indep_loop_p): Rename to ...
21949         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
21950         maintaining a more fine-grained cache.
21951         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
21952         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
21953
21954 2013-03-22  Richard Biener  <rguenther@suse.de>
21955
21956         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
21957         (struct mem_ref): Make accesses_in_loop a vec of a vec of
21958         aggregate mem_ref_loc.
21959         (free_mem_ref_locs): Inline into ...
21960         (memref_free): ... this and adjust.
21961         (mem_ref_alloc): Adjust.
21962         (mem_ref_locs_alloc): Remove.
21963         (record_mem_ref_loc): Adjust.
21964         (get_all_locs_in_loop): Rewrite into ...
21965         (for_all_locs_in_loop): ... this iterator.
21966         (rewrite_mem_ref_loc): New functor.
21967         (rewrite_mem_refs): Use for_all_locs_in_loop.
21968         (sm_set_flag_if_changed): New functor.
21969         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
21970         (ref_always_accessed): New functor.
21971         (ref_always_accessed_p): Use for_all_locs_in_loop.
21972
21973 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
21974
21975         * tree-pass.h (PROP_gimple_lvec): New.
21976         * passes.c (dump_properties): Handle PROP_gimple_lvec.
21977         (init_optimization_passes): Move pass_lower_vector.
21978         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
21979         PROP_gimple_lvec.
21980         (pass_lower_vector): Provide PROP_gimple_lvec.
21981         (pass_lower_vector_ssa): Likewise.
21982         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
21983
21984 2013-03-21  Mark Wielaard  <mjw@redhat.com>
21985
21986         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
21987
21988 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
21989
21990         * config/i386/i386.md (*movdi_internal): Disparage slightly
21991         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
21992         conversion alternatives.
21993
21994 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
21995
21996         PR middle-end/48087
21997         * diagnostic.def (DK_WERROR): New kind.
21998         * diagnostic.h (werrorcount): Define.
21999         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22000         promoted to DK_ERROR, increment DK_WERROR counter instead of
22001         DK_ERROR counter.
22002         * toplev.c (toplev_main): Call print_ignored_options even if
22003         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22004         even if just werrorcount is non-zero.
22005
22006         PR debug/55608
22007         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22008         on failure.
22009         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22010         (string_cst_pool_decl): New function.
22011         (optimize_one_addr_into_implicit_ptr): New function.
22012         (resolve_addr_in_expr): Optimize DWARF location expression
22013         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22014         which doesn't live in memory, but has DW_AT_location or
22015         DW_AT_const_value, or refers to a string literal, into
22016         DW_OP_GNU_implicit_pointer.
22017         (optimize_location_into_implicit_ptr): New function.
22018         (resolve_addr): If removing DW_AT_location of a variable because
22019         it was DW_OP_addr of address of the variable, but the variable doesn't
22020         live in memory, try to emit const value attribute for the initializer.
22021
22022 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22023
22024         * tree.h (VECTOR_TYPE_P): New macro.
22025         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22026         TYPE_MODE): Use it.
22027         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22028         VEC_COND_EXPR cannot be lvalues.
22029         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22030
22031 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22032
22033         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22034         Restrict the transformation to equal modes.
22035
22036 2013-03-21  Richard Biener  <rguenther@suse.de>
22037
22038         PR tree-optimization/39326
22039         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22040         (MEM_ANALYZABLE): Adjust.
22041         (record_mem_ref_loc): Move bitmap ops ...
22042         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22043         unanalyzable refs, do not record locations for it.
22044         (analyze_memory_references): Allocate ref zero as shared
22045         unanalyzable ref.
22046         (refs_independent_p): Do not test for unanalyzed mems here.
22047         (ref_indep_loop_p_1): Special-case disambiguation against
22048         the unanalyzed ref.
22049         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22050
22051 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22052
22053         * config/arm/arm-protos.h (tune_params): Add
22054         prefer_neon_for_64bits field.
22055         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22056         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22057         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22058         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22059         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22060         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22061         (arm_option_override): Handle -mneon-for-64bits new option.
22062         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22063         (prefer_neon_for_64bits): Declare new variable.
22064         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22065         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22066         (arch_enabled): Handle new arch types. Remove support for onlya8
22067         and nota8.
22068         (one_cmpldi2): Use new arch names.
22069         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22070         * config/arm/arm.opt (mneon-for-64bits): Add option.
22071         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22072         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22073         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22074         of onlya8.
22075         * doc/invoke.texi (-mneon-for-64bits): Document.
22076
22077 2013-03-21  Richard Biener  <rguenther@suse.de>
22078
22079         PR tree-optimization/39326
22080         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22081         (sort_bbs_in_loop_postorder_cmp): New function.
22082         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22083
22084 2013-03-21  Richard Biener  <rguenther@suse.de>
22085
22086         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22087         (vect_insert_into_interleaving_chain): Likewise.
22088         (vect_drs_dependent_in_basic_block): Inline ...
22089         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22090         split out from ...
22091         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22092         (vect_check_interleaving): Simplify.
22093         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22094         (vect_slp_analyze_data_ref_dependences): ... this new function.
22095         (dr_group_sort_cmp): New function.
22096         (vect_analyze_data_ref_accesses): Compute data-reference groups
22097         here instead of in vect_analyze_data_ref_dependence.  Use
22098         a more efficient algorithm.
22099         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22100         vect_slp_analyze_data_ref_dependences.  Call
22101         vect_analyze_data_ref_accesses earlier.
22102         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22103         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22104         (vect_slp_analyze_data_ref_dependences): New prototype.
22105
22106 2013-03-21  Richard Biener  <rguenther@suse.de>
22107
22108         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22109         ref is stored in the loop.
22110         (find_refs_for_sm): Walk only over all stores.
22111         (store_motion_loop): Allocate from lim_bitmap_obstack.
22112         (store_motion): Likewise.
22113
22114 2013-03-21  Richard Biener  <rguenther@suse.de>
22115
22116         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22117         Update virtual SSA form.
22118
22119 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22120
22121         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22122         * configure: Regenerate.
22123         * config.in: Regenerate.
22124         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22125         if !HAVE_LD_EH_FRAME_CIEV3.
22126
22127 2013-03-21  Richard Biener  <rguenther@suse.de>
22128
22129         * tree-cfg.c (verify_expr_no_block): New function.
22130         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22131         nor DECL_VALUE_EXPR have locations with associated blocks.
22132         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22133         (clear_unused_block_pointer): Remove code dealing with
22134         blocks in DECL_DEBUG_EXPR locations.
22135
22136 2013-03-21  Richard Biener  <rguenther@suse.de>
22137
22138         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22139         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22140         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22141         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22142         instead of DECL_DEBUG_EXPR_IS_FROM.
22143         * gimplify.c (gimplify_modify_expr): Likewise.
22144         * tree-cfg.c (verify_expr_location_1): Likewise.
22145         * tree-complex.c (create_one_component_var): Likewise.
22146         * tree-sra.c (create_access_replacement): Likewise.
22147         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22148         (clear_unused_block_pointer): Likewise.
22149         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22150         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22151         * var-tracking.c (var_debug_decl): Likewise.
22152         (track_expr_p): Likewise.
22153         * tree-inline.c (add_local_variables): Likewise.  Set
22154         DECL_HAS_DEBUG_EXPR_P after copying it.
22155         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22156         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22157
22158 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22159
22160         PR bootstrap/56656
22161         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22162         * configure: Regenerate.
22163         * config.in: Regenerate.
22164         * config/i386/i386.md (*movdf_internal): Use
22165         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22166         movd instead of movq mnemonic for interunit moves.
22167         (*movdi_internal): Ditto.
22168
22169 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22170
22171         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22172         (abd<mode>_3): New pattern.
22173         (aba<mode>_3): New pattern.
22174         (fabd<mode>_3): New pattern.
22175
22176 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22177
22178         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22179         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22180         occurrence of REGISTER_PREFIX as its empty string.
22181
22182 2013-03-20  Jeff Law  <law@redhat.com>
22183
22184         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22185         addititional equivalences for equality comparisons between an SSA_NAME
22186         and a constant where the SSA_NAME was set from a widening conversion.
22187
22188 2013-03-20  Walter Lee  <walt@tilera.com>
22189
22190         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22191
22192 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22193
22194         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22195         depending on type attribute.
22196         (*movti_internal): Ditto.
22197         (*movtf_internal): Ditto.
22198         (*movxf_internal): Ditto.
22199         (*movdf_internal): Ditto.
22200         (*movsf_internal): Ditto.
22201
22202 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22203
22204         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22205         maybe_vex for sselog1 and ssemov types.
22206         (*movdi_internal): Reorder operand constraints.
22207         (*movsi_internal): Ditto.  Set prefix attribute to
22208         maybe_vex for sselog1 and ssemov types.
22209         (*movtf_internal): Set prefix attribute to maybe_vex
22210         for sselog1 and ssemov types.
22211         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22212         DImode ssemov types.  Reorder operand constraints.
22213         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22214         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22215         attribute for SImode ssemov types.  Reorder operand constraints.
22216
22217 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22218
22219         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22220         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22221
22222 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22223
22224         * config/rs6000/predicates.md (indexed_address, update_address_mem
22225         update_indexed_address_mem): New predicates.
22226         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22227         attribute for load/store instructions.
22228         * config/rs6000/dfp.md (movsd_store): Likewise.
22229         (movsd_load): Likewise.
22230         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22231         (unnamed HI->DI extend define_insn): Likewise.
22232         (unnamed SI->DI extend define_insn): Likewise.
22233         (unnamed QI->SI extend define_insn): Likewise.
22234         (unnamed QI->HI extend define_insn): Likewise.
22235         (unnamed HI->SI extend define_insn): Likewise.
22236         (unnamed HI->SI extend define_insn): Likewise.
22237         (extendsfdf2_fpr): Likewise.
22238         (movsi_internal1): Likewise.
22239         (movsi_internal1_single): Likewise.
22240         (movhi_internal): Likewise.
22241         (movqi_internal): Likewise.
22242         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22243         attribute for load/store instructions.
22244         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22245         instructions.
22246         (mov<mode>_softfloat): Likewise.
22247         (mov<mode>_hardfloat32): Likewise.
22248         (mov<mode>_hardfloat64): Likewise.
22249         (mov<mode>_softfloat64): Likewise.
22250         (movdi_internal32): Likewise.
22251         (movdi_internal64): Likewise.
22252         (probe_stack_<mode>): Likewise.
22253
22254 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22255
22256         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22257         floating point, and decimal floating point to reload iterator.
22258
22259         * config/rs6000/constraints.md (wl constraint): New constraints to
22260         return FLOAT_REGS if certain options are used to reduce the number
22261         of separate patterns that exist in the file.
22262         (wx constraint): Likewise.
22263         (wz constraint): Likewise.
22264
22265         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22266         -mdebug=reg, print wg, wl, wx, and wz constraints.
22267         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22268         Initialize the reload functions for 64-bit binary/decimal floating
22269         point types.
22270         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22271         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22272         create the buffer on the stack to overcome not having a 32-bit
22273         load and store.
22274         (rs6000_emit_move): Likewise.
22275         (rs6000_secondary_memory_needed_rtx): Likewise.
22276         (rs6000_alloc_sdmode_stack_slot): Likewise.
22277         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22278         via xxlxor, just like DFmode 0.0.
22279
22280         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22281         define as 1 if we are running on a power7 or newer.
22282         (enum r6000_reg_class_enum): Add new constraints.
22283
22284         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22285         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22286         with other moves by using conditional constraits (wg).  Use LFIWZX
22287         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22288         (movsd splitter): Likewise.
22289         (movsd_hardfloat): Likewise.
22290         (movsd_softfloat): Likewise.
22291
22292         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22293         binary and decimal floating point moves.
22294         (fmove_ok): New attributes to combine binary and decimal floating
22295         point moves, and to combine power6x (mfpgpr) moves along normal
22296         floating moves.
22297         (real_value_to_target): Likewise.
22298         (f32_lr): Likewise.
22299         (f32_lm): Likewise.
22300         (f32_li): Likewise.
22301         (f32_sr): Likewise.
22302         (f32_sm): Likewise.
22303         (f32_si): Likewise.
22304         (movsf): Combine binary and decimal floating point moves.  Combine
22305         power6x (mfpgpr) moves with other moves by using conditional
22306         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22307         (mov<mode> for SFmode/SDmode); Likewise.
22308         (SFmode/SDmode splitters): Likewise.
22309         (movsf_hardfloat): Likewise.
22310         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22311         (movsf_softfloat): Likewise.
22312         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22313
22314         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22315         wx and wz constraints.
22316
22317         * config/rs6000/constraints.md (wg constraint): New constraint to
22318         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22319
22320         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22321         constraint.
22322
22323         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22324         -mdebug=reg, print wg, wl, wx, and wz constraints.
22325         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22326         Initialize the reload functions for 64-bit binary/decimal floating
22327         point types.
22328         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22329         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22330         create the buffer on the stack to overcome not having a 32-bit
22331         load and store.
22332         (rs6000_emit_move): Likewise.
22333         (rs6000_secondary_memory_needed_rtx): Likewise.
22334         (rs6000_alloc_sdmode_stack_slot): Likewise.
22335         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22336         via xxlxor, just like DFmode 0.0.
22337
22338         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22339         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22340         with other moves by using conditional constraits (wg).  Use LFIWZX
22341         and STFIWX for loading SDmode on power7.
22342         (movdd splitters): Likewise.
22343         (movdd_hardfloat32): Likewise.
22344         (movdd_softfloat32): Likewise.
22345         (movdd_hardfloat64_mfpgpr): Likewise.
22346         (movdd_hardfloat64): Likewise.
22347         (movdd_softfloat64): Likewise.
22348
22349         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22350         64-bit binary and decimal floating point moves.
22351         (FMOVE64X): Likewise.
22352         (movdf): Combine 64-bit binary and decimal floating point moves.
22353         Combine power6x (mfpgpr) moves with other moves by using
22354         conditional constraits (wg).
22355         (mov<mode> for DFmode/DDmode): Likewise.
22356         (DFmode/DDmode splitters): Likewise.
22357         (movdf_hardfloat32): Likewise.
22358         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22359         (movdf_softfloat32): Likewise.
22360         (movdf_hardfloat64_mfpgpr): Likewise.
22361         (movdf_hardfloat64): Likewise.
22362         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22363         (movdf_softfloat64): Likewise.
22364         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22365         (reload_<mode>_load): Move to later in the file so they aren't in
22366         the middle of the floating point move insns.
22367         (reload_<mode>_store): Likewise.
22368
22369         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22370         constraint.
22371
22372         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22373         constraint if -mdebug=reg.
22374         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22375         Enable using dd reload support if needed.
22376
22377         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22378         binary and decimal floating point moves in rs6000.md.
22379         (movtd_internal): Likewise.
22380
22381         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22382         decimal floating point moves.
22383         (movtf): Likewise.
22384         (movtf_internal): Likewise.
22385         (mov<mode>_internal, TDmode/TFmode): Likewise.
22386         (movtf_softfloat): Likewise.
22387         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22388
22389         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22390         movdi_internal64, using wg constraint for move direct operations.
22391         (movdi_internal64): Likewise.
22392
22393         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22394         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22395         the various virtual registers. Use GPR/FPR first/last values,
22396         instead of hard coding the register numbers.  Print which modes
22397         have reload functions registered.
22398         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22399         settings before/after setting cpu, target and subtarget settings.
22400         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22401         and for secondary reload failures in rs6000_secondary_reload_inner.
22402         (rs6000_secondary_reload_fail): Likewise.
22403         (rs6000_secondary_reload_inner): Likewise.
22404
22405         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22406         macros for first/last GPR and FPR registers.
22407         (LAST_GPR_REGNO): Likewise.
22408         (FIRST_FPR_REGNO): Likewise.
22409         (LAST_FPR_REGNO): Likewise.
22410
22411         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22412         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22413         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22414         (vcond<mode><mode>): Likewise.
22415         (vcondu<mode><mode>): Likewise.
22416         (vector_gtu<mode>): Likewise.
22417         (vector_gte<mode>): Likewise.
22418         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22419         to prevent the compiler from converting DImode operations to TImode.
22420         (ior<mode>3): Likewise.
22421         (and<mode>3): Likewise.
22422         (one_cmpl<mode>2): Likewise.
22423         (nor<mode>3): Likewise.
22424         (andc<mode>3): Likewise.
22425
22426         * config/rs6000/constraints.md (wt constraint): New constraint
22427         that returns VSX_REGS if TImode is allowed in VSX registers.
22428
22429         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22430         constant under VSX.
22431
22432         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22433         similar to TImode, but it is restricted to being in the GPRs.
22434
22435         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22436         TImode to occupy a single VSX register.
22437
22438         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22439         -mvsx-timode for power7/power8.
22440         (power7 cpu): Likewise.
22441         (power8 cpu): Likewise.
22442
22443         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22444         sure that TFmode/TDmode take up two registers if they are ever
22445         allowed in the upper VSX registers.
22446         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22447         registers.
22448         (rs6000_init_hard_regno_mode_ok): Likewise.
22449         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22450         constraint.  Print if LRA is turned on.
22451         (rs6000_option_override_internal): Give an error if -mvsx-timode
22452         and VSX is not enabled.
22453         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22454         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22455         to reg+offset addressing.  Use PTImode when checking offset
22456         addresses for validity.
22457         (reg_offset_addressing_ok_p): Likewise.
22458         (rs6000_legitimate_offset_address_p): Likewise.
22459         (rs6000_legitimize_address): Likewise.
22460         (rs6000_legitimize_reload_address): Likewise.
22461         (rs6000_legitimate_address_p): Likewise.
22462         (rs6000_eliminate_indexed_memrefs): Likewise.
22463         (rs6000_emit_move): Likewise.
22464         (rs6000_secondary_reload): Likewise.
22465         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22466         reloads to fpr registers to continue to use reg+offset addressing,
22467         but 64-bit reloads to altivec registers need reg+reg addressing.
22468         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22469         it.  Treat LO_SUM like a PLUS operation.
22470         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22471         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22472         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22473         registers to share a register with a smaller sized type, since VSX
22474         puts scalars in the upper 64-bits.
22475         (print_operand): Add support for PTImode.
22476         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22477         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22478         registers, but don't have arithmetic support.
22479         (rs6000_memory_move_cost): Add test for VSX.
22480         (rs6000_opt_masks): Add -mvsx-timode.
22481
22482         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22483         for TImode.
22484         (VSs): Likewise.
22485         (VSr): Use wt constraint for TImode.
22486         (VSv): Drop TImode support.
22487         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22488         (vsx_movti_64bit): Likewise.
22489         (vsx_movti_32bit): Likewise.
22490         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22491         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22492         one '?' on the appropriate output constraint.  Do not allow TImode
22493         logical operations on 32-bit systems.
22494         (vsx_ior<mode>3): Likewise.
22495         (vsx_xor<mode>3): Likewise.
22496         (vsx_one_cmpl<mode>2): Likewise.
22497         (vsx_nor<mode>3): Likewise.
22498         (vsx_andc<mode>3): Likewise.
22499         (vsx_concat_<mode>): Likewise.
22500         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22501
22502         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22503         OPTION_MASK_VSX_TIMODE.
22504         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22505         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22506
22507         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22508         (TI2 iterator): New iterator for TImode, PTImode.
22509         (wd mode attribute): Add values for vector types.
22510         (movti_string): Replace TI move operations with operations for TImode
22511         and PTImode.  Add support for TImode being allowed in VSX registers.
22512         (mov<mode>_string, TImode/PTImode): Likewise.
22513         (movti_ppc64): Likewise.
22514         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22515         (TI mode splitters): Likewise.
22516
22517         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22518         constraint.
22519
22520 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22521
22522         PR tree-optimization/56355
22523         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22524         Also handle integers with undefined overflow.
22525
22526 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22527             Maciej W. Rozycki  <macro@codesourcery.com>
22528             Tom de Vries  <tom@codesourcery.com>
22529             Nathan Sidwell  <nathan@codesourcery.com>
22530             Iain Sandoe  <iain@codesourcery.com>
22531             Nathan Froyd  <froydnj@codesourcery.com>
22532             Chao-ying Fu  <fu@mips.com>
22533
22534         * doc/extend.texi: (micromips, nomicromips, nocompression):
22535         Document new function attributes.
22536         * doc/invoke.texi (minterlink-compressed, mmicromips,
22537         m14k, m14ke, m14kec): Document new options.
22538         (minterlink-mips16): Update documentation.
22539         * doc/md.texi (ZC, ZD): Document new constraints.
22540         * configure.ac (gcc_cv_as_micromips): Check if linker
22541         supports the .set micromips directive.
22542         * configure: Regenerate.
22543         * config.in: Regenerate.
22544         * config/mips/mips-tables.opt: Regenerate.
22545         * config/mips/micromips.md: New file.
22546         * constraints.md (ZC, ZD): New constraints.
22547         * config/mips/predicates.md (movep_src_register): New predicate.
22548         (movep_src_operand): New predicate.
22549         (non_volatile_mem_operand): New predicate.
22550         * config/mips/mips.md (multimem): New type.
22551         (length): Differentiate between 17-bit and 18-bit branch offsets.
22552         (MOVEP1, MOVEP2): New mode iterator.
22553         (mov_<load>l): Use ZC constraint.
22554         (mov_<load>r): Likewise.
22555         (mov_<store>l): Likewise.
22556         (mov_<store>r): Likewise.
22557         (*branch_equality<mode>_inverted): Add microMIPS support.
22558         (*branch_equality<mode>): Likewise.
22559         (*jump_absolute): Likewise.
22560         (indirect_jump_<mode>): Likewise.
22561         (tablejump_<mode>): Likewise.
22562         (<optab>_internal): Likewise.
22563         (sibcall_internal): Likewise.
22564         (sibcall_value_internal): Likewise.
22565         (prefetch): Use constraint ZD.
22566         * config/mips/mips.opt (minterlink-compressed): New option.
22567         (minterlink-mips16): Now an alias for minterlink-compressed.
22568         (mmicromips): New option.
22569         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22570         (compare_and_swap_12): Likewise.
22571         (sync_add<mode>): Likewise.
22572         (sync_<optab>_12): Likewise.
22573         (sync_old_<optab>_12): Likewise.
22574         (sync_new_<optab>_12): Likewise.
22575         (sync_nand_12): Likewise.
22576         (sync_old_nand_12): Likewise.
22577         (sync_new_nand_12): Likewise.
22578         (sync_sub<mode>): Likewise.
22579         (sync_old_add<mode>): Likewise.
22580         (sync_old_sub<mode>): Likewise.
22581         (sync_new_add<mode>): Likewise.
22582         (sync_new_sub<mode>): Likewise.
22583         (sync_<optab><mode>): Likewise.
22584         (sync_old_<optab><mode>): Likewise.
22585         (sync_new_<optab><mode>): Likewise.
22586         (sync_nand<mode>): Likewise.
22587         (sync_old_nand<mode>): Likewise.
22588         (sync_new_nand<mode>): Likewise.
22589         (sync_lock_test_and_set<mode>): Likewise.
22590         (test_and_set_12): Likewise.
22591         (atomic_compare_and_swap<mode>): Likewise.
22592         (atomic_exchange<mode>_llsc): Likewise.
22593         (atomic_fetch_add<mode>_llsc): Likewise.
22594         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22595         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22596         (umips_save_restore_pattern_p): Likewise.
22597         (umips_load_store_pair_p): Likewise.
22598         (umips_output_load_store_pair): Likewise.
22599         (umips_movep_target_p): Likewise.
22600         (umips_12bit_offset_address_p): Likewise.
22601         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22602         (mips_base_mips16): Rename this...
22603         (mips_base_compression_flags): ...to this. Update all uses.
22604         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22605         (mips_mips16_decl_p): Delete.
22606         (mips_nomips16_decl_p): Delete.
22607         (mips_get_compress_on_flags): New function.
22608         (mips_get_compress_off_flags): New function.
22609         (mips_get_compress_mode): New function.
22610         (mips_get_compress_on_name): New function.
22611         (mips_get_compress_off_name): New function.
22612         (mips_insert_attributes): Support multiple compression types.
22613         (mips_merge_decl_attributes): Likewise.
22614         (umips_12bit_offset_address_p): New function.
22615         (mips_start_function_definition): Emit .set micromips directive.
22616         (mips_call_may_need_jalx_p): New function.
22617         (mips_function_ok_for_sibcall): Add microMIPS support.
22618         (mips_print_operand_punctuation): Support short delay slots and
22619         compact jumps.
22620         (umips_swm_mask, umips_swm_encoding): New.
22621         (umips_build_save_restore): New function.
22622         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22623         (was_mips16_p): Remove.
22624         (old_compression_mode): New.
22625         (mips_set_compression_mode): New function.
22626         (mips_set_current_function): Add microMIPS support.
22627         (mips_option_override): Likewise.
22628         (umips_save_restore_pattern_p): New function.
22629         (umips_output_save_restore): New function.
22630         (umips_load_store_pair_p_1): New function.
22631         (umips_load_store_pair_p): New function.
22632         (umips_output_load_store_pair_1): New function.
22633         (umips_output_load_store_pair): New function.
22634         (umips_movep_target_p) New function.
22635         (mips_prepare_pch_save): Add microMIPS support.
22636         * config/mips/mips.h (TARGET_COMPRESSION): New.
22637         (TARGET_CPU_CPP_BUILTINS): Update macro
22638         to use new compression flags and to support microMIPS.
22639         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22640         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22641         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22642         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22643         (ASM_SPEC): Support mmicromips and mno-micromips.
22644         (M16STORE_REG_P): New macro.
22645         (MIPS_CALL): Support TARGET_MICROMIPS.
22646         (MICROMIPS_J): New macro.
22647         (mips_base_mips16): Rename this...
22648         (mips_base_compression_flags): ...to this.
22649         (UMIPS_12BIT_OFFSET_P): New macro.
22650         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22651         (MULTILIB_DIRNAMES): Likewise.
22652 2013-03-20  Richard Biener  <rguenther@suse.de>
22653
22654         PR tree-optimization/56661
22655         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22656         the result does not have to be distinct.
22657
22658 2013-03-20  Richard Biener  <rguenther@suse.de>
22659
22660         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22661         remap_gimple_op_r.
22662
22663 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22664             Steven Bosscher  <steven@gcc.gnu.org>
22665
22666         PR rtl-optimization/56605
22667         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22668
22669 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22670
22671         PR bootstrap/56656
22672         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22673         that require movd instead of movq.
22674
22675 2013-03-20  Richard Biener  <rguenther@suse.de>
22676
22677         * tree-ssa-structalias.c (struct variable_info): Add pointer
22678         to the first field of an aggregate with sub-vars.  Make
22679         this and the pointer to the next subfield its ID.
22680         (vi_next): New function.
22681         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22682         storedanything_id, integer_id): Increment by one.
22683         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22684         get_call_clobber_vi): Adjust.
22685         (solution_set_expand): Simplify and speedup.
22686         (solution_set_add): Inline into ...
22687         (set_union_with_increment): ... this.  Adjust accordingly.
22688         (do_sd_constraint): Likewise.
22689         (do_ds_constraint): Likewise.
22690         (do_complex_constraint): Simplify.
22691         (build_pred_graph): Adjust.
22692         (solve_graph): Likewise.  Simplify and speedup.
22693         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22694         get_constraint_for_component_ref, get_constraint_for_1,
22695         first_vi_for_offset, first_or_preceding_vi_for_offset,
22696         create_function_info_for, create_variable_info_for_1,
22697         create_variable_info_for, intra_create_variable_infos): Adjust.
22698         (init_base_vars): Push NULL for ID zero.
22699         (compute_points_to_sets): Adjust.
22700
22701 2013-03-20  Richard Biener  <rguenther@suse.de>
22702
22703         * cfgloop.c (verify_loop_structure): Streamline and avoid
22704         ICEing on corrupt loop tree.
22705         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22706         loop tree.
22707
22708 2013-03-20  Richard Biener  <rguenther@suse.de>
22709
22710         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22711         check whether an SSA update is needed.
22712
22713 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22714
22715         * config/mips/constraints.md (T): Rename to...
22716         (Yf): ...this.
22717         (U): Rename to...
22718         (Yd): ...this.
22719         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22720         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22721
22722 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22723
22724         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22725         (*subsi3_carryin_uxtw): Likewise.
22726
22727 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22728
22729         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
22730         (*rorsi3_insn_uxtw): Likewise.
22731
22732 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22733
22734         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
22735         (*extrsi5_insn_uxtw): Likewise.
22736
22737 2013-03-19  Richard Biener  <rguenther@suse.de>
22738
22739         PR tree-optimization/56273
22740         * passes.c (init_optimization_passes): Move second VRP after DOM.
22741
22742 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
22743
22744         * config/i386/i386.md (*movti_internal): Merge from
22745         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
22746         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
22747         nox64 isa attributes.
22748
22749 2013-03-18  Richard Biener  <rguenther@suse.de>
22750
22751         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
22752         (unite): Likewise.
22753         (merge_node_constraints): Likewise.
22754         (build_succ_graph): Likewise.
22755         (valid_graph_edge): Inline into single caller.
22756         (unify_nodes): Likewise.  Use bitmap_set_bit return value
22757         and cache varinfo.
22758         (scc_visit): Fix formatting and variable use.
22759         (do_sd_constraint): Use gcc_checking_assert.
22760         (do_ds_constraint): Likewise.
22761         (do_complex_constraint): Likewise.
22762         (condense_visit): Likewise.  Cleanup.
22763         (dump_pred_graph): New function.
22764         (perform_var_substitution): Dump the pred-graph before
22765         variable substitution.
22766         (find_equivalent_node): Use gcc_checking_assert.
22767         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
22768
22769 2013-03-18  Richard Biener  <rguenther@suse.de>
22770
22771         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
22772         Remove cond_expr_stmt_list argument and do not gimplify the
22773         built expression.
22774         (vect_loop_versioning): Adjust.
22775         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
22776         Cleanup to use less temporaries.
22777         (vect_create_data_ref_ptr): Cleanup.
22778
22779 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
22780
22781         PR tree-optimization/56635
22782         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
22783         require types_compatible_p types.
22784
22785 2013-03-18  Nick Clifton  <nickc@redhat.com>
22786
22787         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
22788         spurious backslash.
22789
22790         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
22791         Add missing line to comment describing function.
22792
22793 2013-03-18  Richard Biener  <rguenther@suse.de>
22794
22795         PR tree-optimization/56210
22796         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22797         Handle string / character search functions.
22798         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
22799
22800 2013-03-18  Richard Biener  <rguenther@suse.de>
22801
22802         PR middle-end/56483
22803         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
22804         and implement properly.
22805         * gimple.h (gimple_cond_single_var_p): Remove.
22806
22807 2013-03-18  Richard Biener  <rguenther@suse.de>
22808
22809         * tree-data-ref.h (find_data_references_in_loop): Declare.
22810         * tree-data-ref.c (get_references_in_stmt): Use a stack
22811         vector pre-allocated in the callers.
22812         (find_data_references_in_stmt): Adjust.
22813         (graphite_find_data_references_in_stmt): Likewise.
22814         (create_rdg_vertices): Likewise.
22815         (find_data_references_in_loop): Export.
22816         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
22817         Compute dependences here...
22818         (vect_analyze_data_refs): ...not here.  When we encounter
22819         a non-vectorizable data reference in basic-block vectorization
22820         truncate the data reference vector.  Do not bother to
22821         fixup data-dependence information for gather loads.
22822         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
22823         of data references, as reported.
22824
22825 2013-03-18  Richard Biener  <rguenther@suse.de>
22826
22827         PR tree-optimization/3713
22828         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
22829         has_constants and expr.
22830         (stmt_has_constants): Properly valueize SSA names when deciding
22831         whether the stmt has constants.
22832
22833 2013-03-18  Richard Biener  <rguenther@suse.de>
22834
22835         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
22836         whole function when there is nothing to do.
22837         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
22838         * tree-vectorizer.c (vectorize_loops): Update virtual and
22839         loop-closed SSA once.
22840         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
22841
22842 2013-03-18  Richard Biener  <rguenther@suse.de>
22843
22844         PR middle-end/56113
22845         * domwalk.c (bb_postorder): New global static.
22846         (cmp_bb_postorder): New function.
22847         (walk_dominator_tree): Replace scheme imposing an order for
22848         visiting dominator sons by one sorting them at the time they
22849         are pushed on the stack.
22850
22851 2013-03-18  Richard Biener  <rguenther@suse.de>
22852
22853         PR tree-optimization/39326
22854         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
22855         (struct mem_ref): Replace mem member with ao_ref typed member.
22856         (MEM_ANALYZABLE): Adjust.
22857         (memref_eq): Likewise.
22858         (mem_ref_alloc): Likewise.
22859         (gather_mem_refs_stmt): Likewise.
22860         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
22861         (execute_sm_if_changed_flag_set): Adjust.
22862         (execute_sm): Likewise.
22863         (ref_always_accessed_p): Likewise.
22864         (refs_independent_p): Likewise.
22865         (can_sm_ref_p): Likewise.
22866
22867 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
22868
22869         PR c/56566
22870         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
22871         return 1 even for !unsignedp.
22872
22873 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22874
22875         * config/i386/i386.md (isa): Add x64 and nox64.
22876         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
22877         (*pushtf): Enable *roF alternative for x64 isa only.
22878         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
22879         mode attribute of integer alternatives to DImode for TARGET_64BIT.
22880         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
22881         (*movtf_internal): Merge from *movtf_internal_rex64 and
22882         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
22883         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
22884         nox64 isa attributes.
22885         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
22886         nox64 isa attributes.
22887         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
22888
22889 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22890
22891         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22892
22893 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
22894
22895         PR target/56640
22896         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
22897         class names.  Remove trailing comma after "ALL_REGS".
22898
22899 2013-03-16  Jan Hubicka  <jh@suse.cz>
22900
22901         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
22902         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
22903         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
22904         of cgraph_get_create_node.
22905         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
22906
22907 2013-03-16  Jason Merrill  <jason@redhat.com>
22908
22909         PR debug/49090
22910         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
22911         with DW_AT_default_value.
22912
22913 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
22914
22915         * BASE-VER: Set to 4.9.0.
22916
22917 2013-03-14  Andi Kleen  <ak@linux.intel.com>
22918
22919         PR target/56619
22920         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
22921         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
22922         Document _x* TSX intrinsics.
22923
22924 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
22925             David Holsgrove  <david.holsgrove@xilinx.com>
22926
22927         * configure.ac: Add MicroBlaze TLS support detection.
22928         * configure: Regenerate.
22929         * config/microblaze/microblaze-protos.h
22930         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
22931         symbol_mentioned_p, label_mentioned_p): Add prototypes.
22932         * config/microblaze/microblaze.c (microblaze_address_type): Add
22933         ADDRESS_TLS and tls_reloc address types.
22934         (microblaze_address_info): Add tls_reloc.
22935         (TARGET_HAVE_TLS): Define.
22936         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
22937          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
22938          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
22939         load_tls_operand,  microblaze_call_tls_get_addr,
22940         microblaze_legitimize_tls_address): New functions.
22941         (microblaze_classify_unspec): Handle UNSPEC_TLS.
22942         (get_base_reg): Use microblaze_tls_symbol_p.
22943         (microblaze_classify_address): Handle TLS.
22944         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
22945         label_mentioned_p and microblaze_tls_referenced_p.
22946         (microblaze_legitimize_address): Handle TLS.
22947         (microblaze_address_insns): Handle ADDRESS_TLS.
22948         (pic_address_needs_scratch): Handle TLS.
22949         (print_operand_address): Handle TLS.
22950         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
22951         (microblaze_expand_move): Handle TLS.
22952         (microblaze_legitimate_constant_p): Check
22953         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
22954         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
22955         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
22956         (PIC_OFFSET_TABLE_REGNUM): Set.
22957         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
22958         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
22959         (addsi3, movsi_internal2, movdf_internal): Update constraints
22960         * config/microblaze/predicates.md (arith_plus_operand): Define
22961         (move_operand): Redefine as move_src_operand,
22962         check microblaze_tls_referenced_p.
22963
22964 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
22965
22966         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
22967         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
22968
22969 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
22970
22971         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
22972         CC mode for AND.
22973
22974 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
22975
22976         PR tree-optimization/53265
22977         * common.opt (Waggressive-loop-optimizations): New option.
22978         * tree-ssa-loop-niter.c: Include tree-pass.h.
22979         (do_warn_aggressive_loop_optimizations): New function.
22980         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
22981         if number_of_latch_executions returned constant.
22982         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
22983         early.  If number_of_latch_executions returned constant, set
22984         nb_iterations_upper_bound back to it.
22985         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
22986         field.
22987         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
22988         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
22989
22990         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
22991         (MULTILIB_OSDIRNAMES): Set.
22992         * genmultilib: If defaultosdirname doesn't start with :: , set
22993         defaultosdirname2 instead, clear it and emit two . multilib_raw
22994         entries instead of just one.
22995
22996 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
22997
22998         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
22999         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23000         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23001         (SUBTARGET_OVERRIDE_OPTIONS): New.
23002
23003 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23004
23005         PR target/49880
23006         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23007         (musermode): Convert to Var(TARGET_USERMODE).
23008         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23009         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23010         * config/sh/sh.c (sh_option_override): Use
23011         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23012         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23013         condition.
23014         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23015         TARGET_SH4.
23016         (udivsi3_i4_single, divsi3_i4_single): Use
23017         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23018
23019 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23020
23021         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23022         default setting.
23023
23024 2013-03-13  Richard Biener  <rguenther@suse.de>
23025
23026         PR tree-optimization/56608
23027         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23028         calls when vectorizing basic-blocks.
23029
23030 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23031
23032         PR plugins/45078
23033         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23034         tm_file.
23035
23036 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23037
23038         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23039
23040 2013-03-11  Jan Hubicka  <jh@suse.cz>
23041
23042         PR lto/56557
23043         * lto-streamer-out.c (output_symbol_p): Skip references from
23044         constructors of external variables.
23045
23046 2013-03-11  Jan Hubicka  <jh@suse.cz>
23047
23048         PR middle-end/56571
23049         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23050         from pseudos.
23051         * emit-rtl.c (verify_rtx_sharing): Likewise.
23052         (copy_insn_1): Likewise.
23053         * rtl.c (copy_rtx): Likewise.
23054
23055 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23056
23057         PR target/56591
23058         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23059         output_operand_lossage message.
23060
23061 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23062
23063         PR target/56470
23064         * arm.c (shift_op): Validate RTL pattern on the fly.
23065         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23066         the RTL.
23067
23068 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23069
23070         PR target/56347
23071         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23072         new call patterns that clobber %fr12.
23073         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23074         split and postreload patterns.
23075         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23076         registers %fr12 and %fr12R as call used.
23077
23078 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23079
23080         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23081         (canon_address, record_store, replace_read, check_mem_read_rtx,
23082         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23083         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23084         rest_of_handle_dse): Likewise.
23085
23086 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23087
23088         PR middle-end/56524
23089         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23090         Add base_optabs.
23091         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23092         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23093         (save_optabs_if_changed): Replace with...
23094         (init_tree_optimization_optabs): ...this.
23095         * optabs.c (save_optabs_if_changed): Rename to...
23096         (init_tree_optimization_optabs): ...this.  Take the optimization node
23097         as argument.  Do nothing if the base optabs are already correct.
23098         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23099         to recompute optabs.
23100         * function.h (function): Remove optabs field.
23101         * function.c (invoke_set_current_function_hook): Call
23102         init_tree_optimization_optabs.  Use the result to initialize
23103         this_fn_optabs.
23104
23105 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23106
23107         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23108         if GTMA_HAS_NO_INSTRUMENTATION.
23109         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23110         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23111         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23112         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23113         GTMA_HAS_NO_INSTRUMENTATION.
23114
23115 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23116
23117         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23118         libasan_preinit.o.
23119
23120 2013-03-08  Marek Polacek  <polacek@redhat.com>
23121             Jakub Jelinek  <jakub@redhat.com>
23122
23123         PR tree-optimization/56478
23124         * predict.c (is_comparison_with_loop_invariant_p): Change the
23125         type of loop_step to tree.
23126         (predict_loops): Adjust.
23127         (predict_iv_comparison): Perform the computations on double_ints.
23128
23129 2013-03-08  Richard Biener  <rguenther@suse.de>
23130
23131         PR tree-optimization/56570
23132         * tree-cfg.c (verify_expr_location_1): Verify locations for
23133         DECL_DEBUG_EXPR.
23134         * tree-sra.c (create_access_replacement): Strip locations
23135         from DECL_DEBUG_EXPRs.
23136
23137 2013-03-08  Richard Biener  <rguenther@suse.de>
23138
23139         * tree-inline.c (expand_call_inline): Do not associate
23140         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23141         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23142
23143 2013-03-08  Richard Biener  <rguenther@suse.de>
23144
23145         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23146         or block changes with -Og.  Fix for location / block encoding
23147         changes and PHI arguments with locations.
23148
23149 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23150
23151         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23152         for all counters.
23153         (struct output_info): Likewise.
23154         (register_overhead): Remove bad gcc_assert.
23155         (bitmap_find_bit): If there is only a single bitmap element, do not
23156         count a miss as a search.
23157         (print_statistics): Update for counter type changes.
23158         (dump_bitmap_statistics): Likewise.  Print headers such that they
23159         are properly lined up with the printed counters.
23160
23161 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23162
23163         PR tree-optimization/56559
23164         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23165         check that it has only a single use.
23166
23167 2013-03-07  Richard Biener  <rguenther@suse.de>
23168
23169         * doc/invoke.texi (fwhole-program): Discourage use in combination
23170         with -flto.
23171
23172 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23173
23174         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23175
23176         PR tree-optimization/56539
23177         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23178         instead of GSI_CONTINUE_LINKING as last argument to
23179         force_gimple_operand_gsi.  Adjust function comment.
23180
23181         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23182         aarch64-cores.def.
23183
23184         PR middle-end/56548
23185         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23186         promoted mode, convert the result back to the original mode.
23187
23188 2013-03-06  Richard Biener  <rguenther@suse.de>
23189
23190         PR middle-end/56294
23191         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23192         (insert_updated_phi_nodes_compare_uids): New function.
23193         (update_ssa): Sort symbols_to_rename after UID before
23194         traversing it to insert PHI nodes.
23195
23196 2013-03-06  Richard Biener  <rguenther@suse.de>
23197
23198         PR middle-end/50494
23199         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23200         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23201
23202         Revert
23203         2013-02-13  Richard Biener  <rguenther@suse.de>
23204
23205         PR lto/50494
23206         * varasm.c (output_constant_def_1): Get the decl representing
23207         the constant as argument.
23208         (output_constant_def): Wrap output_constant_def_1.
23209         (make_decl_rtl): Use output_constant_def_1 with the decl
23210         representing the constant.
23211         (build_constant_desc): Optionally re-use a decl already
23212         representing the constant.
23213         (tree_output_constant_def): Adjust.
23214
23215 2013-03-06  Joey Ye  <joey.ye@arm.com>
23216
23217         PR lto/50293
23218         * gcc.c (convert_white_space): New function.
23219         (main): Handles white space in function name.
23220
23221 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23222
23223         PR target/56529
23224         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23225         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23226         to SH_DIV_CALL_TABLE for TARGET_SH2.
23227         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23228         list.
23229         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23230         call-table options.
23231
23232 2013-03-05  Sterling Augustine  <saugustine@google.com>
23233             Cary Coutant  <ccoutant@google.com>
23234
23235         PR debug/55364
23236         * dwarf2out.c (resolve_addr): Don't call
23237         remove_loc_list_addr_table_entries a second time for the same
23238         expression.
23239
23240 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23241
23242         PR debug/56510
23243         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23244         (avoid_complex_debug_insns): New function.
23245         (expand_debug_locations): Call it.
23246
23247         PR rtl-optimization/56484
23248         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23249         lifetimes of hard registers on small register class machines.
23250
23251 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23252
23253         * config/microblaze/microblaze-protos.h: Rename
23254         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23255         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23256         fast_interrupt.
23257         (microblaze_fast_interrupt_function_p): New function.
23258         (microblaze_is_interrupt_handler): Rename to
23259         microblaze_is_interrupt_variant and add fast_interrupt check.
23260         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23261         (save_restore_insns): Likewise.
23262         (compute_frame_size): Likewise.
23263         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23264         (microblaze_globalize_label): Likewise.
23265         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23266         * config/microblaze/microblaze.md: Use wrapper
23267         microblaze_is_interrupt_variant.
23268
23269 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23270
23271         * sdbout.c (sdbout_one_type): Switch to current function's section
23272         supporting cold/hot.
23273
23274 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23275
23276         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23277         -mxl-reorder.
23278
23279 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23280
23281         PR middle-end/56461
23282         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23283         if VALGRIND_GET_VBITS is defined, temporarily make object
23284         memory all defined, and restore previous valgrind addressability
23285         and definability afterwards.  Free this_object at the end.
23286
23287         PR middle-end/56461
23288         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23289         right before calling lra_create_live_ranges, also call it
23290         when clearing live_p.  Only call lra_clear_live_ranges
23291         at the end if live_p.
23292
23293         PR middle-end/56461
23294         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23295
23296 2013-03-05  Richard Biener  <rguenther@suse.de>
23297
23298         PR tree-optimization/56521
23299         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23300         value-id.
23301
23302 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23303
23304         PR c++/55135
23305         * except.h (remove_unreachable_eh_regions): New prototype.
23306         * except.c (remove_eh_handler_splicer): New function, split out
23307         of remove_eh_handler.
23308         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23309         warning about running it on many EH regions one at a time.
23310         (remove_unreachable_eh_regions_worker): New function, walk the
23311         EH tree in depth-first order and remove non-marked regions.
23312         (remove_unreachable_eh_regions): New function.
23313         * tree-eh.c (mark_reachable_handlers): New function, split out
23314         from remove_unreachable_handlers.
23315         (remove_unreachable_handlers): Use mark_reachable_handlers and
23316         remove_unreachable_eh_regions.
23317         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23318         and remove_unreachable_eh_regions.
23319
23320 2013-03-05  Richard Biener  <rguenther@suse.de>
23321
23322         PR middle-end/56525
23323         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23324         not freeing them until the end.
23325
23326 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23327
23328         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23329
23330 2013-03-05  Richard Biener  <rguenther@suse.de>
23331
23332         PR tree-optimization/56270
23333         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23334         of loads after scheduling an SLP instance.
23335
23336 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23337
23338         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23339         tic6x.exp.
23340         (check_gcc_parallelize): Run guality.exp as a separate job from
23341         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23342         struct-layout-1.exp with stackalign.exp.
23343
23344         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23345
23346         PR middle-end/56461
23347         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23348         load_index sbitmap even if some bit in it isn't set.
23349
23350         PR middle-end/56461
23351         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23352         (discover_iteration_bound_by_body_walk): Change queues to
23353         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23354         spelling in comment.  Call safe_push on queues[bound_index] directly.
23355         Release queues[queue_index] in every iteration unconditionally.
23356         Release bounds vector.
23357
23358         PR middle-end/56461
23359         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23360         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23361         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23362         inner_phis vector.
23363
23364 2013-03-05  Richard Biener  <rguenther@suse.de>
23365
23366         PR lto/56515
23367         * tree-inline.c (remap_blocks_to_null): New function.
23368         (expand_call_inline): When expanding a call stmt without
23369         an associated block inline remap all callee blocks to NULL.
23370
23371 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23372
23373         PR rtl-optimization/56494
23374         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23375         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23376         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23377
23378         PR middle-end/56461
23379         * sel-sched-ir.c (free_sched_pools): Release
23380         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23381         if succs_info_pool.max_top isn't -1.
23382
23383         PR bootstrap/56509
23384         * opts.c (opts_obstack, opts_concat): Moved to...
23385         * opts-common.c (opts_obstack, opts_concat): ... here.
23386
23387 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23388
23389         PR middle-end/56461
23390         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23391
23392 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23393
23394         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23395         all appropriate places.
23396
23397 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23398
23399         PR tree-optimization/56424
23400         * ipa-split.c (split_function): Do not set the RSO flag if result is
23401         not by reference and its type is a register type.
23402
23403 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23404
23405         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23406         (microblaze_legitimate_pic_operand): Likewise
23407         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23408         new function microblaze_legitimate_pic_operand
23409         * config/microblaze/microblaze-protos.h
23410         (microblaze_legitimate_pic_operand): Declare.
23411
23412 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23413
23414         * config/microblaze/predicates.md (call_insn_simple_operand):
23415         New predicate for supported rtx code types.
23416         * config/microblaze/microblaze.md (call_internal1): Use
23417         call_insn_simple_operand predicate.
23418
23419 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23420
23421         PR middle-end/56461
23422         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23423         partitions.ordered_remove.
23424
23425         PR middle-end/56461
23426         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23427         vec_oprnds0.create (1) for modifier == NONE.
23428
23429         PR middle-end/56461
23430         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23431         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23432         vec_oprnds1 right before pushing anything to it for
23433         scalar_shift_arg.
23434
23435         PR middle-end/56461
23436         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23437         set nbbs to 0 instead of having separate code path.
23438         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23439         instead of false as last argument if returning NULL.
23440
23441 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23442
23443         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23444         the attribute is now called "target" instead of "option".
23445         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23446         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23447         attribute/pragma name for TARGET_OPTION_VALID_P and
23448         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23449         * doc/tm.texi: Regenerated.
23450
23451 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23452
23453         * config/microblaze/microblaze.c:
23454         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23455         * config/microblaze/microblaze.h: Add -mxl-reorder to
23456         DRIVER_SELF_SPECS.
23457         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23458         instructions emitted if TARGET_REORDER.
23459         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23460         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23461         separately.
23462
23463 2013-03-01  Xinliang David Li  <davidxl@google.com>
23464
23465         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23466         walk length.
23467
23468 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23469
23470         PR middle-end/56461
23471         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23472         vector even when returning true.  Fix up function comment formatting.
23473
23474         PR middle-end/56461
23475         * ira-build.c (ira_loop_nodes_count): New variable.
23476         (create_loop_tree_nodes): Initialize it.
23477         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23478
23479         PR middle-end/56461
23480         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23481         method on dr_chain and result_chain.
23482         * tree-vect-stmts.c (vectorizable_store): Only call
23483         result_chain.create if j == 0.
23484
23485         PR middle-end/56461
23486         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23487         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23488         before overwriting it.
23489
23490 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23491
23492         * doc/extended.texi (C Extensions): Change order in @menu
23493         to match @node.
23494         (Other MIPS Built-in Functions): Move last MIPS entry before
23495         "picoChip Built-in Functions".
23496         (SH Built-in Functions): Move after RX Built-in Functions.
23497         * doc/gcc.texi (Introduction): Change order in @menu
23498         to match @node.
23499         * doc/md.texi (Constraints): Ditto.
23500         * gty.texi (Type Information): Ditto.
23501         (User-provided marking routines for template types): Make
23502         subsection.
23503         * doc/invoke.texi (AArch64 Options): Move before
23504         "Adapteva Epiphany Options".
23505
23506 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23507             Jakub Jelinek  <jakub@redhat.com>
23508
23509         PR sanitizer/56454
23510         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23511         no_address_safety_analysis attribute.
23512         * doc/extend.texi (no_address_safety_attribute): Rename to
23513         no_sanitize_address attribute, mention no_address_safety_analysis
23514         attribute as deprecated alias.
23515
23516 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23517
23518         PR middle-end/56461
23519         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23520         type to vec<vec<tree> > *.
23521         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23522         to be vec<tree> instead of vec<tree> *, set vec_defs
23523         to vNULL and call vec_defs.create (number_of_vects), adjust other
23524         uses of vec_defs.
23525         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23526         vectorizable_condition): Adjust vect_get_slp_defs callers.
23527
23528 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23529
23530         * config/aarch64/aarch64.c
23531         (aarch64_float_const_representable): Remove unused variable.
23532
23533 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23534
23535         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23536
23537 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23538
23539         * config/aarch64/aarch64-builtins.c
23540         (aarch64_init_simd_builtins): Make static.
23541
23542 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23543
23544         * config/aarch64/aarch64.c
23545         (aarch64_simd_make_constant): Make static.
23546
23547 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23548
23549         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23550         with no initialization to the RHS of debug statements.
23551
23552 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23553
23554         PR tree-optimization/56294
23555         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23556         Adjust dumping.
23557         (get_access_replacement): Do not call create_access_replacement.
23558         Assert a replacement exists.
23559         (get_repl_default_def_ssa_name): Create the replacement declaration
23560         itself.
23561
23562 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23563
23564         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23565         final_end_function.
23566
23567 2013-02-28  Marek Polacek  <polacek@redhat.com>
23568
23569         PR rtl-optimization/56466
23570         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23571         if we're changing a loop.
23572         (peel_loops_completely): Likewise.
23573
23574 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23575
23576         PR c++/55813
23577         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23578
23579 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23580
23581         PR target/56445
23582         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23583         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23584         INTX_FTYPE_FX, FX_FTYPE_INTX.
23585         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23586
23587 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23588
23589         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23590         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23591         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23592         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23593         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23594         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23595         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23596         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23597         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23598         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23599         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23600         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23601         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23602         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23603         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23604         (avrxmega6): Increase max flash segments from 5 to 6.
23605         * config/avr/t-multilib: Regenerate.
23606         * config/avr/avr-tables.opt: Regenerate.
23607         * doc/avr-mmcu.texi: Regenerate.
23608
23609 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23610
23611         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23612         (avr_device_to_arch): Rename to avr_device_to_ld.
23613         (avr_device_to_as): New prototype.
23614         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23615         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23616         * config/avr/driver-avr.c (avr_device_to_as): New.
23617         (avr_device_to_arch): Rename to avr_device_to_ld.
23618
23619 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23620
23621         PR middle-end/56461
23622         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23623         method on dr_chain and result_chain.
23624
23625         PR middle-end/56461
23626         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23627         pointer_set_destroy on not_executed_last_iteration.
23628
23629         PR middle-end/56461
23630         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23631
23632         PR middle-end/56461
23633         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23634         FOR_EACH_DEFINED_FUNCTION when freeing state.
23635
23636         PR middle-end/56461
23637         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23638         pool_free.
23639         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23640         overwriting it.
23641
23642         PR middle-end/56461
23643         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23644         known_aggs[i].items and release known_aggs vector.
23645
23646         PR middle-end/56461
23647         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23648
23649 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23650
23651         * config/microblaze/microblaze.c (microblaze_emit_compare):
23652         Use xor for EQ/NE comparisions.
23653         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23654         (cbranchsf4): Adjust operator to comparison_operator.
23655
23656 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23657
23658         PR middle-end/56461
23659         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23660         vector.
23661         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23662         vec_safe_push, always update *slot.
23663         (redirect_edge_var_map_clear): Use vec_free.
23664         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23665         (free_var_map_entry): Use vec_free.
23666         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23667         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23668
23669 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23670
23671         PR middle-end/45472
23672         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23673         when the may_trap_p bit of the exprs being merged differs.
23674         Reorder tests for speculativeness in the logical and operator.
23675
23676 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23677
23678         * incpath.c (add_standard_paths): Use reconcat instead of concat
23679         where appropriate and avoid leaking memory.
23680
23681         * opts.h: Include obstack.h.
23682         (opts_concat): New prototype.
23683         (opts_obstack): New declaration.
23684         * opts.c (opts_concat): New function.
23685         (opts_obstack): New variable.
23686         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23687         (finish_options): Use opts_concat instead of concat
23688         and XOBNEWVEC instead of XNEWVEC.
23689         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23690         generate_option): Likewise.
23691         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23692         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23693
23694         PR target/56455
23695         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23696         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23697
23698 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23699
23700         PR middle-end/56461
23701         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23702
23703 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23704
23705         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23706         (arm_block_move_unaligned_straight): Likewise.
23707         (arm_adjust_block_mem): Likewise.
23708
23709 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23710
23711         PR target/48901
23712         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23713         temp, cond and label.
23714         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23715
23716         PR target/52500
23717         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23718         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23719
23720         PR target/52501
23721         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23722         of prologue/epilogue functions.
23723
23724         PR target/52550
23725         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23726         Remove unused variable cfa_offset.
23727         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
23728
23729         PR target/54639
23730         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
23731         type promotion to unsigned.
23732
23733         PR target/54640
23734         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
23735         for HOST_WIDE_INT of 32 bit / same size as int.
23736         (arm_block_move_unaligned_straight): Likewise.
23737         (arm_adjust_block_mem): Likewise.
23738
23739         PR target/54662
23740         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
23741         ALL_CFLAGS.
23742
23743 2013-02-26  Marek Polacek  <polacek@redhat.com>
23744
23745         PR tree-optimization/56426
23746         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
23747
23748 2013-02-26  Richard Biener  <rguenther@suse.de>
23749
23750         PR target/56444
23751         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
23752         unused variable loops.
23753
23754 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23755
23756         PR tree-optimization/56448
23757         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
23758         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
23759         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
23760         later operands of the references, or even first operand for
23761         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
23762
23763         PR tree-optimization/56443
23764         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
23765         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
23766         to type_for_mode langhook.
23767
23768 2013-02-25  Matt Turner  <mattst88@gmail.com>
23769
23770         * doc/invoke.texi: Document r4700.
23771
23772 2013-02-25  Richard Biener  <rguenther@suse.de>
23773
23774         PR tree-optimization/56175
23775         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
23776         split out from ...
23777         (simplify_bitwise_binary): ... here.  Also guard the conversion
23778         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
23779
23780 2013-02-25  Catherine Moore  <clm@codesourcery.com>
23781
23782         Revert:
23783         2013-02-24  Catherine Moore  <clm@codesourcery.com>
23784             Maciej W. Rozycki  <macro@codesourcery.com>
23785             Tom de Vries  <tom@codesourcery.com>
23786             Nathan Sidwell  <nathan@codesourcery.com>
23787             Iain Sandoe  <iain@codesourcery.com>
23788             Nathan Froyd  <froydnj@codesourcery.com>
23789             Chao-ying Fu  <fu@mips.com>
23790
23791         * doc/extend.texi: (micromips, nomicromips, nocompression):
23792         Document new function attributes.
23793         * doc/invoke.texi (minterlink-compressed, mmicromips,
23794         m14k, m14ke, m14kec): Document new options.
23795         (minterlink-mips16): Update documentation.
23796         * doc/md.texi (ZC, ZD): Document new constraints.
23797         * configure.ac (gcc_cv_as_micromips): Check if linker
23798         supports the .set micromips directive.
23799         * configure: Regenerate.
23800         * config.in: Regenerate.
23801         * config/mips/mips-tables.opt: Regenerate.
23802         * config/mips/micromips.md: New file.
23803         * constraints.md (ZC, AD): New constraints.
23804         * config/mips/predicates.md (movep_src_register): New predicate.
23805         (movep_src_operand): New predicate.
23806         (non_volatile_mem_operand): New predicate.
23807         * config/mips/mips.md (multimem): New type.
23808         (length): Differentiate between 17-bit and 18-bit branch offsets.
23809         (MOVEP1, MOVEP2): New mode iterator.
23810         (mov_<load>l): Use ZC constraint.
23811         (mov_<load>r): Likewise.
23812         (mov_<store>l): Likewise.
23813         (mov_<store>r): Likewise.
23814         (*branch_equality<mode>_inverted): Add microMIPS support.
23815         (*branch_equality<mode>): Likewise.
23816         (*jump_absolute): Likewise.
23817         (indirect_jump_<mode>): Likewise.
23818         (tablejump_<mode>): Likewise.
23819         (<optab>_internal): Likewise.
23820         (sibcall_internal): Likewise.
23821         (sibcall_value_internal): Likewise.
23822         (prefetch): Use constraint ZD.
23823         * config/mips/mips.opt (minterlink-compressed): New option.
23824         (minterlink-mips16): Now an alias for minterlink-compressed.
23825         (mmicromips): New option.
23826         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23827         (compare_and_swap_12): Likewise.
23828         (sync_add<mode>): Likewise.
23829         (sync_<optab>_12): Likewise.
23830         (sync_old_<optab>_12): Likewise.
23831         (sync_new_<optab>_12): Likewise.
23832         (sync_nand_12): Likewise.
23833         (sync_old_nand_12): Likewise.
23834         (sync_new_nand_12): Likewise.
23835         (sync_sub<mode>): Likewise.
23836         (sync_old_add<mode>): Likewise.
23837         (sync_old_sub<mode>): Likewise.
23838         (sync_new_add<mode>): Likewise.
23839         (sync_new_sub<mode>): Likewise.
23840         (sync_<optab><mode>): Likewise.
23841         (sync_old_<optab><mode>): Likewise.
23842         (sync_new_<optab><mode>): Likewise.
23843         (sync_nand<mode>): Likewise.
23844         (sync_old_nand<mode>): Likewise.
23845         (sync_new_nand<mode>): Likewise.
23846         (sync_lock_test_and_set<mode>): Likewise.
23847         (test_and_set_12): Likewise.
23848         (atomic_compare_and_swap<mode>): Likewise.
23849         (atomic_exchange<mode>_llsc): Likewise.
23850         (atomic_fetch_add<mode>_llsc): Likewise.
23851         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
23852         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
23853         (umips_save_restore_pattern_p): Likewise.
23854         (umips_load_store_pair_p): Likewise.
23855         (umips_output_load_store_pair): Likewise.
23856         (umips_movep_target_p): Likewise.
23857         (umips_12bit_offset_address_p): Likewise.
23858         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
23859         (mips_base_mips16): Rename this...
23860         (mips_base_compression_flags): ...to this. Update all uses.
23861         (mips_attribute_table): Add micromips, nomicromips and nocompression.
23862         (mips_mips16_decl_p): Delete.
23863         (mips_nomips16_decl_p): Delete.
23864         (mips_get_compress_on_flags): New function.
23865         (mips_get_compress_off_flags): New function.
23866         (mips_get_compress_mode): New function.
23867         (mips_get_compress_on_name): New function.
23868         (mips_get_compress_off_name): New function.
23869         (mips_insert_attributes): Support multiple compression types.
23870         (mips_merge_decl_attributes): Likewise.
23871         (umips_12bit_offset_address_p): New function.
23872         (mips_start_function_definition): Emit .set micromips directive.
23873         (mips_call_may_need_jalx_p): New function.
23874         (mips_function_ok_for_sibcall): Add microMIPS support.
23875         (mips_print_operand_punctuation): Support short delay slots and
23876         compact jumps.
23877         (umips_swm_mask, umips_swm_encoding): New.
23878         (umips_build_save_restore): New function.
23879         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
23880         (was_mips16_p): Remove.
23881         (old_compression_mode): New.
23882         (mips_set_compression_mode): New function.
23883         (mips_set_current_function): Add microMIPS support.
23884         (mips_option_override): Likewise.
23885         (umips_save_restore_pattern_p): New function.
23886         (umips_output_save_restore): New function.
23887         (umips_load_store_pair_p_1): New function.
23888         (umips_load_store_pair_p): New function.
23889         (umips_output_load_store_pair_1): New function.
23890         (umips_output_load_store_pair): New function.
23891         (umips_movep_target_p) New function.
23892         (mips_prepare_pch_save): Add microMIPS support.
23893         * config/mips/mips.h (TARGET_COMPRESSION): New.
23894         (TARGET_CPU_CPP_BUILTINS): Update macro
23895         to use new compression flags and to support microMIPS.
23896         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
23897         (MIPS_ARCH_FLOAT_SPEC): Likewise.
23898         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
23899         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
23900         (ASM_SPEC): Support mmicromips and mno-micromips.
23901         (M16STORE_REG_P): New macro.
23902         (MIPS_CALL): Support TARGET_MICROMIPS.
23903         (MICROMIPS_J): New macro.
23904         (mips_base_mips16): Rename this...
23905         (mips_base_compression_flags): ...to this.
23906         (UMIPS_12BIT_OFFSET_P): New macro.
23907         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
23908         (MULTILIB_DIRNAMES): Likewise.
23909
23910 2013-02-25  Tom de Vries  <tom@codesourcery.com>
23911
23912         PR rtl-optimization/56131
23913         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
23914         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
23915         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
23916
23917 2013-02-25  Tobias Burnus  <burnus@net-b.de>
23918
23919         * doc/invoke.texi (-fsanitize=): Move from optimization
23920         to debugging options.
23921
23922 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
23923
23924         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
23925
23926 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
23927             Alexander Monakov  <amonakov@ispras.ru>
23928
23929         PR middle-end/56077
23930         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
23931         flush pending lists also on non-jumps.  Adjust comment.
23932
23933 2013-02-24  Catherine Moore  <clm@codesourcery.com>
23934             Maciej W. Rozycki  <macro@codesourcery.com>
23935             Tom de Vries  <tom@codesourcery.com>
23936             Nathan Sidwell  <nathan@codesourcery.com>
23937             Iain Sandoe  <iain@codesourcery.com>
23938             Nathan Froyd  <froydnj@codesourcery.com>
23939             Chao-ying Fu  <fu@mips.com>
23940
23941         * doc/extend.texi: (micromips, nomicromips, nocompression):
23942         Document new function attributes.
23943         * doc/invoke.texi (minterlink-compressed, mmicromips,
23944         m14k, m14ke, m14kec): Document new options.
23945         (minterlink-mips16): Update documentation.
23946         * doc/md.texi (ZC, ZD): Document new constraints.
23947         * configure.ac (gcc_cv_as_micromips): Check if linker
23948         supports the .set micromips directive.
23949         * configure: Regenerate.
23950         * config.in: Regenerate.
23951         * config/mips/mips-tables.opt: Regenerate.
23952         * config/mips/micromips.md: New file.
23953         * constraints.md (ZC, AD): New constraints.
23954         * config/mips/predicates.md (movep_src_register): New predicate.
23955         (movep_src_operand): New predicate.
23956         (non_volatile_mem_operand): New predicate.
23957         * config/mips/mips.md (multimem): New type.
23958         (length): Differentiate between 17-bit and 18-bit branch offsets.
23959         (MOVEP1, MOVEP2): New mode iterator.
23960         (mov_<load>l): Use ZC constraint.
23961         (mov_<load>r): Likewise.
23962         (mov_<store>l): Likewise.
23963         (mov_<store>r): Likewise.
23964         (*branch_equality<mode>_inverted): Add microMIPS support.
23965         (*branch_equality<mode>): Likewise.
23966         (*jump_absolute): Likewise.
23967         (indirect_jump_<mode>): Likewise.
23968         (tablejump_<mode>): Likewise.
23969         (<optab>_internal): Likewise.
23970         (sibcall_internal): Likewise.
23971         (sibcall_value_internal): Likewise.
23972         (prefetch): Use constraint ZD.
23973         * config/mips/mips.opt (minterlink-compressed): New option.
23974         (minterlink-mips16): Now an alias for minterlink-compressed.
23975         (mmicromips): New option.
23976         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
23977         (compare_and_swap_12): Likewise.
23978         (sync_add<mode>): Likewise.
23979         (sync_<optab>_12): Likewise.
23980         (sync_old_<optab>_12): Likewise.
23981         (sync_new_<optab>_12): Likewise.
23982         (sync_nand_12): Likewise.
23983         (sync_old_nand_12): Likewise.
23984         (sync_new_nand_12): Likewise.
23985         (sync_sub<mode>): Likewise.
23986         (sync_old_add<mode>): Likewise.
23987         (sync_old_sub<mode>): Likewise.
23988         (sync_new_add<mode>): Likewise.
23989         (sync_new_sub<mode>): Likewise.
23990         (sync_<optab><mode>): Likewise.
23991         (sync_old_<optab><mode>): Likewise.
23992         (sync_new_<optab><mode>): Likewise.
23993         (sync_nand<mode>): Likewise.
23994         (sync_old_nand<mode>): Likewise.
23995         (sync_new_nand<mode>): Likewise.
23996         (sync_lock_test_and_set<mode>): Likewise.
23997         (test_and_set_12): Likewise.
23998         (atomic_compare_and_swap<mode>): Likewise.
23999         (atomic_exchange<mode>_llsc): Likewise.
24000         (atomic_fetch_add<mode>_llsc): Likewise.
24001         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24002         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24003         (umips_save_restore_pattern_p): Likewise.
24004         (umips_load_store_pair_p): Likewise.
24005         (umips_output_load_store_pair): Likewise.
24006         (umips_movep_target_p): Likewise.
24007         (umips_12bit_offset_address_p): Likewise.
24008         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24009         (mips_base_mips16): Rename this...
24010         (mips_base_compression_flags): ...to this. Update all uses.
24011         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24012         (mips_mips16_decl_p): Delete.
24013         (mips_nomips16_decl_p): Delete.
24014         (mips_get_compress_on_flags): New function.
24015         (mips_get_compress_off_flags): New function.
24016         (mips_get_compress_mode): New function.
24017         (mips_get_compress_on_name): New function.
24018         (mips_get_compress_off_name): New function.
24019         (mips_insert_attributes): Support multiple compression types.
24020         (mips_merge_decl_attributes): Likewise.
24021         (umips_12bit_offset_address_p): New function.
24022         (mips_start_function_definition): Emit .set micromips directive.
24023         (mips_call_may_need_jalx_p): New function.
24024         (mips_function_ok_for_sibcall): Add microMIPS support.
24025         (mips_print_operand_punctuation): Support short delay slots and
24026         compact jumps.
24027         (umips_swm_mask, umips_swm_encoding): New.
24028         (umips_build_save_restore): New function.
24029         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24030         (was_mips16_p): Remove.
24031         (old_compression_mode): New.
24032         (mips_set_compression_mode): New function.
24033         (mips_set_current_function): Add microMIPS support.
24034         (mips_option_override): Likewise.
24035         (umips_save_restore_pattern_p): New function.
24036         (umips_output_save_restore): New function.
24037         (umips_load_store_pair_p_1): New function.
24038         (umips_load_store_pair_p): New function.
24039         (umips_output_load_store_pair_1): New function.
24040         (umips_output_load_store_pair): New function.
24041         (umips_movep_target_p) New function.
24042         (mips_prepare_pch_save): Add microMIPS support.
24043         * config/mips/mips.h (TARGET_COMPRESSION): New.
24044         (TARGET_CPU_CPP_BUILTINS): Update macro
24045         to use new compression flags and to support microMIPS.
24046         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24047         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24048         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24049         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24050         (ASM_SPEC): Support mmicromips and mno-micromips.
24051         (M16STORE_REG_P): New macro.
24052         (MIPS_CALL): Support TARGET_MICROMIPS.
24053         (MICROMIPS_J): New macro.
24054         (mips_base_mips16): Rename this...
24055         (mips_base_compression_flags): ...to this.
24056         (UMIPS_12BIT_OFFSET_P): New macro.
24057         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24058         (MULTILIB_DIRNAMES): Likewise.
24059
24060 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24061
24062         PR target/52555
24063         * target-globals.c (save_target_globals): For init_reg_sets and
24064         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24065
24066 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24067
24068         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24069         * config/aarch64/t-aarch64
24070         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24071
24072 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24073
24074         PR inline-asm/56148
24075         * lra-constraints.c (process_alt_operands): Reload operand
24076         conflicting with earlier clobber only if no more other conflicting
24077         operands.
24078
24079 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24080
24081         PR sanitizer/56393
24082         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24083         if not linking a shared library.
24084
24085 2013-02-22  Seth LaForge  <sethml@google.com>
24086
24087         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24088
24089 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24090
24091         * config/arm/arm.md (split for extendsidi): Update condition.
24092         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24093         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24094         (qhs_zextenddi_cstr): Likewise.
24095
24096 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24097
24098         PR middle-end/56420
24099         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24100         avoid signed wrapping.
24101         (expand_mult): Handle properly multiplication by
24102         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24103         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24104         in the compiler if coeff is HOST_WIDE_INT_MIN.
24105         (expand_divmod): Don't make ext_op1 static, change it's type to
24106         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24107
24108         PR rtl-optimization/50339
24109         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24110         field.
24111         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24112         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24113         into splitting_ashiftrt field.
24114         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24115         ASHIFTRT.
24116         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24117         choices.
24118
24119 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24120
24121         PR middle-end/56108
24122         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24123         are sure to go irrevocable.
24124
24125 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24126
24127         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24128         scalars are valid operands.
24129
24130 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24131
24132         PR tree-optimization/56310
24133         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24134         only matching indices and non-negative final offsets.
24135         (intersect_aggregates_with_edge): Pass src_idx to
24136         agg_replacements_to_vector.  Pass src_idx insstead of index to
24137         intersect_with_agg_replacements.
24138
24139 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24140
24141         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24142         instead of hard-wired defaults.
24143
24144 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24145
24146         * doc/invoke.texi (MIPS Options): Update documentation of the
24147         floating-point multiply-accumulate instruction restrictions.
24148
24149 2013-02-21  Kostya Serebryany  <kcc@google.com>
24150
24151         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24152         asan_shadow_offset on x86_64 linux.
24153
24154 2013-02-21  Richard Biener  <rguenther@suse.de>
24155
24156         PR tree-optimization/56415
24157         Revert
24158         2013-02-11  Richard Biener  <rguenther@suse.de>
24159
24160         PR tree-optimization/56273
24161         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24162         first VRP run.
24163
24164 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24165
24166         PR bootstrap/56258
24167         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24168         instead of @itemx.
24169
24170         PR inline-asm/56405
24171         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24172         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24173
24174 2013-02-20  Jan Hubicka  <jh@suse.cz>
24175
24176         PR tree-optimization/56265
24177         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24178         when target is referenced for first time.
24179
24180 2013-02-20  Richard Biener  <rguenther@suse.de>
24181
24182         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24183         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24184         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24185         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24186         not return anything.
24187         (rename_ssa_copies): Do not remove unused locals.
24188         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24189         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24190         * passes.c (execute_function_todo): Do not schedule unused locals
24191         removal if cleanup_tree_cfg did something.
24192         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24193         about the number of removed locals.
24194
24195 2013-02-20  Richard Biener  <rguenther@suse.de>
24196
24197         PR tree-optimization/56398
24198         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24199
24200 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24201
24202         PR tree-optimization/55334
24203         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24204         restricted pointers to arrays.
24205
24206 2013-02-20  Richard Biener  <rguenther@suse.de>
24207             Jakub Jelinek  <jakub@redhat.com>
24208
24209         PR tree-optimization/56396
24210         * tree-ssa-ccp.c (n_const_val): New static variable.
24211         (get_value): Return NULL for SSA names we don't have a lattice
24212         entry for.
24213         (ccp_initialize): Initialize n_const_val.
24214         * tree-ssa-copy.c (n_copy_of): New static variable.
24215         (init_copy_prop): Initialize n_copy_of.
24216         (get_value): Return NULL_TREE for SSA names we don't have a
24217         lattice entry for.
24218
24219 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24220
24221         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24222
24223 2013-02-20  Richard Biener  <rguenther@suse.de>
24224
24225         * genpreds.c (write_lookup_constraint): Do not compare first
24226         letter of the constraint again.
24227
24228 2013-02-20  Richard Biener  <rguenther@suse.de>
24229
24230         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24231         and ceil_log2.
24232         (get_use_iv_cost): Terminate hashtable walk when coming across
24233         an empty entry.
24234
24235 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24236
24237         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24238         reassociation for avx2 targets.
24239
24240 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24241
24242         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24243         Add version check for v8.10.a to enable microblaze_has_clz
24244         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24245         version and TARGET_PATTERN_COMPARE check
24246         * config/microblaze/microblaze.md: New clzsi2 instruction
24247
24248 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24249
24250         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24251         function before branching.
24252
24253 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24254
24255         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24256         DUMP_INSN_RTX_UID.
24257         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24258
24259 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24260
24261         PR middle-end/55889
24262         * sel-sched.c: Include ira.h.
24263         (implicit_clobber_conflict_p): New function.
24264         (moveup_expr): Use it.
24265         * Makefile.in (sel-sched.o): Depend on ira.h.
24266
24267 2013-02-19  Richard Biener  <rguenther@suse.de>
24268
24269         PR tree-optimization/56384
24270         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24271         (vn_hash_type): Split out from ...
24272         (vn_hash_constant_with_type): ... here.
24273         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24274         (vn_phi_eq): Compare types from vn_phi_s structure.
24275         (vn_phi_lookup): Populate vn_phi_s type.
24276         (vn_phi_insert): Likewise.
24277
24278 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24279
24280         PR tree-optimization/56350
24281         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24282         if haven't found reduction or nested cycle operand, rather than
24283         asserting we must find it.
24284
24285         PR tree-optimization/56381
24286         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24287         to fold_build3.
24288
24289 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24290             Jakub Jelinek  <jakub@redhat.com>
24291
24292         PR target/52555
24293         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24294         (swap_optab_enable): Same.
24295         (init_all_optabs): Use argument instead of global.
24296         * tree.h (struct tree_optimization_option): New field target_optabs.
24297         * expr.h (init_all_optabs): Add argument to prototype.
24298         (TREE_OPTIMIZATION_OPTABS): New.
24299         (save_optabs_if_changed): Protoize.
24300         * optabs.h: Declare this_fn_optabs.
24301         * optabs.c (save_optabs_if_changed): New.
24302         Declare this_fn_optabs.
24303         (init_optabs): Add argument to init_all_optabs() call.
24304         * function.c (invoke_set_current_function_hook): Handle per
24305         function optabs.
24306         * function.h (struct function): New field optabs.
24307         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24308         optimization_current_node has changed.
24309         * target-globals.h (save_target_globals_default_opts): Protoize.
24310         * target-globals.c (save_target_globals_default_opts): New.
24311
24312 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24313
24314         PR target/56347
24315         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24316         registers %fr12 and %fr12R as call used.
24317
24318         PR target/56214
24319         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24320         and HImode, require all displacements to be an integer multiple of
24321         their mode size.
24322         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24323         only allow QImode and HImode when reload is in progress and strict is
24324         true.  Likewise for symbolic addresses.  Use base14_operand to check
24325         displacements in REG+BASE addresses.
24326
24327 2013-02-18  Richard Biener  <rguenther@suse.de>
24328
24329         PR tree-optimization/56366
24330         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24331         handle sign-conversion of outer-loop initial induction value.
24332
24333 2013-02-18  Richard Biener  <rguenther@suse.de>
24334
24335         PR middle-end/56349
24336         * cfghooks.c (merge_blocks): If we merge a latch into another
24337         block adjust references to it.
24338         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24339         (verify_loop_structure): Verify that a recorded latch is in fact
24340         a latch.
24341
24342 2013-02-18  Richard Biener  <rguenther@suse.de>
24343
24344         PR tree-optimization/56321
24345         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24346         order SSA name release and virtual operand unlinking.
24347
24348 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24349
24350         * config/microblaze/microblaze.md (save_stack_block): Define.
24351         (restore_stack_block): Likewise.
24352
24353 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24354
24355         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24356         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24357         * config/microblaze/microblaze.c (microblaze_option_override):
24358         Bail out early for PIC modes when target does not support PIC.
24359
24360 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24361
24362         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24363         Replace with a microblaze version.
24364         (microblaze_trampoline_init): Adapt for microblaze.
24365         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24366         microblaze.
24367
24368 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24369             Dodji Seketeli  <dodji@redhat.com>
24370
24371         PR asan/56330
24372         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24373         (instrument_mem_region_access): Do not forget to always put
24374         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24375         0) statement, even for cases where either 'base' or 'base + len'
24376         are not instrumented -- because they have been previously
24377         instrumented.  Simplify the logic by putting all the statements
24378         instrument 'base + len' inside a sequence, and then insert that
24379         sequence right before the current insertion point.  Then, to
24380         instrument 'base + len', just get an iterator on that statement.
24381         And do not forget to update the pointer to iterator the function
24382         received as argument.
24383
24384 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24385
24386         PR rtl-optimization/56348
24387         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24388
24389 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24390
24391         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24392         (clean_graph_dump_file): Pass base to start_graph_dump.
24393
24394 2013-02-14  Richard Henderson  <rth@redhat.com>
24395
24396         PR target/55941
24397         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24398
24399 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24400
24401         * collect2-aix.h: Define F_LOADONLY.
24402
24403 2013-02-14  Richard Biener  <rguenther@suse.de>
24404
24405         PR lto/50494
24406         * varasm.c (output_constant_def_1): Get the decl representing
24407         the constant as argument.
24408         (output_constant_def): Wrap output_constant_def_1.
24409         (make_decl_rtl): Use output_constant_def_1 with the decl
24410         representing the constant.
24411         (build_constant_desc): Optionally re-use a decl already
24412         representing the constant.
24413         (tree_output_constant_def): Adjust.
24414
24415 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24416
24417         Fix an asan crash
24418         * asan.c (instrument_builtin_call):  Really put the length of the
24419         second source argument into src1_len.
24420
24421 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24422
24423         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24424         argument.  If it is false, don't create edge from then_bb to
24425         fallthru_bb.
24426         (insert_if_then_before_iter): Pass true to it.
24427         (build_check_stmt): Pass false to it.
24428         (transform_statements): Flush hash table only on extended basic
24429         block boundaries, rather than at the beginning of every bb.
24430         Don't flush hash table on nonfreeing_call_p calls.
24431         * tree-flow.h (nonfreeing_call_p): New prototype.
24432         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24433
24434 2013-02-13  David S. Miller  <davem@davemloft.net>
24435
24436         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24437
24438 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24439
24440         PR target/56184
24441         * ira.c (max_regno_before_ira): Move from ...
24442         (ira): ... here.
24443         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24444         vec_safe_length.
24445
24446 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24447
24448         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24449
24450 2013-02-13  Richard Biener  <rguenther@suse.de>
24451
24452         PR lto/56295
24453         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24454         globals in MEM_REFs.
24455
24456 2013-02-13  Richard Biener  <rguenther@suse.de>
24457
24458         * loop-init.c (loop_optimizer_init): Clear loop state when
24459         re-initializing preserved loops.
24460         * loop-unswitch.c (unswitch_single_loop): Return whether
24461         we unswitched the loop.  Do not verify loop state here.
24462         (unswitch_loops): When we unswitched a loop discover new loops.
24463
24464 2013-02-13  Kostya Serebryany  <kcc@google.com>
24465
24466         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24467         on x86_64 linux.
24468         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24469
24470 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24471
24472         Avoid instrumenting duplicated memory access in the same basic block
24473         * Makefile.in (asan.o): Add new dependency on hash-table.h
24474         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24475         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24476         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24477         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24478         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24479         (get_mem_ref_of_assignment): New functions.
24480         (get_mem_refs_of_builtin_call): Extract from
24481         instrument_builtin_call and tweak a little bit to make it fit with
24482         the new signature.
24483         (instrument_builtin_call): Use the new
24484         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24485         of is_gimple_builtin_call.
24486         (instrument_derefs, instrument_mem_region_access): Insert the
24487         instrumented memory reference into the hash table.
24488         (maybe_instrument_assignment): Renamed instrument_assignment into
24489         this, and change it to advance the iterator when instrumentation
24490         actually happened and return true in that case.  This makes it
24491         homogeneous with maybe_instrument_assignment, and thus give a
24492         chance to callers to be more 'regular'.
24493         (transform_statements): Clear the memory reference hash table
24494         whenever we enter a new BB, when we cross a function call, or when
24495         we are done transforming statements.  Use
24496         maybe_instrument_assignment instead of instrumentation.  No more
24497         need to special case maybe_instrument_assignment and advance the
24498         iterator after calling it; it's now handled just like
24499         maybe_instrument_call.  Update comment.
24500
24501 2013-02-13  Richard Biener  <rguenther@suse.de>
24502
24503         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24504         Fix loop discovery code.
24505
24506 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24507
24508         PR inline-asm/56148
24509         * lra-constraints.c (process_alt_operands): Match early clobber
24510         operand with itself.  Check conflicts with earlyclobber only if
24511         the operand is not reloaded.  Prefer to reload conflicting operand
24512         if earlyclobber and matching operands are the same.
24513
24514 2013-02-12  Richard Biener  <rguenther@suse.de>
24515
24516         PR lto/56297
24517         * lto-streamer-out.c (write_symbol): Do not output symbols
24518         for hard register variables.
24519
24520 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24521
24522         PR target/54222
24523         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24524         (umulsidi3_insn, mulsidi3_insn): New insns.
24525
24526 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24527
24528         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24529         (struct tune_params): Add vec_costs field.
24530         * config/arm/arm.c (arm_builtin_vectorization_cost)
24531         (arm_add_stmt_cost): New functions.
24532         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24533         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24534         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24535         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24536         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24537         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24538         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24539
24540 2013-02-12  Richard Biener  <rguenther@suse.de>
24541
24542         PR lto/56295
24543         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24544         decls again if possible.
24545
24546 2013-02-12  Richard Biener  <rguenther@suse.de>
24547
24548         PR middle-end/56288
24549         * tree-ssa.c (verify_ssa_name): Fix check, move
24550         SSA_NAME_IN_FREE_LIST check up.
24551
24552 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24553             Steven Bosscher   <steven@gcc.gnu.org>
24554
24555         PR rtl-optimization/56151
24556         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24557         equal to op0 or op1, and last_insn pattern is CODE operation
24558         with MEM dest and one of the operands matches that MEM.
24559
24560 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24561
24562         * doc/extend.texi: Document Function Multiversioning and "default"
24563         parameter string to target attribute.
24564         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24565         target attribute parameter is "default".
24566         (ix86_compare_version_priority): Remove checks for target attribute.
24567         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24568         Remove check for target attribute equal to NULL. Add assert.
24569         (ix86_generate_version_dispatcher_body): Change error to sorry.
24570
24571 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24572             Jack Howarth  <howarth@bromo.med.uc.edu>
24573             Patrick Marlier  <patrick.marlier@gmail.com>
24574
24575         PR libitm/55693
24576         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24577         define ENDFILE_SPEC as TM_DESTRUCTOR.
24578         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24579
24580 2013-02-11  Alexander Potapenko  <glider@google.com>
24581             Jack Howarth  <howarth@bromo.med.uc.edu>
24582             Jakub Jelinek  <jakub@redhat.com>
24583
24584         PR sanitizer/55617
24585         * config/darwin.c (cdtor_record): Rename ctor_record.
24586         (sort_cdtor_records): Rename sort_ctor_records.
24587         (finalize_dtors): New routine to sort destructors by
24588         priority before use in assemble_integer.
24589         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24590
24591 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24592
24593         PR rtl-optimization/56275
24594         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24595         offset is non-negative and less than cmode size before
24596         calling simplify_subreg.
24597
24598 2013-02-11  Richard Biener  <rguenther@suse.de>
24599
24600         PR tree-optimization/56264
24601         * cfgloop.h (fix_loop_structure): Adjust prototype.
24602         * loop-init.c (fix_loop_structure): Return the number of
24603         newly discovered loops.
24604         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24605         are discovered, do a full loop-closed SSA rewrite.
24606
24607 2013-02-11  Richard Biener  <rguenther@suse.de>
24608
24609         PR tree-optimization/56273
24610         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24611         first VRP run.
24612         (check_array_ref): Fix missing newline in dumps.
24613         (search_for_addr_array): Likewise.
24614
24615 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24616
24617         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24618
24619 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24620
24621         PR target/56256
24622         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24623
24624 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24625
24626         PR rtl-optimization/56246
24627         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24628         reload pseudo.
24629         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24630         constraints are satisfied.
24631
24632 2013-02-08  Jeff Law  <law@redhat.com>
24633
24634         PR debug/53948
24635         * emit-rtl.c (reg_is_parm_p): New function.
24636         * regs.h (reg_is_parm_p): New prototype.
24637         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24638         callee-clobbered registers.
24639
24640 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24641
24642         PR target/56043
24643         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24644         If there is no implicit builtin declaration, just return NULL.
24645
24646 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24647
24648         * config/i386/sse.md (FMAMODEM): New mode iterator.
24649         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24650         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24651
24652 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24653
24654         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24655         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24656         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24657
24658 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24659
24660         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24661         (microblaze*-*-elf): Likewise.
24662         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24663         LINK_SPEC.
24664         * config/microblaze/microblaze-c.c: Add builtin defines for
24665         _LITTLE_ENDIAN and _BIG_ENDIAN.
24666         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24667         add to TARGET_DEFAULT flags.
24668         Expand ASM_SPEC and LINK_SPEC.
24669         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24670         * config/microblaze/microblaze.md: Update extendsidi2 and
24671         movdi_internal instructions to use low-order / high-order reg
24672         print_operands.
24673         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24674         options and inversemask / mask of LITTLE_ENDIAN.
24675         * config/microblaze/t-microblaze: Expand multilib options to
24676         include mlittle-endian (le) and update exceptions patterns.
24677
24678 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24679
24680         PR rtl-optimization/56195
24681         * lra-constraints.c (get_reload_reg): Don't reuse regs
24682         if they have smaller mode than requested, if they have
24683         wider mode than requested, try to return a SUBREG.
24684
24685         PR tree-optimization/56250
24686         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24687         if type is unsigned and code isn't MULT_EXPR.
24688
24689 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24690
24691         PR tree-optimization/56064
24692         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24693         bits according to mode.
24694         * fixed-value.h (fixed_from_double_int)
24695         (const_fixed_from_double_int): Adjust comments.
24696
24697 2013-02-08  Richard Biener  <rguenther@suse.de>
24698
24699         PR lto/56231
24700         * lto-streamer.h (struct data_in): Remove current_file, current_line
24701         and current_col members.
24702         * lto-streamer-out.c (lto_output_location): Stream changed bits
24703         en-block for efficiency.
24704         * lto-streamer-in.c (clear_line_info): Remove.
24705         (lto_input_location): Cache current file, line and column
24706         globally via local statics.  Read changed bits en-block.
24707         (input_function): Do not call clear_line_info.
24708         (lto_read_body): Likewise.
24709         (lto_input_toplevel_asms): Likewise.
24710
24711 2013-02-08  Michael Matz  <matz@suse.de>
24712
24713         PR tree-optimization/52448
24714         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24715         (nt_call_phase): New static.
24716         (add_or_mark_expr): Only mark accesses with newer phase than any
24717         call seen.
24718         (nonfreeing_call_p): New.
24719         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24720         (nt_fini_block): Keep blocks marked as visited.
24721         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24722
24723 2013-02-08  Richard Biener  <rguenther@suse.de>
24724
24725         * ira.c (ira): Free broken dominator information.
24726
24727 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24728
24729         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
24730
24731 2013-02-08  Marek Polacek  <polacek@redhat.com>
24732
24733         * cfgloop.c (verify_loop_structure): Add more checking of headers.
24734
24735 2013-02-08  Richard Biener  <rguenther@suse.de>
24736
24737         PR middle-end/56181
24738         * cfgloop.h (flow_loops_find): Adjust.
24739         (bb_loop_header_p): Declare.
24740         * cfgloop.c (bb_loop_header_p): New function split out from ...
24741         (flow_loops_find): ... here.  Adjust function signature,
24742         support incremental loop structure update.
24743         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
24744         * cfgloopmanip.c (fix_loop_structure): Move ...
24745         * loop-init.c (fix_loop_structure): ... here.
24746         (apply_loop_flags): Split out from ...
24747         (loop_optimizer_init): ... here.
24748         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
24749         in incremental mode, only remove dead loops here.
24750
24751 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24752
24753         PR target/54222
24754         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
24755         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
24756         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
24757         (*round<mode>3.libgcc): New insns for fixed-modes.
24758         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
24759         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
24760         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
24761         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
24762         implementations.  Define to __builtin_avr_absFX,
24763         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
24764         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
24765         __builtin_avr_countlsFX, respectively.
24766         * config/avr/avr-c.c (target.h): Include it.
24767         (enum avr_builtin_id): New enum.
24768         (avr_resolve_overloaded_builtin): New static function.
24769         (avr_register_target_pragmas): Use it to set
24770         targetm.resolve_overloaded_builtin.
24771         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
24772         tree nodes used by DEF_BUILTIN.
24773         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
24774         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
24775         <AVR_BUILTIN_xxBITS>: Same.
24776
24777 2013-02-08  Richard Biener  <rguenther@suse.de>
24778
24779         * cfgloop.c (verify_loop_structure): Properly handle
24780         a loop exiting to another loop header.
24781         * ira-int.h (ira_loops): Remove.
24782         * ira.c (ira_loops): Remove.
24783         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
24784         (do_reload): Use loop_optimizer_finalize.
24785         * ira-build.c (create_loop_tree_nodes): Use get_loops and
24786         number_of_loops to access the loop tree.
24787         (more_one_region_p): Likewise.
24788         (finish_loop_tree_nodes): Likewise.
24789         (rebuild_regno_allocno_maps): Likewise.
24790         (mark_loops_for_removal): Likewise.
24791         (mark_all_loops_for_removal): Likewise.
24792         (remove_unnecessary_regions): Likewise.
24793         (ira_build): Likewise.
24794         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
24795
24796 2013-02-08  Richard Biener  <rguenther@suse.de>
24797
24798         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
24799         * ipa-pure-const.c (analyze_function): Avoid calling
24800         mark_irreducible_loops twice.
24801         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
24802
24803 2013-02-07  David S. Miller  <davem@davemloft.net>
24804
24805         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
24806         on 'reg'.
24807         * var-tracking.c (vt_add_function_parameter): Test the presence of
24808         HAVE_window_save properly and do not remap argument registers when
24809         we have a leaf function.
24810
24811 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
24812
24813         PR bootstrap/56227
24814         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
24815         instead of "ll".
24816         * config/i386/i386.c (ix86_print_operand): Ditto.
24817
24818 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
24819
24820         * lra-constraints.c (process_alt_operands): Fix recently added comment.
24821
24822 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
24823
24824         PR rtl-optimization/56225
24825         * lra-constraints.c (process_alt_operands): Check that reload hard
24826         reg can hold value for strict_low_part.
24827
24828 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
24829
24830         PR debug/56154
24831         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
24832         dwarf2out_end_function.
24833         (in_first_function_p, maybe_at_text_label_p,
24834         first_loclabel_num_not_at_text_label): New variables.
24835         (dwarf2out_var_location): In the first function find out
24836         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
24837         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
24838         functions.
24839
24840 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
24841
24842         PR rtl-optimization/56178
24843         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
24844         SUBREG of a register.  Tidy up related block of code.
24845         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
24846         note if the source is a register or a SUBREG of a register.
24847
24848 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
24849
24850         PR target/56228
24851         * config/rs6000/rs6000.md (ptrm): New mode attr.
24852         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
24853         call_value_indirect_aix<pttrsize>,
24854         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
24855         m in constraints.
24856
24857 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
24858
24859         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
24860         if -bnortl. Convert to strcmp and strncmp.
24861
24862 2013-02-07  Alan Modra  <amodra@gmail.com>
24863
24864         PR target/54009
24865         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
24866         addresses won't wrap when offsetting.
24867         (rs6000_secondary_reload): Provide secondary reloads needed for
24868         wrapping LO_SUM addresses.
24869
24870 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
24871
24872         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
24873         MACH, just __MACH__.
24874
24875 2013-02-06  Richard Biener  <rguenther@suse.de>
24876
24877         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
24878         instead of calling fix_loop_structure.
24879
24880 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
24881
24882         PR middle-end/56217
24883         * omp-low.c (use_pointer_for_field): Return false if
24884         lower_send_shared_vars doesn't generate any copy-out code.
24885
24886 2013-02-06  Tom de Vries  <tom@codesourcery.com>
24887
24888         PR rtl-optimization/56131
24889         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
24890         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
24891         of the label is NULL.  Add comment.
24892
24893 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
24894
24895         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
24896
24897         PR sanitizer/55374
24898         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
24899         (STATIC_LIBTSAN_LIBS): Likewise.
24900         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
24901         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
24902         is defined, don't add anything else beyond that.
24903         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
24904         (LINK_COMMAND_SPEC): Use them.
24905
24906         PR tree-optimization/56205
24907         * tree-stdarg.c (check_all_va_list_escapes): Return true if
24908         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
24909         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
24910
24911 2013-02-05  Richard Biener  <rguenther@suse.de>
24912
24913         PR tree-optimization/53342
24914         PR tree-optimization/53185
24915         * tree-vectorizer.h (vect_check_strided_load): Remove.
24916         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
24917         not disallow peeling for vectorized strided loads.
24918         (vect_check_strided_load): Make static and simplify.
24919         (vect_analyze_data_refs): Adjust.
24920         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
24921         correctly when vectorizing strided loads.
24922
24923 2013-02-05  Richard Biener  <rguenther@suse.de>
24924
24925         * doc/install.texi: Refer to ISL, not PPL.
24926
24927 2013-02-05  Jan Hubicka  <jh@suse.cz>
24928
24929         PR tree-optimization/55789
24930         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
24931
24932 2013-02-05  Jan Hubicka  <jh@suse.cz>
24933
24934         PR tree-optimization/55789
24935         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
24936         the dead call anyway.
24937
24938 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
24939
24940         PR sanitizer/55374
24941         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
24942
24943 2013-02-04  Alexander Potapenko  <glider@google.com>
24944             Jack Howarth  <howarth@bromo.med.uc.edu>
24945             Jakub Jelinek  <jakub@redhat.com>
24946
24947         PR sanitizer/55617
24948         * config/darwin.c (sort_ctor_records): Stabilized qsort
24949         on constructor priority by using original position.
24950         (finalize_ctors): New routine to sort constructors by
24951         priority before use in assemble_integer.
24952         (machopic_asm_out_constructor): Use finalize_ctors if needed.
24953
24954 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
24955
24956         PR libstdc++/54314
24957         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
24958         about visibility on artificial decls.
24959         * config/sol2.c (solaris_assemble_visibility): Likewise.
24960
24961 2013-02-04  Kai Tietz  <ktietz@redhat.com>
24962
24963         PR target/56186
24964         * config/i386/i386.c (function_value_ms_64): Add additional valtype
24965         argument and improve checking of return-argument types for 16-byte
24966         modes.
24967         (ix86_function_value_1): Add additional valtype argument on call
24968         of function_value_64.
24969         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
24970         handling infunction_value_64 function.
24971
24972 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
24973
24974         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
24975
24976 2013-02-04  Richard Biener  <rguenther@suse.de>
24977
24978         PR tree-optimization/56188
24979         * tree-ssa-structalias.c (label_visit): Consider case with
24980         initially non-empty points-to set.
24981         (perform_var_substitution): Dump node mapping and clean up.
24982
24983 2013-02-04  Richard Guenther  <rguenther@suse.de>
24984
24985         PR lto/56168
24986         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
24987         node prevail as last resort.
24988         (lto_symtab_merge_decls): Remove guard on LTRANS here.
24989         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
24990
24991 2013-02-04  Richard Biener  <rguenther@suse.de>
24992
24993         PR tree-optimization/56113
24994         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
24995         Merge into ...
24996         (equiv_class_lookup_or_add): ... this.
24997         (label_visit): Adjust and fix error in previous patch.
24998         (perform_var_substitution): Adjust.
24999
25000 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25001
25002         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25003         * config/sh/sh4-300.md: Likewise.
25004         * config/sh/sh4a.md: Likewise.
25005         * config/sh/constraints.md: Likewise.
25006         * config/sh/sh.md: Likewise.
25007         * config/sh/netbsd-elf.h: Likewise.
25008         * config/sh/predicates.md: Likewise.
25009         * config/sh/sh-protos.h: Likewise.
25010         * config/sh/ushmedia.h: Likewise.
25011         * config/sh/linux.h: Likewise.
25012         * config/sh/sh.c: Likewise.
25013         * config/sh/superh.h: Likewise.
25014         * config/sh/elf.h: Likewise.
25015         * config/sh/sh4.md: Likewise.
25016         * config/sh/sh.h: Likewise.
25017
25018 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25019
25020         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25021         constraint to match_test floating_point_store_memory_operand().
25022         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25023         (base14_operand): New.
25024         (floating_point_store_memory_operand): New.
25025         (integer_store_memory_operand): Revise to use base14_operand and
25026         reg_plus_base_memory_operand.
25027         (move_dest_operand): Allow symbolic_memory_operands.
25028         (symbolic_memory_operand): Check for LO_SOM.
25029         (symbolic_operand): Change default case to break.
25030         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25031         CONST_DOUBLE values to be reloaded by putting them into memory when
25032         the destination is a floating point register.
25033         (movdf): Remove code to handle CONST_DOUBLE.
25034         (movsf): Likewise.
25035         (reload_indf_r1): New.
25036         (reload_insf_r1): New.
25037         Consistently use "Q" and "T" constraints with integer and floating
25038         point move instructions, respectively.
25039         (movdi): Remove FAIL.
25040         Change predicate for source operand unamed DImode move from
25041         general_operand to move_src_operand.
25042         (umulsidi3): Change predicate for destination operand to
25043         register_operand.
25044         Likewise for similar unamed patterns.
25045         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25046         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25047         (hppa_legitimize_address): Simplify mask calculation.
25048         (pa_emit_move_sequence): Revised handling of secondary reloads from
25049         REG+D addresses for floating point loads and stores.  Directly handle
25050         loading CONST0_RTX (mode) to a floating point register.
25051         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25052         to floating point registers.  Don't restrict secondary reloads to
25053         floating point registers to integer modes.  Revise some comments and
25054         cleanup some code.
25055         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25056         (pa_legitimate_address_p): New.
25057         (pa_legitimize_reload_address): New.
25058         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25059         (STRICT_REG_OK_FOR_BASE_P): New.
25060         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25061         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25062
25063 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25064             Andrew Dixie  <andrewd@gentrack.com>
25065
25066         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25067         flag set.
25068
25069 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25070
25071         * expmed.c (extract_bit_field_1): Pass the full width of the
25072         structure to get_best_reg_extraction_insn.
25073
25074 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25075
25076         PR target/54601
25077         * configure.ac (use_cxa_atexit): Add AIX.
25078         * configure: Regenerate.
25079
25080         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25081
25082 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25083
25084         PR debug/54793
25085         * final.c (need_profile_function): New variable.
25086         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25087         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25088         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25089         notes, targetm.asm_out.function_prologue doesn't emit anything,
25090         HAVE_prologue and profiler should be emitted before prologue,
25091         set need_profile_function instead of emitting it.
25092         (final_scan_insn): If need_profile_function, emit
25093         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25094         NOTE_INSN_FUNCTION_BEG note.
25095
25096 2013-02-01  Richard Henderson  <rth@redhat.com>
25097
25098         * config/rs6000/rs6000.md (smulditi3): New.
25099         (umulditi3): New.
25100
25101         * config/alpha/alpha.md (umulditi3): New.
25102
25103 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25104
25105         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25106         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25107
25108 2013-02-01  Richard Biener  <rguenther@suse.de>
25109
25110         PR tree-optimization/56113
25111         * tree-ssa-structalias.c (label_visit): Reduce work for
25112         single-predecessor nodes.
25113
25114 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25115
25116         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25117         range isn't testing for zero.
25118
25119 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25120
25121         PR middle-end/56113
25122         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25123
25124 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25125             Nick Clifton  <nickc@redhat.com>
25126
25127         * config/v850/constraints.md (Q): Define as a memory constraint.
25128         * config/v850/predicates.md (label_ref_operand): New predicate.
25129         (e3v5_shift_operand): New predicate.
25130         (ior_operator): New predicate.
25131         * config/v850/t-v850: Add e3v5 multilib.
25132         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25133         (v850_gen_movdi): Prototype.
25134         * config/v850/v850.c: Add support for e3v5 architecture.
25135         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25136         TARGET_V850E_UP.
25137         (construct_save_jarl): Add e3v5 long JARL support.
25138         (v850_adjust_insn_length): New function.  Adjust length of call
25139         insns when using e3v5 instructions.
25140         (v850_gen_movdi): New function: Generate instructions to move a
25141         DImode value.
25142         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25143         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25144         (TARGET_USE_FPU): Enable for e3v5.
25145         (CONST_OK_FOR_W): New macro.
25146         (ADJUST_INSN_LENGTH): Define.
25147         * config/v850/v850.md (UNSPEC_LOOP): Define.
25148         (attr cpu): Add v850e3v5.
25149         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25150         (movdi): New pattern.
25151         (movdi_internal): New pattern.
25152         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25153         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25154         (cstoresf4): Likewise.
25155         (cstoredf4): Likewise.
25156         (insv): New pattern.
25157         (rotlso3_a): New pattern.
25158         (rotlsi3_b): New pattern
25159         (rotlsi3_v850e3v5): New pattern.
25160         (doloop_begin): New pattern.
25161         (fix_loop_counter): New pattern.
25162         (doloop_end): New pattern.
25163         (branch_normal): Add e3v5 long branch support.
25164         (branch_invert): Likewise.
25165         (branch_z_normal): Likewise.
25166         (branch_z_invert): Likewise.
25167         (branch_nz_normal): Likewise.
25168         (branch_nz_invert): Likewise.
25169         (call_internal_short): Add e3v5 register-indirect JARL support.
25170         (call_internal_long): Likewise.
25171         (call_value_internal_short): Likewise.
25172         (call_value_internal_long): Likewise.
25173         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25174         (mloop): New option.
25175         * config.gcc: Add support for configuring v840e3v5 target.
25176         * doc/invoke.texi: Document new v850 specific command line options.
25177
25178 2013-01-31  Paul Koning  <ni1d@arrl.net>
25179
25180         PR debug/55059
25181         PR debug/54508
25182         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25183         children if parent is a class.
25184         (prune_unused_types_prune): Don't add DW_AT_declaration.
25185
25186 2013-01-31  Richard Biener  <rguenther@suse.de>
25187
25188         PR tree-optimization/56157
25189         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25190         match up operand with SLP child.
25191
25192 2013-01-31  Jason Merrill  <jason@redhat.com>
25193
25194         PR debug/54410
25195         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25196         parameters the first time.
25197         (gen_scheduled_generic_parms_dies): Check completeness here.
25198
25199 2013-01-31  Richard Biener  <rguenther@suse.de>
25200
25201         PR middle-end/53073
25202         * common.opt (faggressive-loop-optimizations): New flag,
25203         enabled by default.
25204         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25205         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25206         infer_loop_bounds_from_undefined by it.
25207
25208 2013-01-31  Richard Biener  <rguenther@suse.de>
25209
25210         PR tree-optimization/56150
25211         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25212         visit virtual operands.
25213         (find_uses_to_rename_bb): Likewise.
25214
25215 2013-01-31  Richard Biener  <rguenther@suse.de>
25216
25217         PR tree-optimization/56150
25218         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25219         mixed store non-store stmts.
25220
25221 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25222
25223         PR sanitizer/55374
25224         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25225         LIBASAN_EARLY_SPEC is defined.
25226         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25227         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25228         before %o.
25229         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25230
25231         PR c++/55742
25232         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25233         invalid args instead of ICEing on it.
25234         (ix86_valid_target_attribute_tree): Return error_mark_node if
25235         ix86_valid_target_attribute_inner_p failed.
25236         (ix86_valid_target_attribute_p): Return false only if
25237         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25238         target("default") attribute.
25239         (sorted_attr_string): Change argument from const char * to tree,
25240         merge in all target attribute arguments rather than just one.
25241         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25242         instead of free.  Avoid using strcat.
25243         (ix86_mangle_function_version_assembler_name): Mangle
25244         target("default") as if no target attribute is present.  Adjust
25245         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25246         instead of xmalloc and XDELETEVEC instead of free.
25247         (ix86_function_versions): Don't return true if one of the decls
25248         doesn't have target attribute.  If they don't and one of the decls
25249         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25250         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25251         (ix86_supports_function_versions): Remove.
25252         (make_name): Fix up formatting.
25253         (make_dispatcher_decl): Remove resolver_name and its initialization.
25254         Avoid leaking memory.
25255         (is_function_default_version): Return true if there is
25256         target("default") attribute rather than no target attribute at all.
25257         (make_resolver_func): Avoid leaking memory.
25258         (ix86_generate_version_dispatcher_body): Likewise.
25259         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25260         * target.def (supports_function_versions): Remove.
25261         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25262         * doc/tm.texi: Regenerated.
25263
25264 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25265
25266         PR rtl-optimization/56144
25267         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25268         for values with side effects.
25269
25270 2013-01-30  Richard Biener  <rguenther@suse.de>
25271
25272         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25273         (sparseset_pop): Likewise.
25274         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25275         to be able to use quick_push in the worker loop.
25276
25277 2013-01-30  Marek Polacek  <polacek@redhat.com>
25278
25279         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25280
25281 2013-01-30  Richard Biener  <rguenther@suse.de>
25282
25283         PR lto/56147
25284         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25285
25286 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25287
25288         PR tree-optimization/56064
25289         * fixed-value.c (fixed_from_double_int): New function.
25290         * fixed-value.h (fixed_from_double_int): New prototype.
25291         (const_fixed_from_double_int): New static inline function.
25292         * fold-const.c (native_interpret_fixed): New static function.
25293         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25294         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25295         (native_encode_fixed): New static function.
25296         (native_encode_expr) <FIXED_CST>: Use it.
25297         (native_interpret_int): Move double_int worker code to...
25298         * double-int.c (double_int::from_buffer): ...this new static method.
25299         * double-int.h (double_int::from_buffer): Prototype it.
25300
25301 2013-01-30  Richard Biener  <rguenther@suse.de>
25302
25303         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25304         New pointer-map and obstack.
25305         (init_alias_vars): Allocate pointer-map and obstack.
25306         (delete_points_to_sets): Free them.
25307         (find_what_var_points_to): Cache result.
25308         (find_what_p_points_to): Adjust for changed interface of
25309         find_what_var_points_to.
25310         (compute_points_to_sets): Likewise.
25311         (ipa_pta_execute): Likewise.
25312
25313 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25314
25315         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25316         * configure: Regenerate.
25317         * config.in: Regenerate.
25318         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25319         #nobits/#progbits if supported.
25320
25321 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25322
25323         PR target/56121
25324         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25325         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25326         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25327
25328 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25329
25330         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25331         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25332
25333 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25334
25335         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25336         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25337
25338 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25339
25340         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25341         declaration.
25342         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25343         * config/arm/cortex-a7.md: New bypasses using
25344         arm_mac_accumulator_is_result.
25345
25346 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25347
25348         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25349         (cortex_a7_neon_mla): Likewise.
25350         (cortex_a7_fpfmad): New reservation.
25351         (cortex_a7_fpmacs): Use ffmas and update required units.
25352         (cortex_a7_fpmuld): Update required units and latency.
25353         (cortex_a7_fpmacd): Likewise.
25354         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25355         (cortex_a7_neon). Likewise.
25356         (bypass) Update participating units.
25357
25358 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25359
25360         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25361         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25362         from fmac to ffma.
25363         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25364         (vfp_fmul): Use ffmad.
25365         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25366         (cortex_r4_fmacd): Use ffmad.
25367         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25368         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25369         (cortex_a9_fmacd): Use ffmad.
25370         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25371         (cortex_a8_vfp_macd): Use ffmad.
25372         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25373         (cortex_a5_fpmacd): Use ffmad.
25374         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25375         (cortex_a15_vfp_macd): Use ffmad.
25376         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25377
25378 2013-01-29  Jason Merrill  <jason@redhat.com>
25379
25380         PR libstdc++/54314
25381         * varasm.c (default_assemble_visibility): Don't warn about
25382         visibility on artificial decls.
25383
25384 2013-01-29  Richard Biener  <rguenther@suse.de>
25385
25386         PR tree-optimization/56113
25387         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25388         the bitmap leader.
25389         (label_visit): Free duplicate bitmaps and record the leader instead.
25390         (perform_var_substitution): Adjust.
25391
25392 2013-01-29  Richard Biener  <rguenther@suse.de>
25393
25394         PR tree-optimization/55270
25395         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25396         the CFG, schedule loops for fixup.
25397
25398 2013-01-29  Nick Clifton  <nickc@redhat.com>
25399
25400         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25401         SP_REG.
25402
25403 2013-01-28  Leif Ekblad  <leif@rdos.net>
25404
25405         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25406         * config/i386/i386.h (TARGET_RDOS): New macro.
25407         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25408         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25409         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25410         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25411         DEFAULT_LARGE_SECTION_THRESHOLD.
25412         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25413         * config/i386/rdos.h: New file.
25414         * config/i386/rdos64.h: New file.
25415
25416 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25417
25418         PR other/54814
25419         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25420         TEST_HARD_REG_BIT.
25421
25422 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25423
25424         PR rtl-optimization/56117
25425         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25426         call cselib_lookup_from_insn on the MEM before calling
25427         add_insn_mem_dependence.
25428
25429 2013-01-28  Richard Biener  <rguenther@suse.de>
25430
25431         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25432         to a stmt that didn't have one.
25433         (copy_phis_for_bb): Likewise for PHI arguments.
25434         (copy_debug_stmt): Likewise for debug stmts.
25435
25436 2013-01-28  Richard Biener  <rguenther@suse.de>
25437
25438         PR tree-optimization/56034
25439         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25440         (partition_builtin_p): Adjust.
25441         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25442         it is the last partition.
25443         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25444         up the vertex for the definition.
25445         (classify_partition): Classify whether a partition is a
25446         PKIND_REDUCTION, thus has uses outside of the loop.
25447         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25448         Merge all PKIND_REDUCTION partitions into the last partition.
25449         (tree_loop_distribution): Seed partitions from reductions as well.
25450
25451 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25452
25453         PR tree-optimization/56125
25454         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25455         pow(x,c) into sqrt(x) * powi(x, n/2) or
25456         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25457         optimizing for size.
25458         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25459         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25460         integer.
25461
25462         PR tree-optimization/56094
25463         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25464         to UNKNOWN_LOCATION while gimplifying expr.
25465
25466 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25467
25468         PR target/56114
25469         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25470         operand 0 in movabs insn template for -masm=intel asm alternative.
25471         (*movabs<mode>_2): Ditto for operand 1.
25472
25473 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25474
25475         PR target/54663
25476         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25477         of microblaze-c.o
25478
25479 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25480
25481         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25482         tm_file.
25483
25484 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25485
25486         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25487         Undef to avoid warning.
25488
25489 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25490
25491         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25492         * configure: Regenerate.
25493
25494 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25495
25496         PR tree-optimization/56098
25497         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25498         for stmts with volatile ops.
25499         (cond_store_replacement): Don't optimize if assign has volatile ops.
25500         (cond_if_else_store_replacement_1): Don't optimize if either
25501         then_assign or else_assign have volatile ops.
25502         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25503         volatile ops.
25504
25505 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25506
25507         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25508
25509 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25510
25511         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25512         missing ':' in asm example.
25513
25514 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25515
25516         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25517         entries into lane and laneq entries.
25518         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25519         Remove AdvSIMD scalar modes.
25520         (aarch64_sq<r>dmulh_laneq<mode>): New.
25521         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25522         modes.
25523         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25524         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25525         * config/aarch64/iterators.md (VCOND): New.
25526         (VCONQ): New.
25527
25528 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25529
25530         PR target/54222
25531         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25532         Add NULL LIBNAME argument to existing definitions.
25533         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25534         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25535         * config/avr/avr.c (DEF_BUILTIN): Same.
25536         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25537         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25538         implementation is available (DECL_ASSEMBLER_NAME is set).
25539         (avr_fold_absfx): New static function.
25540         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25541         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25542         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25543         AVR_BUILTIN_ABSLLK.
25544         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25545         (abshk, absk, abslk, absllk): Provide as static inline functions.
25546
25547 2013-01-25  Marek Polacek  <polacek@redhat.com>
25548
25549         PR tree-optimization/56035
25550         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25551
25552 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25553
25554         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25555         (*movtf_internal_rex64): Add (!o,C) alternative
25556         (*movxf_internal_rex64): Ditto.
25557         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25558
25559 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25560
25561         * doc/invoke.texi: fix typo.
25562         * doc/objc.texi: fix typo.
25563
25564 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25565
25566         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25567         for the first two alternatives.
25568
25569 2013-01-24  Diego Novillo  <dnovillo@google.com>
25570
25571         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25572         (ggc-zone.o): Remove.
25573         * configure.ac: Remove option --with-gc.
25574         * configure: Re-generate.
25575         * doc/install.texi: Remove documentation for --with-gc.
25576         * gengtype.c (write_enum_defn): Remove.  Update all users.
25577         (write_Types_process_field): Remove generation of gt_e_* argument.
25578         (output_type_enum): Remove.  Update all users.
25579         (write_enum_defn): Remove.  Update all users.
25580         (enum alloc_zone): Remove.  Update all users.
25581         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25582         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25583         Update all callers.
25584         (struct ptr_data): Remove field TYPE.  Update all users.
25585         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25586         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25587         Update all users.
25588         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25589         (struct alloc_zone): Remove.
25590         (ggc_internal_alloc_zone_stat): Remove.
25591         (ggc_internal_cleared_alloc_zone_stat): Remove.
25592         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25593         (ggc_pch_count_object): Remove last argument.  Update all users.
25594         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25595         (struct alloc_zone): Remove.
25596         * ggc-zone.c: Remove.
25597         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25598         (struct alloc_zone): Remove.
25599         (ggc_alloc_typed_stat): Remove.
25600         (ggc_alloc_typed): Remove.
25601         (ggc_splay_alloc): Remove first argument.
25602         (rtl_zone): Remove.  Update all users.
25603         (tree_zone): Remove.  Update all users.
25604         (tree_id_zone): Remove.  Update all users.
25605         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25606         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25607         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25608         * tree-ssanames.c: Remove references to zone allocator in comments.
25609
25610 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25611
25612         * config/avr/avr.c (avr_out_fract): Make register numbers that
25613         might be outside of source operand signed.
25614
25615 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25616
25617         * config/i386/constraints.md (Yf): New constraint.
25618         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25619         of f constraint to conditionaly disable x87 register preferences.
25620         (*movdf_internal): Ditto.
25621         (*movsf_internal): Ditto.
25622
25623 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25624
25625         PR inline-asm/55934
25626         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25627         that have operands with impossible constraints.
25628         Add a FIXME for a speed-up opportunity.
25629         * lra-constraints.c (process_alt_operands): Verify that a class
25630         selected from constraints on asms is valid for the operand mode.
25631         (curr_insn_transform): Remove incorrect comment.
25632
25633 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25634
25635         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25636         TOC operand is a valid symbol ref in the constant pool.
25637
25638 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25639
25640         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25641
25642 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25643
25644         PR target/54222
25645         * config/avr/stdfix.h: New file.
25646         * t-avr (stdfix-gcc.h): New rule to build it.
25647         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25648
25649 2013-01-23  Kostya Serebryany  <kcc@google.com>
25650
25651         * config/darwin.h: remove dependency on
25652         CoreFoundation (asan on Mac OS).
25653
25654 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25655
25656         PR target/49069
25657         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25658         instead of cmpdi_operand for first comparison operand.
25659         Don't assert that comparison operands aren't both constants.
25660
25661 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25662
25663         * doc/install.texi (Downloading the Source): Update references to
25664         downloading separate components.
25665
25666 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25667
25668         * doc/extend.texi (__int128): Improve grammar.
25669
25670 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25671
25672         PR target/56028
25673         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25674         alternative to (o,r).
25675         (*movdi_internal_rex64): Remove (!o,n) alternative.
25676         (DImode immediate->memory splitter): Remove.
25677         (DImode immediate->memory peephole2): Remove.
25678         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25679         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25680         alternative to (!o,*r).
25681         (*movtf_internal_sse): New pattern.
25682         (*movxf_internal_rex64): New pattern.
25683         (*movxf_internal): Disable for TARGET_64BIT.
25684         (*movdf_internal_rex64): Remove (!o,F) alternative.
25685
25686 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25687
25688         PR middle-end/56074
25689         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25690         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25691         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25692         stmt locations where LOCATION_LOCUS of the stmt location is
25693         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25694
25695         PR target/55686
25696         * config/i386/i386.md (UNSPEC_STOS): New.
25697         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25698         *strsetqi_1): Add UNSPEC_STOS.
25699
25700 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25701
25702         PR c++/56067
25703         * doc/invoke.texi: Remove left over -Wsynth example.
25704
25705 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25706
25707         PR tree-optimization/56051
25708         * fold-const.c (fold_binary_loc): Don't fold
25709         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25710         a narrowing conversion, or widening conversion from signed
25711         to unsigned.
25712
25713 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25714
25715         PR rtl-optimization/56023
25716         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25717         dependent on debug instruction.
25718
25719 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25720
25721         PR middle-end/56022
25722         * function.c (allocate_struct_function): Call
25723         invoke_set_current_function_hook earlier.
25724
25725 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25726
25727         * reload1.c (init_reload): Only initialize reload_obstack
25728         during the first call.
25729
25730 2013-01-21  Marek Polacek  <polacek@redhat.com>
25731
25732         * cfgloop.c (verify_loop_structure): Fix up grammar.
25733
25734 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
25735
25736         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
25737         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
25738
25739 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25740
25741         PR target/56058
25742         * config/arm/marvell-pj4.md: Update copyright year.
25743         Fix up use of alu to alu_reg and simple_alu_imm.
25744
25745 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25746
25747         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
25748
25749 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
25750
25751         PR target/55433
25752         * lra-constraints.c (curr_insn_transform): Don't reuse original
25753         insn for secondary memory move when memory mode should be different.
25754
25755 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25756
25757         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
25758         atomic_storedi_1): New patterns.
25759
25760 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
25761
25762         btver2 pipeline descriptions.
25763         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
25764         descriptions.
25765         * config/i386/i386.md (btver2_decode): New type attributes.
25766         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
25767         type attributes.
25768         * config/i386/btver2.md: New file describing btver2 pipelines.
25769
25770 2013-01-19  Andrew Pinski  <apinski@cavium.com>
25771
25772         PR tree-optimization/52631
25773         * tree-ssa-sccvn (visit_use): Before looking up the original
25774         statement, try looking up the simplified expression.
25775
25776 2013-01-19  Anthony Green  <green@moxielogic.com>
25777
25778         * config/moxie/moxie.c (moxie_expand_prologue): Set
25779         current_function_static_stack_size.
25780
25781 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
25782
25783         PR tree-optimization/56029
25784         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
25785         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
25786
25787 2013-01-18  Sharad Singhai  <singhai@google.com>
25788
25789         PR tree-optimization/55995
25790         * dumpfile.c (dump_loc): Print location only if available.
25791         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
25792
25793 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
25794
25795         PR target/55433
25796         * lra-constraints.c (curr_insn_transform): Reuse original insn for
25797         secondary memory move.
25798         (inherit_reload_reg): Use rclass instead of cl for
25799         check_secondary_memory_needed_p.
25800
25801 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
25802
25803         PR middle-end/56015
25804         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
25805         the case where writing real complex part of target modifies op1.
25806
25807 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
25808
25809         * config/aarch64/aarch64-simd.md
25810         (aarch64_vcond_internal<mode>): Handle unordered cases.
25811         * config/aarch64/iterators.md (v_cmp_result): New.
25812
25813 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
25814             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25815
25816         * config/arm/marvell-pj4.md: New file.
25817         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
25818         * config/arm/arm.md (generic_sched): Add marvell_pj4.
25819         (generic_vfp): Likewise.
25820         * config/arm/arm-cores.def: Add marvell-pj4.
25821         * config/arm/arm-tune.md: Regenerate.
25822         * config/arm/arm-tables.opt: Regenerate.
25823         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
25824         * doc/invoke.texi: Document marvell-pj4.
25825
25826 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
25827
25828         * config/aarch64/arm_neon.h: Map scalar types to standard types.
25829
25830 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25831
25832         PR debug/54114
25833         PR debug/54402
25834         PR debug/49888
25835         * var-tracking.c (negative_power_of_two_p): New.
25836         (global_get_addr_cache, local_get_addr_cache): New.
25837         (get_addr_from_global_cache, get_addr_from_local_cache): New.
25838         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
25839         heading comment.
25840         (vt_stack_offset_p): Remove.
25841         (vt_canon_true_dep): Always canonicalize loc's address.
25842         (clobber_overlapping_mems): Make sure we have a MEM.
25843         (local_get_addr_clear_given_value): New.
25844         (val_reset): Clear local cached entries.
25845         (compute_bb_dataflow): Create and release the local cache.
25846         Disable duplicate MEMs clobbering.
25847         (emit_notes_in_bb): Clobber MEMs likewise.
25848         (vt_emit_notes): Create and release the local cache.
25849         (vt_initialize, vt_finalize): Create and release the global
25850         cache, respectively.
25851         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
25852
25853 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25854
25855         PR libmudflap/53359
25856         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
25857         not found in the symtab.
25858
25859 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
25860
25861         PR debug/56006
25862         PR rtl-optimization/55547
25863         PR rtl-optimization/53827
25864         PR debug/53671
25865         PR debug/49888
25866         * alias.c (offset_overlap_p): New, factored out of...
25867         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
25868         the conservative special case for symbolic constants.  Don't
25869         adjust zero sizes on alignment.
25870
25871 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
25872
25873         PR rtl-optimization/52573
25874         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
25875         REG_UNUSED for the same register.
25876
25877 2013-01-17  Richard Biener  <rguenther@suse.de>
25878             Marek Polacek  <polacek@redhat.com>
25879
25880         PR rtl-optimization/55833
25881         * loop-unswitch.c (unswitch_loops): Move loop verification...
25882         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
25883         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
25884         Set it to true when we're removing a loop from hierarchy tree in
25885         an irreducible region.
25886         (fix_bb_placements): Adjust caller.
25887         (fix_loop_placements): Likewise.
25888
25889 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
25890
25891         * config/avr/builtins.def (DEF_BUILTIN): Factor out
25892         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
25893         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
25894         Remove ID.  Adjust comments.
25895         * config/avr/avr-c.c (avr_builtin_name): Remove.
25896         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
25897         * config/avr/avr.c (avr_tolower): New static function.
25898         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
25899         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
25900         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
25901         default expansion.
25902
25903 2013-01-17  Jan Hubicka  <jh@suse.cz>
25904
25905         PR tree-optimization/55273
25906         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
25907
25908 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
25909
25910         PR target/55981
25911         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
25912         store through atomic_store<mode>_1.
25913         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
25914
25915 2013-01-17  Martin Jambor  <mjambor@suse.cz>
25916
25917         PR tree-optimizations/55264
25918         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
25919         for virtual methods.
25920         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
25921         virtual methods before inlining is over.
25922         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
25923         virtual functions.
25924         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
25925         non-virtual.
25926
25927 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
25928
25929         PR rtl-optimization/56005
25930         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
25931         pending reads for prefetch.
25932
25933 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
25934
25935         * config/aarch64/aarch64.md
25936         (*cstoresi_neg_uxtw): New pattern.
25937         (*cmovsi_insn_uxtw): New pattern.
25938         (*<optab>si3_uxtw): New pattern.
25939         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
25940         (*<optab>si3_insn_uxtw): New pattern.
25941         (*bswapsi2_uxtw): New pattern.
25942
25943 2013-01-16  Richard Biener  <rguenther@suse.de>
25944
25945         * tree-inline.c (tree_function_versioning): Remove set but
25946         never used variable.
25947
25948 2013-01-16  Richard Biener  <rguenther@suse.de>
25949
25950         PR tree-optimization/55964
25951         * tree-flow.h (rename_variables_in_loop): Remove.
25952         (rename_variables_in_bb): Likewise.
25953         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
25954         (copy_loop_before): Adjust and delete update-ssa status.
25955         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
25956         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
25957         (rename_variables_in_loop): Remove.
25958         (slpeel_update_phis_for_duplicate_loop): Likewise.
25959         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
25960         use available cfg machinery instead of duplicating it.
25961         Update PHI nodes and perform poor-mans SSA update here.
25962         (slpeel_tree_peel_loop_to_edge): Adjust.
25963
25964 2013-01-16  Richard Biener  <rguenther@suse.de>
25965
25966         PR tree-optimization/54767
25967         PR tree-optimization/53465
25968         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
25969         (vrp_visit_phi_node): For PHI arguments coming via backedges
25970         drop all symbolical range information.
25971         (execute_vrp): Compute backedges.
25972
25973 2013-01-16  Richard Biener  <rguenther@suse.de>
25974
25975         * doc/install.texi: Update CLooG and ISL requirements to
25976         0.18.0 and 0.11.1.
25977
25978 2013-01-16  Christian Bruel  <christian.bruel@st.com>
25979
25980         PR target/55301
25981         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
25982         (broken_move): Handle UNSPECV_SP_SWITCH_B.
25983         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
25984
25985 2013-01-16  DJ Delorie  <dj@redhat.com>
25986
25987         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
25988         (UNSPECV_SP_SWITCH_E): New.
25989         (sp_switch_1): Change to an unspec.
25990         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
25991         replace $r15.
25992
25993 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
25994
25995         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
25996         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
25997         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
25998         (expand_mem_thread_fence): Ditto.
25999         (expand_mem_signal_fence): Ditto.
26000         (expand_atomic_load): Ditto.
26001         (expand_atomic_store): Ditto.
26002
26003 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26004
26005         PR rtl-optimization/55547
26006         PR rtl-optimization/53827
26007         PR debug/53671
26008         PR debug/49888
26009         * alias.c (memrefs_conflict_p): Set sizes to negative after
26010         AND adjustments.
26011
26012 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26013
26014         PR target/55940
26015         * function.c (thread_prologue_and_epilogue_insns): Always
26016         add crtl->drap_reg to set_up_by_prologue.set, even if
26017         stack_realign_drap is false.
26018
26019 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26020
26021         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26022         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26023         *call): Fix indention.
26024
26025 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26026
26027         PR target/55876
26028         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26029         Update comment.
26030
26031 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26032
26033         PR rtl-optimization/55153
26034         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26035
26036 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26037
26038         PR tree-optimization/55920
26039         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26040         accesses as grp_to_be_debug_replaced.
26041
26042 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26043
26044         PR tree-optimization/55920
26045         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26046         there is non-useless type conversion needed from debug rhs to lhs,
26047         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26048
26049 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26050             Mikael Pettersson  <mikpe@it.uu.se>
26051
26052         PR target/43961
26053         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26054         Thumb.
26055         (ASM_OUTPUT_CASE_LABEL): Remove.
26056         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26057         * final.c (shorten_branches): Update alignment of labels before
26058         jump tables if CASE_VECTOR_SHORTEN_MODE.
26059
26060 2013-01-15  Richard Biener  <rguenther@suse.de>
26061
26062         PR bootstrap/55961
26063         * system.h: Do not include gmp.h for building host tools.
26064
26065 2013-01-15  Richard Biener  <rguenther@suse.de>
26066
26067         PR middle-end/55882
26068         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26069         account for bitpos when computing alignment.
26070
26071 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26072
26073         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26074         (ix86_target_macros_internal): Likewise.
26075
26076         * config/i386/i386.c (m_CORE2I7): Removed.
26077         (m_CORE_HASWELL): New macro.
26078         (m_CORE_ALL): Likewise.
26079         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26080         (initial_ix86_arch_features): Likewise.
26081         (processor_target_table): Initializations for Core avx2.
26082         (cpu_names): New names "core-avx2".
26083         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26084         PROCESSOR_CORE_HASWELL.
26085         (ix86_issue_rate): New case.
26086         (ia32_multipass_dfa_lookahead): Likewise.
26087         (ix86_sched_init_global): Likewise.
26088
26089         * config/i386/i386.h (TARGET_HASWELL): New macro.
26090         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26091         (processor_type): New PROCESSOR_HASWELL.
26092
26093 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26094
26095         PR tree-optimization/55955
26096         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26097         *SHIFT_EXPR and *ROTATE_EXPR codes.
26098
26099         PR tree-optimization/48766
26100         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26101         -ftrapv disable -fwrapv.
26102
26103 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26104
26105         PR target/55974
26106         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26107         etc. to 1 and not to __flash.
26108         Use LL suffix for __INT24_MAX__ with -mint8.
26109         Use ULL suffix for __UINT24_MAX__ with -mint8.
26110
26111 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26112
26113         * config/avr/avr-arch.h
26114         (struct base_arch_s): Use typedef avr_arch_t instead.
26115         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26116         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26117         * config/avr/avr.c: Same.
26118         * config/avr/avr-devices.c: Same.
26119         * config/avr/driver-avr.c: Same.
26120         * config/avr/gen-avr-mmcu-texi.c: Same.
26121         * config/avr/avr-mcus.def: Adjust comment.
26122
26123 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26124
26125         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26126         * config/aarch64/iterators.md (VALLDI): New.
26127
26128 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26129             Andi Kleen  <ak@linux.intel.com>
26130
26131         PR target/55948
26132         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26133         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26134         memmodel flag.
26135
26136 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26137
26138         * config/avr/avr-stdint.h: Remove trailing blanks.
26139         * config/avr/avr-log.h: Same.
26140         * config/avr/avr-arch.h: Same.
26141         * config/avr/avr-devices.c: Same.
26142         * config/avr/avr-dimode.md: Same.
26143         * config/avr/predicates.md: Same.
26144         * config/avr/avr-c.c: Same.  And fix typo.
26145
26146         * config/avr/avr-protos.h: Same.  And:
26147         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26148         (init_cumulative_args): Rename to avr_init_cumulative_args.
26149         (expand_prologue): Rename to avr_expand_prologue.
26150         (expand_epilogue): Rename to avr_expand_epilogue.
26151         (adjust_insn_length): Rename to avr_adjust_insn_length.
26152         (notice_update_cc): Rename to avr_notice_update_cc.
26153         (final_prescan_insn): Rename to avr_final_prescan_insn.
26154         * config/avr/avr.c: Same.
26155         * config/avr/avr.h: Same.
26156         * config/avr/avr.md: Remove trailing blanks.
26157         (prologue): Use avr_expand_prologue.
26158         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26159
26160 2013-01-14  Richard Biener  <rguenther@suse.de>
26161
26162         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26163         verify_location, collect_subblocks): New functions.
26164         (verify_gimple_in_cfg): Verify that locations only reference
26165         BLOCKs in the functions BLOCK tree.
26166
26167 2013-01-14  Richard Biener  <rguenther@suse.de>
26168
26169         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26170         PHI argument.
26171         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26172         unshare reference.
26173         (insert_out_of_ssa_copy_on_edge): Likewise.
26174         (rewrite_close_phi_out_of_ssa): Likewise.
26175         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26176         debug expressions.
26177         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26178         propagated constants.
26179         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26180         can not be shared.
26181
26182 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26183
26184         * config/avr/avr-modes.def: Add GPL copyright notice.
26185
26186 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26187
26188         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26189         MEMMODEL_MASK to determine memory model.
26190         (atomic_store<mode>): Ditto from operands[2].
26191         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26192
26193 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26194
26195         PR fortran/55935
26196         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26197         (fold_gimple_assign): Don't call unshare_expr here.
26198         (fold_ctor_reference): Call unshare_expr.
26199
26200 2013-01-13  Terry Guo  <terry.guo@arm.com>
26201
26202         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26203         * doc/fragments.texi: Document MULTILIB_REUSE.
26204         * gcc.c (multilib_reuse): New internal spec.
26205         (set_multilib_dir): Also search multilib from multilib_reuse.
26206         * genmultilib (tmpmultilib3): Refactor code.
26207         (tmpmultilib4): Ditto.
26208         (multilib_reuse): New multilib argument.
26209
26210 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26211
26212         * Makefile.in: Update copyright.
26213
26214 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26215
26216         PR middle-end/55890
26217         * calls.c (expand_call): Check if arg_nr is valid.
26218
26219 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26220
26221         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26222         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26223         documentation.  Add missing '__' in front of
26224         __builtin_ia32_packssdw256.
26225
26226 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26227
26228         PR target/55719
26229         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26230         NO_REGS for larl operands.
26231         (s390_reload_larl_operand): Use s390_load_address instead of
26232         emit_move_insn.
26233
26234 2013-01-11  Richard Biener  <rguenther@suse.de>
26235
26236         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26237         (verify_node_sharing): ... here.
26238         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26239
26240 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26241
26242         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26243         Substitute TREECHECKING.
26244         * configure: Regenerate.
26245         * Makefile.in (TREECHECKING): New.
26246
26247 2013-01-11  Richard Guenther  <rguenther@suse.de>
26248
26249         PR tree-optimization/44061
26250         * tree-vrp.c (extract_range_basic): Compute zero as
26251         value-range for __builtin_constant_p of function parameters.
26252
26253 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26254
26255         Update copyright years.
26256
26257 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26258
26259         PR rtl-optimization/55672
26260         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26261         const to be eliminable.
26262
26263 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26264
26265         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26266         * configure: Regenerate.
26267
26268 2013-01-10  Richard Biener  <rguenther@suse.de>
26269
26270         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26271
26272 2013-01-10  Richard Biener  <rguenther@suse.de>
26273
26274         PR bootstrap/55792
26275         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26276         locations for virtual PHI arguments.
26277         (rewrite_update_phi_arguments): Likewise.
26278
26279 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26280
26281         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26282         on to assembler.
26283
26284 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26285
26286         PR tree-optimization/55921
26287         * tree-complex.c (expand_complex_asm): New function.
26288         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26289
26290 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26291
26292         PR target/55718
26293         * config/s390/s390.c (s390_symref_operand_p)
26294         (s390_loadrelative_operand_p): Merge the two functions.
26295         (s390_check_qrst_address, print_operand_address): Add parameters
26296         to s390_loadrelative_operand_p invokation.
26297         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26298         (s390_reload_larl_operand, s390_secondary_reload): Use
26299         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26300         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26301
26302 2013-01-09  Mike Stump  <mikestump@comcast.net>
26303
26304         * dse.c (record_store): Remove unnecessary assert.
26305
26306 2013-01-09  Jan Hubicka  <jh@suse.cz>
26307
26308         PR tree-optimization/55569
26309         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26310         * cfgloop.h (scale_loop_profile): Likewise.
26311
26312 2013-01-09  Jan Hubicka  <jh@suse.cz>
26313
26314         PR lto/45375
26315         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26316         functions.
26317         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26318
26319 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26320
26321         PR middle-end/55114
26322         * expr.h (maybe_emit_group_store): Declare.
26323         * expr.c (maybe_emit_group_store): New function.
26324         * builtins.c (expand_builtin_int_roundingfn): Call it.
26325         (expand_builtin_int_roundingfn_2): Likewise.
26326
26327 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26328
26329         PR rtl-optimization/55829
26330         * lra-constraints.c (match_reload): Add code for absent output.
26331         (curr_insn_transform): Add code for reloads of matched inputs
26332         without output.
26333
26334 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26335
26336         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26337         attribute of movddup insn to DF.
26338         (*vec_interleave_lowv2df): Ditto.
26339         (vec_dupv2df): Ditto.
26340
26341 2013-01-09  Jan Hubicka  <jh@suse.cz>
26342
26343         PR tree-optimiation/55875
26344         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26345         EVERY_ITERATION parameter.
26346         (number_of_iterations_exit): Check if exit is executed every iteration.
26347         (idx_infer_loop_bounds): Similarly here.
26348         (n_of_executions_at_most): Simplify
26349         to only test for cases where statement is dominated by the
26350         particular bound; handle correctly the "postdominance" test.
26351         (scev_probably_wraps_p): Use max loop iterations info
26352         as a global bound first.
26353
26354 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26355             Nick Clifton  <nickc@redhat.com>
26356
26357         * config/v850/v850.md (cbranchsf4): New pattern.
26358         (cstoresf4): New pattern.
26359         (cbranchdf4): New pattern.
26360         (cstoredf4): New pattern.
26361         (movsicc): Disallow floating point comparisons.
26362         (cmpsf_le_insn): Fix order of operators.
26363         (cmpsf_lt_insn): Likewise.
26364         (cmpsf_eq_insn): Likewise.
26365         (cmpdf_le_insn): Likewise.
26366         (cmpdf_lt_insn): Likewise.
26367         (cmpdf_eq_insn): Likewise.
26368         (cmpsf_ge_insn): Use LE comparison.
26369         (cmpdf_ge_insn): Likewise.
26370         (cmpsf_gt_insn): Use LT comparison.
26371         (cmpdf_gt_insn): Likewise.
26372         (cmpsf_ne_insn): Delete pattern.
26373         (cmpdf_ne_insn): Delete pattern.
26374         * config/v850/v850.c (v850_gen_float_compare): Use
26375         gen_cmpdf_eq_insn for NE comparison.
26376         (v850_float_z_comparison_operator)
26377         (v850_float_nz_comparison_operator): Move from here ...
26378         * config/v850/predicates.md: ... to here.  Move GT and GE
26379         comparisons into v850_float_z_comparison_operator.
26380         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26381         Delete prototype.
26382         (v850_float_nz_comparison_operator): Likewise.
26383
26384 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26385
26386         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26387         with calls to gen_insvsi/gen_insvdi.
26388
26389 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26390
26391         * config/i386/i386.c (initial_ix86_tune_features): Set up
26392         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26393
26394 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26395             Jakub Jelinek  <jakub@redhat.com>
26396
26397         PR tree-optimization/48189
26398         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26399         If nitercst is 0, don't predict the exit edge.
26400
26401 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26402
26403         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26404         in asm_fprintf with reg_names.
26405         (aarch64_print_operand_address): Likewise.
26406         (aarch64_return_addr): Likewise.
26407         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26408
26409 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26410
26411         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26412         (INT_U6_BITS): Likewise.
26413         * config/pa/predicates.md (uint6_operand): New predicate.
26414         (shift5_operand, shift6_operand): Likewise.
26415         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26416         arith32_operand.
26417         (lshrdi3): Use shift6_operand.
26418         (shrpsi4, shrpdi4): New insn patterns.
26419         (extzv): Delete expander.
26420         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26421         predicates in unamed zero extract patterns.  Tighten common constraint.
26422         (extv): Delete expander.
26423         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26424         predicates in unamed sign extract patterns.  Tighten common constraint.
26425         (insv): Delete expander.
26426         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26427         predicates in unamed insert patterns.  Tighten common constraint.
26428         Change uint32_operand predicate to uint6_operand predicate in unamed
26429         DImode pattern to insert constant values of type 1...1xxxx.
26430
26431 2013-01-04  Jan Hubicka  <jh@suse.cz>
26432
26433         PR tree-optimization/55823
26434         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26435         issue.
26436
26437 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26438             Uros Bizjak  <ubizjak@gmail.com>
26439
26440         PR rtl-optimization/55845
26441         * df-problems.c (can_move_insns_across): Stop scanning at
26442         volatile_insn_p source instruction or give up if
26443         across_from .. across_to range contains any volatile_insn_p
26444         instructions.
26445
26446 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26447
26448         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26449         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26450         Declare.
26451         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26452         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26453
26454 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26455
26456         PR fortran/55341
26457         * asan.c (asan_clear_shadow): New function.
26458         (asan_emit_stack_protection): Use it.
26459
26460 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26461
26462         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26463         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26464         with tab instead of space.
26465
26466 2013-01-08  Nick Clifton  <nickc@redhat.com>
26467
26468         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26469         register bank 0 at the start of an interrupt handler.
26470         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26471         MDBH registers.
26472
26473 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26474
26475         * config/aarch64/aarch64-simd.md
26476         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26477         (aarch64_simd_bsl): Likewise.
26478         (aarch64_vcond_internal<mode>): Likewise.
26479         (vcond<mode><mode>): Likewise.
26480         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26481         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26482
26483 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26484
26485         * config/aarch64/aarch64-builtins.c
26486         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26487
26488 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26489
26490         PR debug/55579
26491         * tree-sra.c (analyze_access_subtree): Return true also after
26492         potentially creating a debug-only replacement.
26493
26494 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26495
26496         PR middle-end/55890
26497         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26498
26499         PR tree-optimization/54120
26500         * tree-vrp.c (range_fits_type_p): Don't allow
26501         src_precision < precision from signed vr to unsigned_p
26502         if vr->min or vr->max is negative.
26503         (simplify_float_conversion_using_ranges): Test can_float_p
26504         against CODE_FOR_nothing.
26505
26506 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26507             Richard Biener  <rguenther@suse.de>
26508
26509         PR middle-end/55851
26510         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26511         types instead of just INTEGER_TYPE types.
26512
26513 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26514
26515         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26516         TF_SIZE): Define.
26517
26518 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26519
26520         PR target/42661
26521         * config/mips/mips.opt: Change mad to mmad to match documentation.
26522
26523 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26524
26525         PR target/55897
26526         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26527         .progmemx.data now.
26528
26529 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26530
26531         PR target/55897
26532         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26533         (avr_addrspace_t): Add .section_name field.
26534         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26535         array size.
26536         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26537         NULL entry.  Put __memx into .progmemx.data.
26538         (progmem_section_prefix): Remove.
26539         (avr_asm_init_sections): No need to initialize progmem_section.
26540         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26541         section name prefix.
26542         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26543         retrieve the progmem section.
26544         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26545         boundary to run over avr_addrspace[].
26546         (avr_register_target_pragmas): Ditto.
26547
26548 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26549
26550         * varasm.c (output_constant_def_contents): For asan_protect_global
26551         protected strings, adjust DECL_ALIGN if needed, before testing for
26552         anchored symbols.
26553         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26554         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26555         normal decls.
26556         (output_object_block): For asan protected decls, emit asan padding
26557         after their contents.
26558         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26559         (asan_finish_file): Test it here instead.
26560
26561 2013-01-07  Nick Clifton  <nickc@redhat.com>
26562             Matthias Klose  <doko@debian.org>
26563             Doug Kwan  <dougkwan@google.com>
26564             H.J. Lu  <hongjiu.lu@intel.com>
26565
26566         PR driver/55470
26567         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26568
26569         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26570
26571         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26572
26573         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26574
26575         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26576
26577 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26578
26579         PR target/54461
26580         * doc/install.texi (Cross-Compiler-Specific Options): Document
26581         --with-avrlibc.
26582
26583 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26584
26585         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26586         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26587         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26588         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26589         vqmovun_high_s64): Fix source operand number and update copyright.
26590
26591 2013-01-07  Richard Biener  <rguenther@suse.de>
26592
26593         PR middle-end/55890
26594         * gimple.h (gimple_call_builtin_p): New overload.
26595         * gimple.c (validate_call): New function.
26596         (gimple_call_builtin_p): Likewise.
26597         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26598         Use gimple_call_builtin_p.
26599         (find_func_clobbers): Likewise.
26600         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26601         (strlen_optimize_stmt): Likewise.
26602
26603 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26604
26605         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26606         (vld1q_dup_*): Likewise.
26607         (vld1_*): Likewise.
26608         (vld1q_*): Likewise.
26609         (vld1_lane_*): Likewise.
26610         (vld1q_lane_*): Likewise.
26611
26612 2013-01-07  Richard Biener  <rguenther@suse.de>
26613
26614         * lto-streamer.h (LTO_minor_version): Bump to 2.
26615
26616 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26617
26618         * config/aarch64/aarch64-protos.h
26619         (aarch64_const_double_zero_rtx_p): Rename to...
26620         (aarch64_float_const_zero_rtx_p): ...this.
26621         (aarch64_float_const_representable_p): New.
26622         (aarch64_output_simd_mov_immediate): Likewise.
26623         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26624         move immediate case.
26625         * config/aarch64/aarch64.c
26626         (aarch64_const_double_zero_rtx_p): Rename to...
26627         (aarch64_float_const_zero_rtx_p): ...this.
26628         (aarch64_print_operand): Allow printing of new constants.
26629         (aarch64_valid_floating_const): New.
26630         (aarch64_legitimate_constant_p): Check for valid floating-point
26631         constants.
26632         (aarch64_simd_valid_immediate): Likewise.
26633         (aarch64_vect_float_const_representable_p): New.
26634         (aarch64_float_const_representable_p): Likewise.
26635         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26636         (aarch64_output_simd_mov_immediate): New.
26637         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26638         (*movdf_aarch64): Likewise.
26639         * config/aarch64/constraints.md (Ufc): New.
26640         (Y): call aarch64_float_const_zero_rtx.
26641         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26642
26643 2013-01-07  Richard Biener  <rguenther@suse.de>
26644
26645         PR tree-optimization/55888
26646         PR tree-optimization/55862
26647         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26648         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26649         not if it is contained therein.
26650
26651 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26652
26653         * config/avr/t-avr: Typo.
26654
26655 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26656
26657         PR55243
26658         * config/avr/t-avr: Don't automatically rebuild
26659         $(srcdir)/config/avr/t-multilib
26660         $(srcdir)/config/avr/avr-tables.opt
26661         $(srcdir)/doc/avr-mmcu.texi
26662         (avr-mcus): New phony target to build them on request.
26663         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26664         * avr/avr-mcus.def: Adjust comments.
26665
26666 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26667
26668         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26669
26670 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26671
26672         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26673
26674 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26675
26676         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26677
26678 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26679
26680         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26681         to generate profiling.
26682         * config/rs6000/aix64.h (LIB_SPEC): Same.
26683
26684 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26685
26686         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26687         New function.
26688         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26689
26690 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26691
26692         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26693         unconditionally.
26694         (ix86_expand_move): Ditto.
26695         (ix86_zero_extend_to_Pmode): Ditto.
26696         (ix86_expand_call): Ditto.
26697         (ix86_expand_special_args_builtin): Ditto.
26698         (ix86_expand_builtin): Ditto.
26699
26700 2013-01-04  Richard Biener  <rguenther@suse.de>
26701
26702         PR tree-optimization/55862
26703         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26704         translating them through PHI nodes.
26705
26706 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26707
26708         PR tree-optimization/55755
26709         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26710         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26711
26712 2013-01-04  Marek Polacek  <polacek@redhat.com>
26713
26714         PR middle-end/55859
26715         * opts.c (default_options_optimization): Clarify error message.
26716
26717 2013-01-04  Richard Biener  <rguenther@suse.de>
26718
26719         PR middle-end/55863
26720         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26721         reassociation.
26722
26723 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26724
26725         PR target/53789
26726         * config/pa/pa.md (movsi): Revert previous change.
26727         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
26728         references.
26729
26730 2013-01-03  Richard Henderson  <rth@redhat.com>
26731
26732         * config/i386/i386.c (ix86_expand_move): Always assign to op1
26733         after eliminating TLS symbols.
26734
26735 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
26736
26737         PR bootstrap/50167
26738         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
26739         * graphite-poly.c (debug_gmp_value): Likewise.
26740
26741 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
26742
26743         PR target/55712
26744         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
26745         selected code model, define __code_mode_small__, __code_model_medium__,
26746         __code_model_large__, __code_model_32__ or __code_model_kernel__.
26747         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
26748         xchg temporary register with %k.  Declare temporary register as
26749         early clobbered.
26750         [__x86_64__]: For medium and large code models, preserve %rbx register.
26751
26752 2013-01-03  Richard Biener  <rguenther@suse.de>
26753
26754         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
26755         (dump_subscript): Adjust.
26756         (finalize_ddr_dependent): Do not dump redundant info.
26757         (analyze_siv_subscript): Adjust.
26758         (subscript_dependence_tester): Likewise.
26759         (compute_affine_dependence): Likewise.
26760
26761 2013-01-03  Richard Biener  <rguenther@suse.de>
26762
26763         Revert
26764         2013-01-03  Richard Biener  <rguenther@suse.de>
26765
26766         PR tree-optimization/55857
26767         * tree-vect-stmts.c (vectorizable_load): Do not setup
26768         re-alignment for invariant loads.
26769
26770         2013-01-02  Richard Biener  <rguenther@suse.de>
26771
26772         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26773         invariant load do not generate a vector load from the scalar location.
26774
26775 2013-01-03  Richard Biener  <rguenther@suse.de>
26776
26777         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
26778         for not vectorizing.
26779         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
26780         not build INDIRECT_REFs, call get_name once only.
26781         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
26782         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
26783
26784 2013-01-03  Richard Biener  <rguenther@suse.de>
26785
26786         PR tree-optimization/55857
26787         * tree-vect-stmts.c (vectorizable_load): Do not setup
26788         re-alignment for invariant loads.
26789
26790 2013-01-03  Richard Biener  <rguenther@suse.de>
26791
26792         PR lto/55848
26793         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
26794         prefer a built-in decl.
26795
26796 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26797
26798         * gcc.c (process_command): Update copyright notice dates.
26799         * gcov.c (print_version): Likewise.
26800         * gcov-dump.c (print_version): Likewise.
26801
26802         PR rtl-optimization/55838
26803         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
26804         iv0.step, iv1.step and step.
26805
26806 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26807             Marc Glisse  <marc.glisse@inria.fr>
26808
26809         PR tree-optimization/55832
26810         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
26811         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
26812         integer_{one,zero}_node.
26813
26814 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
26815
26816         PR debug/54402
26817         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
26818         * var-tracking.c (reverse_op): Don't add reverse ops to
26819         VALUEs that have already
26820         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
26821
26822 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
26823
26824         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
26825
26826 2013-01-02  Teresa Johnson  <tejohnson@google.com>
26827
26828         * dumpfile.c (dump_loc): Print filename with location.
26829         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
26830         new location_t parameter to emit complete unroll message with
26831         new dump framework.
26832         (canonicalize_loop_induction_variables): Compute loops location
26833         and pass to try_unroll_loop_completely.
26834         * loop-unroll.c (report_unroll_peel): New function.
26835         (peel_loops_completely): Use new dump format with location
26836         for main dumpfile message, and invoke report_unroll_peel on success.
26837         (decide_unrolling_and_peeling): Ditto.
26838         (decide_peel_once_rolling): Remove old dumpfile message subsumed
26839         by report_unroll_peel.
26840         (decide_peel_completely): Ditto.
26841         (decide_unroll_constant_iterations): Ditto.
26842         (decide_unroll_runtime_iterations): Ditto.
26843         (decide_peel_simple): Ditto.
26844         (decide_unroll_stupid): Ditto.
26845         * cfgloop.c (get_loop_location): New function.
26846         * cfgloop.h (get_loop_location): Declare.
26847
26848 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
26849
26850         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
26851         NULL.
26852
26853 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26854
26855         PR middle-end/55198
26856         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
26857         BLKmode objects when EXPAND_MEMORY is specified.
26858
26859 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
26860
26861         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
26862         in loop predicate.
26863         (fold_builtin_cpu): Do not share cpu model decls across statements.
26864
26865 2013-01-02  Jason Merrill  <jason@redhat.com>
26866
26867         PR c++/55804
26868         * tree.c (build_array_type_1): Revert earlier change.
26869
26870 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
26871
26872         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
26873         "cortex-a57".
26874         * config/aarch64/aarch64-tune.md: Re-generate.
26875
26876 2013-01-02  Richard Biener  <rguenther@suse.de>
26877
26878         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26879         invariant load do not generate a vector load from the scalar location.
26880
26881 2013-01-02  Richard Biener  <rguenther@suse.de>
26882
26883         PR bootstrap/55784
26884         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
26885         * configure: Regenerate.
26886
26887 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
26888
26889         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
26890         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
26891         (expand_builtin_int_roundingfn_2): Keep the original target around
26892         for the fallback case.
26893
26894 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
26895
26896         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
26897         to be clear for sign changes.
26898
26899 2013-01-01  Jan Hubicka  <jh@suse.cz>
26900
26901         * ipa-inline-analysis.c: Fix formatting.
26902
26903 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
26904
26905         PR tree-optimization/55831
26906         * tree-vect-loop.c (get_initial_def_for_induction): Use
26907         gsi_after_labels instead of gsi_start_bb.
26908 \f
26909 Copyright (C) 2013 Free Software Foundation, Inc.
26910
26911 Copying and distribution of this file, with or without modification,
26912 are permitted in any medium without royalty provided the copyright
26913 notice and this notice are preserved.